Proceeding 3rd ICNERE and 8th EECCIS 2017

Joint International Conference
3rd ICNERE and 8th EECCIS

The 3rd International Conference on Nano Electronics Research and Education
The 8th International Conference on Electrical, Electronics, Communications, Controls and Informatics System
October 31 – November 2, 2016

Batu, Indonesia
PROCEEDING
Organized by: Supported by:

Proceeding
Joint International Conference
3rd ICNERE and 8th EECCIS

The 8th International Conference on Electrical, Electronics, Communications, Controls and
Informatics System
October 31 –November 2, 2016
Batu - Indonesia
Published by:
Jurusan Teknik Elektro
Fakultas Teknik
Universitas Brawijaya
Indonesia
Proceeding Joint International Conference 3rd ICNERE and 8th EECCIS 2016
ISBN :
978-602-8692-32-8
Editor :
Primatar Kuswiradyo
Publisher :
Electrical Engineering Department
Engineering Faculty
Brawijaya University
JL. MT. Haryono 167 Malang – East Java - Indonesia
phone:+62-341- 554166
facsimile:+62-341- 551430
elektro@ub.ac.id
eeccis@ub.ac.id
Published for:
rd
The 3 International Conference on Nano Electronics Research and Education
The 8th International Conference on Electrical, Electronics, Communications, Controls and
Informatics System
Malang- Indonesia
All articles in this book have been selected by the reviewers and the technical committee. All authors have signed a
publishing copyrighted paper. No reproduction, downloading, clipping, publish, transfer, into another form with a
specific purpose, except with the permission of the author and demonstrate its contribution to the author. The
publisher does not increase or decrease the content and is not responsible for any errors in this book
Copyright © Electrical Engineering Department, Enginering Faculty,

2016
Joint International Conference - The 3rd International Conference on Nano Electronics Research and Education &
th
The 8 Inter. Conf. on Electrical, Electronics, Communications, Controls and Informatics System (ICNERE-EECCIS 2016).
Royal Orchids Garden Hotel, Batu, Indonesia. October 31 – November 2, 2016.
Joint International Conference:

(ICNERE 2016)
The 8th International Conference on Electrical, Electronics, Communications, Controls and Informatics
System
(EECCIS 2016)
"Green Technology towards a Better Environment”

Batu- Indonesia
Organized by :
Supported by :
Fakultas Teknik
Universitas Brawijaya PSeeMRG
©ICNERE-EECCIS 2016 iii

th
TABLE OF CONTENTS
TABLE OF CONTENTS .................................................................... iv

ABOUT THE CONFERENCE .............................................................. v
FOREWORD ................................................................................... vi
WELCOMING MESSAGE ............................................................... vii
ORGANIZING COMMITTEE .......................................................... viii
KEYNOTE SPEAKER ......................................................................... x
PROGRAM SCHEDULE ................................................................. xvi
PAPER COMPILATION..................................................................... 1
©ICNERE-EECCIS 2016 iv
th
ABOUT THE CONFERENCE
Universitas Indonesia, Shizuoka University, Universitas Brawijaya and Universiti Tun Hussein Onn Malaysia
(UTHM) are collaborating to organize and pleased to announce the joint International Conference of
the 3rd International Conference on Nano Electronics Research Education 2016 (ICNERE 2016) and
8th International Conference on Electrical Power, Electronics, Communications, Controls and Informatics
System 2016 (EECCIS 2016). The first ICNERE were held in Bali, Indonesia in 2012 and the 2nd ICNERE
were held in Hamamatsu, Japan in 2014. EECCIS is a biannually conference held since 2000 until 2014. The
conference will be held in Malang, a second largest city in East Java Indonesia famous as “Paris of East Java”,
from 31 October – 2 November 2016.
The objective of this joint conference is to provide opportunities for researchers, scientists, engineers and
vendors working in the area of electric power, electronic, telecommunication, control and informatics to
exchange their ideas and also to present and discuss latest results and future directions of research and
application, and developments in the converging technology and educations of electrical and electronics
engineering. All accepted and presented papers will be published in one of the following publications:
 Proceeding ICNERE - EECCIS 2016

 Makara Journal of Technology (accredited by DIKTI)
ISSN : (P) 2355-2786, (E) 2356-4539
 Jurnal Teknologi (Scopus indexed)
ISSN : 2180-3722
This joint conference topics are include, however not limited to :

1. Nano electronics and nano photonic devices
2. Nano and micro electro mechanical systems (NEMS/MEMS)
3. Nano materials
4. Quantum information and computation
5. Nano technologies for medical applications
6. Power and energy, power electronics, renewable energy source and system
7. Electronics circuits, instrumentation, optoelectronics, FPGA & embedded system
8. Fiber optic, wireless & sensor networks, microwave, antenna and signal processing
9. Advance control system, automation, robotics and autonomous system
10. Innovative teaching and learning mechanism in electrical & electronics engineering education
11. Image processing, artificial intelligent, pattern recognition, knowledge discovery and security.
©ICNERE-EECCIS 2016 v
th
FOREWORD
Assalamu’alaikum warahmatullahi wabarakatuh
First of all, I would like to express my acknowledgement to the whole parties, lecturers, students, and all other
people impossible to cite individually, for having involved in the good achievement of the organization of the
Joint International Conference on the 3rd International Conference on Nano Electronics Research and
Education (ICNERE) and the 8th International Conference on Electrical, Electronics, Communications,
Controls, and Informatics System (EECCIS) 2016, collaboration between Brawijaya University, Universiti
Tun Hussein Onn Malaysia (UTHM), University of Indonesia and Shizuoka University- Japan.
I also would like to express my gratitude to Prof. Robert Rieger, Ph.D (National Sun Yat Sen University,
Taiwan), Prof. Masaaki Nagatsu (Shizuoka University, Japan) and Assoc. Prof. Dr. Nafarizal Nayan
(Universiti Tun Hussein Onn Malaysia - UTHM) for having accepted to become the keynote speakers of this
Joint International Conference ICNERE-EECCIS 2016.
The EECCIS 2016 Seminar follows the success of the previously held seminars of EECCIS 2000, 2004, 2006,
2008, 2010, 2012 and 2014. It becomes a part of scientific activity programmes in the Faculty of Engineering
to contribute to the creation of Brawijaya University as a research and entrepreneurial university.
As a part of the Brawijaya University, civitas academica of the Faculty of Engineering play a very strategic
and active role in producing a tight link to industry and society in general. It is hoped that through the Joint
International Conference ICNERE-EECCIS 2016 the tight link could be maintained and improved either
nationally or internationally, so that the scientific culture among the research and education institutions as
well as its link-and-match to industry could bring out the welfare of the Indonesian society, and humanity in
general.
Wassalamu’alaikum warahmatullahi wabarakatuh
Dr. Ir. Pitojo Tri Juwono, MT.

Dean of Engineering Faculty
©ICNERE-EECCIS 2016 vi
th
WELCOMING MESSAGE
It is our pleasure to invite you to the 3rd International Conference on Nano Electronics Research and
Education or ICNERE 2016, a collaborating event between Universitas Indonesia and Shizuoka University,
Japan. This conference will be held in Malang, East Java Indonesia from October 31 to November 2,
2016. This International event is being held in conjunction with the 8th International Conference on
Electric Power, Electronic, Communication, Control, and Informatic Systems (EECCIS), a collaborating
event between Universitas Brawijaya and Universiti Tun Hussein Onn Malaysia (UTHM).
As always, the conference will provide an important forum and opportunities for researchers and engineers
from the industries, and professors as well as graduate students from academia to establish a scientific network
and to discuss new ideas and developments in the converging technology and educations of electrical and
electronics engineering. By attending the conference, you will become a part of the development of Electrical
Engineering and Nano Electronics Research and Education.
We hope to welcome you in our midst of our friendly people and one of the very exotic place since centuries.
We look forward to seeing you at the conference!
Prof. Dr. Harry Sudibyo S, DEA

General Chair
©ICNERE-EECCIS 2016 vii

th
ORGANIZING COMMITTEE
International Advisory Board

 Dean of Faculty of Engineering, Universitas Indonesia
 Dean of Graduate School of Science and Technology, Shizuoka University, Japan
 Dean Faculty of Engineering, Universitas Brawijaya, Indonesia
 Dean of Faculty of Electrical and Electronics Engineering, Universiti Tun Hussein Onn Malaysia
Steering Committee
 Djoko Hartanto (UI-Indonesia)
 Eko Tjipto Rahardjo (UI-Indonesia)
 Gunawan Wibisono (UI-Indonesia)
 Sholeh Hadi Pramono (UB-Indonesia)
 M Aziz Muslim (UB-Indonesia)
 Hadi Suyono (UB-Indonesia)
 Michiharu Tabe (Shizuoka Univ – Japan)
 Hidenori Mimura (Shizuoka Univ – Japan)
 Kazuhiko Hara (Shizuoka Univ – Japan)
 Datuk Mohd Noh Bin Dalimin (UTHM-Malaysia)
 Afandi bin Ahmad (UTHM-Malaysia)
 Indra Surjati (Trisakti University – Indonesia)
 Mudrik Alaydrus (Mercubuana Univ. -Indonesia)
 Djamhari Sirat (Mpu Tantular Univ. -Indonesia)
 Baskoro Abie Pandowo (ISTN – Indonesia)
Organizing Committee
 General Chair : Prof. Harry Sudibyo (UI-Indonesia)
o EECCIS
 Chair : Panca Mudjirahardjo (UB-Indonesia)
 Vice Chair : Mohd Khairul bin Ahmad (UTHM-Malaysia)
o ICNERE
 Chair : Arief Udhiarto (UI-Indonesia)
 Vice Chair : Masaaki Nagatsu (Shizuoka Univ – Japan)
 Secretary :
o Catur Apriono (UI-Indonesia)
o Primatar Kuswiradyo (UB-Indonesia)
 Treasurer :
o Rahmadwati (UB-Indonesia)
 Publication :
o Ardiansyah (UI-Indonesia)
o I Gde Dharma Nugraha (UI-Indonesia)
o Raden Arief Setyawan (UB-Indonesia)
©ICNERE-EECCIS 2016 viii

th
o Adharul Muttaqin (UB-Indonesia)

o Eka Maulana (UB-Indonesia)
 Local Organizing Committee :
o Ajib Setyo Arifin (UI-Indonesia)
o Arief Noor Rahman (UI-Indonesia)
o Akhmad Zainuri (UB-Indonesia)
o Sapriesty Naini Sari (UB-Indonesia)
o Rudy Yuwono (UB-Indonesia)
o Teguh Utomo (UB-Indonesia)
o Eni Suwartini (UB-Indonesia)
o Rakhmad Romadhoni (UB-Indonesia)
o Indra Setyawan (UB-Indonesia)
 Sponsorship :
o M. Suryanegara (UI-Indonesia)
o Sigit Kusmaryanto (UB-Indonesia)
©ICNERE-EECCIS 2016 ix
th
KEYNOTE SPEAKERS
Student centered design of CMOS programmable biosignal recorders
Robert Rieger
National Sun Yat Sen University, Taiwan
Abstract: Sensing the environment by means of very small and low power electronic circuits is an
accelerating trend which paves the way for advanced applications in areas ranging from healthcare to the
Internet-of-Things. This talk presents certain essential circuits for the multichannel monitoring of biomedical
signals using integrated CMOS technology, including the configurable gain amplifier and analog-to-digital
converter. The complexity of the designs combined with the rapid pace of product and innovation cycles
frequently conflicts with the requirements in education and research. This speech presents front-end circuits
which aim at providing a low-complexity analog implementation enhanced by digital control and processing
lowering the entrance barrier for students when designing advanced integrated recording systems. This
approach promises the implementation of first-time-right designs and minimizes the system design time by
circuit reuse. Circuit examples are provided which are part of programmable monitoring systems recently
realized in the presenter’s research lab.
Prof. Robert Rieger, Ph.D

National Sun Yat Sen University, Taiwan
Biography
Robert Rieger was born in Duesseldorf, Germany, in 1975. He received the intermediate diploma in electrical
engineering and the B.Eng. degree in electrical and electronics engineering from Chemnitz University of
Technology, Germany, and the University of Kent, U.K., in 1998 and 2000 respectively.
He earned the Ph.D. degree in electronic and electrical engineering at University College London (UCL),
U.K. in 2004. From 2004 to 2005 he was a Design Engineer with the Industry & Medical business unit of
austriamicrosystems AG, Rapperswil, Switzerland (now AMS), where he was concerned with the design of
high-reliability integrated analog circuits in CMOS technology.
Since an internship with Xanavi Informatics Co., Japan, in 1998, which was supported by a grant from the
German Academic Exchange Service (DAAD), he became interested in locating to Asia. Hence, in February
2006 he joined the Electrical Engineering Department of National Sun Yat-sen University, Kaohsiung,
Taiwan, where he is currently a Professor leading the Bionics Integrated Systems Laboratory.
©ICNERE-EECCIS 2016 x
th
Dr. Rieger is a Senior Member of the IEEE and the founding officer of the IEEE Engineering in Medicine and
Biology Society (EMBS) Tainan Chapter. He is a member of the IEEE Technical Committees on Biomedical
and Life Science Circuits and Systems (BioCAS), VLSI Systems and Applications (VTA), and Circuits &
Systems Education and Outreach (CASEO). He is also a member of the Taiwan Chip Implementation Center
(CIC) Peer-Review Committee and Technical Committee Member of the Bureau of Standards, Metrology &
Inspection (BSMI) of the Taiwan Ministry of Economic Affairs. He also serves as an Associate Editor for
IEEE Transactions on Biomedical Circuits & Systems (TBCAS) and for IEEE Transactions on Circuits &
Systems I (TCAS-I). In 2005 he was also a Guest Editor for the special issue on Bioelectronic Circuits for
Enhanced Diagnosis and Therapy of the IEEE Transactions on Circuits and Systems II (TCAS2). He was
awarded with the 2014 IEEE Tainan Section Macronix Award ‘in recognition of outstanding contributions to
advances of circuit design for wearable and implantable biomedical applications.’ His research interests are
in the area of integrated electronics for biomedical application and low-power mixed-signal circuits.
©ICNERE-EECCIS 2016 xi
th
The Future of Nanotechnology for Medical Applications
Masaaki Nagatsu
Shizuoka University
Research Institute of Electronics, Japan
Abstract: In this study we will present our recent experimental results on the advanced plasma nanotechnology
for medical application. The virus and bacteria detection system using the plasma surface-functionalized
carbon-encapsulated magnetic nano-particles (MNPs) fabricated by dc arc discharge and carbon
nanotube(CNT) dot-array prepared with a combined thermal and plasma CVD device are presented. With
antibody-immobilized carbon-encapsulated MNPs, we confirmed that influenza A type H1N1 virus
concentration of 17.3-fold and Salmonella concentration of 70.9-fold were successfully achieved. These
results were better than those obtained using commercially available magnetic beads. The feasibility of CNT
dot-array as a microarray biosensor has been successfully demonstrated by maskless functionalization of
amino (-NH2) and carboxyl (-COOH) groups onto CNTs by using a microcapillary atmospheric pressure
plasma jet.
Keywords: carbon-encapsulated magnetic nanoparticles; plasma surface functionalization; carbon nanotube

array; virus detection; atmospheric pressure plasma jet
Prof. Masaaki Nagatsu

Shizuoka University, Japan
Biography
Masaaki Nagatsu was born in 1952. He earned the Ph.D. from Department of Electrical and Electronics
Engineering, Graduate School of Engineering, Nagoya University, Japan, in 1985. From 1986 to 1988 he was
a Visiting Researcher at University of California, Los Angeles (USA). From 1989 to 1991 he was a Lecturer
at Nagoya University, Japan. From 1991 to 2001 he was an Associate Professor at Nagoya University, Japan.
From 2001 to 2006, he was a Professor at Dept. of Electrical & Electronics Eng. Faculty of Eng., Shizuoka
Univ. From 2006 to 2015, he was a Professor at Graduate School of Science and Technology, Shizuoka Univ.
And from 2015 – present, he was a Professor, Research Institute of Electronics, Shizuoka Univ.
Some his professional activities are in 2008 as an Honorable Guest Professor in Institue of Plasma Physics,
Chinese Academy of Science (China). From 2008 to 2015 he was Dean of Graduate School of Science and
Technology, Shizuoka University. In 2014, he was an Adjunct Professor in University of Indonesia
(Indonesia). In 2015, he was an Honorary Professor in Alexandru Ioan Cuza University (Romania).
©ICNERE-EECCIS 2016 xii

th
His awards are in 2011 The 16th JSPF Award for Notable Contribution to Technology, The Japan Society of
Plasma Science and Nuclear Fusion Research. In 2011, he got Takayanagi Award, Hamamatsu Foundation for
the Promotion of Electronic Science. In 2012, he got The 10th Plasma Electronics Award, Division o0f Plasma
Electronics, Japan Society of Applied Physics. In 2013, he got The 15th Award of Plasma Materials Science
(Fundamental section), The 153rd Committee Plasma Materials Science, Japan Society for the Promotion of
Science. In 2014, he got 8th JSAP Fellow Award, Japan Society of Applied Physics. In 2016, he got Prizes for
Science and Technology(Research Category) The Commendation for Science and Technology by the Minister
of Education, Culture, Sports, Science and Technology. His research interests are on medical and biological
applications of plasma science and technology. He had published total 211 papers publication with peer
review.
©ICNERE-EECCIS 2016 xiii

th
Plasma processing technology for future solar device fabrication
N. Nafarizal
Universiti Tun Hussein Onn Malaysia
Abstract: Plasma processing technology has been a major contributor to the development of front-end of
current microelectronic industries. More than 50% of the processes in microelectronic industries require
plasma processing technology. Therefore, we will not be able to develop nano-device and other advanced and
sophisticated device without understanding the plasma processes. Plasmas have been used to deposit and etch
various type of materials. During the fabrication of first generation silicon solar cell, plasma enhanced
chemical vapor deposition (PE-CVD) have been used to deposit amorphous silicon film. Then, in the second
generation of a solar cell, magnetron sputtering plasmas have been used to deposit various type of materials
in thin film solar cells such as CIGS and Cd-Te. Recently, plasmas have been used to deposit materials for
the third generation of solar cells at lower cost and higher efficiency. Plasmas have been used to alter the
surface reaction of subsequence processes. The plasma process is a very useful tool since the process involves
physical and chemical reactions. In the past, plasma equipment consists of the bulky vacuum chamber and
pumping system. However, in recent technology, plasma system produces at atmospheric pressure condition
and it has been explored for superhydrophobic treatment and thin film deposition.
Ass. Prof. Nafarizal Nayan, D. Eng.

Biography
Nafarizal Nayan received his B. Eng., M. Eng. and D. Eng. degrees from Nagoya University, Japan in 2003,
2005 and 2008, respectively. In 2005, he joined Universiti Tun Hussein Onn Malaysia (UTHM) as Lecturer
during his Master Degree study at Nagoya University. He then, in 2008, start his actual work with Universiti
Tun Hussein Onn Malaysia, as Senior Lecturer. His permanent position is at Department of Electronic
Engineering, Faculty of Electrical and Electronic Engineering.
He headed Microelectronic Laboratory and Microelectronic and Nanotechnology – Shamsuddin Research

Centre (MiNT-SRC) at UTHM in 2009-2013. He was promoted to Associate Professor in 2012. He has been
engaged in the research of the characterization and design of low temperature plasma system using vacuum
and non-vacuum methods. His research interest is on plasma diagnostics, plasma processing and thin film
deposition. From 2013 to 2015, he visited the Quantum Science Engineering, Hokkaido University for visiting
research on Cu2ZnSnS4 thin film fabrication using sputtering technique and synthesis and diagnostics of metal
©ICNERE-EECCIS 2016 xiv

th
nanoparticles. He is the author of more than 20 reviewed-journal papers, 100 conference papers including
international invited talks.
©ICNERE-EECCIS 2016 xv
th
PROGRAM SCHEDULE
ICNERE_EECCIS 2016 SCHEDULE

Venue : Hotel Royal Orchid Garden, Batu - East Java
Parallel Session (Monday - October 31, 2016)

08:00 - 14:30 REGISTRATION
ROOM 1 Session Chair : Lukman Audah; Anton Widarta

Time Paper ID Title presenter
09:00 - 09:15 1570280520 A Review on the Achievable Sum Rate in Massive MIMO Audah
for 5G Wireless Cellular Networks
Adeeb Salh and Lukman Audah (UTHM); Nor Shahida
Mohd Shah (UTHM); Shipun Anuar Hamzah (UTHM);
Jiwa Abdullah (UTHM)
09:15 - 09:30 1570298511 Performance Evaluation of Hypertext Transfer Protocol Audah

Over Terrestrial and Satellite Networks
Liyana Zakaria (TM Berhad, Malaysia); Lukman Audah
and Ansar Jamil (UTHM); Jiwa Abdullah (UTHM)
09:30 - 09:45 1570293703 Power Optimization for Tunable Microwave Generation Yulianto
Based on Mixing Two DFB Laser Using Polarizing Effect
Nursidik Yulianto (Indonesia Institute of Sciences (LIPI),
Indonesia); Purnomo Sidi Priambodo (UI); Bambang
Widiyatmoko (Indonesian Institute of Sciences, Indonesia)
09:45 - 10:00 1570293901 Preliminary Studies on Establishment of Malaysia's Selamat
National Standard of RF Attenuation
Arshad Selamat (National Metrology Institute of Malaysia);
Anton Widarta (National Metrology Institute of Japan,
NMIJ/AIST, Japan); Mohd Nasir Zainal Abidin (National
Metrology Institute of Malaysia, Malaysia)
10:00 - 10:15 1570294220 A Review on the Japan National Standards of Radio Widarta
Frequency and Microwave Attenuation
Anton Widarta (National Metrology Institute of Japan,
NMIJ/AIST, Japan)
©ICNERE-EECCIS 2016 xvi

th
ROOM 2 Session Chair: Angger Abdul Razak; Yuning Widiarti

09:00 - 09:15 1570303446 Bandstop Filter for Radar Application with L Yansyah
Resonator
Dian Astuti (Universitas Mercu Buana, Indonesia);
Ferdi Yansyah (University of Mercu Buana,
Indonesia); Mudrik Alaydrus (Universitas Mercu
Buana, Jakarta, Indonesia)
09:15 - 09:30 1570292291 Joint Passive Time-Reversal and Adaptive Equalizer Widiarti
for Underwater Communication
Yuning Widiarti (Institut Teknologi Sepuluh
Nopember & Politeknik Perkapalan Negeri
Surabaya, Indonesia)
09:30 - 09:45 1570303931 Numerical Examination on Transmission Properties of Abdul
FBG with GPU Support Razak
Angger Abdul Razak and Mitsuhiro Yokota
(University of Miyazaki, Japan)
09:45 - 10:00 1570305750 Air Pollution Monitoring Integrated on Smart Home Samijayani
System Using ZigBee Wireless Sensor Networks
Octarina Nur Samijayani (University of Al Azhar
Indonesia, Indonesia)
10:00 - 10:15 1570305780 Directed Diffusion Based Routing Protocol Wireless Yasri
Sensor Networks for A Forest Preventive Fire System
At Tesso Nilo National Park
Indra Yasri (Universitas Riau, Indonesia)
©ICNERE-EECCIS 2016 xvii

th
ROOM 3 Session Chair: Rudy Yuwono; Ucuk Darusalam.

09:00 - 09:15 1570306345 Log Periodic Microstrip Antenna for Digital Video Alam
Broadcasting Application
Indra Surjati and Yuli Kurnia Ningsih (Trisakti University,
Indonesia); Syah Alam (Universitas 17 Agustus 1945
Jakarta, Indonesia)
09:15 - 09:30 1570306914 Mitigating BER of FSO Communications by Minimizing the Darusalam
Temporal Fluctuation in Signal Spectral
Ucuk Darusalam (Universitas Nasional, Indonesia)
09:30 - 09:45 1570306965 A Paddy Soil Monitoring System Based on Wireless Sensor Putra
Networks
Emansa Hasri Putra (Politeknik Caltex Riau, Indonesia)
10:00 - 10:15 1570308472 Star Patch Antenna for ISM Band Applications Yuwono
Rudy Yuwono and Faisal Ramadhan (University of
Brawijaya, Indonesia); Fauziahanim Che Seman (Universiti
Tun Hussein Onn Malaysia, Malaysia)
©ICNERE-EECCIS 2016 xviii

th
ROOM 4 Session Chair: Yus Natali; Ruth Anindita

09:00 - 09:15 1570316087 Drivetest Methods for 3G Network Quality Analysis Kusmaryanto
Sigit Kusmaryanto (Universitas Brawijaya, Indonesia)
09:30 - 09:45 1570309328 Singly-fed Circularly Polarized Triangular Microstrip Fauzan
Antenna with Truncated-tip for CP-SAR Application
Muhammad Fauzan Edy Purnomo (UB); Hadi Suyono
(UB); Dwi Kurniawan, Ali Mustofa and Sholeh Pramono
(UB); Rahmadwati Rahmadwati (UB)
09:45 - 10:00 1570315073 Requirement Analysis on Received RF Voltage Drive for Natali
Electro-Optic Modulator in Distributed Antenna System
Yus Natali (Akademi Teknik Telekomunikasi Sandhy
Putra Jakarta, Indonesia); Purnomo Priambodo
(Universitas Indonesia, Indonesia)
10:00 - 10:15 1570315264 DESIGN OF E-SLOT MICROSTRIP ANTENNA FOR Anindita
WLAN 2.4 GHz
Ruth Anindita (University of Brawijaya, Indonesia);
Onny Setyawati (Brawijaya University, Indonesia);
Rahmadwati Rahmadwati (University of Brawijaya,
Indonesia)
©ICNERE-EECCIS 2016 xix

th
Program Session (Monday - October 31, 2016)

10:15 - 10:30 COFFEE BREAK
10:30 - 10:45 Dance performance

10:45 - 11:00 Opening speech General chair
11:00 - 11:15 Welcoming Speech Dean of Faculty of
Engineering, Universitas
Brawijaya
11:15 - 11:45 Keynote Speaker 1 : Prof. Robert Rieger, Ph.D.
"Student centered design of CMOS programmable
biosignal recorders"
11:45 - 13:15 LUNCH
13:15 - 13:45 Keynote Speaker 2 : Prof. Dr. Masaaki Nagatsu.

"The Future of Nanotechnology for Medical
Applications"
13:45 - 14:15 Keynote Speaker 3 : Assoc. Prof. Dr. Nafarizal bin
Nayan
"Plasma processing technology for future solar device
fabrication"
©ICNERE-EECCIS 2016 xx
th
Parallel Session (Monday - October 31, 2016)

ROOM 1 Session Chair: Y Simamora; Andriani Parastiwi.
14:30 - 14:45 1570289325 Study of Solar Electricity Powered Air Conditioner in a Tarigan
Tropical Hot Climate
Elieser Tarigan (Electrical Department and Center for
Renewable Energy Studies, PSET, University of
Surabaya)
14:45 - 15:00 1570293989 Phase Shifted Boost-Cuk Converter for Renewable Energy Parastiwi
Applications
Alfredo Medina, Alam Salguero and Taufik Taufik
(California Polytechnic State University, San Luis Obispo,
USA); Andriani Parastiwi (State Polytechnics of Malang,
Indonesia)
15:00 - 15:15 1570294013 Integration of Distributed Generation in Radial Nadhiroh

Distribution Systems
Nuha Nadhiroh (INSTITUT TEKNOLOGI SEPULUH
NOPEMBER, Indonesia); Ontoseno Penangsang (Institut
Teknologi Sepuluh Nopember, Indonesia)
15:15 - 15:30 1570294109 Estimation Losses in Radial Distribution Network Under Simamora
Unbalanced Loading Condition Considering Time
Varying Load
Y Simamora (INSTITUT TEKNOLOGI SEPULUH
NOPEMBER, Indonesia); S Suyanto and O Penangsang
(Lecturer, Indonesia)
15:30 - 15:45 1570303295 Influence of Load Properties on the Performance of Low- Santoso
Speed Single-Phase Self-Excited Induction Generator
Hari Santoso (Brawijaya University, Indonesia); Rini Nur
Hasanah (Brawijaya University & Faculty of
Engineering, Indonesia); I Nyoman Gde Wardana and
Budiono Mismail (Brawijaya University, Indonesia)
15:45 - 16:00 1570303539 Power System Stabilizer Optimization Based on Modified Duanaputri
Differential Evolution Algorithm and Network Reduction
Considering Network Losses
Rohmanita Duanaputri and Adi Soeprijanto (ITS,
Indonesia); Ardyono Priyadi (ITS, Indonesia); Dheny
Ashari Hasan (ITS, Indonesia)
©ICNERE-EECCIS 2016 xxi

th
ROOM 2 Session Chair: Dikpride Despa; Ida Bagus Fery Citarsa

14:30 - 14:45 1570303833 Optimum Operation Stand-Alone Microgrid Using PSO Kusuma
Algorithm to Considering Life Time Battery
Vicky Kusuma (Jl. Raya ITS Keputih, Surabaya, Jawa
Timur, Indonesia)
14:45 - 15:00 1570304628 Implementing of Lyapunov Function to PID Voltage Control Zulkifli
for Single Phase Rectifier Using Low Cost Microcontroller
Shamsul Zulkifli (Universiti Tun Hussein Onn Malaysia,
Malaysia)
15:00 - 15:15 1570306762 Analysis of Bidirectional VSI as Regenerative Braking Andrean
Converter for BLDC Motor
Rini Nur Hasanah (UB, Indonesia); Victor Andrean,
Soeprapto Soeprapto and Hadi Suyono (UB, Indonesia)
15:15 - 15:30 1570307040 Comparison Analysis of Phase-shifted and Level-Shifted Citarsa
Carrier-Based PWM on the Five-Level Single- Phase
Cascaded H-Bridge (CHB) Multilevel Inverter
Ida Bagus Fery Citarsa (University of Mataram, Indonesia)
15:30 - 15:45 1570307053 Multi-Area Smart Monitoring of Electrical Quantities Based Despa
on Mini Single Board Computer BCM 2835
Dikpride Despa, Mardiana Rendra and Gigih Forda Nama
(University of Lampung, Indonesia); Michael Bernard
(University of Alberta - Edmonton, Alberta, Canada);
Muhamad Komarudin (University of Lampung, Indonesia)
15:45 - 16:00 1570308005 Schottky Diodes Comparison for Differentially-Driven Zulkifli
Rectifier
Fitri Yuli Zulkifli, Amirsyah Rayhan Mubarak and Basari
Basari (Universitas Indonesia, Indonesia); Eko Tjipto
Rahardjo (University of Indonesia, Indonesia)
©ICNERE-EECCIS 2016 xxii

th
ROOM 3 Session Chair Hadi Suyono; Setyo Nugroho.

14:45 - 15:00 1570308946 A Numerical Study on Twist-Tappered Blade for Low- Nugroho
Speed Horizontal Wind Turbine
Joke Pratilastiarso and Setyo Nugroho (Politeknik
Elektronika Negeri Surabaya, Indonesia); Teguh Hady
Ariwibowo and Akhmad Khilmy (Electronic Engineering
Polytechnic Institute of Surabaya, Indonesia)
15:00 - 15:15 1570309200 Injection Impact of Hybrid Embedded Generation in Suyono
Distribution System
Hadi Suyono (UB, Indonesia); Rini Nur Hasanah (UB,
Indonesia); Panca Mudjirahardjo (UB, Indonesia);
Muhammad Fauzan Edy Purnomo (UB, Indonesia)
15:15 - 15:30 1570311715 The Effect of Position of Anti Notch on the Surface of Stator Wahab
Teeth and Permanent Magnet to Cogging Torque Reduction
Herlina Wahab (Sriwijaya University & University of
15:45 - 16:00 1570321666 Design of Floating Waterwheel for Harnessing River-Flow Suparman
Surface-Stream Energy
Suparman Suparman and Hadi Suyono (Brawijaya
University, Indonesia); Rini Nur Hasanah (Brawijaya
University & Faculty of Engineering, Indonesia)
©ICNERE-EECCIS 2016 xxiii

th
ROOM 4 Session Chair: Catur Apriono; Nabihah Ahmad.

14:30 - 14:45 1570320416 Design of an Automation System for Hydroponic Plantation Wahab
Based on the Ebb & Flow Method
Wahidin Wahab (University of Indonesia, Indonesia);
Andhika Djaffri and Winarsita Sotyaji (Universitas
14:45 - 15:00 1570313969 Data-Logging in an Acidity Measurement System Using pH- Julius
Sensor with Circular Nickel-Wire Electrodes
M. Julius St. (UB, Indonesia); Sudjito Suparman, Eko
Siswanto and Denny Widhiyanuriyawan (UB, Indonesia)
15:15 - 15:30 1570287943 A Conceptual Framework of High-Dimensionality CAP Saon
Modulation on Programmable Chip for Indoor VLC Network
Sharifah Saon, Abd Kadir Mahamad and Maisara Othman
(Universiti Tun Hussein Onn Malaysia, Malaysia)
15:30 - 15:45 1570306032 Dual-Port SRAM Design with Low Power Technique Ahmad
Nabihah Ahmad (Universiti Tun Hussein Onn Malaysia
& Faculty of Electrical and Electronic Engineering,
Malaysia); Nurdyana Mat Jainudin (Universiti Tun Hussein
Onn Malaysia, Malaysia)
18:30 - 20:30 GALA DINNER
©ICNERE-EECCIS 2016 xxiv

th
Parallel Session (Tuesday - November 1, 2016)

08:00 - 13:00 REGISTRATION
ROOM 1 Session Chair: Gunawan Wibisono; Anak Agung Ngurah Gde Sapteka.
09:00 - 09:15 1570306266 Design and Implementation of M2M Technology in Wibisono
Smart Street Light System Node with RS 485
Communication and Hybrid Power Source
Gunawan Wibisono (University of Indonesia, Indonesia);
Antonius Putranto (Universitas Indonesia, Indonesia)
09:30 - 09:45 1570308088 A Strategy to Reduce Stray Capacitance on Electrical Yusuf
Capacitance Tomography Hardware
Arbai Yusuf (Universitas Indonesia & C-Tech Labs
Edwar Technology, Indonesia)
09:45 - 10:00 1570309216 A Study of Electric Field of Gaussian Silicon P-N Sapteka
Junction
Anak Agung Ngurah Gde Sapteka, Anak Agung Ngurah
Made Narottama, I Nyoman Sukarma and I Gusti Putu
Mastawan Eka Putra (Politeknik Negeri Bali, Indonesia);
Mangaraja Longgam Saragi (Uniiversitas Mpu Tantular,
Indonesia)
10:00 - 10:15 1570301615 The Development of Kartini Reactor Data Acquisition Abimanyu
System to Support Nuclear Training Centre (NTC)
Adi Abimanyu (Center for Science and Technology
Accelerator, Indonesia); Syarip Syarip and Elisabeth
Supriyatni (Center for Accelerator Science and
Technology, Indonesia)
©ICNERE-EECCIS 2016 xxv

th
ROOM 2 Session Chair: Onny Setyawati; Retno Purnamaningsih

09:00 - 09:15 1570309326 Design of a Low-Cost Microcontroller-Based ECG Setyawati;
Signal Generator Setyawan
Nanang Sulistiyanto (UB, Indonesia); Mochammad
Rif'an (UB, Indonesia); Raden Arief Setyawan and
Onny Setyawati (UB, Indonesia)
09:15 - 09:30 1570313974 The Evaluation of Tropical Tuber Properties Using Setyawati
Far Field Method
Onny Setyawati and Nanang Sulistiyanto
(Brawijaya University, Indonesia)
09:30 - 09:45 1570317031 Modelling of a Four Branch Power Splitter Using Purnamaningsih
III-nitrides Semiconductor for Long Wavelength
Optical Telecommunication
Retno Purnamaningsih and Nji Poespawati
10:00 - 10:15 1570303963 ScanCart: Android Barcode Scanner for E Saon

Commerce Stock Inventory
Sharifah Saon (Universiti Tun Hussein Onn
Malaysia, Malaysia)
ROOM 3 Session Chair : Agung Darono; Mochammad Rusli

09:00 - 09:15 1570291090 Use of IT in Indonesian Tax Administration System: Darono
Duality of Tax Technology Perspective
Agung Darono (Tax Education and Training Center
& Ministry of Finance, Indonesia)
09:15 - 09:30 1570303820 Flight Control System Improvement for Quadrotor Using Adriansyah
Particle Swarm PID (PS-PID)
Andi Adriansyah, Anwar Minarso and Eko Ihsanto
(Universitas Mercu Buana, Indonesia)
09:45 - 10:00 1570309202 Design of Adaptive Sliding Mode Control for DC Motor Subroto
Speed Control
Ramadhani Kurniawan Subroto (Brawijaya University,
Indonesia)
10:00 - 10:15 1570311549 Study of Design of A Double-Sided LIM (DSLIM) with Rusli
Ladder Secondary for Slow Speed Purpose
Mochammad Rusli (Brawijaya University, Indonesia)
©ICNERE-EECCIS 2016 xxvi

th
ROOM 4 Session Chair :Muhammad Aziz Muslim; Youness Lahdili

09:00 - 09:15 1570308165 Conception of an Autonomous Driving of Unmanned Lahdili
Vehicles Using Stereo Vision
Youness Lahdili (UTHM); Afandi Bin Ahmad (UTHM)
09:15 - 09:30 1570308096 Injection Duration Control in a Four Strokes Engine Based Aziz
on Throttle Position Muslim
Muhammad Aziz Muslim, Goegoes Dwi Nusantoro, Dwi
Fadilla Kurniawan and Ganda Lesmana (Brawijaya
University, Indonesia)
09:30 - 09:45 1570316300 Ignition Timing Prediction of a Four Strokes Engine Using Aziz
ANFIS Predictor Muslim
Muhammad Aziz Muslim, Dwi Fadilla Kurniawan and
Goegoes Dwi Nusantoro (UB, Indonesia)
09:45 - 10:00 1570319166 Traffic Monitoring System for Wireless Network Using Nurwarsito
Embedded System
Heru Nurwarsito and Gembong Setyawan (UB, Indonesia);
Kautsarani Alam (UB, Indonesia)
10:00 - 10:15 1570305967 Design of Sensor Reading Visualization on Line Follower Zainuri
Robot
Akhmad Zainuri (Brawijaya University, Indonesia)
©ICNERE-EECCIS 2016 xxvii

th
ROOM 1 Session Chair : Nafarizal Nayan; Fajar Setiawan

10:30 - 10:45 1570293958 Nitrogen Gas Temperature Evaluation in Pulse- Nayan
Modulated RF Magnetron Sputtering Plasmas
Nafarizal Nayan (UTHM; Microelectronic and
Nanotechnology - Shamsuddin Research Centre
(MiNT-SRC), Malaysia); Soo Ren How (UTHM,
Malaysia); Mohd Khairul Ahmad (UTHM, Malaysia);
Mohd Zainizan Sahdan (UTHM, Malaysia)
10:45 - 11:00 1570294130 Surface Characterization of Glass Substrate Exposed by Nayan
Atmospheric Pressure Plasma Needle
Elfa Rizan Rizon (UTHM, Malaysia); Nafarizal Nayan
(UTHM; Microelectronic and Nanotechnology -
Shamsuddin Research Centre (MiNT-SRC), Malaysia);
Chin Fhong Soon (UTHM; Microelectronic and
Nanotechnology-Shasmsuddin Research Center,
Malaysia); Mohd Khairul Ahmad (UTHM, Malaysia);
Mohd Zainizan Sahdan (UTHM, Malaysia)
11:00 - 11:15 1570294215 Investigation of Gas Temperature Evaluation in Nayan
Magnetron Sputtering Plasma Using Two Different
Resolution of Spectroscopy
Soo Ren How (UTHM, Malaysia); Nafarizal Nayan
(UTHM; Microelectronic and Nanotechnology -
Shamsuddin Research Centre (MiNT-SRC), Malaysia);
Jais Lias (UTHM, Malaysia)
11:30 - 11:45 1570304085 Silicon Single Boron Transistor Ono
Yukinori Ono (Shizuoka University, Japan)
©ICNERE-EECCIS 2016 xxviii

th
ROOM 2 Session Chair : Murakami Kenji; Hiroshi Inokawa.

10:30 - 10:45 1570303961 Growth of Vertically Aligned ZnO Nanorods Array and Kenji
Its Application to Perovskite-type Solar Cells
Murakami Kenji, Albertus Bramantyo and Masayuki
Okuya (Shizuoka University, Japan); Nji Poespawati
and Arief Udhiarto (Universitas Indonesia, Indonesia)
10:45 - 11:00 1570280335 A Framework for Optimum Contour Detection Crysdian
Cahyo Crysdian (UIN MALIKI Malang, Indonesia);
Bayu Adhi Nugroho (Universitas Islam Negeri (UIN)
Sunan Ampel Surabaya, Indonesia)
11:00 - 11:15 1570303995 Comparative Study on Metal Resistor and MOSFET- Inokawa
Based THz Bolometers
Hiroshi Inokawa, Ajay Tiwari, Takeo Ueta and Hiroaki
Satoh (Shizuoka University, Japan); Catur Apriono
(Universitas Indonesia, Indonesia); Eko Tjipto Rahardjo
(University of Indonesia, Indonesia); Norihisa Hiromoto
(Shizuoka University, Japan)
11:30 - 11:45 1570309197 Non Invasive Sensors for Bearing Damage Aswin
Identification Using Sound Signal Frequency Analysis
Muhammad Aswin (Brawijaya University - Malang,
Indonesia)
©ICNERE-EECCIS 2016 xxix

th
ROOM 3 Session Chair Mohd Khairul Ahmad; Yoshimasa Kawata.

10:30 - 10:45 1570304409 Improvement in Film Quality of Hexagonal Boron Nitride Hara
Grown by Chemical Vapor Deposition
Kazuhiko Hara, Naoki Umehara, Atsushi Masuda, Takaki
Shimizu, Tetsuya Kouno and Hiroko Kominami (Shizuoka
University, Japan)
10:45 - 11:00 1570304881 The Effect of Titanium Dioxide Nanorods to the Ahmad
Performance of Dye-Sensitized Solar Cell
Mohd Khairul Ahmad (UTHM, Malaysia); Nurul Syafiqah
Mohamed Mustakim (UTHM, Malaysia); Chin Fhong Soon
(UTHM; Microelectronic and Nanotechnology-Shasmsuddin
Research Center, Malaysia); Nafarizal Nayan (UTHM;
Microelectronic and Nanotechnology - Shamsuddin Research
Centre (MiNT-SRC), Malaysia); Rahmat Sanudin (UTHM,
Malaysia); Suriani Abu Bakar and Azmi Mohamed
(Universiti Pendidikan Sultan Idris, Malaysia); Murakami
Kenji and Shimomura Masaru (Shizuoka University, Japan);
Mohamad Hafiz Mamat and Mohd Firdaus Malek (Universiti
Teknologi MARA, Malaysia)
11:00 - 11:15 1570309187 Label-Free Cellular Structure Imaging with High Resolution Kawata
Using an Electron-Beam Excitation- Assisted Optical
Microscope
Yoshimasa Kawata (Shizuoka University, Japan)
11:15 - 11:30 1570303932 Response of Microcantilever Vibration Due to CO Gas in Aprilia

Vacuum Condition
Lia Aprilia (Shizuoka University, Japan); Ratno Nuryadi
(BPPT, Indonesia); Djoko Hartanto (University of Indonesia,
Indonesia); Makoto Hosoda, Yoichiro Neo and Hidenori
Mimura (Shizuoka University, Japan)
11:30 - 11:45 1570304627 Minimizing Tower of Base Tranceiver Station with Rofii
Considering Coverage Area Using Fuzzy Clusterring Means
and Particle Swarm Optimization (FCM-PSO) - Faqih Rofii,
MT (Universitas Widyagama Malang, Indonesia); Diky
Siswanto (University of Leeds, United Kingdom); Fachrudin
Hunaini (Widya Gama University, Indonesia); Ahmad Rijal
Kafy (Universitas Widyagama Malang, Indonesia)
©ICNERE-EECCIS 2016 xxx

th
ROOM 4 Session Chair Ihsan Ibrahim; Muhammad Muzakkir Mohd Nadzri.

10:30 - 10:45 1570303962 Single and Coupled-Dopants as Quantum Dots for Moraru
Room Temperature Single-Electron Tunneling
Daniel Moraru (Shizuoka University & Research
Institute of Electronics, Japan); Arup Samanta and
Michiharu Tabe (Shizuoka University, Japan)
10:45 - 11:00 1570294232 Design & Implementation of Wildlife (Tiger) Mohd
Surveillance System to Prevent Poaching Nadzri
Muhammad Muzakkir Mohd Nadzri (UTHM,
Malaysia); Afandi Bin Ahmad (UTHM; Faculty of
Electrical and Electronic Engineering, Malaysia);
Abbes Amira (Qatar University, Qatar)
11:00 - 11:15 1570294267 Conpot Analysis: Low-Interaction Honeypot as Attack Ibrahim
Detection System on Industrial Network Control
System / SCADA
Anak Agung Putri Ratna (UI, Indonesia); Mahardianto
Yudha Bestari, Ihsan Ibrahim, Prima Dewi Purnamasari
and Fransiskus Astha Ekadiyanto (UI, Indonesia);
Muhammad Salman (UI, Indonesia)
11:15 - 11:30 1570294275 NOAA Satellite Based with OpenCV Library Weather Ibrahim
Monitoring System Software Development
Ihsan Ibrahim (UI, Indonesia); Gunawan Wibisono (UI,
Indonesia); Prima Dewi Purnamasari (UI, Indonesia)
11:30 - 11:45 1570295207 Application of Microcapillary Atmospheric Pressure Abuzairi
Plasma Jet as a Patterned Surface Modification of
Carbon Nanotube Microarray
Tomy Abuzairi (Shizuoka University, Japan;
Universitas Indonesia, Indonesia)
11:45 - 13:00 LUNCH
©ICNERE-EECCIS 2016 xxxi

th
ROOM 1 Session Chair Fransiskus Astha Ekadiyanto; Farid P Putera.

13:00 - 13:15 1570307008 Greedy Algorithm and FIS Sugeno for Determining Aziz
Evacuation Routes in Eruption of Merapi Mountain in
Indonesia
Abdul Aziz (Universitas Sebelas Maret, Indonesia);
Ananto Setyadi (Sebelas Maret University, Indonesia);
Esti Suryani (University of Sebelas Maret, Indonesia)
13:15 - 13:30 1570309233 Predict Drug Addict Using Linear Discriminant Analysis Priambodo
Based on Eyes and Faces Picture
Bagus Priambodo (University of Mercu Buana)
13:30 - 13:45 1570317579 A Performance Analysis of Online Video Classification Ekadiyanto
System Based on Tag Using MapReduce for Internet
Content Profiling
Fransiskus Astha Ekadiyanto and Adhika Putra
13:45 - 14:00 1570318821 3D Human Brain Tumor Detection from MRI Image Putera
Using K-Means Clustering, Non-Local Means,
Thresholding and Marching Cube
Dodi Sudiana, Farid P Putera and Basari Basari
14:00 - 14:15 1570320423 APT Detection Design Using DNS Traffic Analysis Manggalanny
Muhammad Manggalanny (UI; ID-SIRTII/CC,
Indonesia); Kalamullah Ramli (Universitas Indonesia,
Indonesia)

21:00 - 22:00 PICK UP AT HOTEL FOR TOUR TO BROMO MOUNT
©ICNERE-EECCIS 2016 xxxii

The 8th Inter. Conf. on Electrical, Electronics, Communications, Controls and Informatics System (ICNERE-EECCIS 2016).
____________________________________________________________________________________________________________
The Future of Nanotechnology for Medical

Applications
Masaaki Nagatsu1,2,3*, Anchu Viswan1, Tomy Abuzairi1,4, Mitsuru Okada2, Kuniaki Sugiura2,
Retno W. Purnamaningsih4, and Nji R. Poespawati4
1
Graduate School of Science and Technology, Shizuoka University, 3-5-1, Johoku, Naka-ku, Hamamatsu 432-8561, Japan
2
Graduate School of Integrated Science and Technology, Shizuoka University, 3-5-1, Johoku, Naka-ku, Hamamatsu 432-8561,
Japan
3
Research Institute of Electronics, Shizuoka University, 3-5-1, Johoku, Naka-ku, Hamamatsu 432-8011, Japan
4
Department of Electrical Engineering, Faculty of Engineering, Universitas Indonesia, Depok 16424, Indonesia
*nagatsu.masaaki@shizuoka.ac.jp
Abstract— In this study we will present our recent experimental materials adsorbed by the MNPs. Here, we use carbon-
results on the advanced plasma nanotechnology for medical encapsulated MNPs fabricated by using dc arc discharge.[3-7]
application. The virus and bacteria detection system using the The encapsulation by carbon layers not only prevents the core
plasma surface-functionalized carbon-encapsulated magnetic metal particles from rapidly oxidizing but also reduces their
nano-particles (MNPs) fabricated by dc arc discharge and
agglomeration after appropriate surface modification.
carbon nanotube(CNT) dot-array prepared with a combined
thermal and plasma CVD device are presented. With antibody-
immobilized carbon-encapsulated MNPs, we confirmed that Nano-drug
influenza A type H1N1 virus concentration of 17.3-fold and (drug delivery
Salmonella concentration of 70.9-fold were successfully achieved. Nano-bioimaging system, nano- Nano-materials
These results were better than those obtained using commercially (fluorescence, capsule, etc.) (nanoparticles,
available magnetic beads. The feasibility of CNT dot-array as a quantum-dot, nanophosphors,
microarray biosensor has been successfully demonstrated by MRI, etc.) CNT, etc.)
maskless functionalization of amino (-NH2) and carboxyl (-
Nanotechnology
COOH) groups onto CNTs by using a microcapillary
atmospheric pressure plasma jet. Nano-robotics Nano-device
(nanoengine, (protein sensor,
nanomachine, virus & bacteria
Keywords— carbon-encapsulated magnetic nanoparticles, Nano-therapy
etc.) detection, etc.)
plasma surface functionalization, carbon nanotube array, virus (cancer, drug,
detection, atmospheric pressure plasma jet nanoparticles,
targeting, etc.)
Fig. 1 Application of nanotechnology in the medical field.

I. INTRODUCTION
So far, the nanotechnology has played an important role in In this study, we aim at developing high sensitive virus and
development of various industrial fields. Especially, in the bacteria detection using plasma- functionalized and antibody-
medical field, the nanotechnology enormously exploited new integrated carbon encapsulated MNPs.[8-10] We have also
research fields, such as nano-materials, nano-devices, nano- carried out the immobilization of biomolecules onto CNT dot
therapy[1], nano-robotics[2], nano-bioimaging, and nano-drug, array functionalized by using an ultrafine atmospheric
etc. (see Fig. 1) Among these nanotechnologies in medical pressure plasma jet (APPJ).[11-12] CNT dot arrays were
application, nano-materials such as magnetic nanoparticles or prepared with a combined thermal and plasma CVD devices
gold nanoparticles have been extensively investigated to and constructed in dot array form for realizing the
improve the efficiencies of drug delivery system or virus and development of biochip device.
bacteria detection system for realizing a safe and secure long-
lived society. As for nano-devices, such as protein sensor or II. EXPERIMENTAL SETUP AND RESULTS
virus detection chip, they are also growing more and more in
demand for rapid detection of infectious diseases and 2.1 VIRUS AND BACTERIA CONCENTRATION BY ANTIBODY
unknown virus. IMMOBILIZED MNPS
As for the nano-material synthesis, magnetic nanoparticles In this study, carbon-encapsulated MNPs were made by the
(MNPs) have been widely studied for biomedical and DC arc discharge method.[3-7] The chamber of arc discharge
environmental applications because of their high specific device was filled with He/CH4 mixture gas and pressure is
surface area and the ease of magnetic collection of target
maintained at 100 Torr. We implanted Fe2O3, graphite
©ICNERE-EECCIS 2016 1
____________________________________________________________________________________________________________
powder and graphite bond at ratio 1:1:2 to graphite rod as Nucleoprotein Antigen Capture ELISA kit in accordance with
discharge electrodes. DC voltage of 20 V was supplied to the manufacturer’s instructions.
electrode and flowing current was about 100 A. Iron and
graphite was evaporated uniformly by arc discharge heat, then BF GrMNPs SP(supernatant)
SP BD
iron particles as core metal are capsulated by graphite layers. Salmonella
The particles diameter is found to be about 10-50 nm in
diameter. The average size is about 20 nm.
Bacteria-infected Incubated with Magnetic Discard supernatant
sample GrMNPs separation and wash
Influenza antibody
Influenza antibody Higher concentration Lower concentration
immobilized MNPs Higher concentration
~105 CFU/ml Lower
~10 concentration
2 CFU/ml
x 70.9
MNP
immobilized MNPs
~105 CFU/ml ~102 CFU/ml
MNP
x 70.9
Antibody x 27.5
MNP specifically- MNP
x 27.5
linked with MNP
antigen of
influenza virus
Virus (or Bacteria)
~105 CFU/ml ~102 CFU/ml

~105 CFU/ml x 5.7 ~102 CFU/ml
Selective virus x 5.7 x 0.97
MN
P + collection and x 0.97
condensation
Influenza antibody Virus
immobilized MNPs (or bacteria) Magnetic separation
Fig. 2 Illustration of the virus or bacteria magnetic concentration procedure Fig. 3 Procedure of Salmonella collection by antibody immobilized MNPs,
using antibody- immobilized MNPs. and comparison of the results between our MNPs and commercially available
ones, for two different concentrations of Salmonella bacteria.[8]
Figure 2 shows the illustration of the virus or bacteria
magnetic concentration procedure using antibody-
immobilized MNPs. To examine the capacity of antibody-
integrated MNPs to capture Salmonella from liquid, direct
culture assay was performed.
Figure 3 shows the comparison of their results between our
MNPs and commercially available ones, for two different
concentrations of Salmonella bacteria.[8] Here, BF, SP and
BD stands for samples before incubation with the MNPs,
supernatant after incubation, and sample after magnetic
collection by MNPs, respectively, and TL means the total
sample containing the same quantity of culture medium as BD.
The present results suggest that the antibody-integrated MNPs
concentrated the Salmonella 70.9-fold in higher concentration
case of ~105 CFU/ml (Fig. 3A) and 27.5-fold in lower
concentration case of ~102 CFU/ml(Fig. 3B), respectively.
We also tested the commercially available anti-Salmonella
MNPs (Dynabeads) using the same direct culture assay. The Fig. 4 Concentration and capture of influenza A virus using antibody-
corresponding concentration factor using magnetic beads for integrated magnetic beads.[9]
different dilutions of Salmonella culture was 5.7 and 0.97-fold
for ~105 CFU/ml (Fig. 3C) and ~102 CFU/ml(Fig. 3D), After magnetic separation, the amount of influenza virus
respectively. recovered by the antibody-integrated magnetic beads (BD)
Furthermore, we have also carried out the experiments of was 72.8 ng/ml. The sample before incubation with the beads
influenza virus concentration using the influenza antibody- (BF) contained 4.2 ng/ml, whereas an aliquot of supernatant
immobilized MNPs. From the experimental results, we after incubation (SP) contained only 4.0 ng/ml (see Fig. 4).
confirmed that influenza A H1N1 virus concentrations of These findings suggest that the antibody-integrated magnetic
17.3-fold and 10.9-fold were achieved by using anti-influenza beads concentrated the influenza virus 17.3-fold, and are
A virus hemagglutinin (HA) antibody [9] and matrix consistent with the results from immunochromatography and
protein(M1) antibody, respectively. RT-PCR. [9] Furthermore, when magnetic beads immobilized
We also examined the efficiency with which influenza with anti-influenza A virus antibody C179 by SPDP at 0.03
virus was concentrated by this method by conducting a μM and 3 μM were used for virus capture, influenza A virus
quantitative analysis with ELISA using an Influenza A virus was concentrated 16.1-fold and 11.3 fold, respectively (data
____________________________________________________________________________________________________________
not shown), further demonstrating the successful capture and were biotinylated by Biotin N-hydroxysuccinimide ester
concentration of influenza A virus by the antibody-integrated (Biotin Sulfo-OSu) in phosphate buffered saline (PBS, pH
magnetic beads. 7.4) solution. To block non-specific binding of untreated
CNT from avidin, the biotinylated CNT samples were added a
Bovine serum albumin (BSA) surfactants in PBS. Finally, the
2.2 CNT DOT ARRAY FUNCTIONALIZED BY USING AN CNT samples were dropped by avidin-fluorescein
ULTRAFINE APPJ FOR BIOCHIP SENSOR[11,12] isothiocyanate (FITC) in PBS and followed by rinsing with
We have also studied the immobilization of biomolecules onto PBS buffer and deionized water, and dried by compressed gas
CNT dot array functionalized by using an ultrafine APPJ. The avidin-FITC connection to the biotinylated CNTs were
CNT dot arrays were synthesized by a combined thermal and evaluated from fluorescence signals recorded using
plasma CVD devices and constructed in dot array form for fluorescent microscope. The biotin-avidin biomolecules are
realizing the development of biochip device. For patterning of chosen because it has very strong interaction between analyte
surface modification onto CNTs, an APPJ with a micro- and ligand, and provides universal platform for patterning
capillary of ~1 µm was utilized. Plasma jet was produced in various biomolecules in the biosensor systems.
the glass tube under the high voltage square-wave pulses of
±7.5 kV at a low frequency of 5 kHz and 50 % duty ratio. The
substrate was scanned automatically by computer-controlled
stage for patterning hundred of CNT dot array, as shown in
Fig. 5.
Capillary APPJ
Y
0 100 200 300
[μm]
CNT
Si substrate X
CNT dot array
Fig. 6 Schematic description of the covalent immobilization of biotin-avidin
movable onto CNT microarray platform.[12]
Fig. 5 Illustration and photo of maskless patterning of CNT dot array by using
a ultrathin APPJ(top). Photos of (a)CNT dot array, and the line patterns of (b)
amino (-NH2) and (c) carboxyl (-COOH) groups onto CNT dot array substrate
confirmed by chemical derivatization technique visualized by fluorescent
microscope.[11]
The patterning of CNT dot-array as a microarray biosensor

has been demonstrated by successful maskless
functionalization of amino (-NH2) and carboxyl (-COOH) Fig. 7 Biotin-avidin patterning on the CNT microarray platform via ultrafine
groups onto CNTs by using an ultrafine APPJ. The APPJ. Dark-field image covered with avidin-FITC: (a) “S” pattern and (b)
experimental results of chemical derivatization with the “U” pattern taken from the initials of “Shizuoka University”.[12]
fluorescent dye showed that the CNT dot-array was not only
Finally, to realize maskless and localized biotin-avidin
successfully functionalized with amino group and carboxyl
patterning on the CNT microarray platform, an ultrafine APPJ
group but was also functionalized without any interference
with the optimum treatment period of 0.2 s for pre-treatment
between functional groups. As shown in Figs. 5b) and (c), the
and 2.5 s for post-treatment was employed to create “S” and
success of maskless functionalization in the line pattern
“U” letter patterns (taken from the initials of “Shizuoka
provides a means of a multi-functionalization CNT dot-array
University”). Figures 7(a) and (b) show the fluorescent images
with direct implications for future application of a microarray
of the biotin-avidin-FITC patterning on the CNT microarray
biosensor.
platform with “S” and “U” letter patterns respectively, without
Following the amino group functionalization by ultrafine
any significant interference between the CNT spot. [12]
APPJ, the immobilization of biotin-avidin system was
performed on the CNT microarray platform as shown in the III. CONCLUSIONS
schematic description (Fig. 6). Initially, the CNT samples
____________________________________________________________________________________________________________
In this study we presented the results on the virus and pressure plasma jet using biotin-avidin system”, Appl. Phys. Lett. 109
(2016) 023701(4 pages).
bacteria detection system using the surface-functionalized
carbon-encapsulated MNPs, and CNT dot-array prepared
treated by low-pressure RF plasma and the ultrafine APPJ,
respectively. The former results using antibody-immobilized
MNPs suggest that the proposed techniques are promising for
the future development for rapid and high sensitive detection
of virus and bacteria.
The selective biomolecules immobilization onto CNT dot
array with the ultrafine APPJs was also confirmed by
chemical derivatization technique with the fluorescent dye
visualized by fluorescent microscope. This methodology
allowed introducing patterned biomolecules onto CNT dot
array for the development of CNT biochip device.
ACKNOWLEDGMENT
This work was supported by Grants-in-Aid for Scientific
Research (A) (No. 25246029) from the Japan Society for the
Promotion of Science.
REFERENCES
[1] V. Wiwanitkit, “Cancer nanotherapy: Concept for design of new drug”,
J. Med. Hypotheses Ideas, 7(1)(2013) 3.
[2] R. A. Freitas, Jr., Nanomedicine, Volume I: Basic Capabilities, Landes
Bioscience, Georgetown, TX (1999); http://www.nanomedicine.com/
NMI.htm.
[3] M. Nagatsu, T. Yoshida, M. Mesko, A. Ogino, T. Matsuda, T. Tanaka,
H. Tatsuoka, K. Murakami, “Narrow multi-walled carbon nanotubes
produced by chemical vapor deposition using catalytic metal-carbon
nanoclusters”, Carbon, 44 (2006) 3336.
[4] T. E. Saraswati, A. Ogino, M. Nagatsu, “ RF Plasma-Activated
Immobilization of Biomolecules onto Graphite-Encapsulated Magnetic
Nanoparticles”, Carbon, 50 (2012) pp.1253-1261.
[5] T. E. Saraswati, S. Tsumura, and M.i Nagatsu, “ High-efficiency
plasma surface modification of graphite-encapsulated magnetic
nanoparticles using a pulsed particle explosion technique”, Jpn. J. Appl.
Phys. 53 (2014) 010205.
[6] E. Yang, H. Chou, S. Tsumura and M. Nagatsu, “Surface Properties of
Plasma Functionalized Graphite-Encapsulated Gold Nanoparticles
Prepared by Direct Current Arc Discharge Method”, J. Phys. D; Appl.
Phys. 49(18) (2016) 185304.
[7] A. Viswan, H. Chou, A. Sakudo, and M. Nagatsu, “Bioconjugation
Efficiency of Plasma-Functionalized Carbon-Encapsulated Iron
Nanoparticles with Biotin-Avidin System”, Biomed. Phys. Eng.
Express 1 (2015) 045104(9 pages).
[8] A. Sakudo, H. Chou, and M. Nagatsu, “Antibody-integrated and
functionalized graphite-encapsulated magnetic beads, produced using
ammonia gas plasma technology, for capturing Salmonella”, Bioorg.
Med. Chem. Lett., 25 (2015) 1012.
[9] A. Sakudo, H. Chou, K. Ikuta, and M. Nagatsu, “Integration of
antibody by surface functionalization of graphite-encapsulated
magnetic beads using ammonia gas plasma technology for capturing
influenza A virus”, Bioorg. Med. Chem. Lett., 25 (2015) 1876.
[10] A. Sakudo, A. Viswan, H. Chou, T. Sasaki, K. Ikuta, and M. Nagatsu,
“Capture of dengue viruses using antibody-integrated graphite
encapsulated magnetic beads produced by gas plasma technology”,
Mol. Med. Reports, 14 (2016) 697-704.
[11] T. Abuzairi, M. Okada, Y. Mochizuki, N. R. Poespawati, R. Wigajatri
and M. Nagatsu, “Maskless functionalization of a carbon nanotube dot
array biosensor using an ultrafine atmospheric pressure plasma jet”,
Carbon, 89 (2015) 208.
[12] T. Abuzairi, M. Okada, R. W. Purnamaningsih, N. R. Poespawati, F.
Iwata, and M. Nagatsu, “Maskless localized patterning of biomolecules
on carbon nanotube microarray functionalized by ultrafine atmospheric
____________________________________________________________________________________________________________
A Practical Analysis of the Baseband Superregenerative

Amplifier for Slow Signals
Nanang Sulistiyanto, Mochammad Rif’an, Robert Rieger and Kuang-Ching Zhang
rrieger@mail.nsysu.edu.tw
Electrical Engineering Department, National Sun Yat-Sen University, 804 Kaohsiung, Taiwan
Abstract– This paper analyzes the performance of Therefore, VGA are useful for biosignal processing.
two different circuit topologies realizing the
One approach to VGA implementation is the
baseband superregenerative amplifier. The first
superregenerative (SR) amplifier [5]. It enables the
circuit is the conventional arrangement and the
realization of digitally assisted analog system by
second circuit is a novel configuration which
providing real-time gain tuning by the adjustment of
employs capacitive load at the output. Smaller
digital control pulses (clock waveforms with a
feedback factors reduce the required OTA slew
defined phase, duty-cyle and frequency) provided by
rate to achieve a gain target at the cost of longer
a simple digital processing unit. This approach is
gain cycles. Measured results using an integrated
especially suitable for the low bandwidth signals
amplifier with external capacitive load are
encountered in biosignal processing (typically <15
reported which confirm the analysis in terms of
kHz). This approach can reduce power as the core
linear range and gain.
amplifier gain-bandwidth requirement is low [6].
I. INTRODUCTION In this paper we provide the detailed analysis of two
SR amplifier configurations for low-frequency
Multiplication of an analog signal by a well defined
application. Section II presents the operating principle
gain factor is a core function for realizing discrete-
of the SR and circuit analysis of the conventional
time signal processing, e.g. to perform linear filtering
implementation. The improved circuit is proposed and
(IIR or FIR), frequency analysis by Fourier transform
analyzed in Section III. Measured results are provided
(FFT), or time-frequency analysis (wavelet
in Section IV followed by conclusions in Section V.
transform). Fully digital implementations of discrete-
time signal processing have been widely implemented II. SR AMPLIFICATION WITH CURRENT FEEDBACK
by using microcomputers or digital signal processors. SR amplification employs the idea that a gain circuit
However, analog signal representation offers can be placed in a feedback loop in which the
potentially smaller and less complex feedback is positive as opposed to negative feedback
implementations, high speed, low power consumption used conventionally to stabilize a gain circuit. By
and reduces ADC resolution requirements. In these applying positive feedback a small initial voltage is
cases, multiplication can be performed by a digitally multiplied very rapidly resulting in the desired high
programmable variable gain amplifier (VGA). gain of the overall circuit. The output voltage is
A further area of application is the acquisition of sampled at a defined time point T with respect to the
biosignals, which are typically small and careful time at which the input signal was applied. As long as
design of the integrated front-end amplifier is very this time point is chosen before the amplifier saturates
important. The amplification has to be large enough the result is a sampled version of the input signal with
for further processing the signal without degradation a gain that is related to and controlled by sample time
of the signal-to-noise (SNR) level but small enough to T. Fig. 1 shows a principle circuit diagram of the SR
avoid saturation. Furthermore, the exact magnitude of amplifier using the topology realized in [5]. An
the electrocardiogram (ECG) or electromyogram operational transconductance amplifier (OTA) is used
(EMG) is influenced by electrode contact impedance, here to provide gain where for simplicity its input and
the type of electrode used, the electrode placement, output impedances are at first presumed large enough
and the general level of signal activity [1]–[4]. to be negligible. The OTA output current iC relates to
This work is supported in part by grant MOST 103-2221-E-110- the voltage at the differential OTA input terminals as
057-MY2. Gm=iC/vOTA. The path for this current is via the
____________________________________________________________________________________________________________
feedback gain block where it is scaled by factor β in a input voltage amplitude and that it depends
range from zero to unity. A most simple exponentially on t which enables variable high gain
implementation is achieved for β=1 where a by choosing an appropriate sampling time t=T at
conducting wire connects the OTA output back to its which the output voltage is read using a succeeding
positive input terminal. Voltage source Voff represents S/H (not shown in Fig. 1). Also the OTA offset
the (constant) offset voltage of the OTA referred to its voltage is amplified exponentially, however
negative input. A switch is used to sample the input multiplied by the further factor 1/β. The gain
voltage Vin onto a storage capacitor C. Operation experienced by the circuit offset thus may be larger
requires that the switch is initially closed and the than the signal gain if a small feedback factor is
capacitor is precharged to Vin. Then the switch is selected. For this reason and, most importantly, since
opened at a time t=0 and the feedback current β.iC realizing arbitrary β by current scaling is difficult to
begins to charge the capacitor. Standard circuit achieve in a practical circuit the practical
analysis in the Laplace domain yields the expression configuration is limited to unity gain feedback by
for the output voltage. wire connection between the output node and the
OTA positive input. Notably, in this circuit
Vout  Vin 
Gm
sC

 Vout  Voff  configuration the input voltage is sampled via the
input switch and so the accuracy of this sampling
s Gm procedure depends on the internal load of the
 Vout  Vin  Voff (1) charging voltage source. Fig. 1 shows the ideal zero-
  Gm    Gm 
s C  s   resistance signal source, in practice however the
C  C  charging time of the sampling capacitor must be taken
The amplitude of input and offset voltage at the into account which complicates the circuit design in
beginning of the charging period t=0 are denoted as [5, 6]. Therefore, here we propose the alternative
and respectively so that it follows configuration of Fig. 2 introduced next.
1 1 III. PROPOSED SR CIRCUIT
Vin  Vˆ  , Voff  V 
s s (2) Here, the input source connects continuously to the
 Gm
  t high-impedance amplifier negative terminal so that
 Vout  Vˆ   1  V  e C   1  V the input source is not significantly loaded, making
the performance independent of the signal source
Similarly, for the current ic one finds
impedance. The exponential charging capacitor is
 Gm
  t moved to the output node of the circuit and the
iC  Vˆ   1  V    Gm  e C (3) feedback is changed from a current feedback to
voltage feedback. This change provides the benefit
Converting (2) into the time domain gives
that different feedback factors β can be realized by
 Gm
  t tapping off a proportional feedback if the load
Vout  Vˆ   1  V  e C   1  V (4) capacitor is realized as a series connection of devices.
Also included in Fig. 2 is a resistor representing the
It shows that the output voltage is proportional to the
Fig. 1: Conventional arrangement of the super-regenerative

amplifier before S&H in which the input signal is sampled on a Fig. 2: Alternative circuit configuration with Vin connected to
capacitor at the input and the sampling switch is opened during the negative OTA input. The resistor represents the output
the amplification phase. resistance of the OTA. The OTA offset is reflected to V in.
____________________________________________________________________________________________________________
finite output resistance of the practical OTA (which The linearity of the circuits is ultimately determined
was neglected for simplicity in Fig. 1). by the linearity of the OTA. It is therefore desirable to
maintain a small differential input voltage to the OTA
The OTA input offset voltage adds to the input
during the exponential charging phase. In this circuit
voltage Vi and is hence not shown separately in the
following analysis. Initially, the switch is closed for a the OTA input voltage Vx is Vout   whereas in the
period which is long enough to initialize the capacitor circuit of Fig. 1 Vx=Vout so that the proposed circuit
with a ‘zero’ voltage. The switch is subsequently may achieve higher linearity for a given output
opened and the amplification phase commences. In voltage. Here, the linearity is affected by the applied
this phase the output voltage is given in the Laplace feedback as smaller feedback factor β yields smaller
domain as OTA input drive and thus higher linearity for a set
gain target.
Gm
Vout  Vin (5) IV. MEASURED RESULTS
 1 Gm 
C  s   
 RC C  A test circuit was built using a push-pull OTA
integrated in CMOS technology (ASIC), loaded by
Substituting Vin=( + )/s and transforming into the external capacitors to implement the circuit of Fig. 2.
time domain yields the output voltage The schematic diagram of the OTA is given in Fig. 3.
  1  Gm 
The output signal is observed using Labview
Vˆ  V    
A0  C 
t
 (National Instruments, Austin, TX). The bias tail
Vout   1  e 
1  
current for this OTA is 5 μA, so that the maximum
  
A0 (6) slew current is also about 5 μA. We choose load
C=12 nF. Transconductance Gm is about 18 μA/V. An
Vˆ   Gm t 
  e C  1  Vˆ  A example measurement with its Labview interface is
   shown in Fig. 4. The input voltage was zero and the
 
recorded exponential is thus due to offset in both the
where A0 is the OTA open-loop voltage gain A0=Gm.R amplifier and test voltage. In this example we record
and A is the effective closed loop gain. Here, the the voltage at a sample time T=1 ms as approximately
circuit gain is determined by the sampling time t=T 72 mV. From (7), the expected linear range before
but also by the hardware realization of β. Moreover, slewing commences is limited to 80 mV at the input,
this circuit yields a negative voltage gain where the or 280 mV at the output since the gain calculated
configuration of Fig. 1 realizes positive gain. Overall, from (6) is 3.5 V/V. This matches well with the
(6) realizes a higher gain than (1) for the same measured result shown in Fig. 5.
charging time which is achieved at the cost of a Here, the DC transfer function was recorded by
higher feedback current consumption. We find for stepping the input voltage and plotting the
this circuit corresponding output voltage, always with a 1 ms
charging time. Within the linear portion of the
 
t
1 A0    1 
ˆ 
ic  VGm 1  e RC   1 (7) transfer curve a gain of 3.5 V/V is measured and
   1   A0  slewing commences at about 250 mV output
amplitude. The input offset is 20 mV in agreement
When appropriate feedback is applied so that 1<<βA0 with the measurement in Fig. 4.
(7) approximates to To demonstrate the principle suitability for biosignal
recording the setup is reconfigured using a 3.2 nF
 t Gm 
ic  VGm  e C  1
ˆ (8) load to yield a gain of 198 V/V. The sample rate is
  reduced to 200 Hz while maintaining a charging time
 
of 1 ms. The exponential amplifier output was
Compared to (3) this current is higher by a factor 1/β. captured using the measurement data acquisition card.
Therefore, a main disadvantage of the class of A prerecorded ECG trace was applied to the amplifier
superregenerative circuits discussed here is the high as test input and the output was observed as shown in
slew rate required to satisfy (8) for large amplitudes. Fig. 6. A visual comparison between signal input and
Optimizing the ic/A ratio requires that β be designed output confirms that the prominent features of the
as small as possible. In a practical circuit the required ECG are successfully amplified.
gain is then set in (6) by choice of C.
____________________________________________________________________________________________________________
lin e a r s le w in g re g io n
O TA s w in g lim it
G a in is 3 .5 V /V
O b s e rv e d s le w in g b e g in s (Vo u t > 2 5 0 m V )
P re d ic te d s le w in g fro m (6 )+ (7 ) (Vo u t> 2 8 0 m V )
Fig. 5: DC sweep measured transfer function (positive

quadrant). Values for gain and slew limited range agree with
Fig. 3: Schematic diagram of the integrated push-pull OTA
the analysis.
used in the measurements. Amplifier Output
0.1
Amplitude [V]
0.05
0
−0.05
−0.1
Amplifier Intput
0.8
Amplitude [mV ]
0.6
0.4
0.2
0
−0. 2
−0. 4
0.5 1 1.5 2 2.5
Time [s]
Fig. 6: Amplification of a prerecorded ECG signal using 200

Hz sample rate and 1 ms charging time with 3.2 nF load
resulting in a gain of 198 V/V.
Fig. 4: Exponential waveform captured using the Labview
interface.
[2] J. Taylor, A. Demosthenous, I. Triantis, R. Rieger, and
V. CONCLUSIONS N. Donaldson, “Design of an Adaptive Interference
Two different circuit topologies for the baseband Reduction System for Nerve Cuff Electrode
Recording,” IEEE Trans. Circuits and Systems I, vol.
superregenerative amplifier have been analyzed. The
51, no. 4, pp. 629-639, 2004.
proposed circuit employs a capacitive output load to [3] M. M. Puurtinen, S. M. Komulainen, P. K. Kauppinen,
achieve exponential charging in a positive feedback J. A. V. Malmivuo, and J. A. K. Hyttinen,
circuit. The gain of this circuit is higher than that of „Measurement of noise and impedance of dry and wet
the conventional structure. However, it requires a textile electrodes, and textile electrodes with
higher maximum feedback current. A main advantage hydrogel,” in Proc. 28th IEEE EMBS Annual
of the proposed topology is the potential for choosing International Conference, 2006, pp. 6012-6015.
different feedback factors via taps on the load [4] P. Laferriere, E. D. Lemaire, and A. D. C. Chan,
capacitor. Smaller feedback factors reduce the “Surface Electromyographic Signals Using Dry
required OTA slew rate (but need a longer charging Electrodes,” IEEE Trans. Instrumentation and
Measurement, vol. 60, no. 10, pp. 3259-3268, 2011.
time to reach the same gain). The presented measured
[5] R. R. Antayhua, G. M. Da Silva, and F. R. de Sousa, “A
results confirm the analysis and show the practicality Duty-Cycle Controlled Variable-Gain Instrumentation
of the proposed topology for the amplification of Amplifier Applied For Two-electrode ECG
slowly varying signals, e.g. the ECG. Measurement,” in Proc. IEEE Int. Instrumentation and
Measurement Technology Conf., 2012, pp. 1270-1274.
REFERENCES [6] P. Palà-Schönwälder, F. X. Moncunill-Geniz, J. Bonet-
Dalmau, F. del-Águila-López, and R. Giralt-Mas,
[1] J. Zariffa, M. K. Nagai, Z. J. Daskalakis, and M. R.
“Baseband Superregenerative Amplification,” IEEE
Popovic, “Influence of the Number and Location of
Trans. Circuits & Systems I, vol. 56, no. 9, pp. 1930-
Recording Contacts on the Selectivity of a Nerve Cuff
1937, 2009.
Electrode,” IEEE Trans. Neural Systems and Rehab.
Engineering, vol. 17, no. 5, pp. 420-427, 2009.
____________________________________________________________________________________________________________
A Framework for Optimum Contour Detection

Cahyo Crysdian#1, Bayu Adhi Nugroho*2
#
Teknik Informatika - Universitas Islam Negeri Maulana Malik Ibrahim Malang, Indonesia
1
cahyo@ti.uin-malang.ac.id
*
Sistem Informasi, Universitas Islam Negeri Sunan Ampel Surabaya, Indonesia
2
bayu@uinsby.ac.id
Abstract— The importance of contour detection have been becomes a vital process for visual data analysis particularly
acknowledged by researchers worldwide, and indeed dozens of for object recognition. Researchers have indeed found that
methods have been introduced. However there is no single contour detection become an important component for many
method suit with various conditions of digital images. Most of the image analysis processes such as medical image analysis,
time, a tedious work to select best method from dozens is
gesture recognition, and land use analysis based on satellite
required only to derive the most appropriate objects contour
from a digital image. Once an object contour is recognized, photo. Since then many methods on detecting contour have
further image analysis process can be computed efficiently. This been proposed up to now (Papari & Petkov, 2011; Verma et al.
condition is in contrast with human visual perception which 2011; Somkantha et al, 2011; Koren & Yitzhaky, 2006;
employs contour detection as a preliminary process with minimal Becerikli & Karan, 2005). However recent applications of
energy consumption before conducting exhaustive visual computer vision show the immaturity of contour detection
analysis. Therefore this research aims to develop a framework to especially to support further image analysis processes. Instead
automatically detecting optimum object contour by selecting the of simplifying the content of image, varied methods for
best method for each condition of input image. Efficient energy contour detection often disrupt or even destroy the structure of
consumption will be achieved by applying mechanism based on
image objects. Evidence show that false object recognition is
multi criteria decision making.
often produced by image analysis due to poor contour
detection (Rodriguez & Shah, 2007; Somkantha et al, 2011;
Keywords— Contour detection, Edge detection, Image analysis, Tong, 2014; Amirgaliyev et al, 2014). Therefore it is
Multi criteria decision making important to measure the quality of contour resulted from
detection methods before it is supplied to further image
analysis processes. This research aims to formulate a
I. INTRODUCTION framework for contour quality measurement in order to find
Visual perception becomes an interesting topic to research the most accurate contour detection given a set of natural
recently due to the development of information technology images.
and the Internet which incorporate visual technology
massively. Nowadays it is common to see varied devices
which are equipped with camera or other visual-related II. RELATED WORK
facilities such as GPS (Global Positioning System), There have been tremendous efforts to research contour
Multimedia Player, and Image Gallery. Therefore the growth detection, and lots of methods have been proposed in this field.
of visual data increases exponentially both in sizes of data that Researchers have widely known that contour of objects
require much bigger storage, as well as in the quality of data contained in digital image can traditionally be derived from
due to the increasing resolution of recent devices. Beside the first-order derivative of the image with regard to edge as
opening chances to explore broader data, this condition computed using Equation 1.
becomes a burden to computation. Bigger size of data means
(1)
the computation to analyze data would demand more time
consumption, while higher resolution means more aspects
Implementation of Equation 1 however restricts vast
must be incorporated into the computation, thus algorithm
variation of neighboring pixels which potentially affect the
complexity would increase as well.
formation of contour. Therefore researchers have proposed
Researchers have been recognizing human visual
other approach based on convolution such as Roberts (1963),
perception as a superb example to conduct visual data analysis
Sobel (1970) and Prewitt (1970) which fulfill Equation 2, with
(Nixon & Aguado, 2008), hence a strategy to deal with the
k is the kernel being convolved to the input image I to obtain
above condition is by mimicking how human analyzing visual
its first-order derivative.
data. Clinical evidence proves that human capability to
develop perception on contour is a vital component towards (2)
objects recognition (Papari & Petkov, 2011), it means the Experiments however show that this approach encounters
absence of contour perception would make a patient difficulties to derive edges of the object in natural images
completely unable to recognize object. Thus contour detection particularly in the noisy condition. Later, researchers proposed
____________________________________________________________________________________________________________
mechanisms that still survive in noisy image such as III. FRAMEWORK DEVELOPMENT
Laplacian of Gaussian (LOG) (Marr & Hildreth, 1980) and The research employs a quantitative approach to identify
Canny (1986). Different with first-order derivative method, the best parameter for contour quality evaluation. Schematic
this approach rather detects zero-crossing of second–order diagram of research design is given in Figure 2. In this
derivative to obtain object contour. Even though this approach research there are four main stages required to complete as
performs better than the previous, however experiments show follows:
that this approach encounter difficulties to respond on vast
variety of natural image condition, especially one with A. Data Collection
spurious edges and texture image. This stage aims to collect data set which represents a wide
The condition above drive the researchers to propose many variety of natural images. It is important to notice that the
other different approaches to detect object contour. Some image being collected shall present a dominant foreground
incorporate even artificial intelligence in their mechanism objects. Therefore images which have no dominant objects
such as Becerikli and Karan (2005) that propose fuzzy-based such as satellite photos are excluded from data set.
edge detection system, and Verma et al (2011) that propose
edge detection based on novel bacterial foraging technique.
Testing results however show that the performances of both Data Alternative
Measurement Output
methods are indifferent to the previous proposed methods Collection Development
such as Canny, LOG, Sobel etc. Meanwhile the approach

proposed by Rodriguez and Shah (2007) to segment human in Performace evaluation
crowded scenes assumes that human is isotropic, therefore the Input Enhancement Contour Edge
proposed method is initialized with a set of posture silhouette. Image 1 Detection 1 Map (1,1)
The results however still deliver low accuracy. Other Post Process: Best
Edge width
approach proposed by Somkantha et al (2011) presents Close loop
Alternative
different mechanism to detect object contour. The method Enhancement Contour Edge
employs edge following algorithm that expands snake-like m Detection n Map (m,n)
mechanism based on intensity gradient and texture gradient
features. The experimental results show that this approach Spatial
Approach:
derives only an object boundary despite the existence of many Area
Boundary
objects contained in digital image. Besides, this approach Centroid
depends greatly on the formation of initial model that must be Distance C-B
located correctly. Hence the approach is prone to error since it
has to follow edge magnitude and edge map in order to Fig. 1 Developed framework for fully automatic contour extraction
develop object contour in which the value of edge magnitude
sometimes is very weak for noisy image. B. Alternatives Development
Due to varied natural image condition spanning from noisy This stage aims to collect and organize as many as possible
and spurious edge to texture image which deliver complexities alternative processes to detect contour of the object contained
for contour detection, a group of researchers in this field then in digital images. The processes are composed by the
afford to measure the quality of derived contour. It was started combination of image enhancement and contour detection.
by Yitzhaky and Peli (2003) and then followed by Koren and Therefore, if there exist m enhancement and n contour
Yitzhaky (2006) which propose objective evaluation on the detection, there would be m x n alternative processes to form
result of edge detection in order to define best parameter the edge map that will produce m x n edge map.
selection supplied to certain edge detection method. This
research however is not sensitive enough to deal with noisy C. Measurement
and texture image. Other approach to measure contour such as This stage aims to measure the performance of each edge
conducted by Wang et al (2006) depends a lot on manual map formulated in the previous stages. Two main processes
ground truth, therefore its applicability is questionable since are held to enable performance measurement i.e. post-
different natural image would require different ground truth. detection process and evaluation process based on spatial
Recently Gimenez et al (2014) propose a set of measurement approach. The former is intended to facilitate the latter process
to evaluate contour quality namely equilibrium index and by enhancing the presentation of contour in term of edge
entropy index. This method employs statistical approach width and closure, while the latter is the evaluation towards
based on Kolmogorov-Smirnov to measure the quality of edge contour quality. It is held using a set of testing parameters
map. However this approach has not considered the use of based on spatial approach such as contour area, boundary,
contour in term of edge presentation particularly to represent centroid, and the distance between centroid and boundary.
object existed in the image for further analysis processes.
Therefore further development is still required to produce D. Output Selection
good contour definition which is useful for image analysis This stage becomes the conclusion of the measurement. It
processes. is settled by selecting the best contour quality based on a set
____________________________________________________________________________________________________________
of testing parameters described in the previous stage. By measurement when it has to process varied condition of digital
recognizing which edge map contain the best contour quality, image.
the best alternative process to detect contour for that image
could easily be defined. IV. EXPERIMENT
Preliminary experiment has been conducted to a set of
E. Performance Evaluation digital images such as shown in Table 1 column 2. The input
This stage aims to measure the accuracy of certain testing data is obtained from the image of some cultural heritage sites.
parameters defined in the Measurement stage. It is held by It is important to note rich noises existed in the input image
modifying a set of testing parameters, and then each which form disturbing pattern to the object. Meanwhile the
performance of testing parameter is measured in term of two result is given in the same table column 3. Although
aspects i.e. its accuracy to obtain best contour and its qualitative measurement on the quality of edge derived from
reliability to handle different condition of natural images. At the proposed algorithm has not been carried out, however
the end of this stage, the best testing parameter would be experimental results show there always edges representing
defined based on accuracy and reliability measurements. objects contained in every image and no open edges produced
1) Contour attributes: Contour attributes are defined as a set by the proposed algorithm. This is a promising output from
of parameters which represent the characteristics of contour. the preliminary experiment.
Acquisition of contour attributes is held by computing edge
map using spatial approach. Currently some parameters have TABLE I
been defined to compose contour attributes namely number of EXPERIMENTAL RESULTS
edge, number of non-edge object, edge length, non-edge area, Index Input Image Output
non-edge boundary, non-edge centroid, and distance between
centroid to boundary. The research aims to define the best
parameters composing contour attributes. Here the term best Data 1
parameters mean the parameters that deliver most significant
impact to the measurement of contour quality. The mechanism
to select the best parameters which represent contour
attributes is held together with the definition of parameters
weight given in the next research instrument. Data 2
2) Parameters weight: Parameters weight defines the role of
each contour attribute to measure contour quality. Therefore it
would be assigned to each parameter composing contour
attributes. To obtain the best parameters weight, an iterative Data 3
mechanism based on multi criteria decision making such as
weighted sum model (WSM) or weighted product model
(WPM) is employed to measure the quality and to match the
result with pre-defined contour quality. Assuming that WSM Data 4
is utilized to compute parameters weight, the mechanism
would fulfil Equation 3, with Q denotes contour quality, w
denotes the weight for each parameter p.
(3)
Data 5
It should be noted that computation in Equation 3 requires no
ground truth to define the best parameters weight. Here the
definition of parameters weight becomes a pre-defined
contour quality measurement.
V. DISCUSSION
3) Accuracy & Reliability: Accuracy and reliability becomes
the last instrument to employ in this research. The intention of Measuring contour quality is a growing recent interest in
these instruments is to disclose the performance of weighted contour detection research because of many factors such as
contour attributes defined by the previous instruments. Here vary digital image conditions, plenty methods have been
accuracy deals with correct decision made by weighted proposed in this field, and more importantly significant role of
contour attributes to measure contour quality and more contour detection to support many processes in the field of
importantly to define the best contour detection mechanism image analysis and image understanding. This research
for any given natural images. Meanwhile reliability means the follows Papari and Petkov (2011) which assumes that there
defined weighted contour attributes still produce good exist plenty available methods for contour detection having
capability of handling different condition of natural images.
However due to more varied conditions of natural images, it is
____________________________________________________________________________________________________________
extremely difficult (if we do not want to mention impossible) is clinically proven that human visual cognition is mainly
to find the correct method without putting a great deal of based on the perception of object contour (Papari & Petkov,
efforts to extract the contour of the object from natural images 2011), developing adaptive contour detection which
such as shown by many experimental results (Rodriguez & recognizes local condition of natural image would enable
Shah, 2007; Somkantha et al, 2011; Tong, 2014; Amirgaliyev automatic analysis on tons of visual data existed today.
et al, 2014). Therefore it is important to develop method to
automatically selecting best contour detection.
Here we argue that certain parameters derived from spatial ACKNOWLEDGMENT
approach would have significant role to measure contour The authors wish to acknowledge Collaborative Research
quality, hence they would be useful to settle this task. These Schemes of UIN Maulana Malik Ibrahim Malang that enables
parameters would disclose the characteristic of contour and research cooperation between Department of Computer
thus they become contour attributes. Each contour attribute Science UIN Maulana Malik Ibrahim Malang and Department
would affect differently to measure contour quality, hence a of Information System UIN Sunan Ampel Surabaya.
weight is assigned for each attribute. Computing the attributes
together with their weight in iterative fashioned utilizing multi REFERENCES
criteria decision making as formulated in Equation 3 would [1] Amirgaliyev B, Kairanbay M, Kenshimov Ch, & Yedilkhan D. (2014).
reveal most significant attributes for contour measurement. Development of automatic number plate recognition system. Computer
Modelling & New Technologies 2014 18(3) 193-197.
The work in this research include extending Gimenez at al [2] Becerikli Y, & Karan TM. (2005). A new fuzzy approach for edge
(2014) by formulating automatic mechanism to select best detection. Springer Verlag Berlin Heidelberg: J. Cabestany, A. Prieto,
contour detection method for each given natural image. and DF. Sandofal (Eds.): IWANN 2005, LNCS 3512, 943-951, 2005.
[3] Canny J. (1986). A computational approach to edge detection. IEEE
Transactions in Pattern Analysis and Machine Intelligence 8(6) 679-
698, Nov 1986.
VI. CONCLUSIONS [4] Gimenez J, Martinez J, & Flesia AG. (2014). Unsupervised edge map
The applications of image processing and computer vision scoring: A statistical complexity approach. Elsevier Computer Vision
have been noted capable of improving human life quality. and Image Understanding 122 (2014) 131-142.
[5] Koren R, & Yitzhaky Y. (2006). Automatic selection of edge detector
Researchers have even believed that image processing and parameters based on spatial and statistical measures. Elesvier
computer vision become the complement of human visual Computer Vision and Image Understanding 102 (2006) 204-213.
perception, as each carry its own superiorities and benefits [6] Marr D, & Hildreth E. (1980). Theory of edge detection. Proceedings
(Nixon & Aguado, 2008). Therefore researchers have been of Royal Society London B 207 (1167) 187-217.
[7] Nixon MS, & Aguado AS. (2008). Feature extraction and image
developing many methods in the field of image processing processing 2nd Ed. Elsevier Academic Press, ISBN: 978-0-12372-538-7.
and computer vision since the early age of computer, [8] Papari G, & Petkov N. (2011). Edge and line oriented contour
particularly in the field of image understanding. Amid the detection: state of the art. Elsevier Image and Vision Computing 29
conviction that computer would help people in understanding (2011) 79-103.
[9] Prewitt JMS. (1970). Object enhancement and extraction. Book: B.
visual data, many research outcomes disclose the difficulties Lipkin, A. Rosenfeld (Eds) Academic, New York, 1970.
and complexities to process visual data due to its discrete [10] Roberts LG. (1963). Machine perception of three-dimensional solids.
nature while in reality any objects recorded in an image are Oustanding Dissertations in the Cmputer Sciences, Garland publishing,
not discrete. They somehow correlated physically. The New York, 1963.
[11] Rodriguez MD & Shah M. (2007). Detecting and Segmenting Humans
complexities of any algorithm developed for image in Crowded Scenes. ACM 978-1-59593-701-8/07/0009, MM’07,
understanding come from the necessities to locate and classify September 23-28, 2007, Augsburg, Bavaria, Germany.
discrete data in term of a set of pixels contained in digital [12] Sobel IE. (1970). Camera models and machine perception. PhD Thesis:
image to follow their physical nature. Hence, up to now it is Stanford University, Standfort, CA, USA, 1970.
[13] Somkantha K, Umpon NT, & Auephanwiriyakul S. (2011). Boundary
difficult to mention any application which successfully detection in medical images using edge following algorithm based on
establishes automatic image understanding. If there is intensity gradient and texture gradient features. IEEE Transactions on
successful application, usually it becomes confidential and Biomedical Engineering, Vol. 58, No. 3, 567-573, March 2011.
used for high priority businesses such as US forensics. [14] Tong C. (2014). A novel edge detection method based on 2-D Gabor
wavelet. Computer Modelling & New Technologies 2014 18(8) 153-
Meanwhile human visual perception has been noted 157.
superior for image understanding. However it is not designed [15] Verma OP, Hanmandlu M, Kumar P, Chhabra S, & Jindal A. (2011). A
to process tons of visual data. On the other hand, novel bacterial foraging technique for edge detection. Elsevier Pattern
overwhelming information through visual data is supplied by Recognition Letters 32 (2011) 1187-1196.
[16] Yitzhaky Y, & Peli E. (2003). A method for objective edge detection
different parties originated from Internet and mobile evaluation and detector parameter selection. IEEE Transactions on
communities. Therefore implementing human visual cognition Pattern Analysis and Machine Intelligence, Vol. 25, No. 8, 1027-1033,
to computer vision becomes the focus of this research. Since it August 2003.
____________________________________________________________________________________________________________
$&RQFHSWXDO)UDPHZRUNRI+LJK'LPHQVLRQDOLW\
&$30RGXODWLRQRQ3URJUDPPDEOH&KLSIRU,QGRRU
9/&1HWZRUN
6KDULIDK6DRQ$EG.DGLU0DKDPDG0DLVDUD2WKPDQ

)DFXOW\RI(OHFWULFDODQG(OHFWURQLF(QJLQHHULQJ8QLYHUVLWL7XQ+XVVHLQ2QQ0DOD\VLD
3DULW5DMD0DOD\VLD

VKDULID#XWKPHGXP\
NDGLU#XWKPHGXP\

PDLVDUD#XWKPHGXP\

$EVWUDFW² :LWK D GUDVWLFDOO\ JOREDO PRELOH GDWD WUDIILF DPSOLWXGHSKDVH &$3 KDYH EHHQ SURSRVHG IRU GLJLWDO
JURZ WKH DGYDQFH DQG SRSXODU PRGXODWLRQ IRUPDW DUH VXEVFULEHUOLQH'6/GXULQJHDUO\DQGPLGVE\WKH%HOO
UHTXLUHG WR ERRVW WKH KLJKFDSDFLW\ VHUYLFHV $V DQ /DEV>@>@0RUHRYHU&$3KDVWKHSRWHQWLDOLQUHGXFLQJWKH
DOWHUQDWLYH UHVXOW D KLJKGLPHQVLRQDO FDUULHUOHVV V\VWHPFRVWGXHWRWKHVLPSOLFLW\RIWKHWUDQVFHLYHUGHVLJQDQG
DPSOLWXGH SKDVH &$3 PRGXODWLRQ VFKHPH IRU LQGRRU WKH DEVHQFH RI WKH ORFDO RVFLOODWRU /2 (YHQWXDOO\ WKH
YLVLEOHOLJKWFRPPXQLFDWLRQV9/&LVFKRRVHDVRQHRIWKH FDSDFLW\ KDV QRW EHHQ LQFUHDVHG KRZHYHU WKH &$3 KDV
DOWHUQDWLYH PRGXODWLRQ WHFKQLTXH FRPSDUHG WR WKH PXOWL SRWHQWLDO IRU SURYLGLQJ PRUH IOH[LELOLW\ LQ LQFUHDVLQJ WKH
FDUULHU V\VWHPV 0RUHRYHU KLJKHU XSVDPSOLQJ IDFWRU RI QXPEHU RI GLPHQVLRQV ZLWK ORZ LPSOHPHQWDWLRQ FRVW WR
KLJKGLPHQVLRQDO &$3 PRGXODWLRQ FDXVHV GHFUHPHQW RI VXSSRUW PXOWLSOH VHUYLFHV IRU QH[W JHQHUDWLRQ DFFHVV DQG
WKH VSHFWUDO HIILFLHQF\ 7KHUHIRUH WKH GHVLJQ DQG LQGRRUQHWZRUNV
LPSOHPHQW RQ RI WKH &$3 WUDQVFHLYHU DUFKLWHFWXUH XVLQJ 'HVSLWH WKHLU PDQ\ DGYDQWDJH KLJKGLPHQVLRQDO &$3
;LOLQ[ 9LUWH[ 8OWUDVFDOH '63 WRRO EDVHG RQ WKH ILHOG UHTXLUHV H[FHVV EDQGZLGWK GXH WR WKH KLJKHU XSVDPSOLQJ
SURJUDPPDEOH JDWH DUUD\ )3*$V NQRZQ DV &$3 ,3 IDFWRU ZKLFK FRQWULEXWH LQ D GHFUHDVHG RI VSHFWUDO HIILFLHQF\
FRUHVLVSURSRVHG >@2QWKHRWKHUKDQGFRQVWUXFWLQJWKHUHDOWLPHV\VWHPDQG
GDWDSURFHVVLQJRQILHOGSURJUDPPDEOHJDWHDUUD\V)3*$VLV
.H\ZRUGV²&$3)3*$9/&%(5 WKHQH[WVWHSLQIXOO\UHDOWLPHV\VWHPLPSOHPHQWDWLRQ>@
7KXVGHVLJQDQGLPSOHPHQWDWLRQRIKLJKGLPHQVLRQDO&$3
, ,1752'8&7,21 PRGXODWLRQ DUFKLWHFWXUH RQ UHFRQILJXUDEOH KDUGZDUH XVLQJ
)3*$VIRU9/&XVLQJWKH;LOLQ['63WRROLVRQHRIWKHEHVW
7KH(OHYHQWK3ODQLVIRUPXODWHGDVDEDVLVRIGHYHORSPHQW
DOWHUQDWLYH WR SURYLGH IXOO UHDOWLPH KLJKFDSDFLW\ EURDGEDQG
HIIRUWVIRUWKH\HDU±2QHRIVL[VWUDWHJLFWKUXVWVLV
QHWZRUNV XVLQJ WKH H[LVWLQJ LOOXPLQDWLRQ GHYLFHV 7KHUHIRUH
³6WUHQJWKHQLQJLQIUDVWUXFWXUHWRVXSSRUWHFRQRPLFH[SDQVLRQ´
WKHUH LV D QHHG IRU KLJK SHUIRUPDQFH V\VWHP ZKLOH NHHSLQJ
*RRG LQIUDVWUXFWXUH LV WKH IRXQGDWLRQ RI VRFLDO LQFOXVLRQ DQG
&$3 DUFKLWHFWXUH IOH[LEOH IRU TXLFN XSJUDGHDELOLW\ WR FRSH
HFRQRPLF JURZWK'LJLWDODFFHVV RILQWHUQHWZLOOEHLQFUHDVHG
ZLWK UHDOWLPH DSSOLFDWLRQV 0RUHRYHU )3*$V DUH KLJKO\
WKURXJK EURDGEDQG LQIUDVWUXFWXUH H[SDQVLRQ DQG RWKHUV
UHFRQILJXUDEOH KDUGZDUH SODWIRUPV ZKLFK SURYLGH WKH
LQLWLDWLYHVWRHQVXUHWKHLUDIIRUGDELOLW\
IOH[LELOLW\WRUHDOL]HGDWDDQGRSWLFDOFRPPXQLFDWLRQV\VWHPV
0XOWLPHGLD EURDGEDQG VHUYLFH SURYLVLRQLQJ ZKLFK LV WKH
ZKLOH SUHVHQWLQJ ORZ SRZHU FRQVXPSWLRQ DQG HQKDQFHG
HPHUJLQJVHUYLFHVWKDWHQGXVHUVDUHGHPDQGLQJVXFKDVKLJK
SHUIRUPDQFH>@
GHILQLWLRQ YLGHR VWUHDPLQJ YLGHRFDOOV DQG DFFHVV WR FORXG
)XUWKHUPRUH DQ H[DPLQDWLRQ RI WKH ELW HUURU UDWH %(5
FRPSXWLQJKDYHSXWVHYHUHSUHVVXUHRQWKHWHOHFRPPXQLFDWLRQ
HUURUYHFWRUPDJQLWXGH(90DQGVLJQDOWRQRLVHUDWLR615
QHWZRUN LQIUDVWUXFWXUH WR SURYLGH KLJK FDSDFLW\
SHUIRUPDQFH RI WKH SURSRVHG V\VWHPV DJDLQVW WKH UHFHLYHG
FRPPXQLFDWLRQ V\VWHPV ,Q >@ WKH HVWLPDWHG ([DE\WH
RSWLFDOSRZHUDQGGLVWDQFHZLOOEHFDUULHGRXW
(% RI GDWD WUDIILF WR EH WUDQVIHUUHG YLD PRELOH QHWZRUNV
7R WKHEHVW RI RXU NQRZOHGJHWKLVLVWKHILUVWLQYHVWLJDWLRQ
HYHU\PRQWKE\
RI &$3 ,3 FRUHV ZKLFK KDV WKH SRWHQWLDO IRU SURYLGLQJ
'LIIHUHQW WUDQVPLVVLRQ DQG PRGXODWLRQ IRUPDW ZRXOG EH
DGGLWLRQDO IOH[LELOLW\ LQ LQFUHDVLQJ WKH QXPEHU RI GLPHQVLRQV
EHQHILFLDO LQ RUGHU WR RYHUFRPH WKLV SUREOHP DQG RIIHU WKH
ZLWK ORZLPSOHPHQWDWLRQFRVWWRVXSSRUW PXOWLSOH VHUYLFHV DV
UHTXLUHGEDQGZLGWK DQG LQFUHDVHIOH[LELOLW\ PXOWLSOH[LQJDQG
RIQH[WJHQHUDWLRQLQGRRU9/&QHWZRUNV
WUDQVSDUHQF\ 0XOWLSRLQW WUDQVSDUHQW DUFKLWHFWXUHV WKDW FDQ

HYHQWXDOO\XVHWKHZDYHOHQJWKGRPDLQWRDFKLHYHPXOWLSOH[LQJ

LV RQH RI WKH RSWLFDO WHFKQRORJLHV VROXWLRQV 7KHUHIRUH PXOWL

OHYHO PXOWLGLPHQVLRQDO KLJKGLPHQVLRQDO FDUULHUOHVV
____________________________________________________________________________________________________________
,, $5(9,(:2)&$321)3*$)259/&1(7:25. DQG RSWLFDO FRPPXQLFDWLRQV ,Q >@ '&$3 /' RYHU
7KH PDLQ SXUSRVH RI WKLV VWXG\ LV WR H[WHQG WKH UHVHDUFK P SRO\PHU RSWLFDO ILEHU 32) KDV EHHQ GHPRQVWUDWHG ZLWK
IRFXVIURPWKHLQYHVWLJDWLRQRI&$3LQZLUHGWUDQVPLVVLRQWR UHVRQDQW FDYLW\ OLJKWHPLWWLQJGLRGHV 5&/('V,Q >@ ILUVW
ZLUHGZLUHOHVV V\VWHPV ,QGRRU QHWZRUNV KDYH EHHQ FKRVHQ H[SHULPHQWDOLQYHVWLJDWLRQVKDVEHHQUHSRUWHGRQWKH'&$3
EHFDXVH RI WKH LQFUHDVLQJ QXPEHU RI HQG XVHUV ZLWK KLJK DQG '&$3 IRU RSWLFDO ILEUH WUDQVPLVVLRQ DW ELW UDWHV XS WR
GHPDQG IRU PXOWLPHGLD VHUYLFHV DQ\WLPH DQ\ZKHUH DQG WR 0EV
RYHUFRPH WKH EDQGZLGWK ERWWOHQHFN SUREOHP LQ WKH QHWZRUN 7KHLPSOHPHQWDWLRQ RI9/& V\VWHPEDVHG RQV\VWHP RQD
&$3KDVEHHQZLGHO\XVHGIRUDV\PPHWULFDOGLJLWDOVXEVFULEHU SURJUDPPDEOHFKLS623&6SDUWDQ()3*$VGHYHORSPHQW
OLQH $'6/ DQG DV\QFKURQRXV WUDQVIHU PRGH $70 ORFDO NLWV ZLWK DFFHSWDEOH 3(5 YDOXH RI ( KDV EHHQ GRQH E\
DUHDQHWZRUNV/$1V>@,QERWKRIWKHVHDSSOLFDWLRQV&$3 7HUUDHWDO>@7KHUHIRUHWKLVUHVHDUFKSURSRVDOLVFRQVLGHUHG
KDV EHHQ DGRSWHG SULPDULO\ EHFDXVH RI LWV KLJK EDQGZLGWK DV WKH ILUVW LQYHVWLJDWLRQ RQ GHVLJQ DQG LPSOHPHQWDWLRQ RI
HIILFLHQF\DQGORZHULPSOHPHQWDWLRQFRVWV KLJKGLPHQVLRQDOLW\ &$3 DUFKLWHFWXUH XVLQJ ;LOLQ[ '63 WRRO
9/& >@ LV RQH RI WKH GDWD DQG RSWLFDO FRPPXQLFDWLRQ RQ )3*$ 7KH SURFHVV ZLOO SURGXFH WKH UHFRQILJXUHG ORJLF
WHFKQRORJ\ LQ ZKLFK EDVHEDQG VLJQDOV DUH PRGXODWHG RQ WKH IXQFWLRQ RU LQWHOOHFWXDO SURSHUW\ ,3 FRUHV RI WKH &$3
LQWHQVLW\ RI WKH OLJKW HPLWWLQJ GLRGH /(' 7KHUHIRUH 9/& DUFKLWHFWXUH
KDV VHYHUDO QXPEHUV RI DGYDQWDJHV ZKLFK DUH QR
,,,$&21&(378$/)5$0(:25.2)5(6($5&+
HOHFWURPDJQHWLF UDGLDWLRQ QR SRWHQWLDO KHDOWK WKUHDWHQV DQG
SURYLGHD VHFXUHZD\ WR WUDQVPLW GDWD ZLWKLQDFORVHGLQGRRU 5HVHDUFK PHWKRGRORJ\ DUH GLYLGHG LQWR PDLQ VWDJHV DV
HQYLURQPHQWVRWKDWWKHGDWDLVGLIILFXOWWREHLQWHUFHSWHGIURP OLVWHG LQ UHVHDUFK VFRSHV ZKLFK DUH FRQVWUXFW WKH &$3 XVLQJ
RXWVLGH 9LVLEOH OLJKW LV OHVV GDQJHURXV IRU KLJKSRZHU )3*$QHWZRUNGHVLJQDQGH[SHULPHQWDOVHWXSIRU9/&
DSSOLFDWLRQV EHFDXVH KXPDQV FDQ QRWLFH LW DQG DFW WR SURWHFW
$ 6WDJH&RQVWUXFWWKH&$3DUFKLWHFWXUHXVLQJ;LOLQ['63
WKHLU H\HV IURP LQMXU\ +XPDQ H\H YHU\ VHQVLWLYH WR
WRRO
ZDYHOHQJWKV IURP DERXW WR QP ,Q IUHTXHQF\ UDQJH
WKLVFRUUHVSRQGVWRDEDQGRIWR7+]0RUHRYHU/(' 2SWLPL]DWLRQ DOJRULWKP 2$ FDQ EH DSSOLHG LQ WKH
DUH VPDOO LQ VL]H FKHDS HQHUJ\ VDYLQJ DQG ORQJOLIHVSDQ VLPXODWLRQ ZRUN WR H[WHQG WKH FRQYHQWLRQDO '&$3 VFKHPH
FRPSDUH WR WKH WUDGLWLRQDO OLJKWV $W WKH VDPH WLPH WKLV WR KLJKHU GLPHQVLRQDOLW\ DQG WR SURPLVH WKH SHUIHFW
WHFKQRORJ\ FRYHU ERWK GDWD WUDQVPLVVLRQ DQG LQGRRU UHFRQVWUXFWLRQRIWKHILOWHUV7KHDGYDQWDJHRIWKLVPRGXODWLRQ
LOOXPLQDWLRQ WHFKQLTXH LV WKDW WKH IUHTXHQF\ PDJQLWXGH UHVSRQVH RI WKH
7KH 9/& WHFKQRORJ\ EDVHG RQ ZKLWH OLJKW /(' DUH WUDQVFHLYHU ILOWHUSRVVLEOH WR EHLGHQWLFDO $GGLWLRQDOO\LWLVD
SURPLVLQJIRUKLJKVSHHG GDWDWUDQVPLVVLRQLQFUHDVHGHQHUJ\ SRLQWWRSRLQW PHWKRG WR HQKDQFH WKH GHVLJQ WR KLJKHU
HIILFLHQF\ DQG SRVVLELOLW\ RI XVLQJ VRXUFHV IRU VLPXOWDQHRXV GLPHQVLRQDOLW\&$3V\VWHPV
LOOXPLQDWLRQ >@>@ 2UWKRJRQDO IUHTXHQF\ GLYLVLRQ 7KHWUDQVPLWWHGVLJQDORI'&$3FDQEHIRUPXODWHGDVLQ
PXOWLSOH[LQJ2)'0LPSOHPHQWDWLRQIRUWKHLQGRRUQHWZRUN HTXDWLRQ
ZKLFK RIIHUV KLJK EDQGZLGWK HIILFLHQF\ DQG DOORZV VLPSOH
¦ >D J W Q7 E J W Q7 @
f
HTXDOL]DWLRQ DW WKH UHFHLYHU ZDV UHSRUWHG DV LQ >@ :KLOH V W
WKHLPSOHPHQWDWLRQRIPXOWLSOHLQSXWPXOWLSOHRXWSXW0,02 Q f
Q L Q T
2)'0IRUZLUHOHVVRSWLFDO9/&SURPLVHWUDQVPLVVLRQUDWHVRI
0EV DW D %(5 RI ( >@ 0HDQZKLOH ILUVW :KHUH DQ DQG EQ DUH WKH UHDO DQG LPDJLQDU\ SDUWV 7 LV WKH
LPSOHPHQWDWLRQRIGLPHQWLRQDO&$3IRU9/&V\VWHPEDVHG SHULRG DQG J L W J W FRV S I F W DQG
RQ D VLQJOH FRPPHUFLDOO\ DYDLODEOH 5*%W\SH /(' KDV EHHQ J T W J W VLQ S I F W DUH WKH LQSKDVH DQG TXDGUDWH SKDVH
UHSRUWHG E\ :X HW DO ZLWK WKH ZDYHOHQJWKGLYLVLRQ ILOWHUUHVSRQVHVUHVSHFWLYHO\:KLOH
PXOWLSOH[LQJ:'0VFKHPHLQRUGHUWRLQFUHDVHWKHFDSDFLW\
XSWR*EVRIGDWDUDWHVZLWKOHVVWKDQ(RIWKH%(5
\ W * >V W K W Q W @
>@>@
+LJKHU RUGHU PRGXODWLRQ IRUPDWV VRFDOOHG &$3 KDV EHHQ
SURSRVHG IRU FRSSHU ZLUHV DV HDUO\ DV &$3 LV D PXOWL 5HFHLYHGVLJQDODVLQHTXDWLRQLVFDSWXUHGDQGVDPSOHG
GLPHQVLRQDO DQG PXOWLOHYHO VLJQDO IRUPDW HPSOR\LQJ ZKHUH * LV D JDLQ DQG LV WKH SKRWRGHWHFWRU UHVSRQVLYLW\
RUWKRJRQDOZDYHIRUPVIRUHDFKGLPHQVLRQ 7KHVHZDYHIRUPV )LJXUH VKRZV D 6LPXOLQN PRGHO RI FRQYHQWLRQDO '&$3
DUH REWDLQHG IURP ILQLWH LPSXOVH UHVSRQVH ),5 ILOWHUV ZLWK DUFKLWHFWXUH >@ &$3 DUFKLWHFWXUH LQ )LJXUH ZLOO EH
RUWKRJRQDO ILOWHU UHVSRQVHV ,Q LWV SULQFLSOH LW LV VLPLODU WR FRQYHUWHGWRKDUGZDUHGHVFULSWLRQODQJXDJH+'/FRGHXVLQJ
TXDGUDWXUH DPSOLWXGH PRGXODWLRQ 4$0 ERWK &$3 DQG UDSLGSURWRW\SLQJ PHWKRG 7KLV PHWKRG LQYROYHG WKH SURFHVV
4$0 VXSSRUWV PXOWLSOH OHYHOV &RPSDUH WR 4$0 KRZHYHU RI FRQYHUWLQJ WKH &$3 6LPXOLQN PRGHO XVLQJ +'/ &RGHU
&$3 GRHV QRW UHTXLUH WKH JHQHUDWLRQ RI VLQXVRLGDO FDUULHUV DW :KLOH +'/ FRGH YHULILFDWLRQ RI &$3 DUFKLWHFWXUH ZLOO EH
WKH WUDQVFHLYHU :KLOH GXH WR WKH SRWHQWLDOO\ KLJK VSHFWUDO FRPSOHWHG E\ XVLQJ )3*$LQWKHORRS %RWK +'/ FRGH
HIILFLHQF\ DQG WKH SRVVLELOLW\ RI JHQHUDWLQJ WKH UHTXLUHG YHULILFDWLRQDQG)3*$LQWKHORRSDUHWRROVIURP0DWK:RUNV
RUWKRJRQDO SXOVHV E\ PHDQV RI WUDQVYHUVDO ILOWHUV &$3
PRGXODWLRQKDVUHFHLYHGDWWHQWLRQLQWKHUHVHDUFKDUHDRIGDWD
____________________________________________________________________________________________________________
% 6WDJH1HWZRUNGHVLJQRILQGRRU9/&QHWZRUNV GLVWDQFH )LJXUH VKRZV WKH EORFN GLDJUDP RI RYHUDOO
,QGRRU 9/& XVLQJ WKH H[LVWLQJ LOOXPLQDWLRQ GHYLFHV /(' H[SHULPHQWDOVHWXS
LQIUDVWUXFWXUHKDVDGYDQWDJHVRIKLJKHIILFLHQF\ORZFRVWDQG *HQHUDWHGVWUHDPLQJFRQWHQWZLOOEHPDSSHGRQWRWKH&$3
ORQJ OLIH 7KH GHYHORSHG &$3 ,3 FRUHV RQ )3*$ LV DQ WUDQVPLWWHU LQ ;LOLQ[ )3*$ 7KHQ JHQHUDWHG VLJQDO IURP
LQWHUIDFLQJGHYLFHVFRQQHFWHGWRH[LVWLQJKRPH/('9LVLEOH ;LOLQ[ )3*$LVLQWHUQDOO\ VKLIWHGE\SUHGHILQHG '&OHYHO WR
OLJKW SRVLWLRQLQJ WUDQVPLWV SRVLWLRQLQJ ,' E\ WKH UHDG\PDGH GULYHWKH/('WRHQVXUHRSHUDWLRQLQOLQHDUUHJLRQ/('GULYHU
/(' ZLWKRXW H[WUD GHYLFH 7KH UHFHLYHU UHFHLYHV WKH VLJQDO FRQYHUWV WKH HOHFWURQLF GDWD LQWR DQ RSWLFDO LQWHQVLW\ DQG
IURP/('DQGLPSOHPHQWVWKHSRVLWLRQLQJDOJRULWKPWRUHDOL]H SURSDJDWHV RYHU WKH FKDQQHO ZLUHOHVVO\ 7KH SKRWRGHWHFWRU
LQGRRUSRVLWLRQLQJZLWKSKRWRGHWHFWRUUHFHLYHUDQGSURFHVVLQJ RXWSXW VLJQDO PD\ FRQVLVW RI VHYHUDO VRXUFHV RI QRLVH
VHJPHQW 7KH GHVLJQ FRQFHSW RI WKH SURSRVHG LQGRRU 9/& LQFOXGLQJ DPELHQW WKHUPDO DQG VKRW QRLVH 7R UHGXFH WKH
QHWZRUNVLVVKRZQLQ)LJXUH VRXUFHVRIDPELHQWQRLVHWKHH[SHULPHQWVKRXOGEHFRQGXFWHG
LQDSHUIHFWGDUNODERUDWRU\ZKLOHWKHWKHUPDODQGVKRWQRLVH
& 6WDJH([SHULPHQWDO6HWXSIRU9/&V\VWHPV VRXUFHV DUH DVVXPHG DV DGGLWLYH ZKLWH *DXVVLDQ QRLVH
7KHGHVLJQ RIWKHH[SHULPHQWDOVHWXSDQGWKHSHUIRUPDQFH $:*1 5HFHLYHG VLJQDO ZLOO EH FDSWXUHG DQG VDPSOHG
FKDUDFWHUL]DWLRQ ZLOO EH H[DFWO\ WKH VDPH DV LQ WKH QXPHULFDO ZKLFK FDQ UHPRYH WKH KLJK IUHTXHQF\ QRLVH FRPSRQHQWV
VLPXODWLRQVHWXSWRYDOLGDWHWKHV\VWHPSHUIRUPDQFHLQWHUPRI WKURXJKWKHORZSDVVILOWHU/3)7RLPSURYHWKH%(5(90
%(5 (90 DQG 615 DJDLQVW UHFHLYHG RSWLFDO SRZHU DQG DQG 615 WKH ILOWHUHG VLJQDO ZLOO EH SDVV WKURXJK WKH GRZQ
VDPSOLQJHTXDOL]HUDQGWKUHVKROGGHWHFWHG

)LJ&$3DUFKLWHFWXUH6LPXOLQN0RGHO

)LJ'HVLJQ&RQFHSWIRU9/&LQWHUIDFLQJV\VWHP
____________________________________________________________________________________________________________

)LJ%ORFN'LDJUDPRI([SHULPHQWDO7HVW6HWXS

>@ ' 7HUUD 1 .XPDU 1 /RXUHQFR / 1 $OYHV DQG 5 / $JXLDU
,9&21&/86,216 ³'HVLJQ GHYHORSPHW DQG SHUIRUPDQFH DQDO\VLV RI '666EDVHG
WUDQVFHLYHUIRU9/&´,(((,QWHUQDWLRQDO&RQIHUHQFHRQ&RPSXWHUDVD
7KLV FRQFHSWXDO IUDPHZRUN RI PXOWLOHYHO PXOWL 7RRO(852&21SS
GLPHQVLRQDO &$3 IRU 9/& V\VWHPV XVLQJ )3*$ H[SHFWHG WR >@ 9LVLEOH/LJKW&RPPXQLFDWLRQ&RQVRUWLXPKWWSZZZYOFFQHW
H[WHQG WKH UHDFK DQG FRYHUDJH DUHD 0XOWLGLPHQVLRQDO >@ 3 + 3DWKDN ; )HQJ 3 +X DQG 3 0RKDSDWUD ³9LVLEOH OLJKW
FRPPXQLFDWLRQ QHWZRUNLQJ DQG VHQVLQJ $ VXUYH\ SRWHQWLDO DQG
PRGXODWLRQ &$3 KDV SRWHQWLDO IRU SURYLGLQJ PRUH IOH[LELOLW\ FKDOOHQJHV´,(((&RPPXQLFDWLRQ6XUYH\VDQG7XWRULDOVYROQR
LQ LQFUHDVLQJ WKH QXPEHU RI GLPHQVLRQV ZLWK ORZ SS
LPSOHPHQWDWLRQ FRVW QHHGHG WR VXSSRUW PXOWLSOH VHUYLFHV IRU >@ 2(UJXO('LQFDQG2%$NDQ³&RPPXQLFDWHWRLOOXPLQDWH6WDWH
QH[WJHQHUDWLRQRILQGRRU9/&QHWZRUN RIWKHDUW DQG UHVHDUFK FKDOOHQJHV IRU YLVLEOH OLJKW FRPPXQLFDWLRQV´
-RXUQDORI3K\VLFDO&RPPXQLDFWLRQYROSS
>@ * 6 6RQWDNOH DQG' 0 %KDOHUDR ³&RPSDULVRQ RI GLIIHUHQW
9 $&.12:/('*(0(17 PRGXODWLRQ WHFKQLTXHV IRU KLJK EDQGZLGWK YLVLEOH OLJKW
7KH ILQDQFLDO VXSSRUW UHFHLYHG IURP WKH )XQGDPHQWDO FRPPXQLFDWLRQ´ ,((( &RQIHUHQFH RQ (OHFWULFDO (OHFWURQLFV 6LJQDOV
5HVHDUFK *UDQW 6FKHPH )5*6 9RW 2IILFH IRU &RPPXQLFDWLRQDQG2SWLPL]DWLRQ((6&2SS
>@ $ + $]KDU7$ 7UDQDQG '2¶%ULHQ ³$ *LJDELWV ,QGRRU :LUHOHVV
5HVHDUFK ,QQRYDWLRQ &RPPHUFLDOL]DWLRQ DQG &RQVXOWDQF\ 7UDQVPLVVLRQ XVLQJ 0,022)'0 9LVLEOH /LJKW &RPPXQLFDWLRQ´
0DQDJHPHQW 25,&& 8QLYHUVLW\ 7XQ +XVVHLQ 2QQ ,(((3KRWRQLFV7HFKQRORJ\/HWWHUVYROQRSS
0DOD\VLDLVJUDWHIXOO\DFNQRZOHGJHG >@ )0:X&7/LQ&&:HL&:&KHQ=<&KHQ+7+XDQJ
DQG 6LHQ &KL ³3HUIRUPDQFH &RPSDULVRQ RI 2)'0 6LJQDO DQG &$3
5()(5(1&(6 6LJQDO 2YHU +LJK &DSDFLW\ 5*%/('%DVHG :'0 9LVLEOH /LJKW
&RPPXQLFDWLRQ´,(((3KRWRQLFV-RXUQDOYROQR
>@ &LVFR 9LVXDO 1HWZRUNLQJ ,QGH[ ³*OREDO PRELOH GDWD WUDIILF IRUHFDVW >@ ) 0 :X & 7 /LQ & & :HL& : &KHQ = < &KHQ DQG + 7
XSGDWH±´:KLWH3DSHU&,6& +XDQJ ³ *EV :'0 9LVLEOH /LJKW &RPPXQLFDWLRQ RI D 6LQJOH
>@ -- :HUQHU³7XWRULDORQFDUULHUOHVV$030 3DUW,)XQGDPHQWDOVRI 5*% /(' (PSOR\LQJ &DUULHUOHVV $PSOLWXGH DQG 3KDVH0RGXODWLRQ´
GLJLWDO&$3WUDQVPLWWHU´$7 7&RQWULEXWLRQWR$16,;77330' 2SWLFDO )LEHU &RPPXQLFDWLRQ &RQIHUHQFH DQG ([SRVLWLRQ DQG WKH
-XQH 1DWLRQDO )LEHU 2SWLF (QJLQHHUV &RQIHUHQFH 2)&1)2(& 7HFKQLFDO
>@ - - :HUQHU ³7XWRULDO RQ FDUULHUOHVV $030 3DUW ,, 3HUIRUPDQFH RI 'LJHVWSS
EDQGZLGWK HIILFLHQW OLQH FRGHV´ $7 7 &RQWULEXWLRQ WR $16, ;7 >@ $QWRQLR &DEDOOHUR 7LHQ 7KDQJ3KDP -HVSHU%-HQVHQ DQG ,GHOIRQVR
7330')HE 7DIXU 0RQUR\ ³&DUULHUOHVV 1'LPHQVLRQDO 0RGXODWLRQ )RUPDW IRU
>@ 0% 2WKPDQ ;X =KDQJ /HL 'HQJ 0 :LHFNRZVNL -%-HQVHQ DQG 0XOWLSOH 6HUYLFH 'LIIHUHQWLDWLRQ LQ 2SWLFDO ,QKRPH 1HWZRUNV´
,7 0RQUR\ ³([SHULPHQWDO LQYHVWLJDWLRQV RI ''&$3 ,QWHUQDWLRQDO3KRWRQLFV&RQIHUHQFH,327X0
PRGXODWLRQ ZLWK GLUHFWO\ PRGXODWHG 9&6(/V´ ,((( 3KRWRQLFV >@ 6 5DZDVKGHK ³$ 6LPXOLQN 0RGHO IRU &DUULHUOHVV $PSOLWXGH 3KDVH
7HFKQRORJ\/HWWHUVYROQRSS 0RGXODWLRQ &$3´ $YDLODEOH RQOLQH
>@ 6=YDQRYHF 3 &KYRMND 3$ +DLJK = *KDVVHPORR\ ³9LVLEOH OLJKW KWWSZZZPDWKZRUNVFRPPDWODEFHQWUDOILOHH[FKDQJH
FRPPXQLFDWLRQWRZDUGV*´5DGLRHQJLQHHULQJYROQRSS FDUULHUOHVVTDPFDSPRGHP

____________________________________________________________________________________________________________
Study of Solar Electricity Powered Air Conditioner

in a Tropical Hot Climate
Elieser Tarigan #1
#
Electrical Department and Center for Renewable Energy Studies, PSET, University of Surabaya
Jl. Raya Kalirungkut, Surabaya 60292, Indonesia
1
elieser@staff.ubaya.ac.id
Abstract— Air conditioners are becoming more common and are technology but provided a detailed insight into the
a major strain on energy demands in tropical climate where air methodology of designing and sizing a solar cooling system.
conditioning is used all year round. This work studied of using Not all of the steps provided can be repeated as the design is
solar electricity produced by photovoltaic (PV) system to power for a solar absorption cooling system, however it provides a
air conditioners in Surabaya, Indonesia. Measurement tests were
conducted on different photovoltaic systems. A grid connect
good guideline to follow when planning a design process.
system provides a better solution with the 800 W PV system Ref.[6] studied the possibility of using photovoltaic panels to
providing 80 % of the 3.5 kWh load of the air conditioner, while produce electricity that is used to power the compressor of an
the other 20 % coming from the grid during periods of low inverter air conditioning unit, without the use of batteries or
irradiation. With the grid-connected system, the PV panel’s any inverter regulators. In the study an inverter air conditioner
production does not need to match the consumption of the air is connected to both the grid and photovoltaic panels. One
conditioner, although a smaller PV array will mean a smaller advantage of using an inverter air conditioner is that it uses
percentage of the load will be covered by PV. direct current components meaning the direct current
produced by the photovoltaics can be used. An inverter is not
Keywords— solar energy, PV system, air conditioner, grid needed to convert the direct current of the photovoltaics to
connected, off grid. alternating current, this reduces cost and increases efficiency.
The peak times for using air conditioning offices is
I. INTRODUCTION commonly during the day when the temperature is the highest.
Renewable energy such as solar can reduce the need to On the other hand PV’s peak times are also during the day
burn fossil fuels and in turn reduce the amount of greenhouse when the solar radiation is highest.
gasses being released into the atmosphere. To promote the use The main objective of present work is to study and suggest
of solar energy new innovative solutions must be created in solutions for a solar PV powered air conditioning system,
both engineering and policy. Air conditioners are becoming under condition of hot climate in Surabaya, Indonesia. The
more common and are a major strain on energy demands possibility of different systems is studied including grid
especially in tropical climate countries like Indonesia where connected and off grid systems.
air conditioning is used all year round. Up to 50% of
buildings energy can be from air conditioning in subtropical II. RESEARCH METHOD
regions [1]. Two set experiment were set, each for off grid tests and
The average solar irradiation in Surabaya, Indonesia is grid connected system tests. A 800Wp photovoltaic system
relatively high over the whole year and is about 4.8 was installed on the roof of an office, intended for experiment
kWh/m².day [2]. Temperatures vary with average of 30 °C in this work. The office with area of 9 m2 is equipped with 0,5
during dry season and 32 °C during wet season [3]. With such PK air conditioner split type, Daikin manufactured. (The
high cooling demands and high solar potential, Indonesia commercially unit power of air conditioner in Indonesia is
particularly Surabaya, is the perfect location to test the commonly in PK, stands for Dutch words Paard kracht, =
possibility of using air conditioners and photovoltaic energy 745,7 Watt).
together. Surabaya is located at the coast on the east side of System components for off grid tests were PV modules,
Java very close to the equator represents a tropical climate solar charge controller, batteries, sine wave inverter, and the
very well, that means the presented system solutions can be air conditioner system. For the grid connected system tests,
adopted to other cities and countries with similar weather. the system components were only PV modules and grid tied
Quite number of previous works on PV powered air inverters in addition to air conditioner system. For grid-
conditioner had been reported. Ref. [4] reported a work on connected system tests the photovoltaic panels are connected
the design and construction of a direct current (DC) air directly to the inverter. The air conditioner is connected to
conditioning system integrated with photovoltaic (PV) both the inverter and the grid through one socket. Experiment
system. No DC system was constructed but it contained a setup and the measurement devices are showed scematically
breakdown of all vital components for both the air conditioner in Figure 1. If power is available from the inverter, this will
and PV system. Ref.[5] studied a different type of cooling be used by the air conditioner. If there is not enough power
____________________________________________________________________________________________________________
available, then the excess power needed will be taken from solar radiation, temperatures inside/outside, air conditioner
the grid. The specification for each of system components for power consumption, and the start-up power of the air
both experiments are shown in Table 1. conditioner.
III. RESUTS AND DISCUSSION

A. Off Grid System
The operations of air conditioner with off-grid PV system
mean the air conditioners is completely independent from the
grid and would be 100 % powered by photovoltaic produced
energy. These tests would show if the air conditioner could
run completely on solar power and if so, what challenges this
type of system would present. Attemp to run the air
conditioner on the off grid system were unsuccessful.
Ultimately the failure of the tests came down to the inverter.
The startup power of the air conditioner was underestimated
and so the inverter (inverter 1) chosen was too small. On other
experiments set up with the startup current the 1000 W
SUONER pure sine wave inverter with a surge power of 2000
W was selected ( inverter 2). When the off grid system was
tested with this inverter it started well. Both air conditioner
EM = Energy meter
and inverter were operational until the air conditioner
compressor started then a fault light on the inverter switched
Fig. 1 Experimen schematic with grid connected system
on and both the air conditioner and inverter turned off. It was
clear the inverter could not meet the startup power of the
TABLE 1 compressor. Initially the blame for the failure was the inverter
TABLE OF SYSTEM COMPONENTS IN OFFICE SYSTEM
and perhaps the 1000 W running power and 2000 W start up
pwer were exaggerated by the inverter manufacturer.
Compone It was not certain that the failed test could be attributed to
No Brand Model Size Other
nt
Air 1.35 kW the inverter so further investigation into the startup power is
Conditione Daikin R15HEV1 cooling COP 3.97 required. The maximum startup current of the compressor
1
r capacity happens over a fraction of a second, to measure this the
200 W Mono measuring device must have a short measuring period. The
Photovolta SPU-
I- Solar 1 Max Crystalline
ic Panels 4 180M
Power Module
air conditioner used was only 0.5 PK, one of the smallest
Load Self- available on the market. The fact that a 1000 W inverter with
Charge
controller 2
S series SC-20 Current 20 consumpti 2000 W surge is not able to run the air conditioner shows how
A on 6 mA important the start-up power is. The large start-up power
No Data needed means the inverter has to be greatly oversized. Not
CT12 12 Volt
Battery CIT BAT Sheet
4 100FR 100 Ah
Found only does the inverter need to be sized for the start-up but it is
1500 W vital that the output wave of the inverter is pure sine wave. A
/Surge lower quality waveform can damage the air conditioner.
3000 W Modified Because the off grid system is completely reliant on PV
Inverter 1 SP-1500
1 Intelligent 24 V DC Sine Wave
=> 220 V power, battery storage is essential in times of low irradiation.
AC At the least one day of autonomy is needed, that means the
1000 W batteries must be able to power the air conditioner for one day
/Surge without any electricity being produce by the photovoltaics.
FPC- 2000 W Pure Sine
Inverter 2
1
SUOER
1000A 12 V DC Wave
Direct current (DC) air conditioning systems are mostly
=> 220 V used in off grid applications. While the air conditioning
AC systems are efficient and the panel size can be reduced, large
1000 W battery banks are needed to maintain the use of the system
20-48V Grid Tied
Inverter 3
1 DC => 220 Inverter
during the night or days of low PV production. Lead acid
V AC batteries are very expensive and usually need to be changed
every 8 years depending on the number of cycles and depth of
For the analysis purposes, the following parameters were discharge [7], making the cost and maintenance level increase
measured: the DC power coming from the PV panels, the DC greatly.
power after the batteries, the AC power after the inverter, the
____________________________________________________________________________________________________________
Regular air conditioning units are fixed speed, meaning  With the outside temperature reaching 32°C a set
that the compressor motor is either on or off. The compressor temperature of 22°C is very low. A 10°C difference
will run until the set temperature is reached and will then turn between ambient and indoor is a lot of work for an air
off until the temperature rises above a certain level where the conditioner especially in rooms with a poor U-value.
compressor will start again until the set temperature is reached When comparing PV production for the whole day to
again. This cycle is repeated over and over as long as the air during office hours (07:30-17:30) the values are very similar.
conditioner is running. This stop start cycle means there is For the whole day an average of 3.6 kWh was produced and
high energy usage in the air conditioner and high fluctuations during office hours it was an average of 3.4 kWh. Out of the
in the temperature [8]. whole day only 0.2kWh was produced outside office hours
meaning nearly all of the electricity produced is available
during office hours. This would be beneficial in countries
B. Grid Connected System with low or no feed in tariff as only a small percentage would
The photovoltaic production can be directly compared to have to be feed into the grid.
irradiation values. The times of high photovoltaic production
are the times of high irradiation and the times of low TABLE 3
photovoltaic production are the times of low irradiation. The SUMMARY OF OFFICE HOUR VALUES
irradiance values were obtained using the “davis”
pyranometer. The experiments were conducted during first Value
Office hours
week of May 2015. The weather was relatively clear sky with (07:30-17:30)
daily radiation was 4,8 kWh/m2 on average. Experiment result Air conditioner
3.45 kWh
for four consecutive days is presented in Table 2. consumption
Grid backup 720 Wh (20%)
TABLE 2
EXPERIMENT RESULTS FOR FOUR CONSECUTIVE DAYS Inverter output used
2.7 kWh (80%)
by AC
Feed in 440 Wh
Day Average Energy Inverter Solar
power demand output fraction
(W) (kWh) useb by (%) Grid backup power has two major functions in this system.
AC
(kWh)
First is to supply power to the air conditioner when there is
low PV production. Backup power from the grid used is an
1 345 3.47 2.76 79.4 average of 20% of the consumption of the air conditioner with
the majority of it used in the final two hours of the day when
2 342 3.44 2.70 78.3 irradiation is low. To decrease the grid use the panels would
have to be oversized but even then from 15:00 onwards the
3 343 3.45 2.82 81.7 irradiation on the panels is below 300 W/m2 and decreasing
fast. Powering of the air conditioner during this time with PV
4 344 3.46 2.65 76.6
would not be possible unless batteries are used to store excess
power during the day but as previously stated this is an
The air conditioner power consumption showed that expensive solution. The second use is to power the air
running at approximately 350 W. The air conditioning system conditioner during startups, eliminating the need to oversize
runs from 07:30 to 17:30 with a set temperature of 22 °C. the inverter like in the off grid system. Using the grid backup
This was designed to replicate the conditions in a typical means the inverter does not need to be oversized to meet the
Indonesian office. A 22 °C set value is very low but it is a startup load. It simplifies the system so that batteries and
usual set temperature in Indonesia. It can often be low as charge controllers do not need to be used. There is a constant
17 °C with people in the room wearing extra sweaters and reliable supply from the grid which may not be the case with
jackets to compensate. But the air conditioner ran constantly batteries if there is a long period of low irradiation.
all day. This can mean a number of things. The 800 W PV system turns out to be well sized for this
 The air conditioner could be undersized for this room a 350 W air conditioner. The PV produced energy covers 80%
more powerful one may be needed to reach the set of the air conditioner load, the grid covers 20% with 440 Wh
temperature. being fed into the grid (see summary in Table 3)
 While it has not been calculated exactly the U-value of the When the grid is available, a considerably best way to
office is poor. There are two ventilation holes in the room connect PVs and air conditioning is through a grid connected
which provide fresh air circulation. Air conditioners work system, with the following avantages:
by recycling the cooled air and cooling it further until the
set temperature is reached. This is not possible when
constant fresh warm air is being circulated into the room.
____________________________________________________________________________________________________________
 It is a simple system: A reduced number of components IV. CONCLUSIONS

means the complexity of the PV system is reduced. Only
PV modules and grid tied inverter are used. When the grid is available, a considerably best way to
 Low cost: Not using batteries reduces the cost connect PVs and air conditioning is through a grid connected
significantly. Using the grid for startups means the system. A grid connection system has a much smaller initial
inverter does not need to be oversized also reducing the investment than an off grid system, it is also less complicated
cost.
system with greater efficiency. Off grid systems require
 Reliable: If there is a long period of low irradiation the oversizing of components where grid connected do not.
grid will still be there where as in a battery system the Additional benefits of the grid connected system is the use of
storage could run out. excess electricity. The Indonesian Government's introduction
 Inverter does not need to be oversized: In the off grid of a feed in tariff has made the use of photovoltaics a more
system the inverter needs to be oversized for the startup appealing prospect, although some questions still remain on
load. In this grid connected system the grid provides the the availability of the tariff.
startup.
 Excess electricity can be used: In an off grid system REFERENCES
excess electricity can be used to charge the batteries but [1] K. F. Fong, C. K. Lee, and T. T. Chow, “Comparative study of solar
once the batteries are fully charged any excess electricity cooling systems with building-integrated solar collectors for use in sub-
is wasted. In the grid connected system all excess tropical regions like Hong Kong,” Appl. Energy, vol. 90, no. 1, pp. 189–
195, 2012.
electricity can be used. [2] PWC Indonesia, “Power in Indonesia - Investment and Taxation Guide,”
 Adaptable: This system set up can be adapted to work for 2013.
different air conditioning units. [3] World Weather Online, “:
http://www.worldweatheronline.com/Surabaya-weather/East-
Java/ID.aspx,” 2015. [Online]. Available:
Despite of the advantages, the grid connected system has the http://www.worldweatheronline.com/Surabaya-weather/East-
limitation, such as: Java/ID.aspx. [Accessed: 08-May-2015].
[4] I. Daut, M. Adzrie, M. Irwanto, P. Ibrahim, and M. Fitra, “Solar Powered
 Not 100 % PV power: In the grid connected system there Air Conditioning System,” Energy Procedia, vol. 36, pp. 444–453, 2013.
is about 20% grid use, in an off grid system it would be [5] T. Tsoutsos, E. Aloumpi, Z. Gkouskos, and M. Karagiorgas, “Design of a
100 % PV powered. solar absorption cooling system in a Greek hospital,” Energy Build., vol.
 Weather dependent: If there is a day with low irradiation 42, no. 2, pp. 265–272, 2010.
[6] F. J. Aguilar, P. V. Quiles, and S. Aledo, “Operation and Energy
then PV cannot power the air conditioner. Whereas an off Efficiency of a Hybrid Air Conditioner Simultaneously Connected to the
grid system has storage to power the air conditioner during Grid and to Photovoltaic Panels,” Energy Procedia, vol. 48, pp. 768–777,
the period of low irradiation. 2014.
[7] T. M. Layadi, G. Champenois, M. Mostefai, and D. Abbes, “Lifetime
 Cannot be used during black outs: If there is a black out
estimation tool of lead–acid batteries for hybrid power sources design,”
which can happen frequently in Surabaya the grid Simul. Model. Pract. Theory, vol. 54, pp. 36–48, 2015.
connected system will not work. [8] Bijli Bachao Team, “Air Conditioners with Inverter technology can help
save electricity,” 2015. [Online]. Available:
http://www.bijlibachao.com/air-conditioners/air-conditioners-with-
inverter-technology-can-help-save-electricity-ac-india.html. [Accessed:
28-May-2015].
____________________________________________________________________________________________________________
Use of IT in Indonesian tax administration system:

duality of tax technology perspective
Agung Darono #1
#1
Tax Education and Training Center – Ministry of Finance
Jalan Sakti Raya No.1, Kemanggisan, West Jakarta, Indonesia
1
agungdarono@kemenkeu.go.id
Abstract— The use of information technology (IT) in tax ―…However, the project called PINTAR (Project for Indonesian
administration from social science perspective can be viewed as Tax Administration Reform) was opposed by the legislators. In fact,
an agent-structure duality. The duality of technology is how with this system, he said, all phases of the work carried out as early
agents make technology as tight-part of their action to achieve identification of the tax authorities, the preliminary findings,
their social purposes. Tax authority is a social agent whose job is related to reporting tax liability to be paid by taxpayets through
to collect tax revenue, for which the support of IT applications is online system... ―
required. This study uses discourse analysis in an attempt to source: http://www.koran-
reveal how the agents in the tax administration play their roles sindo.com/news.php?r=2&n=0&date=2015-12-19 (translated by
so that the use of IT in tax administration is consistent with its author)
purpose and functions. The findings of the study are: (1)
Such situations indicating, more or less, that the development
structuration of the use of IT in tax administration makes core
tax application system as the organizational main feature of the
and implementation of IT-based system is not solely a
Directorate General of Tax as Indonesian tax authority; (2) the technical-computing issue. There are some other things
agents use various modalities when they attempt to structure thebeyond; those include the social structure in which it is
use of IT as the organizational property. located..
This study departs from the framework that technological
Keywords— implementation, duality, tax administration implementation should not simply regarded as purely
technical issues (technological determinism). The application
I. INTRODUCTION of technology in public administration is inseparable from the
With the rise of digital age, the use of information social construction of the environment and the users [4; 5; 6].
technology (IT) in tax administration and in public The question, then, is how the stakeholders associated with
administration in general is indispensable. The use of IT in the system of the technology are aware of and then take
public administration basically aims to achieve a service advantage of the features in the existing social structure of the
excellence. IT is one of the tools to achieve the efficiency of organization in support for achieving the goal of the system
services on the one side and to support effective decision- implementation.
making at every level of government managerial on the other This paper sees itself as a study of the social aspects of the
side [1; 2]. IT implementation as part of the tax administration in
In terms of tax administration, it is interesting that Kadin Indonesia. The analytical framework used is the duality of
(the Indonesian Chamber of Commerce and Industry) once technology as proposed by [7]. This framework posits that the
stated that tax administration had not been adequately updated. use of technology is both constituted and constitutive. By
The statement was then responded by the Directorate General using a constructivist-interpretive paradigm, this research
of Taxation (DGT), the tax authority in Indonesia, by aims to gain a thorough and deep understanding (verstehen)
revealing that continuous improvement is underway, for into the research problem. The writer sought to experience
example, by revamping various aspects of IT that required the and to understand as the actors involved in the social
involvement of international financial institutions to support interaction do. The benefits of such understanding would be
the funding [3]. the accumulated knowledge for readers so that it could be
However, the initiatives did not seem to get adequate taken as a reference in decision-making or action. In other
support from the Parliament, at least from what Minister of words, the research seeks to make contribution by providing
Economy Darmin Nasution once stated: additional reference originating from experience and
knowledge of the actors involved in the use of IT in the tax
administration in Indonesia [8; 9; 10].
II. RESEARCH APPROACH

_________________________________________________ Creswell [11] formulates research approach as plans and
The views expressed in this paper are those of author and do not represent procedures of a study related to the choice of philosophical
policy of the institution with which the author is affiliated
assumptions, procedures of studies (research design) and
research method based on research questions and objectives.
____________________________________________________________________________________________________________
In practice, this study formulates its research approach as report their own payable taxes. For this flexibility, the tax
follows: it has constructivist philosophy as its assumption, its authority is entitled to recalculate the amount of tax payable
design research is discourse analysis, and its method is based on comparable data that it has. The use of IT for tax
documentation or library research. This study uses discourse administration, in essence, aims to manage all the information
analysis framework as proposed by [12] and [13]. The that describes the exercises of taxpayers’ rights and
documentation method used in the study is as described by obligations, based on either documents or particular events,
[14] and [15]. including the process to correct the amount of payable tax as
This study seeks to explore and explain how the human it should be.
agents use the surrounding structure to achieve their goals, The use of IT for such objective began in the 1980s and it
namely, the establishment of a tax administration system keeps evolving. The most crucial development took place in
capable of managing all stages of business process in taxation. 1999 with the Y2K problems, in which all computers with
By using discourse analysis, this study carries out a kind of various systems without local area network were then
―archaeology of knowledge.‖ That is, the research will upgraded to a single-platform system (Sistem Informasi
examine the various documents that have been collected to Perpajakan/Taxation Information System, or ―SIP‖ in short)
obtain an understanding that has created certain discursive using local area network. Since then, both tax offices and
practices or has made certain practices dominant [12]. regional offices have been connected to the central office
using communication technologies [17].
III. DUALITY OF TECHNOLOGY In 1995, the DGT carried out a roll out (the term was
Giddens proposed the duality of agent-structure concept as replication) to SIP as the core application system which then
a framework to analyze social problems. The concept is not includes several modules for supporting applications that feed
on a dualism of agent-structure in the sense that there is no data into it including MPN (“Modul Penerimaan
primacy or a dominant one over another but rather on the Negara”/state revenue module), e-SPT (electronic tax return
duality that agent and structure are two things that mutually application), e-Registration (online taxpayer registtration) and
cover each other’s shortfall at a certain time and place to e-Filing (online tax return submission) [18]. SIP was then
overcome social problems. Therefore, the agent-structure upgraded to ―SIPMod‖, before being replaced by ―SIDJP‖,
duality is the agents’ action in specific space and time that, to which was launched simultaneously with the 2008
a certain extent, is influenced by the structure but in another administrative reform. Table 1 presents the milestone in the
situation, influence the structure. The structure is not always a development of IT implementation in the tax authority until
constraint; it is also an enabler of the agents’ action in the 2015.
resulting social situations. In other words, the structure is both TABLE I
the medium and the result of the actions [16]. MILESOTNE OF INFORMATION SYSTEM DEPLOYMENT IN INDONESIAN TAX
Later on, Orlikowski [6; 7] adopted Gidden’s concept by ADMINISTRATION
proposing an analytical framework called duality of Timeframe Description
technology as a model for how technology is structured until circa Due to separated tax function assigned to
within an organization through the agent-structure duality 2002 different type of operational tax office, there are
scheme. Technology will become the organization’s property at least three separated core tax application
through a series of actions that the agents take using such system: ―SIP‖ in district tax office, ―SISMIOP‖ in
modalities as signification, domination and legitimation. Each district land & building tax office, ―Otomasi‖ in
of the three modalities is not discrete but one stands out at the tax audit office
time when the agents are carrying out the structuration
Tax audit used audit software (such as Microsoft
process. Signification operates using shared knowledge as an
Excel, Microsoft Access, IDEA) to deploy data
integral part of communicative encounter, while domination extraction and analysis techniques in order to
reigns by using the power to transform the social and material gathered data in audit planning, fieldwork, as well
world, and legitimation works through interaction within as as reporting
setting moral order [4].
By KUP Law, there was still no legal support
IV. IT-BASED APPLICATION SYSTEM IN INDONESIAN TAX regarding with used of electronic data as legal
ADMINISTRATION: CONTEXT OF THE CASE evidence in term of taxation matters
In general, the term ―tax‖ in Indonesia refers to the tax
2002 Development some of supporting application that
managed by the central government. Unless specifically stated,
will feed data to core application: ―MP3‖ (host to
the term used in this paper also refers to the central host online payment system), ―e-SPT‖
government tax. Actually, in line with the decentralization, (distributed application to taxpayer as electronic
there are also several types of local taxes managed by local tax return generator), ―e-Filing‖ (electronic tax
governments (provinces or regencies/cities). return submission system through
According to KUP Law No. 6/1983 and its amendments, internet/application service provider), ―e-
Indonesian tax system deploy self-assessment regime. This Registration‖ (taxpayer registration through
system provides the opportunity for taxpayers to calculate and internet)
____________________________________________________________________________________________________________
Timeframe Description V. DISCUSSION

From the perspective of the duality of technology and
Initial effort to develop integrated core tax
based on the research data, this section poses at least two
application system, called ―SAPT‖, deployed in
some offices. The rests still use ―SIP‖, that later subjects that will hopefully bring us to the understanding of
modified with some features, called ―SIPMod‖ in the situation in as the purpose of research states. First is a
until 2008 discussion on how the structure-agent duality affects the core
tax application system and how the application system relates
2007 Those three kind of existed operational tax office to the support systems. Second, how the various modalities
(district tax office, district land & building tax mediating the duality could also affect the position occupied
office, tax audit & investigation office) merged by the agent in interacting with the structure.
into single one (district tax office), inline with
this, deployment of single core tax application A. Core tax system application as new institutional property
system, called ―SIDJP‖ to server new
Referring to the description in the context of the case, the
organization structure.
study found that the most genuine efforts the agents made in
KUP Law had enacted some provision regarding the use of IT for tax administration is the development of core
with the used of electronic data as legal evidence tax system application. The most crucial shift from SIP to
in term of taxation matters SIDJP application system is the creation of a feature called
workflow/case management system. It was held that::
―MP3‖ became ―MPN‖ and operated by state ― … IT base business processes based is characterized by the
treasury authority, still only receive tax payment application of the system of "workflow" and "Case
from assigned-bank’s teller. Tax payment have to Management" in the Sistem Informasi Direktorat Jenderal Pajak
settled only through assigned-bank connected to (―SIDJP‖). ... The tax administration becomes increasingly
MPN, its receipt stamped with state-revenue- accountable for determining the start and end of a case to be
transaction-number (―NTPN/Nomor Transaksi generated by the system so it can not be manipulated by humans.
Penerimaan Negara‖) recognized as official In such systems can also be known stages of the process in a
payment receipt transparent manner, so that in case of delay, the system is easy to
detect those who are responsible ...… ―
―PINTAR‖ procurement started, but never reveal source:
who is the selected vendor to complete this http://www.pajak.go.id/sites/default/files/buletin%20XVI%202013
project. This project had been ended before it -Final%20160413.pdf (translated by author)
really kicked-off (see for example Darmin As such, it was still inadequate as SIDJP still has things to
Nasution comment about PINTAR mentioned
earlier)
fix so that the tax authority initiated the PINTAR project,
aiming at a better core application encompassing the entire
2014 - ―e-NOFA‖, online electronic tax invoice number business process of tax administration. That is, agents are
2015 generator. In order to have his/her tax invoice constantly trying to use the structure as a medium, even
legally-authorized, taxpayer have to have his/her though (at least until the time of this writing) such efforts
tax office generated tax invoice number (only the have not achieved the results as expected yet. The bottom line,
number, not the whole tax invoice). however, is how the agents seek to carry out various measures
to realize an integrated IT-based tax administration system
―e-Faktur‖, improvement of ―e-NOFA‖, taxpayer and thus making it part of the institutional property. This is a
have to have his/her tax office generated tax
finding that can be used as references transferable to another
invoice (the whole tax invoice, not only the
number). fully implemented to whole taxpayer by situation (research) in different organization, especially in
July 2016. discussing the implementation of IT.
―e-Filing‖, online tax return submission directly

B. Type of modality in duality of tax technology
without application service provider As an analytical framework, duality of technology offers a
intermedation, still limited to some taxpayers way to understand how technology is structured through
(i.e.: non-business person type of taxpayer) certain types of modalities. This means that a technology can
become an institutional property after the agents chose their
improvement of ―MPN‖ function throuh so-called modality and then carry out the structuration. Table 2 presents
―mini-ATM‖, actually this is a method of
the author's opinion about how the agents choose their
payment by using assigned-banks’ electronic data
capture (EDC) device that located in some tax modalities so that the application system can be structured.
offices. Extension of payment method, not only
through assigned-bank’s teller/counter.
source: author’s analysis
____________________________________________________________________________________________________________
TABLE 2 [2] Z. A. Hasibuan and H. B. Santoso, ―Standardisasi Aplikasi E-

STRUCTURATION OF APPLICATION SYSTEM IN TERM OF ITS MODALITY Government untuk Instansi Pemerintah,‖ presented at the
Modality Application systems Prosiding Konferensi Nasional Teknologi Informasi dan
Komunikasi Indonesia, ITB, 2005, pp. 42–48.
Signification though until 2007 there was no legal
[3] DJP, ―Ditjen Pajak sempurnakan sistem administrasi
support for this matters, tax authority
perpajakannya,‖ Direktorat Jenderal Pajak (DJP), Jakarta,
especially audit department used audit
Press Release, 2011.
software to gathere electronic data as tax
[4] W. J. Orlikowski and D. Robey, ―Information technology and
audit evidence, the objective is more on
the structuring of organizations,‖ Inf. Syst. Res., vol. 2, no. 2,
audit efficiency
pp. 143–169, 1991.
[5] R. Kling, ―What is Social Informatics and Why Does it
Domination enforce assigned-bank that receive state
Matter?,‖ -Lib Mag., vol. 5, 1999.
revenue to ―MPN‖
[6] W. J. Orlikowski, ―Using Technology and Constituting
enforce to taxpayer use electronic tax return
Structures: A Practice Lens for Studying Technology in
(e-SPT) application developed and
Organizations,‖ Organ. Sci., vol. 11, 2000.
distributed by DJP to install in taxpayer
[7] W. J. Orlikowski, ―The Duality of Technology: Rethinking the
computing environment
Concept of Technology in Organizations,‖ Organ. Sci., vol. 3,
pp. 398–427, 1992.
Legitimation efforts to establish some application system
[8] I. Wahyudi, ―Symbolism, rationality and myth in
that should make tax administration
organizational control systems: an ethnographic case study of
operated in ―service excellence‖ mode: e-
PBS Jakarta Indonesia,‖ Ph.D thesis, School of Accounting
Registration, SIDJP, plan to develop
and Finance, University of Wollongong, Wollongong,
PINTAR
Australia, 2004.
[9] L. M. Given, Ed., The Sage encyclopedia of qualitative
source: author’s analysis research methods. Thousand Oaks, California: SAGE
Publications, Inc., 2008.
VI. CONCLUSIONS [10] J. W. Creswell, Research design: Qualitative, quantitative, and
The use of IT in tax administration is a social activity. It is mixed methods approache. Thousand Oaks, California: SAGE
intensely influenced by the surrounding social structure and Publications, Inc., 2009.
the actions of the actors involved in it. Using the perspective [11] J. W. Creswell, Research Design: Qualitative, Quantitative,
of duality of technology, this study concluded that: (1) and Mixed Methods Approaches. SAGE Publications, 2013.
[12] P. Jones, Pengantar Teori-teori Sosial: dari Fungsionalisme
structuration in the use of IT in tax administration made the hingga Post-modernisme. diterjemahkan oleh Achmad Fedyani
core tax application system as the main organizational Saifuddin. Jakarta: Yayasan Pustaka Obor Indonesia, 2003.
property of DGT and of taxation system in general; (2) the [13] C. Willig, ―Discourses andcDiscourse Analysis,‖ in The SAGE
agents use different modalities when they attempt to structure Handbook of Qualitative Data Analysis, U. Flick, Ed. 2014.
the use of IT into the organization's property. [14] G. A. Bowen, ―Document Analysis as a Qualitative Research
This study did not attempt further to explore issues relating Method,‖ Qual. Res. J., vol. 9, no. 2, pp. 27–40, 2009.
to the failure in the development of PINTAR. Further studies [15] J. M. Henslin, Sociology: A Down-to-Earth Approach. Pearson
using the duality of technology analytical framework can be Higher Education, 2010.
carried out to examine the issue and to obtain an [16] A. Giddens, The Constitution of Society: Outline of the Theory
of Structure. California: University of California Press, 1984.
understanding of why the development of the system failed. [17] DJP, ―SE - 05/PJ.9/1999 tentang antisipasi kegagalan
REFERENCES komputer tahun 2000 (contingency plan y2k).‖ Direktorat
Jenderal Pajak (DJP), 1999.
[1] UN, Government Information Systems: A Guide to Effective
[18] DJP, ―keputusan direktur jenderal pajak nomor kep -
Use of Information Technology in the Public Sector of
29/pj./1995 tentang replikasi sistem informasi perpajakan.‖
Developing Countries. New York: Division of Public
Direktorat Jenderal Pajak (DJP), 1995.
Administration and Development Management, Department
for Development Support and Management Services, United
Nations, 1995.
____________________________________________________________________________________________________________
Joint Passive Time-Reversal and Adaptive Equalizer

for Underwater Acoustic Communication
Yuning Widiarti#*1 , Suwadi#2, Wirawan#3
#
Department of Electrical Engineering, Institut Teknologi Sepuluh Nopember (ITS)
Sukolilo, Surabaya, Indonesia 60111
1
yuning.widiarti15@mhs.ee.its.ac.id
2
suwadi@ee.its.ac.id
3
wirawan@ee.its.ac.id
*
Politeknik Perkapalan Negeri Surabaya (PPNS)
Jl. Teknik Kimia, Sukolilo, Surabaya, Indonesia 60111
Abstract— The problems most difficult in shallow underwater bit rate and the worsening of channel characteristics,
acoustic communication is multipath propagation varies with equalizers become significantly complex, so they are
time to produce a negative impact on the data rates. At high data prevented from using in high data rate underwater acoustic
rates intersymbol interference (ISI) requires adaptive algorithms communication.
on the receiver side with intensive computing systems and
Time reversal (TR) is a relatively simple approach and the
complicated. A new technique called acoustic time reversal. This
technique can adapt to the effects of propagation in complex computational complexity is low, so it has attracted
medium. Furthermore, he can focus more energy on the target considerable attention in underwater communications in
and a specific depth. Passive time reversal is one of interesting recent years [2]. The properties of time reversal include:
method as a computationally inexpensive means for eliminating temporal focusing can reduce ISI and spatial focus can reduce
the ISI introduced by channel. In this paper we combine time- channel fading and improve signal-to-noise ratio (SNR). The
reversal technique with least-mean-square equalizer (LMS). preliminary concepts and experiments for active TR
Spatial focus property possessed by time reversal able to communications have been demonstrated [3]. The preliminary
overcome channel fading and produces the signal-to-noise-ratio concepts and experiments for passive TR communication have
(SNR) at the receiver (array). While the temporal focus property
been demonstrated [4]. To overcome the problem of residual
is able to reduce the ISI significantly. We show that the
performance of time-reversal communication can be improved ISI, the adaptive channel equalization techniques combined
significantly by using quadrature phase-shift keying (QPSK) with time reversal. Performance of time reversal
modulation and cascading the received time series with an LMS communication can be improved using single channel
equalizer to reduce ISI. equalizer [5].
Passive TR combined with adaptive multichannel
Keywords— passive time-reversal, LMS equalizer, ISI,
combining is proposed [6]. The performance of passive TR
underwater acoustic communication communications can be improved and computational intensity
I. INTRODUCTION of communications is significantly decreased through the
Underwater sound propagation is mainly influenced by combination. In the experiment of [7], the maximum number
transmission loss, noise, reverberation, and temporal and of receiver elements is 10 and the multipath propagation
spatial variability of the underwater acoustic channel. Noise which varies from 5 symbol intervals to 15 symbol intervals is
and transmission loss limit the available signal-to-noise ratio, not severely, so taps of adaptive multichannel combiner are
while signal processing and design influenced by the time- relatively few. This paper is motivated by the work of Zhang
varying multipath. Multipath propagation contributes to signal [6-7] is that a single channel equalizer is used between passive
fading, and causes inter symbol interference in a digital TR and adaptive multichannel combining in each spatial
communication system. diversity channel. When multipath propagation is relatively
The bandwidth-limited underwater acoustic (UWA) severe and the number of receiver elements is few, an adaptive
channel is a time-varying multipath environment. Inter symbol multipath combiner that has few taps can’t achieve good
interference (ISI) caused by multipath propagation distorts the performance. Performance of passive TR communication can
signal and degrades the performance of high data rate be improved by combining it with adaptive multichannel. In
communications. Adaptive channel equalization is one of the this paper the spatial diversity equalization refers to the
effective methods which can reduce the effect of ISI in combination of adaptive multichannel.
underwater acoustic communications [1]. As the increasing of
____________________________________________________________________________________________________________
The signals from all the elements of the receiver are combined
into:
() () [ ( ) ( )]
= ) ) (5)
N is the number of receiver elements and the q(t) function

indicates the sum of the autocorrelation of each channel
impulse response [8]. Passive TR communication
Fig. 1 Passive Reversal Time System and Adaptive Equalizer Model [9] performance depends on the behavior of the q (t) function.
The passive TR communication process doesn’t include the
This paper is organized as follows: In section II, the effect of noise. Elements of the receiver record both signal
theory of passive TR communications is outlined. The method and noise. However, noise is incoherent so noise is not retro-
of passive TR communications combined with adaptive focused through a process of passive TR.
equalization and simulation results are presented section III. A. Joint Passive Time Reversal And Adaptive Equalization
The last section presents our conclusions. This section presents the time reversal passive
communication combined with adaptive equalization as
II. PASSIVE TIME REVERSAL THEORY shown in Figure 1. Referring to Figure 1, a sequence of data
In this section we present the theory of passive time symbols In is sent out by a source at a symbol rate 1 / T. The
reversal communication. Passive time reversal communication impulse response of the transmitter filter G0 (f) made fix and
is one-way communication process. Matched filter element is power spectral density of uncorrelated noise w i (t) is Sn (f) =
implemented in the receiver using the channel impulse No, i = 1,2, ..., N. From the perspective of passive TR
response obtained by sending a signal probe [4]. When a processing, the impulse response of the ith receive filter is
known probe signal p(t) is sent out by a source, the signal Gi(f)= GT*(f)Hi*(f), where Hi(f), i = 1,2, ..., N is the channel
received by the ith receiver element can be written as follows: impulse response of the ith channel. Samplers following
receive filters that operate at symbol rate 1 / T and a sampler
output signals are sent to the adaptive equalizer.
() () () (1) B. Process of Passive Time Reversal Communication
TR combined with the adaptive equalizer in passive time
where ( = 1,2, ..., N) shows the channel impulse response reversal communication process, the format data can be
between the source and the ith receiver element, while * shown in figure 2. In figure 2 probe signal is used to estimate
denotes convolution operation. On the receiving element the channel impulse response. The data sequence is comprised
probe signal p(t) is used as a matched filter to process yp (t), of training sequence and information. Training sequence is
the output signal of ith matched filter is: used to determine the tap coefficients. There is a guard delay
between probe signal and the data sequence to accept all
() () ( ) multipath propagation of probe signal and estimate the
=[ ( ) ( )] () channel impulse response correctly.
= ( ) () (2) The process of passive TR communication can be shown
in Figure 3. Referring to Figure 3 modulated data signals
transmitted by the source after the shaping impulse. Each
where Rp ( ) is the autocorrelation function p(t), channel receiver element is implemented in passive TR to the received
impulse response h i (t) can be estimated correctly using data signals using the estimated channel impulse response.
equation (2). After multipath arrivals are clear, the data signal The signals processed by the passive TR are then sent to the
s (t) is transmitted and the signal received by the ith receiver adaptive equalizer. After the demodulation process and
element is: decision-making, estimation data is acquired.
() () () (3) Probe signal Guard delay Training Information

sequence sequence
Fig. 2 The transmitted data format
At each element of the receiver array, the obtained channel
impulse response hi (t) is used as a matched filter to process C. Channel Model
the received data signal, then the output of the ith matched A shallow water channel model has relatively severe
filter is: multipath propagation is built based on transmitter-receiver
geometry, environmental conditions and system parameters. The
() () [ () ( )] (4)
____________________________________________________________________________________________________________
parameter measurements have been performed in a concrete

towing tank with dimensions 12 m x 200 m, and 6m in depth.
There were no waves and sources of transient noise. The
measurement were carried out in uniform medium conditions.
The generated sound is chirp signal with frequency sweep
from 200 Hz to 17000 Hz, in 2 sec duration. The chirp signal
was used in the experiment as in the following equation:
()= (2 ( )+ ) (6)
(b)
Fig. 5.Received signals (a) Channel 1, (b) Channel 2
III. ANALYSIS OF SIMULATION RESULTS
In this session we present the simulation results of passive
time reversal communication for the two conditions: time
reversal without equalization and time reversal combined with
adaptive channel equalizer.
A. Chirp signal
Simulation results are implemented based on time-
unvarying shallow water channel model. The chirp signal was
used in the results as in the figure 4.
Fig. 6.Channel impulse response for channel of 103-m depth.
Channel impulse response

Sound speed is c = 1500m/s. A source is deployed at 3 m
depth. A 2-element vertical receiver array is deployed
spanning the water column with 50 cm space. The range
between the source and the array is 103 m. The multipath
Fig. 3.Passive TR Communication Process arrival structures obtained in shallow water channel at each
sensor can be shown in figure 5. By using matched filtering,
we can calculate the multipath arrival structure. The result can
be obtained by taking the envelope of the cross correlation
sequence of received signal with the transmitted signal.
A sequence of impulses obtained from each echo in
received waveform. They provide a visualization of the
impulse response. The channel impulse response shows the
variation in the multipath structure can be shown in figure 6
B. Performance of Communication Using Passive Time-
Reversal and Linear Equalizer
Passive TR communication without equalization is simulated in
the shallow water channel model. The probe signal p0(t) is a
linear frequency modulation (LMF) chirp with the effective time
of probe signal is 2 sec. Quadrature phase-shift keying (QPSK)
is used to encode the data sequence. The symbol rate is
R=1/T=500 symbols/sec.
Fig. 4.Spectrogram of the chirp signal
(a)
(a)
____________________________________________________________________________________________________________
using TR processing, there always is residual ISI. In order to

reduce the effect of residual ISI and improve the performance of
communications, adaptive equalization is used in passive TR
communications. The results demonstrate that the performance of
passive TR combined with the adaptive equalizer is superior to
that of communications using passive TR alone.
ACKNOWLEDGMENT
This work was supported by BPPS grant fiscal year 2016.
(b)
We are grateful to Laboratorium Hidrodinamika Indonesia
(LHI) for their support to carry out the measurements, so the
research could proceed smoothly.
REFERENCES
[1] M.Stojanovic, J.Capitovic and J.Proakis, "Adaptive
multichannel combining and equalization for underwater
acoustic communications" J.Acoust. Soc. Am., vol. 94, pp.
1621-1631, Sep. 1993.
(c) [2] H. Song, W. Hodgkiss, W. Kuperman, M. Stevenson, and
Fig. 7 Performance of QPSK with various SNR.(a)SNR=10 dB (b)SNR=20
dB (c)40 dB T. Akal, "Improvement of time reversal communications
using adaptive channel equalizers," J. Ocean. Eng., vol.
Experiments of passive TR communications combined 31, pp. 487-496, Apr. 2006.
with linear equalization are implemented in the same shallow [3] W. A. Kuperman, W. S. Hodgkiss, H. C. Song, T. Akal,
water channel. The probe signal, carrier signal, modulation C. Ferla, and D.Jackson, "Phase conjugation in the ocean:
mode and symbol rate are same as the experiments of Experimental demonstration of an acoustic time-reversal
communications using passive TR alone. The data sequence mirror," J. Acoust. Soc. Am., vol. 103, pp. 25-40, Jan.
contains 5000 symbols and 200 symbols ahead are used as 1998.
training symbols. The receiver elements are selected [4] D. Rouseff, D. Jackson, W. Fox, C. Jones, J. Ritcey, and
sequentially from the surface. The number of receiver D. Dowling, "Underwater acoustic communications by
elements used in experiments is 2. The number of taps used is passive-phase conjugation: Theory and experimental
13 with convergence factor µ=0.001. results," J. Ocean. Eng., vol. 26, pp. 821-831, Oct. 2001.
In order to compare, experiments of passive TR alone and [5] H. Song, P. Roux, W. Hodgkiss, W. Kuperman, T. Akal,
passive TR communications combined with a single channel and M. Stevenson, "Multiple-input-Multiple-output
linear equalizer are conducted. Figure 7 shows the coherent time reversal communications in a shallow-
constellation diagrams of passive TR alone (left side) and water acoustic channel," J. Ocean. Eng., vol. 31, pp. 170-
passive TR combined with linear equalizer (right side). 178, Jan. 2006.
The performance of time-reversal communication is [6] G. S. Zhang, and H. F. Dong, "Spatial diversity in
shown in figure 7(a). The result in a symbol error rate (SER) multichannel processing for underwater acoustic
was 0.0205 and the symbol SNR was 10 dB at the receiver. communications," J. Ocean Engineering, vol. 38, pp.
Because of many received multipath wave, the better 1611-1623,Oct. 2011.
performance at the SNR has not been reached. There is a [7] G. S. Zhang, and H. F. Dong, "Joint passive-phase
residual ISI that degrades the performance. Figure 7(b) and conjugation with adaptive multichannel combining for
7(c) shows an improved performance by cascading an coherent underwater acoustic communications," J.
adaptive LMS equalizer where the scattered symbols are Applied Acoustics, vol. 73, pp. 433-439,Apr. 2012.
tightened. At SNR = 20dB and SNR=40 dB, passive TR [8] T. Yang, "Temporal resolutions of time-reversed and
communication performance with adaptive equalizer TR passive-phase conjugation for underwater acoustic
yields a better performance compared to the TR communications," J. Oceanic Eng.,vol. 28, pp. 229-245,
communication alone. This looks at SNR = 20 dB, scatter plot Apr.2003.
produces SER at 0.0014 and at SNR = 40 dB, SER equals 0. [9] H. Song, W. Hodgkiss, W. Kuperman, W. Higley, K.
In general, the results of scatter plot show that the Raghukumar and T. Akal , "Spatial diversity in passive
communication time reversal combined with the equalizer can time reversal communications," J. Acoust. Soc. Am., vol.
yield a better performance due to the increasing value of SNR. 120,pp.2067-2076,Oct.2006
IV. CONCLUSION
Time reversal method with the spatial and temporal
focusing properties offer a potential application to underwater
acoustic communication. While ISI is reduced significantly
____________________________________________________________________________________________________________
Preliminary Studies on Establishment of Malaysia’s

National Standard of RF Attenuation
Arshad Selamat#1, Anton Widarta*2, Mohd Nasir Zainal Abidin#3
#
National Metrology Institute of Malaysia (NMIM)
SIRIM Berhad, Lot 4803, BBST, 43900 Sepang, Malaysia
1
arshads@sirim.my
3
drnasir@sirim.my
*
National Metrology Institute of Japan (NMIJ)
Advanced Industrial Science and Technology (AIST), Tsukuba, Japan
2
anton-widarta@aist.go.jp
Abstract— This paper describes the preliminary studies on the uses an inductive voltage divider (IVD) as the reference
establishment of a precision attenuation measurement system as standard which can be operated at selective frequency of 1
a national primary standard in Malaysia. The system is based on kHz [1] or 10 kHz [6]. The system has been chosen not only
the system developed by NMIJ that uses dual channel IF for its precision, but also due to the least amount of capital
substitution method operating at 30 MHz frequency. A traceable
required.
inductive voltage divider (IVD) works at 1 kHz is used as the
reference standard and is employed by the system. The
performance of the system has been evaluated and the obtained
data are compared with the previous published results.
Keywords— Nonlinearity, IVD, Attenuation
I. INTRODUCTION
Attenuation measurement is essential as the frequency goes
higher up to 110 GHz. Losses occur and attenuate some
signals in microwave system due to imperfection matching of
the transmission line impedance. This improper matching
introduces mismatch between connections in the system and
reflects some incident signals especially at radio frequency
(RF) spectrum. At microwave level, power losses due to noise
and leakage take place. These losses need to be addressed as
they are the major contributors to the reduction of radiated
power in RF and microwave system.
There are several methods available to measure attenuation.
These include power ratio method, RF substitution method,
AF substitution method, swept frequency method, and method
using VNA. In Malaysia, the VNA has long been used to
measure attenuation. However, VNA has many disadvantages
as the attenuation dynamic range is very limited. Furthermore,
Fig. 1 Block diagram of precision attenuation measurement using IVD
the VNA uncertainties are normally big depending on the
VNA performance. Although there are many precision
attenuation measurement systems available today, their
limitations on certain ranges of operating frequency made II. MEASUREMENT SYSTEM
them less suitable as standard. For example, some difficulties The attenuation measurement system, as shown in Fig. 1,
occurred when using the waveguide below-cutoff (WBCO) can be divided into two sections (RF and IF). The basic
below 100 MHz frequency [1]. principle of the system is the RF attenuation of the DUT will
In this paper, studies of NMIJ’s precision attenuation be converted to IF attenuation by the IVD using the dual
measurement system intended to be established as Malaysia’s channel IF substitution method, where the IVD will be the
national primary standard will be discussed. The system uses reference standard. RF signal will flow from signal source (in
IF substitution for high accuracy attenuation measurement. It RF section) to a directional coupler and divides the signal to
has been selected to work at the frequency of 30 MHz and two different paths.The first path carries the measurement
____________________________________________________________________________________________________________
signal through the tuning box and goes straight to the DUT, -30 to -40 dBm -5.4 x 10
-5
while the second path flows down directly to a reference -5

mixer to channel a reference signal to the system. The -40 to -50 dBm -9.3 x 10
measurement signal will be mixed with the LO signal in the -50 to -60 dBm -1.3 x 10
-4
main mixer to produce the IF signal. The dotted line in red -60 to -70 dBm -3.4 x 10
-4
colour represents the undesired signal flow which is also -3
called internal leakage. Isolators are installed between the -70 to -80 dBm -1.0 x 10
main and reference mixers to prevent the internal leakage. In -80 to -90 dBm -8.2 x 10
-3
the IF section, the IVD is then making a ratio of the IF signal -3

-90 to -100 dBm -3.4 x 10
incident to it by initial and final setting of the DUT at the RF -2
section. The IF attenuation produced by the IVD ratio is -100 to -110 dBm -3.3 x 10
proportional to the RF attenuation of the DUT. The reference
signal is employed to achieve a high precision measurement
B. RF Leakage
by using a null balance method, where the IF signals are kept
in balance by adjusting the IVD setting and the phase shifter As mentioned earlier, the leakage is caused by two main
before and after setting of the DUT. Thus the attenuation is sources. The first source is from internal leakage that
determined by the ratio of the IVD values at initial and final corresponds to isolation performance of the system. The
settings. isolation of the attenuation system is expected to be larger
than 180 dB, thus the internal leakage is 80 dB lower for the
measurement up to 100 dB (ratio of 0.0001). The second
III. MEASUREMENT RESULTS source of the leakage comes from external leakage. This
A linearity measurement has been conducted at nominal 10 leakage is due to some signal flowing outside of devices in the
dB attenuations with 10 dB intervals at the frequency of 30 system. The signal will bypass the DUT and reach the main
MHz and the results are provided in Fig. 2. The horizontal mixer. From the results obtained from the nonlinearity
axis indicates the input level of main mixer in dBm unit. The measurement, it is observed that the leakage is estimated to be
vertical axis indicates the deviation of attenuation in dB unit. ±0.001 dB up to 70 dB.
Based from the results obtained, the system is expected to
have an accuracy of 0.001 dB for a dynamic range of over 50 C. Stability
dB, while for over 80 dB dynamic range the accuracy is about
0.01 dB. Limits of the nonlinearity can be found form Table 1. The stability of the system was observed for about 10
minutes based from its IF output signal. The IF frequency was
set at 1 kHz where the input power level feed to the main
A. Nonlinearity mixer remained at -30 dBm. The system has also been set to
10 dB step setting using a band gap attenuator. Fig. 3 shows
the plot of the results obtained, indicating that the stability of
the 1 kHz IF output signal is less than 0.0003 dB.
0.0002 dB
Fig. 2 Nonlinearity of the system
TABLE I
LIMITS OF NONLINEARITY
Attenuation Range Limit

-4
-10 to -20 dBm 6.9 x 10 Fig. 3 Stability of IF output response at 1 kHz
-6
-20 to -30 dBm 2.0 x 10
____________________________________________________________________________________________________________
D. Type A
Repeatability measurement of the attenuation system was TABLE IV
performed by making five measurements and the values taken PRELIMINARY UNCERTAINTY ESTIMATIONS FOR ATTENUATION
are tabulated in Table II (i.e at 10 dB). The estimated MEASUREMENTS AT 30 MHZ
uncertainty for the repeatability is ±0.000199 dB.
TABLE II
TYPE A
R1 R2 Attenuation
0.001012 0.003196 9.993686
0.001012 0.003196 9.992531
0.0010116 0.003196 9.992789
0.0010116 0.003196 9.992881
0.0010117 0.003196 9.992735 IV. CONCLUSIONS
The preliminary study of the precise attenuation system
Average = 9.992925, based on NMIJ’s developed system employing dual-channel
Standard deviation = 0.000445 IF substitution method and a traceable IVD as the reference
standard operating at 30 MHz has been concluded. The
estimated expanded uncertainty of 0.034 dB for attenuation
range up to 80 dB obtained from this study shows a marked
E. Uncertainty Table improvement and can be developed further to make the
system a reference standard.
The combined standard uncertainty is calculated using the
following equation:
ACKNOWLEDGMENT
2 2 2 2 2 2 2
Uc = U1 + U2 + U3 + U4 + U5 + U6 (1) The authors would like to thank the staff of Radio
Frequency Group at NMIJ for all the technical supports
while the expanded uncertainty is determined using the provided.
below stated expression:
U=k.Uc. (2) REFERENCES

[1] A. Widarta and T. Kawakami, "Attenuation measurement system in the
where k is the coverage factor and Un is the standard frequency range of 10 to 100 MHz,” IEEE Trans. Instrum. Meas., vol.
52, no. 2, pp. 302-306, Apr 2003.
uncertainty component. Detailed preliminary estimations for [2] T. Kawakami, A. Nagatuka and I. Yokoshima, “30-MHZ attenuation
the measurement uncertainty are shown in Table III, whereas standard using binary phase modulation method,” IEEE Trans. Instru.
the summary of the attenuation from 0 to 80 dB according to Meas., vol. 38, pp. 448-450, Apr. 1989.
their measurements ranges are given in Table IV. [3] T. Kawakami, A. Nagatuka, M. Maeda and S. Igarashi, “RF attenuation
measurement system with 1-kHz voltage ratio standard,” IEEE Trans.
Instru. Meas., vol. 42, pp. 1014-1019, Dec. 1993.
TABLE III [4] D. H. Russell, “The waveguide below-cutoff attenuation standard,”
PRELIMINARY UNCERTAINTY ESTIMATIONS FOR 10 DB ATTENUATION AT 30 IEEE Trans. Microwave Theory Tech., vol. 45, pp. 2408-2413, Dec.
MHZ (-10 DBM TO -50 DBM) 1997.
[5] Y. Y. Shan and S. W Chua, “A dual channel af substitution microwave
attenuation measurement system for precision attenuator calibration,”
NCSL International. Workshop and Symposium., 2007.
[6] A. Widarta H. Iida, and T. Kawakami, "Attenuation-measurement
system in the frequency range of 18-40 GHz,” IEEE Trans. Instrum.
Meas., vol. 56, no. 2, pp. 641-645, Apr 2007.
[7] W. Ruhle, D. Stumpe, "Voltage ratio technique used with the national
rf standard attenuation measuring equipment at the PTB,” IEEE Trans.
Instrum. Meas., vol. 42, pp. 1009-1013, Jun. 1993.
[8] I. A. Harris and F. L. Warner, “Re-examination of mismatch
uncertainty when measuring microwave power and attenuation,” Porc.
Inst, Electr. Eng., vol. 128, no.1, pp. 35-41, Feb. 1981.
____________________________________________________________________________________________________________
Nitrogen gas temperature evaluation in pulse-

modulated rf magnetron sputtering plasmas
Nafarizal Nayan#1, Soo Ren How#2, Mohd Khairul Ahmad#, Mohd Zainizan Sahdan#
#
Microelectronic and Nanotechnology-Shamsuddin Research Centre (MiNT-SRC)&
Faculty of Electrical and Electronic Engineering,
Universiti Tun Hussein Onn Malaysia 86400 Parit Raja, Johor, Malaysia
1
nafa@uthm.edu.my
2
lenz5351@gmail.com
Abstract— In the present work, we evaluated the nitrogen gas spectroscopy (OES) system. The sputtering plasma
temperature in pulse-modulated radio frequency (rf) magnetron
sputtering plasmas. The main objective of this work is to control was produced by a 13.56 MHz rf magnetron
the gas temperature using pulse modulating system. We found discharges with an automated matching network.
that at low discharge pressure, the pulse duty ratio plays an The sputter target was made of 3 inches pure
important role in the nitrogen gas temperature. On the other
hand, at high discharge pressure, the pulse frequency plays an (99.99% purity) titanium target. The sputtering gas,
important role in the nitrogen gas temperature. In addition, the argon and reactive gas, nitrogen was introduced
gas temperature was found significantly low at high discharge into the chamber by using mass flow controller that
pressure condition.
was attached to the top of the chamber. The argon
and the nitrogen flow rate were fixed at 100 sccm
Keywords— magnetron sputtering plasma, optical emission
spectroscopy, gas temperature evaluation, thin film and 10 sccm, respectively. The discharge pressures
were 10 and 100 mTorr. The optical emission
I. INTRODUCTION intensities of excited atomic species in the plasma
Magnetron sputtering plasma is widely used in to were collected by the optical fiber that was
deposit uniform and high quality thin films[1]–[3]. connected to monochromator from Princeton
In the case of compound thin films deposition, radio Instruments. Emission light from the plasma was
frequency (rf) 13.56 MHz is used as power supply detected through the side view port of the
since most of the time compound sputter target is magnetron sputtering system as shown in Fig. 1.
nonconductive. RF magnetron sputtering plasma is The view port windows were made of quartz glass.
also known as reactive magnetron sputtering when
the reactive gas such as oxygen and nitrogen are
introduced in the chamber to deposit oxide and
nitride thin films[4]–[7]. One of the major
advantages of rf magnetron sputtering plasma
deposition is that it deposit film at relatively low
temperature. Recently, the rf magnetron sputtering
plasma was used to deposit thin film on plastic
surface[8]–[10]. In such deposition, the gas
temperature during the film deposition is very
important. In the present work, we propose the use
of pulse-modulated rf magnetron sputtering plasmas
as a new technique to deposit films at very low
temperature.
II. EXPERIMENTAL SETUP
Figure 1 show the experimental setup using rf Fig. 1 Magnetron sputtering plasmas and optical emission spectroscopy
magnetron sputtering system and optical emission measurement using high-resolution monochromator.
____________________________________________________________________________________________________________
III. RESULT AND DISCUSSION

We operated the rf power supply in pulsing mode
and capture the optical emission spectroscopy from
the plasmas. In the present study, we measured
nitrogen emission at the range of 773-776 nm,
which is the rotational subband of the first positive
(2-0) system of nitrogen emission spectrum. The
spectrometer has a resolution of better than 0.05 nm.
As shown in Fig. 2, nitrogen emission was
successfully captured at various duty ratio and pulse
frequency of rf power supply. This nitrogen
emission is corresponding to emission from first
positive of molecular nitrogen as reported by N.
Britun. et. al. [11], [12]. The rf power and discharge
pressure were fixed at 150 W and 10 mTorr,
respectively. Figure 2 (a) shows significant
influence of duty ratio toward the nitrogen emission.
On the other hand, figure 2(b) shows that the
nitrogen emission was almost the same at various
pulsing frequency. Since the spectrum was captured
using high-resolution spectrometer, the spectrum
profile is more detail than that reported by N.
Britun and other researchers[13]–[15]. Then we
repeated the experiment with higher discharge
pressure.
Fig. 3 shows the nitrogen emission when the rf
Fig. 2 Nitrogen emission at various (a) duty ratio and (b) pulse frequency.
power and discharge pressure were fixed at 150 W The rf power supply and the discharge pressure were fixed at 150 W and 10
and 100 mTorr, respectively. By comparing the mTorr, respectively.
peak intensity values in Fig. 2(a) and 3(a), we

understood that the emission intensity increased Nitrogen emission spectrum in Fig. 2 and 3 were
with the discharge pressure that is commonly used to evaluate the gas temperature inside the
understood when we introduce larger volume of plasmas at various duty ratios and pulse frequencies.
nitrogen molecules. Fig. 3(a) shows a significant The evaluation technique is a simple calculation by
influence of duty ratio on the nitrogen emission using the peak intensity of nitrogen emission at
spectrum. The emission intensity increased with the 774.0 (name as P1) and 775.4 (name as P2) nm.
duty ratio, as shown in Fig. 3(a) and 2(a). In The gas temperature or also known as rotational
addition, the effect of pulse frequency was slightly temperature can be calculated from
clearer at high discharge pressure as shown in Fig. Trot(K)=195/[(P2/P1)-0.52]. N. Britun has proposed
3(b). this formula from his previous work on nitrogen
plasma in ICP discharges[11].
____________________________________________________________________________________________________________
Fig. 4 Gas temperature as a function of duty ratio in 10 and 100 mTorr

discharge pressure of magnetron sputtering plasmas.
Fig. 3 Nitrogen emission at various (a) duty ratio and (b) pulse frequency.
The rf power supply and the discharge pressure were fixed at 150 W and 100
mTorr, respectively.
Fig. 5 Gas temperature as a function of pulse frequency in 10 and 100 mTorr

Gas temperatures at various duty ratio and pulse discharge pressure of magnetron sputtering plasmas.
frequency were evaluated from the nitrogen

emission spectrum. Fig. 4 shows the gas intensity of nitrogen spectrum in Fig. 3(a) increased
temperature as a function of duty ratio for 10 and with the duty ratio, the gas temperature at high
100 mTorr of discharge pressure. We chose 10 and discharge pressure was almost unchanged even
100 mTorr for sake of comparison between low and when the duty ratio increased.
high discharge pressure, respectively. We use the The effect of pulse frequency on the nitrogen gas
term “high discharge pressure” here, since in a temperature is shown in Fig. 5. The influence of
conventional magnetron sputtering deposition, the pulse frequency was not significant at low
discharge pressure is generally at below than 20 discharge pressure, as shown by filled rectangular.
mTorr. At low discharge pressure, nitrogen gas The nitrogen gas temperature was almost the same
temperature increased with the duty ratio. On the at 420 K at low discharge pressure. On the other
other hand, at high discharge pressure, the nitrogen hand, the nitrogen gas temperature was significantly
gas temperature was almost the same. At high lower when the discharge pressure changed to 100
discharge pressure, the atoms and molecules in mTorr. At high discharge pressure, the nitrogen gas
discharge chamber collide between each other more temperature slightly decreases with the pulse
frequent and transfer their energy rapidly[16]–[18]. frequency, as shown by filled circles in Fig. 5.
As a result, we can see that although the emission
____________________________________________________________________________________________________________
Therefore, it is understood that at high discharge as protective coating on bell-metal by reactive sputtering in
cylindrical magnetron device,” Appl. Surf. Sci., vol. 254, no. 18, pp.
pressure, the mean free path of atoms and 5760–5765, Jul. 2008.
molecules in the plasmas is small and contribute to [6] C.-C. Li, D.-H. Kuo, P.-W. Hsieh, and Y.-S. Huang, “Thick In x
Ga1−x N Films Prepared by Reactive Sputtering with Single
rapid collision rate within particles and chamber Cermet Targets,” J. Electron. Mater., vol. 42, no. 8, pp. 2445–2449,
wall and finally reduced the gas temperature. [7]
Jun. 2013.
R. U. Snyders, M. Wautelet, R. Gouttebaron, J. P. Dauchot, and M.
Results in Fig. 4 and 5 clearly show that the gas Hecq, “Correlation between the gas composition and the
temperature can be easily control by the discharge stoichiometry of SnO x films prepared by DC magnetron reactive
sputtering,” pp. 187–191, 2001.
pressure, the duty ratio and the pulse frequency. [8] M. A. Surmeneva, A. I. Tyurin, T. M. Mukhametkaliyev, T. S.
Pirozhkova, I. A. Shuvarin, M. S. Syrtanov, and R. A. Surmenev,
“Enhancement of the mechanical properties of AZ31 magnesium
alloy via nanostructured hydroxyapatite thin films fabricated via
IV. CONCLUSIONS radio-frequency magnetron sputtering.,” J. Mech. Behav. Biomed.
Mater., vol. 46, pp. 127–36, Jun. 2015.
In this work we have successfully determined the [9] K.-I. Park, S. Xu, Y. Liu, G.-T. Hwang, S.-J. L. Kang, Z. L. Wang,
nitrogen emission spectrum in pulse-modulated rf and K. J. Lee, “Piezoelectric BaTiO₃ thin film nanogenerator on
magnetron sputtering plasmas using high-resolution plastic substrates.,” Nano Lett., vol. 10, no. 12, pp. 4939–43, Dec.
2010.
spectrometer. The profile of each spectrum was [10] A. N. Banerjee, C. K. Ghosh, K. K. Chattopadhyay, H. Minoura, A.
clearly observed and distinguished. Then we used a K. Sarkar, A. Akiba, A. Kamiya, and T. Endo, “Low-temperature
deposition of ZnO thin films on PET and glass substrates by DC-
simple formula to evaluate the nitrogen gas sputtering technique,” Thin Solid Films, vol. 496, no. 1, pp. 112–
temperature from the peak intensity of nitrogen 116, Feb. 2006.
[11] N. Britun, M. Gaillard, A Ricard, Y. M. Kim, K. S. Kim, and J. G.
emission at 774 and 775 nm. We found significant Han, “Determination of the vibrational, rotational and electron
influence of discharge pressure, pulse duty ratio and temperatures in N 2 and Ar–N 2 rf discharge,” J. Phys. D. Appl.
Phys., vol. 40, no. 4, pp. 1022–1029, Feb. 2007.
pulse frequency on the nitrogen gas temperature. In [12] N. Kang, N. Britun, S. Oh, F. Gaboriau, and A. Ricard,
recent progress of magnetron sputtering plasma “Experimental study of Ar and Ar–N 2 afterglow in a pulse-
modulated ICP discharge: observation of highly excited Ar(6d)
deposition, low temperature deposition on plastic afterpeak emission,” J. Phys. D. Appl. Phys., vol. 42, no. 11, p.
and flexible substrate is required for future 112001, Jun. 2009.
[13] X.-M. Zhu and Y.-K. Pu, “Optical emission spectroscopy in low-
technology. Therefore, we proposed that pulse- temperature plasmas containing argon and nitrogen: determination
modulated rf magnetron sputtering plasma is useful of the electron temperature and density by the line-ratio method,” J.
Phys. D. Appl. Phys., vol. 43, no. 40, p. 403001, Oct. 2010.
for such applications. [14] N. Britun, T. Godfroid, S. Konstantinidis, and R. Snyders, “Time-
resolved gas temperature evolution in pulsed Ar–N[sub 2]
ACKNOWLEDGMENT microwave discharge,” Appl. Phys. Lett., vol. 98, no. 14, p. 141502,
2011.
This work is partially supported by the [15] J. Zhang, L. Liu, T. Ma, and X. Deng, “Rotational temperature of
Exploratory Research Grant Scheme ERGS E025 nitrogen glow discharge obtained by optical emission
spectroscopy.,” Spectrochim. Acta. A. Mol. Biomol. Spectrosc., vol.
and Fundamental Research Grant Scheme 1207 of 58, no. 9, pp. 1915–22, Jul. 2002.
Ministry of Higher Education Malaysia. [16] P. Saikia and B. Kakati, “Effect of process parameters on properties
of argon–nitrogen plasma for titanium nitride film deposition,” J.
REFERENCES Vac. Sci. Technol. A Vacuum, Surfaces, Film., vol. 31, no. 6, p.
061307, 2013.
[1] I. Safi, “Recent aspects concerning DC reactive magnetron [17] J. M. Palomares, W. A. A. D. Graef, S. Hübner, and J. J. A. M. van
sputtering of thin films: a review,” Surf. Coatings Technol., vol. 127, der Mullen, “Time resolved laser induced fluorescence on argon
no. 2–3, pp. 203–218, May 2000. intermediate pressure microwave discharges: Measuring the
[2] J. T. Gudmundsson, N. Brenning, D. Lundin, and U. Helmersson, depopulation rates of the 4p and 5p excited levels as induced by
“High power impulse magnetron sputtering discharge,” J. Vac. Sci. electron and atom collisions,” Spectrochim. Acta Part B At.
Technol. A Vacuum, Surfaces, Film., vol. 30, no. 3, p. 030801, 2012. Spectrosc., vol. 88, pp. 156–166, Oct. 2013.
[3] A. Anders, “A review comparing cathodic arcs and high power [18] A. Palmero, E. D. van Hattum, W. M. Arnoldbik, A. M. Vredenberg,
impulse magnetron sputtering (HiPIMS),” Surf. Coatings Technol., and F. H. P. M. Habraken, “Characterization of the plasma in a
vol. 257, pp. 308–325, Sep. 2014. radio-frequency magnetron sputtering system,” J. Appl. Phys., vol.
[4] Z. Li and W. Gao, “ZnO thin films with DC and RF reactive 95, no. 12, p. 7611, 2004.
sputtering,” Mater. Lett., vol. 58, no. 7–8, pp. 1363–1370, Mar.
2004.
[5] S. M. Borah, A. R. Pal, H. Bailung, and J. Chutia, “Optimization of
plasma parameters for high rate deposition of titanium nitride films
____________________________________________________________________________________________________________
Phase Shifted Boost-Cuk Converter for Renewable

Energy Applications
Alfredo Medina, Alam D. Salguero, and Taufik Andriani Parastiwi

Electrical Engineering Department Electrical Engineering Department
California Polytechnic State University Polytechnic State of Malang
San Luis Obispo, CA 93407, USA Malang, Indonesia
Abstract—The proliferating need for renewable energy sources number of coupled inductor phases. The increase in current
demands highly efficient conversions to maximize power ripple frequency reduces magnetic component size, which yields
extraction. Renewable energy sources, such as wind and solar an improvement relative to isolated topologies. Coupled
produce low DC voltages that must be converted to high AC inductors demonstrate improvements in topologies which
voltages for transmission. The conversion is a two-step process require two inductors, such as the cuk and SEPIC converters;
requiring a DC/DC converter to increase the low DC voltage to a however, coupled inductors in single inductor topologies, such
high DC voltage and an inverter to convert the high DC voltage to as boost and buck-boost, result in increased magnetic
high AC voltage. This paper focuses on the prior, a step-up DC/DC component size and cost. Furthermore, [4] demonstrates input
converter requiring a high conversion gain, low input current
current increases in high power applications for coupled
ripple, and low output voltage ripple. A novel topology consisting
inductor topologies due to leakage currents. Overall
of a parallel input - series output boost and cuk converter is
proposed to increase conversion gain. Furthermore, a phase shift
performance of coupled inductor technologies offer high voltage
control strategy allows for reduction in input current ripple and gain at reduced component size and voltage stress relative to
output voltage ripple. The results demonstrate the topology is ideal isolated topologies [2].
for multi-level inverter applications, where a split supply is Three level converters are a possible solution to reducing
typically required and must be regulated from a single supply size, weight, and volume of the filter inductor due to doubling
input. of the voltage gain [5]; however, this topology demonstrates
Keywords— Boost converter, cuk converter, dual voltage supply
large switch voltage stress. The multi-level topology presented
in [6] employs a single switch and input inductor to achieve a
high conversion gain. The conversion gain of [6] is greater than
I. INTRODUCTION the topology proposed in [5] with minimized switch voltage
Increasing utilization of renewable energy technologies pose stress. Although benefits are realized by the large voltage gain
issues regarding electronics that govern effective and efficient and minimal switch stress voltage, regulation of the output
conversion. Typically, renewable energies supply low DC voltage is impractical since a single switch cannot
voltage, which must be converted to high DC voltage for simultaneously regulate both output voltage rails. Additionally,
inversion and transmission [1][2][3]. Simple solutions employ the voltage rails design in [6] are inequivalent in magnitude,
isolated DC/DC converter topologies by selecting appropriate which limits the feasibility of use in inverter applications.
transformer turns ratio for high voltage gain; however, these In addition to high voltage gain, step-up DC-DC converters
topologies are less efficient, have bulky transformers, and in PV application must be able to supply multiple regulated
incorporate complex isolated controls [4]. Alternatively, non- output voltages for high performance multi-level inverters. The
isolated topologies exclude transformers; thus, exhibit higher benefits of multi-level inverters include reduced switch voltage
efficiencies with reduced costs. Traditional non-isolated DC stress, reduced output harmonic content, which reduces the size
incrementing topologies such as boost, buck-boost, cuk, Zeta, and cost of filtering components [8]. The drawbacks of the
and SEPIC demonstrate theoretically high gains; however, multi-level inverter topology include increased component
practical applications limit allowable voltage gain, due to duty count and system complexity [8]. Additionally, a balanced
cycle pulse width constraints [5]. The literature demonstrates neutral point voltage is achieved using split capacitors; however,
extensive efforts to achieve high voltage gain, high efficiency, this results in voltage balancing issues [8][9]. A low frequency
and low costs using augmented non-isolated topologies. current ripple is revealed in the analysis of split capacitor voltage
A promising technology for high voltage gain involves balancing in [8]. Moreover, voltage imbalances of even order
implementing coupled inductors [2][4][5]. Coupled inductors harmonics are observed in [9]. Capacitor voltage balancing via
achieve high voltage gain by exploiting the magnetizing current modulation techniques are demonstrated in [10]; however, using
of the primary inductor of most non-isolated DC incrementing modulation to achieve voltage balancing limits the performance
topologies [5]. Coupled inductors result in higher efficiencies of the multi-level inverter, since modulation strategies may be
without the need of isolated controls. Additionally, [2] allocated to provide improved performance. Topologies to
demonstrates inductor current ripple frequency increases by the achieve voltage balancing via DC/DC converters are commonly
____________________________________________________________________________________________________________
isolated [11]; however, as previously explained these topologies The input inductor of both the boost and cuk converter gives
are less efficient and more costly. the advantage of reduced input current ripple. By controlling
To achieve voltage balancing and maintain high conversion the input current ripple phase of each individual converter, it is
ratio, a parallel input series output DC/DC converter composed possible to reduce the total input current ripple from the input
of boost and cuk are proposed. Similar topologies exist such as source; thus, achieving greater efficiencies.
those demonstrated in [6]. The topology proposed in [6] does not
contain equivalent magnitude voltages at the output; thus,
capacitor balancing cannot be achieved. In addition, the single B. Phase Shift Control
switch limits simultaneous regulation of both output voltage The duty cycles of the boost and cuk topologies are offset,
rails. In addition, the proposed topology utilizes phase shifting such as to achieve minimum on-time overlap within one
to effectively reduce input current ripple and output ripple switching period. The cuk converter switch is on at the
voltage. Analysis of the proposed inverter is discussed in this beginning of each switch period and the boost converter switch
paper. is on at the end of each switching period. A sample waveform
of the control signals is demonstrated in Figure 2.
II. DERIVATION OF DUAL OUTPUT CONVERTER
A. Topology Description
The proposed topology is composed of a cuk and boost

converter connected in parallel from a common input source
and series connected outputs. The proposed converter topology
is shown in Figure 1.
Fig 2. Cuk (top) and boost (bottom) phase shift controls
Figure 2 demonstrates the typical controls for the buck and

boost for a 100kHz switching frequency. The effects of phase
shifting are observed for the input supply current ripple and
output voltage ripple.
III. INPUT CURRENT RIPPLE AND OUTPUT VOLTAGE RIPPLE
A. Design Equations
Fig 1. The Boost-Cuk Converter Topology
Equation 4 demonstrates the input current ripple in the cuk
and the boost. This design equation is used to calculate the input
Equation (1) demonstrates the boost converter conversion
inductor needed to achieve the desired input current ripple for
gain, where D is the duty cycle. Notice that for duty cycles less
the boost and Cuk converters.
than 1 the boost converter will step up the input voltage.
𝑉𝑉𝑖𝑖𝑖𝑖 𝐷𝐷𝐷𝐷 (4)
𝑉𝑉𝑜𝑜𝑜𝑜𝑜𝑜 1 (1) ∆𝐼𝐼 =
= 𝐿𝐿
𝑉𝑉𝑖𝑖𝑖𝑖 1 − 𝐷𝐷
Equation 5 shows the output voltage ripple for the cuk
Equation (2) demonstrates the Cuk converter conversion gain, converter. The output capacitor for the cuk converter can be
which exhibits negative polarity relative to the input voltage calculated by selecting a desired voltage ripple.
when the duty cycle is greater than 0.5.
𝑉𝑉𝑖𝑖𝑖𝑖 𝐷𝐷 (5)
𝑉𝑉𝑜𝑜𝑜𝑜𝑜𝑜 𝐷𝐷 (2) ∆𝑉𝑉𝐶𝐶𝐶𝐶𝐶𝐶 =
=− 8𝑓𝑓 2 𝐶𝐶𝐶𝐶
𝑉𝑉𝑖𝑖𝑖𝑖 1 − 𝐷𝐷
Equation 6 demonstrate the output voltage ripple of the boost
Equation (3) demonstrates the overall conversion gain of the converter. The output capacitor for the boost converter can be
Boost-Cuk converter. calculated by selecting the desired voltage ripple.
𝑉𝑉𝑜𝑜𝑜𝑜𝑜𝑜 1 𝐷𝐷𝐶𝐶𝐶𝐶𝐶𝐶 (3) 𝑉𝑉𝑜𝑜𝑜𝑜𝑜𝑜 𝐷𝐷
=� + � (6)
𝑉𝑉𝑖𝑖𝑛𝑛 1 − 𝐷𝐷𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 1 − 𝐷𝐷𝐶𝐶𝐶𝐶𝐶𝐶 ∆𝑉𝑉𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 =
𝑅𝑅𝑅𝑅𝑅𝑅
____________________________________________________________________________________________________________
B. Simulation Parameters voltage ripple. These two parameters are expected to vary with
Simulation and analysis is performed via Orcad Capture and respect to duty cycle; thus, an analysis is performed to obtain
Pspice, respectively. The simulation schematic for the proposed the optimum conversion gain that will result in both minimal
converter is shown in Figure 3. Control of the converter is input current and output voltage ripple.
realized using a pulse waveform voltage source. The switch is C. Simulation Results and Analysis
modeled by Sbreak with an on resistance of 10mΩ. Parasitic
Figure 4 shows the input current waveform for the boost, cuk,
elements such as inductance, capacitance, and resistance were
and the source. Figure 4 demonstrates that when the boost input
neglected for all other components. Table 1 shows the inductor is charging the Cuk input inductor is discharging; thus,
parameters used for simulation of the boost-cuk converter. achieving ripple cancellation via phase shifting. Without phase
TABLE I. PARAMETERS USED IN SIMULATION
shifting, the input current ripple will be the combination of each
converters respective input current ripple, which is 200mA.
Parameter Value
Boost input current ripple 100 mA
Cuk input current ripple 100 mA
Boost output voltage ripple 200 mV
Cuk output voltage ripple 200 mV
Frequency 100 kHz

Fig 4. Boost (top) and Cuk (bottom) input current waveforms
Output current 1A
Input voltage 48 V
Switch ON resistance 10 mΩ
Fig 5. Input current ripple vs. Voltage conversion ratio
Figure 5 shows that as the voltage conversion ratio increases

the total input current ripple increases. Increase in conversion
gain rises duty cycle, which in turn increases the charging time
Fig 3. Simulation schematic for the proposed converter
for the input inductor of the boost and Cuk. As a result, the
charging time of the boost and Cuk overlap resulting in an
The conversion gain magnitude is adjusted from 2 to 9 for increase of the total input current ripple. Total input current
each converter. Note that the conversion gain will be of ripple is minimized for small conversion gains; however, this is
opposite polarity, due to the inverting cuk converter; thus, the ineffective since large conversion gains are required for inverter
conversion gain magnitude of each converter is always applications.
equivalent in order to achieve balanced output voltage rails. The Figure 6 shows the output voltage waveforms for the boost,
output current is kept constant at 1A for each converter to cuk, and the boost-cuk. Similar to the total input current ripple,
minimize; thus, the output power varies as the output voltage the output voltage ripple is also affected by the phase offset
varies, due to adjustment of the conversion gain. The output between the boost and cuk; however, output voltage ripple is
voltage ripple for each converter is designed to 200mV and the minimized when the individual boost and cuk output voltages
input current ripple is designed to 100mA. are in phase. Without phase shifting the maximum total output
To achieve different conversion gains, the duty cycle must be voltage is expected to be the sum of the voltage ripple in each
adjusted; thus, exhibiting a phase shift on both the output converter, which is a total of 400mV.
voltage ripple and input current relative to each converter. Each Figure 7 shows that the minimum total output voltage ripple
converter is designed to maintain an input current ripple of is at voltage conversion ratio of 3.5. Due to the phase shift the
100mA and an output voltage ripple of 200mV; however, phase total output voltage ripple is minimized as the duty cycle of both
shifting will influence total input current ripple and total output converter increases. It is important to mention that the total
____________________________________________________________________________________________________________
output voltage ripple start increasing after the voltage is large enough to achieve a high voltage output, due to a limited
conversion ratio of 3.5. conversion gain.
The results of the total input current ripple and total output Improvements to the proposed topology are realized by
voltage ripple demonstrate an optimum conversion gain of 3.5, achieving greater conversion gain. The single switch topology
since the output voltage ripple is minimal, while maintaining a proposed in [6] utilizes multiplication cells to increase the
reduced total input current ripple. The optimum conversion gain conversion gain. Further developments may be conducted to
demonstrates a reduction in output voltage ripple by 79.4% and utilize the multiplication cells to achieve greater conversion
a reduction of input current ripple of 34%. gain; however, the topology adds a switch and inductor to allow
for regulation of both rail voltages. Other improvements to the
topology include replacing the boost converter with the cuk
converter in non-inverting mode [13]. The non-inverting cuk
topology proposed in [13], further reduces the output voltage
ripple with reduced capacitor sizing. These improvements
demonstrate potential for a high efficiency high power DC/DC
converter suited for PV and inverter applications.
REFERENCES
[1] Ho, C. N, H Breuninger, S. Pettersson, G. Escobar, L. A Serpa, and A
Fig 6. Boost (top), Cuk (middle), and Boost-Cuk (bottom) output voltage Coccia. "Practical Design and Implementation Procedure of an
waveforms Interleaved Boost Converter Using SiC Diodes for PV Applications."
IEEE Transactions on Power Electronics, 27.6 (2012): 2835-2845.
[2] Torrico-Bascope, R. P, L. F Costa, and G. V Torrico-Bascope.
"Generation of New Nonisolated High Voltage Gain DC-DC Converters."
2011 IEEE 33rd International Telecommunications Energy Conference
(INTELEC), (2011): 1-8.
[3] Tomaszuk, A, and A Krupa. "High Efficiency High Step-up DC/DC
Converters - a Review."Bulletin of the Polish Academy of Sciences:
Technical Sciences, 59.4 (2011): 475-483
[4] Muhammad, M., Armstrong, M., and Elgendy. M. A., "A Nonisolated
Interleaved Boost Converter for High-Voltage Gain Applications." IEEE
Journal of Emerging and Selected Topics in Power Electronics, 4.2
(2016): 352-362.
[5] Tofoli, FL, DD Pereira, WJ de Paula, and DD Oliveira. "Survey on Non-
Fig. 7. Output voltage ripple vs. Voltage conversion ratio isolated High-voltage Step-up Dc-dc Topologies Based on the Boost
Converter." Iet Power Electronics, 8.10 (2015): 2044-2057.
[6] Ranjana, M. S. B., Reddy, N. S., and Kumar R. K. P. "Non-isolated Dual
IV. CONCLUSION Output Hybrid DC-DC Multilevel Converter for Photovoltaic
Applications." 2014 2nd International Conference on Devices, Circuits
Investigation of the parallel input series output DC/DC cuk and Systems (ICDCS), (2014): 1-6.
and boost based converter demonstrates improved performance [7] Silveira, C., Luiz, G., Bezerra, S., Torrico-Bascope R.P., and Tofoli, F.L.,
"DC-DC Nonisolated Boost Converter with High Voltage Gain Adequate
using phase shift control. An optimum conversion gain of 3.5 for Split-capacitor Inverter Applications." 2013 Brazilian Power
results in a 34% input current ripple reduction and 79.4% Electronics Conference, (2013): 58-65.
reduction in total output ripple voltage. The diminished input [8] Celanovic, N, and D Boroyevich. "A Comprehensive Study of Neutral-
current ripple yields larger efficiencies from the supply. In point Voltage Balancing Problem in Three-level Neutral-point-clamped
Voltage Source PWM Inverters."IEEE Transactions on Power
addition, the reduced output voltage ripple results in larger Electronics, 15.2 (2000): 242-249.
efficiencies for the load to be drive. [9] Pou, J, D Boroyevich, and R Pindado. "Effects of Imbalances and
The reduced input current ripple is ideal for PV Nonlinear Loads on the Voltage Balance of a Neutral-point-clamped
Inverter." IEEE Trans. on Power Electronics, 20.1 (2005): 123-131.
applications, since AC losses compromise PV efficiency as [10] Gayathri Devi, K. S, S Arun, and C Sreeja. "Comparison of Capacitor
demonstrated in [12]. Moreover, line losses are minimized in Voltage Balancing Techniques in Multilevel Inverters." 2014 Annual
the case of long distance PV transmission. In addition, the International Conference on Emerging Research Areas: Magnetics,
reduced output voltage ripple results in greater efficiencies for Machines and Drives (AICERA/iCMMD), (2014): 1-6.
[11] Kang, J.L, C.W. Roh, G.W. Moon, and M.J. Youn. "Phase-shifted
inverter applications, since ripple voltage may contribute Parallel-input/series-output Dual Convertor for High-power High-output
significant losses when bulk capacitors are used at the input of Voltage Applications", International Journal of Electronics, 89.8 (2002):
the inverter. Although benefits of the proposed converter are 603-624.
promising in PV applications, the limited conversion range [12] El Khateb, Ahmad, Nasrudin Abd Rahim, Jeyraj Selvaraj, and Barry W
Williams. "The Effect of Input Current Ripple on the Photovoltaic Panel
impacts usefulness in inverter applications. Typically high Efficiency." 2013 IEEE Conference on Clean Energy and Technology
conversion gains are required to increases DC/DC converter (CEAT), (2013): 478-481.
efficiency, while supplying high voltages to inverters for high [13] Zhang, Zhe, and S Cuk. "A High Efficiency 500 W Step-up Cuk
voltage transmission; however, a large input voltage may Converter." Proceedings IPEMC 2000. Third International Power
Electronics and Motion Control Conference (IEEE Cat. No.00EX435), 2
achieve a output high voltage. Overall, the results demonstrate (2000): 909-914 vol.2.
utility in both PV and inverter applications if the input voltage
____________________________________________________________________________________________________________
Integration of Distributed Generation in Radial

Distribution Systems
Nuha Nadhiroh1, Ontoseno Penangsang2, Rony Seto Wibowo3
#
Department of Electrical Engineering, Institut Teknologi Sepuluh Nopember
ITS Sukolilo, Surabaya, Indonesia
1nuha.nadhiroh14@mhs.ee.its.ac.id,
2zenno_379@yahoo.com,
3ronyseto@ee.its.ac.id
Abstract— With the growth of distributed generations (DGs) distribution systems considering DG units as PQ or PV nodes.
penetration, load flow analysis of distribution system needs This paper propose the combination of direct-approach
special models and algorithm. DGs are generally modeled as PV technique and bisection method. Test result show that the
or PQ nodes. This paper introduces a three phase unbalaced load proposed method can be used to analyse the penetration of DG
flow method, which is simple and fast and can handle DG units
in both PV and PQ mode. The proposed method combines the
to distribution systems.
direct-approach technique and bisection method. An bisection
approach which can be easily combined with a direct-approach
II. POWER FLOW ALGORITHM
technique which fully exploits network characteristic of The proposed method can be devided into two process.
distribution feeder is presented. Modified IEEE 34-node test First is constructing three phase radial distribution load flow
feeder to validate the proposed method. The different mode and using the direct-approach technique that consist of bus-
penetration level of DG is analyzed and the results are presented. injection to branch-current (BIBC) and branch-current to bus-
voltage (BCBV) matrices. Second is combining the direct-
Keywords— Distributed generation (DG), load flow, radial approach technique with bisection method for handling DG as
distribution system PV node.
I. INTRODUCTION A. First process: Constructing three phase radial distribution

load flow using the direct-approach technique
Distributed generations (DGs) have become an important
part of electrical generation in many countries. The rise in The relationship between the bus current injections and
power demand has led to installation of small power units or branch current can be expressed as BIBC matrix. This matrix
DGs. With the installation of DG within power system, has is an upper triangular matrix that only contain values of 0 and
altered the distribution network from a passive network to an +1. The relationship between branch currents and bus voltages
active one. The alteration impact the overall system losses and can be expressed as BCBV matrix. This matrix is a lower
voltage profile of the system. Distribution networks have triangular matrix that has non-zero entries consisted by line
characteristic such as high resistance-to-reactance (R/X) ratio, impedance values [6].
radial or weakly-mesh topology, large number of branches The building algorithm for BIBC matrix can be described
and nodes and unbalance distributed load. Therefore the as follows:
existing power flow methods might need to be modified. [B]=[BIBC][I] (1)
Its is to be noted that determining a suitable model for each 1. The dimension of BIBC matrix is , where m
DG unit requires knowing the DG operation and the type of its is the number of branch sections and n is the number of
connection to the grid (direct or indirect). Many studies are buses in a distribution system.
performed on distribution power flow with presence of DG [1- 2. If a line section is located between bus and bus ,
3]. Reference [4] used the Gauss implicit Z-matrix method to copy the column of the th bus of BIBC matrix to the
solve the three-phase load flow problem. The Gauss method is column of the th bus and fill to the position of the
traditionally used for the load flow solution of general meshed th line row and the th bus column.
networks; however, distribution systems typically have a 3. Repeat the step two until all line sections are included in
radial or weakly meshed structure. Reference [5] proposed a the BIBC matrix. The relation between branch currents
compensation-based technique, where the forward/backward and bus voltages can be expressed as
sweep algorithm was adopted in the solution scheme. A (2)
direct-approach technique for solving the three-phase
(3)
distribution power flow which fully exploits network
Where [V] and [V0] are the vectors of bus voltages and
characteristic of distribution feeder was proposed in [6].
no-load bus voltages, respectively. The building
In general, the aim of these studies is concentrated in
algorithm for BCBV matrix can be described as follows:
handling an three phase load flow in passive and active
____________________________________________________________________________________________________________
4. The dimension of BCBV matrix is , where

m is the number of branch sections and n is the number
of buses in a distribution system.
5. If a line section is located between bus and bus ,
copy the row of the th bus of BCBV matrix to the row
of the th bus and fill the line impedance (Zij) to the
position of the th bus row and the th line column.
6. Repeat step five until all line sections are included in the
BCBV matrix.
Note that for radial distribution networks, . In
additional, for multiple line sections and buses, the
algorithm can easily be expanded. For example, if the
line section between bus and bus is a three-phase line
section, the corresponding branch current Bl will be a
vector, the in the BIBC matrix becomes a
identity matrix and the Zij in the BCBV matrix is a
impedance matrix as shown
(4)
Where is the impedance between phases x and y with
the ground effect included.
B. Second process: Combining the direct-approach technique
with bisection method for handling DG as PV node
Modelling DG in power flow studies depends on the
control method that used in the converter control circuit. In
case the converter control circuit is designed to control P and
V independently, the DG can be modelled as a PV node and
when the converter control circuit is designed to control P and Fig. 1 Flowchart of reactive power calculation for PV model
Q independently, the DG can be modelled as a PQ node.
The DG units, which are modelled as PQ nodes can be
handled as negative PQ loads without any problem. However,
handling PV nodes requires some additional process. In this
paper, the value of reactive power is calculted using bisection
method.
The bisection method in mathematics is a root-finding
method that repeatedly bisects an interval and then selects a
subinterval in which a root must lie for further processing. It is
a very simple and robust method. Bisection method is slightly
modified to find the value of reactive power in PV bus. The
flowchart of reactive power calculation is shown in Fig. 1.
III. RESULTS AND DISCUSSION

The modified IEEE 34-node test feeder [7] was studied
under different DG modeling and varying penetration. The
single line diagram of the test feeder is shown in Fig. 2. This
study helps to demonstrate how the percentage of the
distributed generation as well as the DG model affect the final
voltage results. This hightlights the variation in results
depending on the DG model.
1. DG is connected to node 29. Considering that node 29
has the lowest bus voltage, due to its topology.
2. DG penetration is defined as follow: Fig. 2 Test feeder: modified IEEE 34-node test feeder
____________________________________________________________________________________________________________
3. Three cases are defined as follow: Case 3: DG was introduced in the modified test feeder
Case 1: the power flow calculation results for modified in node 29 and modeled as PV node. The results are
IEEE 34-node test feeder with no DG integration are shown in Fig. 3. It is seen that the voltage at node 29
shown in Fig. 3. It is seen that the voltage profile is not is maintained at 1 p.u. Also the voltages of all nodes
desireable. have improved
Case 2: DG was introduced in the modified test feeder Fig. 3 shows the voltage profile of above cases with
in node 29 and modelled as PQ node. The results are different DG penetration level (10% and 50%).
shown in Fig. 3. It is seen that the voltage profile is
improved due to the presence of DG.
Fig. 3 Comparison of voltage profile of modified IEEE 34-node test feeder in three cases
proposed algorithm can be used to analyse the effect of DG

IV. CONCLUSIONS models and the impact of DG penetration percentage to the
Modelling DG in power flow studies depends on the voltage profile of distribution systems.
control method that used in the converter control circuit. The
DG shall be a PQ node, in case the P and Q are controlled ACKNOWLEDGMENTS
independently, and the DG shall be a PV node, in case the P The author gratefully acknowledge the support of Indonesia
and V are controlled independently. Moreover, a three-phase Endowment Fund for Education (LPDP) in this research.
distribution load flow program which can handle load flow in
passive network and active network with DG in both PV and REFERENCES
PQ node was proposed. This algorithm that combined the [1] E. M. S.M. Moghaddas-Tafreshi, “Distributed generation modelong for
direct-approach technique and bisection method was tested on power flow studies and a three-phase unbalaced power flow solution
for radial distribution systems considering distributed generation,”
IEEE 34-node test feeder. The test results show that the Electric Power System Research, vol. 79, no. 4, pp. 680-686, 2009. . M.
____________________________________________________________________________________________________________
[2] S. Khushalani, J. M. Solanki and N. N. Schulz, “Development of [5] C. S. Cheng and D. Shirmohammadi, “A three-phase power flow
Three-Phase Unbalanced Power Flow Using PV and PQ Models for method for real-time distribution system analysis,” IEEE Transactions
Distributed Generation and Study of the Impact of DG Models,” IEEE on Power Systems, vol. 10, no. 2, pp. 671 - 679, 1995.
Transactions on Power Systems, vol. 22, no. 3, pp. 1019 - 1025, 2007. [6] J.-H. Teng, “A direct approach for distribution system load flow
[3] J. -H. Teng, “Modelling distributed generations in three-phase solutions,” IEEE Transactions on Power Delivery, vol. 18, no. 3, pp.
distribution load flow,” IET Generation, Transmission & Distribution, 882 - 887, 2003.
vol. 2, no. 3, pp. 330 - 340, 2008. [7] Distribution System Analysis Subcommittee, IEEE 34 Node Feeder,
[4] T. -H. Chen, M. -S. Chen, T. Inoue, P. Kotas and E. A. Chebli, Radial Test feeder. [online] Available:
“ Three-phase cogenerator and transformer models for distribution http://ewh.ieee.org/soc/pes/dsacom/testfeeders/index.html.
system analysis,” IEEE Transactions on Power Delivery, vol. 6, no. 4,
pp. 1671 - 1681, 1991.
____________________________________________________________________________________________________________
Estimation Losses in Radial Distribution Network

Under Unbalanced Loading Condition Considering
Time Varying Load
Yoakim Simamora#1, Suyanto#2, Ontoseno Penangsang*3, Rony Seto Wibowo*4
Electrical Engineering Department, Institute of Technology Sepuluh November
Kampus ITS, Keputih-Sukolilo, Surabaya-60111
1yoakim.simamora14@mhs.ee.its.ac.id
2suyantotf2@gmail.com
3zenno_379yahoo.com
4ronyseto@ee.its.ac.id
Abstract — Distribution system is a part of the electrical power every hour in a certain period, sometimes the demand
distribution that has different parameters and characteristics becomes even bigger in a certain moment.[2]
with transmission system. This is because of the distribution The use of electrical power nowadays becomes wider, either
system is directly connected to the load. Different types of load it is for household purpose, street lighting, business, or
can make the analysis of load flow in distribution become more
industries. With the use of electric power with different
difficult. These differences can create a unique character for
each point of load (distribution substations). Each load point has purposes there are various groups of electric load with
different character according to the type of load. In order to be different characters.[3]
able to know the characteristic of each load point, identification The large number of loads that spread widely will make the
of character of each load that connected with the load point is measurement in certain time becomes, more difficult (real
required. The result of this identification can be used to estimate time condition). Besides the large number, load also has
the load at a certain time and also can make a distinctive different types, either from the characteristic of the load or
characteristic curve for each load point. This paper uses three time usage that cause each load has different load curve
phase balanced and unbalanced 20 KV distribution system in characteristic, so it would be more difficult for calculating
Medan. To estimate the losses, K-Matrix and ZBR method load
energy in a network
flows are used to obtain losses in each interval of 24 time
intervals considering time varying of the load. Then, to reduce This paper identifies the characteristic of load types on a
the power loss, network reconfiguration is used by applying the distribution network, to estimate the distribution transformer
method of Binary Particle Swarm Optimization (BPSO) without loading. The estimation results can used to calculate losses of
changing the number of lines. The simulation result shows that a distribution network using power flow methods, and there
reconfiguration with BPSO method is able to minimize the active are methods to minimize losses of the network.
power loss in interval 3, namely 19.5% (80.5 % reduction of There are methods that generally used in power flow
losses) and the average losses reduction in all 24 time interval is calculation, in this paper K-matrix and Zbr methods are used
32.85 %. by forming an impedance matrix obtained by tracing the path
of distribution channel. These methods use few formulas with
Keywords — load flow analysis, loading estimation, loss of
distribution network simple calculation so that can reach the convergent point in a
short and fast iteration time.[4]
I. INTRODUCTION The proposed method to minimize the losses is distribution
Distribution system is one part of the electrical power network reconfiguration by applying Binary Particle Swarm
system that has certain characteristics. Different from Optimization (BPSO) method in one feeder. Reconfiguration
transmission line, the distribution line uses medium voltage is carried out by attempting several new bus connections to
level only so that voltage drop is relatively large, and is establish new configuration itself in the distribution network,
generally radial. [1] without changing the position of distribution transformers, the
Distribution network is complicated with many branches number of lines, and the structure of radial networks.
and types of loads that generally connected in to a feeder.
Each load has different load curve characteristic. A II. LOAD AND ITS CHARACTERISTICS
distribution transformer is loaded with different types of loads,
so that when a measurement is performed to a distribution Distribution loads can be grouped into a few categories,
transformer at a certain time it will show a unique load curve they are residential, commercial, public and industrial load.
characteristic on each distribution transformer. Load that used The identification of characteristic started from
by a consumer creates a varying electrical power demand identifying the type of load in each distribution transformer.
____________________________________________________________________________________________________________
Some transformers are used to supply a certain type of load. where :

But, there is a possibility that some distribution transformers Pnref_res’ = Power reference of residential load in 1–24 time
used to supply more than one type of load. For example, the intervals
residential and commercial load. Pref_res = Power reference of residential load from
measurement
A. Measurement Data and Reference Data pfnref_res = Power factor reference of residential load in 1–24
The data used in this paper are collected by PT. Hagler time intervals (%)
Bailly Indonesia, and also by PT. PLN Persero. Measurement kVArating = Transformer rating
data of distribution substation are also used as preliminary lf = Percentage of transformer loading (%)
data to do the identification. These data include load lcres = Composition of Residential Load (%)
measurement; it is also supported by data on nameplate n = 1, 2, 3, … , 24
transformers and time when taking the measurements.
Information about these measurements are very important to Commercial, public, and industrial loading estimations are
be used as a reference of base load measurement. calculated by using the above equation. The difference is only
Consumer data are used to determine the percentage of on reference power, reference power factor, and composition
each load on each transformer because there is a possibility of loading that indicates the type of load to be estimated.[5]
that a transformer is used to supply a few types of load.
Channel data and bus data are used to analyze the power flow C. Loading Distribution Transformer
and estimate the losses in the next process. In reality transformer can be loaded until a few percent
The survey results from PT. Hagler Bailly Indonesia are above its capacity, but its only apply for a few moments. The
daily load curves of loads. These curves include curves on transformer can work optimally if the loading condition is
weekdays and curves on weekends. Fig. 1 is a residential daily about 80% of its capacity.
TABLE I
load curve at low voltage side on weekdays.[5] Standard limit over loading on distribution transformers
Next, loading value from this curve in every hour is used as
reference database in doing the identification. This database Load
% limit and overload length of time (hours)
will determine curve profile and load volume in a distribution 2 4 6 8 10 12
Factor
hours hours hours hours hours hours
system load. The curve that used in this paper is restricted, 0.6 23 20 17 14 10 6
they are the daily load curve of low voltage side on weekdays 0.7 17.5 15 12.5 10 7.5 5
and weekends and constant power factor of each load. There 0.75 14 12 10 8 6 4
are several plants at PT. PLN (Persero) Pancur Batu, Medan 0.8 11.5 10 8.5 7 5.5 3
0.85 8 7 6 4.5 3 2
Tuntungan. One of them is the plant NR7 which is used in this 0.9 4 3 2 - - -
paper.
D. Total Daily Load Curve

After the daily load curve in each distribution transformer
has been obtained , then the total daily load curve on a feeder
is obtained by summing the loads of all transformers in the
same time interval (range between 1 to 24 ). It can be
expressed by equation (2).
jum _ trafo
Fig. 1 Daily curve of low voltage side weekday residential area of Medan
 Pn m (MW) (2)
Pntotal_ penyulang  m 1
1000
B. Estimation Daily Load Curve Process
The estimation process of total load curve on a feeder,
started by determining daily load curve of each distribution
transformer. Load curve of each transformer can be estimated
based on measurement data, the percentage of customer and
data base of the daily load from each load. Load estimation for
each residential transformer is expressed by the following
equation (1).
 Pnref _ res ' 
Pnres   .(kVArating.lf  lc res  pfnref _ res ) (1)
 P 
 ref _ res  Fig . 2 Total Daily Load Curve of Feeder NR7
____________________________________________________________________________________________________________
Figure 2 (Total daily load curve of feeder NR7) is similar III. SIMULATION AND ANALYSIS
to the residential load characteristic curve, because the
majority of load on the distribution transformers along feeder A. Power Loss Calculation
NR7 are residential loads. From the estimated loading of each The result of load estimation from each distribution
load point, it can be estimated power loss in each interval of transformer in a certain time interval is used in this calculation.
time. The next step to reduce power losses in the feeder NR7. Simulation and analysis in this paper use two scenarios, they
The method used to reduce losses is feeder reconfiguration are the balanced loading and unbalanced loading by 20%
using Binary Particle Swarm Optimization Algorithm. between phase.
Calculation of power flow is done by calculating the value
of voltages in each bus and network losses. Figure 3 shows
E. Distribution Network Reconfiguration For Loss Reduction the comparison of initial losses in balance condition and
by Binary Particle Swarm Optimization unbalance condition using BIBC-BCBV load flow method.
Distribution network reconfiguration is done in order to
reduce losses, which can be done by changing the status of tie
switches, or by changing the topology structure of the existing
network.
A feeder of distribution network generally has a radial
network structure. In radial distribution network, total losses
of one feeder is calculated as follows,
= (3)
where Ploss(i,i+1) expresses one part of line losses between bus

i and i+1, and is expressed as :
= (4) Fig. 3 Comparison Curve of Losses in Balance and Unbalance Condition
and constraint functions : B. Loss Reduction

Distribution network reconfiguration is done by changing
≤ ≤ (5) the status of tie switches, or by changing the topology
structure of the network to reduce power losses, load
≤ ≤ (6) balancing, and stabilize the voltage for normal operating
condition. Reconfiguration is done by considering the
PSO standard is designed to solve continues function operation constraints such as operating voltage limits,
optimization problem. It is not used to solve discrete function maximum current flowing through the lines and still maintain
optimization [6]. That is why, Kennedy and Eberhart propose a network structure in normal operating conditions.
modification version of PSO called Binary Particle Swarm Reconfiguration process is done by adding 14 new lines.
Optimization (BPSO) that can be used to solve problems of Considerations that are used in the selection of a new line is
discrete function optimization. [7] the line must have a pathway that connects buses adjacent
In Binary Particle Swarm Optimization (BPSO) Pi and Pg of with the path that is not above the housing. Figure 5 is a
a swarm is updated in PSO standard. The main difference single line diagram of the feeder NR7 with 14 new lines.
between PSO standard and BPSO is in the interpretation of
velocity. In BPSO, the velocity is limited on the range [0,1].
The similarity of velocity is :
1 (7)
vij' (t )  sig ((t ))   v (t )
1  e ij
Equation (7) is used to renew particle velocity vector. And,

the new particle position is obtained by using the similarity
(8).
 1 if rij  sig (vij (t  1))
xij (t  1)  
0 otherwise (8)
____________________________________________________________________________________________________________
1
2 Substation
84 3
Bus
4
5 6 Sectionalizing
7
Switch
9
8 Tie Switch
10 85
11
12
14
13
17 15 16
86
19
18
20
21
22
87 Fig. 7 Comparison of Minimum Voltage initial configuration &
23
reconfiguration using BPSO
24
26 25 27
IV. CONCLUSION
29
28 The estimation results show that the characteristic of load
30
curve on feeder NR7 is similar to the residential load curve.
88
31
89 Power loss of unbalance condition is greater than that of
32
33
balance condition. Network reconfiguration on feeder NR7
35 using BPSO Algorithm can reduce losses and improve the
34
36 voltage profile in each time interval. The simulation results
37
38
40 41 42 43 44
show that the greatest reduction of losses is in the third
39
interval (80.5% reduction). The average reduction of losses
45
46
for all 24 intervals is 32.85 %.
90
47 48 49
59
51
52
91
References
53
56
[1] Civanlar, S., Grainger, J.J., Yin H., dan Lee, S.S.H., (1988),
55 “Distribution Feeder Reconfiguration for Loss Reduction”, IEEE Trans.
54
57 Power Del., Vol.3, No.3, hal.1217-1223
58
92
[2] L. W. Manning, “Electric Utility Engineering Reference
59
61 Book Volume 3, Distribution Systems”, inWestinghouse
60
93
Electric Company, Pennsylvania, 1965.
62
94 95 [3] Mulyo Darminto, “Ramalan Beban Jangka Pendek Untuk
63 74
96 97 Pengoperasian Sistem Interkonekasi Se-Jawa Area IV”, Tugas Akhir
64 66
73
Teknik Elektro-ITS, 1987
65
[4] G. W. Chang, S. Y. Chu, ” An Improved Backward/Forward Sweep
72
67
80 82 84
Load Flow Algorithm for Radial Distribution Systems”, IEEE
68 Transactions On Power Systems, Vol. 22, No. 2, May 2007
[5] M. Zaini Rohmawan, “Identifikasi Karakteristik Jenis Beban Pada
69 70
Jaringan Distribusi Radial Untuk Estimasi Pembebnanan
Transformator Distribusi”, Tugas Akhir Teknik Elektro-ITS, Juli 2006.
Fig. 5 Single Line Diagram of Feeder NR7 [6] J. Kennedy and R. Eberhart, “Particle Swarm Optimization”, In IEEE
Int. Conf on Neural Networks, Perth, Australia, 1942-1948, 1995
Figure 6 and Figure 7 are a comparison of losses and [7] Xiaoling Jin., Jianguo Zhao., Ying Sun., Kejun Li., Bqin Zhang.,
“Distribution Network Reconfiguration for Load Balancing Using
minimum voltage in the initial configuration and Binary Particle Swarm Optimization”, International Conference on
reconfiguration using BPSO. Power System Technology-POWERCON, November 2004
. [8] T.-H. Chen, N.-C.Yang, “Three-phase power-flow by direct Zbr
method for unbalanced radial distribution systems”, IET
Gener.Transm.Distrib., 2009,Vol.3, Iss.10,pp.903-910
[9] Turan Gonen, “Electric Power Distribution System Engineering”,
McGrow-Hill Book Company, 1987.
[10] Saadat, Hadi. “Power System Analysis (Second Edition)”, McGraw-
Hill Education (Asia). Singapore, 2004.
[11] redusction in Three Phase Distribution System Under Unbalanced
Loading Condition”. IEEE Confrence Aug. 31 2010-Sept. 3 2010.
[12] G. W. Chang, S. Y. Chu, ” An Improved Backward/Forward Sweep
Load Flow Algorithm for Radial Distribution Systems”, IEEE
Transactions On Power Systems, Vol. 22, No. 2, May 2007
Fig. 6 Comparison of Losses initial configuration & reconfiguration using

BPSO
____________________________________________________________________________________________________________
Surface Characterization of Glass Substrate Exposed

by Atmospheric Pressure Plasma Needle
R. R. Elfa#1, M. K. Ahmad#, C. F. Soon#
M. Z. Sahdan*, and N. Nafarizal*2
Faculty of Electrical and Electronic Engineering,

#
Universiti Tun Hussein Onn Malaysia (UTHM)

Batu Pahat, 86400 Johor, Malaysia
*Microelectronic and Nanotechnology- Shamsuddin Research Centre (MiNT-SRC)

Batu Pahat, 86400 Johor, Malaysia
1 elfa.rizan.r@gmail.com
2nafa@uthm.edu.my
Abstract— Atmospheric pressure plasma needle using argon gas treatment is on the substrate sample to avoid contamination of
discharge and inspired from dielectric barrier discharge is the sample before undergoing another process and the
presented in this paper. The plasma needle is applied for the alteration of substrate sample to have another surface property
surface modification application. A microscope glass is used as can lead to another invention due to different surface property
testing substrate based on huge glass application in various
[9],[10]. The previous study on five different type of polymers
industries. Low adhesion and highly liquid spread rate are
common issues in industrial application. Hence to provide the have shown that there is an increasing on surface adhesion and
solution, atmospheric pressure plasma needle treatment is adhesive bonding on the polymers surface after treatment [11].
applied to the glass substrate with the different period of Meanwhile, the surface topography of stainless steel shows
treatment. Then the glass surface is characterized using atomic only slightly decreased in the roughness of the surface [12].
force microscope (AFM) and water contact angle measurement. However, there is still an urge of demand to have a deeper
From the AFM result the surface roughness become at the range understanding of this surface treatment by APPN. In the
of 3.5 nm when exposed 30-450 s with plasma needle. In addition, application of semiconductor fabrication the research on this
the water contact angle on glass has superhydrophilic property aspect still narrow. Therefore, in this study is starting from the
at 7.26° when exposed for 450 s, indicating that the surface have
development of custom made APPN investigate the surface
low surface adhesion due high rate of water spreading.
property of the pre-treated sample that is microscope glass.
Keywords— atmospheric pressure plasma jet, surface
The analysis of surface property of the pre-treated sample
modification, surface topography, atomic force microscope
measured by atomic force microscope (AFM) and water
analysis
contact angle measurement to give a better understanding of
I. INTRODUCTION surface topography and surface characteristic of the sample
after expose to the plasma treatment.
From the invention of dielectric barrier discharge,
atmospheric pressure plasma needle (APPN) have become a II. EXPERIMENTAL
hot topic to be discussed because of its wide application in
engineering and medical field [1]-[6]. This is due to APPN The development of the atmospheric pressure plasma jet is
characteristics that able to create plasma discharge at set up at open space with the rectangular board as simple
atmospheric pressure. Whereby, the drawback of low pressure confinement of the equipment. The APPN is placed in upside
plasma have been overcome by this technique as the system position so that the gas will flow in the turbulent direction as
do not need any special chamber, rotary pump and etc. Even shown in Fig. 1. The high purity Argon (Ar) gas as the
though the APPN is operating at atmospheric pressure, it still reactive gas is directly attached to the input of the tube glass
can give an output of the plasma discharge that contains high of the APPN with continuously flowing at 50 l/min
energetic reactive gasses that able to modify and clean the pre- throughout the experiment. The input of the gas is controlled
treated surface [7]. by the manual mass flow controller. The tube glass or body of
The advantage of the APPN to clean the surface have the APPN act as the dielectric material and the copper wire
attracted researcher to dig in more information to identify the that is inserted into the tube act as powered electrode. The
limitation and advance effect of the pre-treated sample. In the diameter of the inner tube is 3 mm and the outer diameter is 5
medical field, the study of the application of the APPN is mm. In addition, the bottom brass plate is connected to a
more focus on the modifying cancer cell and remove ground of the circuit. In order to have high frequency and high
microorganism [8]. Meanwhile, in engineering the APPN voltage power supply, the input signal from function generator
____________________________________________________________________________________________________________
is mix with voltage booster using digital mixer amplifier that and the test is conducted right after plasma expose for each of
has the output power of 50 watts. The distance from the period treatment. The comparison of before and after
substrate to exit nozzle is kept on constant at 5 mm. The treatment shows that the roughness of the glass is increasing
output discharge of the plasma plume is approximately 15 mm for after treatment sample as shown in Table 1. Before
as shown in Fig. 2. The typical cleaning method of the treatment, the roughness of the glass is 1.56 nm for root mean
substrate is skipped for this experiment to prove that without square roughness (Rq) mean square of the sample. However,
cleaning process the APPN able to modify the substrate to it seems that the roughness is not very significant for each of
have low adhesion surface. The treatment time is varied for 30 treatment period. For 30 s, 60 s, 150 s and 300 s treatment
s, 60 s, 150 s, 300 s and 450 s. The treatment process is done time, the roughness is 3.69 nm, 3.68 nm, 3.54 nm and 3.58 nm
at ambient temperature. Then, the analysis for the surface respectively. The root mean square roughness of 450 s is 3.32
topography of sample is taken from AFM (XE-100, Park nm. In contrary, the water contact angle shows a linearly
systems, Korea). Meanwhile, the water contact angle decreased degree in water contact angle measurement with an
measurement is taken from contact angle analyzer (VCA increased time of treatment as shown in Fig. 3. Whereby, the
optima, USA). optimum time of treatment that is 450 s shows incredibly
result in the range of superhydrophilic surface.
TABLE I
SURFACE ROUGHNESS OF THE GLASS AFTER TREATMENT AND EACH PERIOD OF
TREATMENT.
Glass substrate Rq (nm) WCA (°)

Before treatment 1.56 36.3
30 s treatment 3.69 27.57
150 s treatment 3.54 19.37
300 s treatment 3.58 15.73
Fig. 1 Experimental set up of needle plasma jet using argon gas.
Fig. 3 The root mean square roughness and water contact angle of sample
exposed by various treatment times.
AFM results of the three-dimensional surface of before

treatment, 30 s, 60 s, 150 s, 300 s and 450 s of treatment
images are shown in Fig. 4. The topography images of sample
Fig. 2 The image and length of the plasma plume. before the plasma treatment shows that the surface of glass
has several bump-like features as shown in Fig. 4(a). Then, as
III. RESULT AND DISCUSSION the treatment time increase the grown of bump-like features is
The surface of the glass before treatment and each of increased. As shown by Fig. 4(b), 4(c) and 4(d), the sample
treatment time is analyzed with AFM in order to compare the for 30 s, 60 s and 150 s of sample show grow of bump-like but
surface morphology and roughness of those treated surface. not uniform. Meanwhile, for 300 s and 450 s sample shows
The scanning area of the AFM’s cantilever tip is 3μm x 3 μm uniform growth bump-like features as shown in Fig. 4(e) and
____________________________________________________________________________________________________________
4(f), respectively. This result suggest the optimum period of Thus, it will give out smoother bump-like feature that will
treatment that able to treat the surface area evenly. The growth enhance the spreading of water droplets once the analysis
of the bump-like features is conjunction with highly plasma water contact angle measurement is done. However, the
discharge and etching process took place leads to cleaning the roughness of each treatment time shows no significant
contamination on the surface [13]. The grown of the bump- difference.
like features will increase the surface area of the particle [14].
(a) (b)
(c) (d)
(e) (f)
Fig. 4 Three-dimensional surface topography images of glass substrate before treatment (a), 30 s treatment (b), 60 s treatment (c), 300 s treatment (d) and 450 s
treatment (e)
____________________________________________________________________________________________________________
[13] S-J. Park and H-Y. Lee, “Effect of Atmospheric-Pressure Plasma on

IV. CONCLUSIONS Adhesion Characteristics of Polyimide Film”, Elsevier Inc,, pp.267-272,
May 2004
The application of surface modification using atmospheric [14] K. K. Samantha, M. Jassal and A. K. Agrawal, “Improvement in water
pressure plasma needle is successfully proved by AFM and and oil absorbency of textile substrate by atmospheric pressure cold
contact angle measurements. The results that show an increase plasma treatment”, Elsevier, Surf. and Coat Tech. 203, pp. 1336-1342,
Oct. 2009
of clean surface particles with exposure time of treatment.
Uniform surface morphology is shown for 450 s treatment as
optimum time for plasma treatment for surface modification.
The contact angle result of 450 s sample of treatment has
superhydrophilic surface. With an increased of clean surface
area, it resulted in the smoother surface of the sample. Thus,
this enhanced the liquid spread rate that gives an advantage in
surface coatings, paintings and microelectronic fabrication.
ACKNOWLEDGMENT
The facility of this study was supported by MiNT-SRC,
Universiti Tun Hussein Onn Malaysia. This work was
partially supported by Fundamental Research Grant Scheme
FRGS/UTHM/1207 and Exploratory Research Grant Scheme
ERGS E025 of Ministry of Education Malaysia.
REFERENCES
[1] J. P. Fernández-Blázquez, D. Fell, E. Bonaccurso and A. D. Campo
“Superhydrophilic and superhydrophobic nanostructured surfaces via
plasma treatment,” Journ.Of Coll. and Inter. Scien. 357, pp. 234-238,
Feb. 2011
[2] Z. Fang, J. Yang, Y. Liu, T. Shao and C. Zhang, “Surface treatment of
polyetylene to terephthalate to improving hydrophilicity using
atmospheric pressure plasma jet,” IEEE Trans. on Plasm. Scien 41, no.
6, pp. 1627-1634, June 2013
[3] X-J. Shao, G-J. Zhang, J-Y. Zhan and G-M. Xu, “Research on surface
modification of argon polytetrafluoroethylene coupled with argon
dielectric barrier discharge plasma jet characteristics,” IEEE Trans. on
Plasm. Scien.. 39, no. 11, pp. 3095-3102, Nov. 2011
[4] M-H. Chiang, K-C. Liao, I-M. Lin, C-C. Lu, H-Y. Huang, C-L. Kuo
and J-S. Wu, “Modification of hydrophilic property of polypropylene
films by a parallel–plate nitrogen-based dielectric barrier discharge
jet,” IEEE Trans. on Plasm. Scien. 38, no. 6, pp. 1489-1498, June 2010
[5] Y. Ren, C. Wang and Y. Qiu, “Aging of surface properties of ultra high
modulus polyethylene fibeers treated with He/O2 atmospheric pressure
plasma jet”, Elsevier, Surf. and coat. tech. 202, pp. 2670-2676, Sept.
2008
[6] K. Koch and W. Barthlott “Superhydrophobic and superhydrophilic
plant surfaces: an inspiration for biomimetic materials,” Phil. Trans. R.
Soc. A 367, pp. 1487-1509, May 2009.
[7] L. Zhang, N. Zhao and J. Xu, “Fabrication and application of
superhydrophilic surfaces: a review,” Adh. Scien. and Tech., pp. 769-
790, 2014
[8] L. F. Gaunt, C. B. Beggs and G. E. Georghiou, “Bacterial action of the
reactive species produced by gas-discharge nonthermal plasma at
atmospheric pressure: A review”, IEEE Trans. 34, pp 1257-1269, Aug.
2006
[9] E. Ueda and P. A. Levkin, “Emerging applications of superhydrophilic-
superhydrophobic micropatterns,” Adv. Mater. 25, pp. 1234-1247,
2013
[10] E. Celia, T. Darmanin, E. T. D. Givenchy, S. Amigoni, F. Guittard,
“Recent advances in designing superhydrophobic surfaces,” Elsevier,
Journ. of Coll. And Inter. Scien., 402, pp. 1-18, July 2013
[11] M. Noeske, J. Degenhardt, S. Strudthoff and U. Lommatzsch, “Plasma
Jet Treatment of Five Polymers at Atmospheric Pressure: Surface
Modifications and The Relevance For Adhesion”, Elsevier, pp. 171-
177, Sept. 2004
[12] M. C. Kim, D. K. Song, H. S. Shin, S-H. Baeng, G. S. Kim, J-H. Boo, J.
G. Han and S. H. Yang, “Surface Modification for Hydrophilic
Property of Stainless Steel Treated by Atmospheric-Pressure Plasma
Jet”, Elsevier Science, pp. 312-316, 2003
____________________________________________________________________________________________________________
Investigation of Gas Temperature Evaluation in

Magnetron Sputtering Plasma using Two Different
Resolution of Spectroscopy
Soo Ren How#1, Nafarizal Nayan*2, Jais Lias*3
#
Microelectronic and Nanotechnology- Shamsuddin Research Centre (MiNT-SRC) &
Faculty of Electrical and Electronic Engineering
Batu Pahat,86400 Johor, Malaysia
1lenz5351@gmail.com.my
2nafa@uthm.edu.my
3jaisl@uthm.edu.my
Abstract— For this experiment, a high-pressure magnetron lines of various species especially to investigate the ion
sputtering plasma was used for plasma discharge with various excitation mechanism in the plasma. The manipulating
flow rates of nitrogen gas. The introduction of nitrogen gas parameters of discharge power and flow rate of nitrogen gas
modifies the discharge leading to modifications of plasma have been studied [7].
parameters and ionization mechanism of transition species.
In previous studies, some of the researchers have reported
Optical emission spectroscopy (OES) was employed as non-
invasive plasma diagnostics method and detect the light emission the application of OES usually collaborated with the low
during plasma discharge. Furthermore, OES results of nitrogen pressure magnetron sputtering plasma [7]. In this report, we
spectrum were used for the calculation of gas rotational investigated the nitrogen gas rotational temperature using a
temperature by using a simple formula and it was done for simple calculation formula in high pressure (approximately
various discharge powers and various nitrogen flow rates. A 100 mTorr) magnetron sputtering plasma. In the same time,
comparative of two different spectrometers resolution have been we investigate the effect of using low and high resolution of
used for the determination of gas temperature. The results spectrometers when calculate the rotational temperature.
showed that the evaluated temperatures were almost the same as
expected in high-pressure magnetron sputtering plasmas even
when using better resolution of spectrometer. II. EXPERIMENTAL SETUP
For this experiment, OES measurement was carried out as
Keywords—magnetron sputtering plasma, optical emission
plasma diagnostics tool in the high-pressure magnetron
spectroscopy, rotational temperature, spectrometer
sputtering plasma. Fig. 1 show the schematic representation of
OES configuration and structure of the magnetron sputtering
I. INTRODUCTION
device with the turbo molecular pump and rotary pump.
Plasma diagnostic is very important to understand the Optical measurements were carried out in two different type
fundamental of plasma processing in plasma engineering. of spectrometers with different resolution. Fig. 2 (a) shows the
Basically study of plasmas processing are concerned with Ocean Optics HR4000 diagram with a low resolution
determining the types of neutrals and ionic species, their properties (0.27 nm) while Fig. 2 (b) shows the Princeton
densities, and the electron-energy distribution functions Instrument diagram with high resolution properties (0.017 nm).
(EEDFs) which can sometimes be characterized by an electron The plasma chambers consists of six viewports which has the
temperature [1], [2]. In fact, there are several of plasma diameter of 16 cm and a height of 18 cm. The magnetron
diagnostics technique have been discovered for many years. sputtering source was made of cylindrical permanent magnets
Optical emission spectroscopy (OES) is one of the plasma which attached to an indirect water cooling system. The
diagnostics tool been discussed among the scientific sputtering plasma was produced by a 13.56 MHz RF
community. Moreover, OES measurement has been magnetron sputtering discharges with automated matching
recognized as one of the suitable plasma diagnostics network. Solid pure titanium target with purity of 99.99 %
techniques due to its simplicity and low cost. Subsequently, was used as sputter target. The plasma chamber was evacuated
benefit of using OES diagnostics able to acquire the data to a base pressure of 10-6 mTorr using the turbo molecular
without perturbing the plasma during the diagnostics process pump (TMP) and rotary pump. Argon, Ar gas used as carrier
[3], [4], [5], [6]. gas whereas nitrogen, N2 gas used as reactive gas in the
This method basically used in the laboratory to describe plasma chamber. Those gases were injected into the chamber
and understand the specific phenomena occurring in various and controlled by using mass flow controller (MFC) where it
plasmas. Besides that, it also able to separate the emission
____________________________________________________________________________________________________________
attached at the top of the chamber. The flow rate of argon gas (a) 14.8 cm
was keep constant at 36 sccm while nitrogen gas were varied
to 10, 20 and 40 sccm throughout the experiment. The rf
discharge power also been varied between 100, 150 and 200W
at different flow rate of nitrogen gas.
The light emitted from the plasma discharge passed
through a quartz window chamber. A quartz optical lens was 10.5 cm
used and placed in front of the window chamber so that the
emission will be more focused from the vacuum chamber
during plasma discharge. The detector with high sensitivity
used to trace the light emission and interface with a
spectrometer via the flexible optical fiber. The spectrometer
able to respond to the wavelength within the range of 200 to
1100 nm. A specific software package was used and interface
with the spectrometer to the computer when acquired the data
during the plasma discharge. The summary of the range
process variable for this study is shown in Table I.
(b) 52.8 cm
TABLE I: The Range Process Variable Varying During Plasma Discharge 27.5 cm
Process variable Range
Solid metal target Titanium (Ti)
Reactive gas Ar and N2
Base pressure (mTorr) 3.5 x 10-6
Vacuum pressure (mTorr) 100
Working discharge power (W) 100, 150, 200
Ar gas flow rate (sccm) 36
N2 gas flow rate (sccm) 10, 20, 40
Fig. 2 The diagram spectrometer of (a) Ocean Optics HR4000 and (b)
Princeton Instrument
III. RESULTS AND DISCUSSIONS

Fig.3 (a) and (b) show the emission spectrum collected
using Ocean Optics HR4000 and Princeton Instrument
spectrometer within the wavelength range of 750 to 777 nm,
respectively. By comparing the Fig. 3 (a) and (b), we can
clearly see the difference between the spectrum that collected
by low resolution and high resolution spectrometers. In Fig. 3
(a), there are few stronger peak spectrum that able been
detected by using the low resolution spectrometer. In contrast,
high resolution spectrometer has ability to detect not only the
prominently peak spectrum as well as weak peak spectrum
throughout the plasma diagnostics process as shown in Fig. 3
(b). As a results, more peak spectrum either weak or strong
could be detected by using high resolution spectrometer.
Fig.1 Configuration of optical emission spectroscopy measurement in high- A spectral region between 750 to 777 nm was utilised and
pressure magnetron sputtering plasma system reveal the important information related to the calculation of
gas rotational temperature. The method of determination gas
rotational temperature in this study is refer to the N2 with
14.8 cm 775.3 nm rotational sub-band of the first positive system of
nitrogen emission spectrum is shown in Fig. 3.
____________________________________________________________________________________________________________
195
(a) Trot [ K ]  (1)
( P1/ P2)  0.52
P2
P1 Fig. 4 show the rotational temperature with the variation
of flow rate of nitrogen gas for 100W, 150W and 200W
respectively. These measurements of rotational temperature
was done under two different resolution of spectrometer. Fig.
4(a) and 4(b) show the line graph of rotational temperature
versus flow rate of N2 gas that was plotted under the condition
of low resolution and high resolution spectrometer,
respectively. From these graph, we could seen that no much
fluatuation occur as the flow rate of N2 increased even though
using better resolution spectrometer. Therefore, the similar
trend for both lines graph were obtained. In other words, by
using better resolution spectrometer to detect these region, it
did not significantly give much different compared to low
resolution spectrometer. In addition, it only very useful when
P2 P1
detect the spectrum that contain of weak peak spectrum in
(b)
certain region which low resolution spectrometer unable to do
so.
(a)
Fig. 3 The spectrum of rotational 775.3 nm sub-band of nitrogen for rotational

temperature evaluation using (a) Ocean Optics HR4000 and (b) Princeton
Instrument spectrometer
Gas rotational temperature measurements are very

important for molecular gases discharges because the
translational temperature inside a plasma and also known as (b)
gas temperature which can be obtained by measuring
rotational temperature, Trot. In previous works, determination
of rotational temperature of N2 can be done by using any
rotational band of the emission spectrum of nitrogen or
another molecular gas [8]. For instance, using either the band
from the first negative system at 391.4 nm or second positive
system bands at 337.4 or 380.4 nm with better resolution
spectrometers.The spectrum calculation procedure for definite
molecular gas is based on the basics principles of optical
emission and requires knowledge of the electronic, vibrational
and rotational constants corresponding to the part of the
emission spectrum under consideration. As discussed before,
data of the N2 spectrum was taken around at 775 nm as shown
in Fig.3 (a) or (b). The rotational temperature can be
calculated using formula after some correlation of data has Fig. 4 The rotational temperature plotted with the variation of flow rate of N 2
gas for 100W, 150W and 200W using (a) Ocean Optics HR4000 and (b)
been done. The formula that proposed by N.Britun et al for his Princeton Instrument specrometer.
previous works [8] and is given by
____________________________________________________________________________________________________________
IV. CONCLUSIONS
In conclusion, plasma diagnostic is a vital instrument to
understand the fundamental of plasma processing in plasma
engineering. In plasma processing technology, optical
diagnostics methods have been implemented by most of the
people either in the applied area or basic science area. In this
experiment, the optical emission have been detected and
performed by different resolution of spectrometer. Variation
of flow rate of N2 gas was used for 100W, 150W and 200W
using low resolution and high resolution spectrometer. The
nitrogen gas temperature evaluated using both spectrometers
were almost the same at 450 K. The influence of discharge
power and nitrogen flow rate is almost negligible. Therefore,
we concluded that using better resolution spectrometer did not
give significantly changes when measuring gas rotational
temperature in certain region.
ACKNOWLEDGMENT
The present work was partially supported by Fundamental
Research Grant Scheme FRGS/UTHM/1207 and Exploratory
Research Grant Scheme ERGS E025 of Ministry of Higher
Education Malaysia.
REFERENCES
[1] R. Pintaske, Th. Welzel, M.Schaller, N.Kahl, J. Hahn, F.Richter,
“Spectroscopic studies of a magnetron sputtering discharge for boron
nitride deposition”, Surf. and Coatings Tech., vol 99, pp.266-273, Sept.
1998.
[2] S.P. Koirala, I. Awaah, S.L. Burkett and M. H. Gordon, “Correlation
of plasma characteristics to etch rate and via sidewall angle in a deep
reactive ion etch system using Langmuir probe and optical emission
spectroscopy”, Jour. of Vac. Sci. and Technol, vol. 29, pp. 1-6, Jan.
2011.
[3] A.J.M. Mackus, S.B.S. Heil, E. Langereis, H.C.M. Knoops, M.C.M.
van de Sanden, and W.M.M. Kessels, “Optical emission spectroscopy
as tool for studying, optimizing and monitoring plasma-assited atomic
layer deposition processes”, Jour. of Vac. Sci. & Technology, vol. 28,
pp. 77-87, Dec. 2009.
[4] M K Sharma and B K Saikia, “Discharge conditions and emission
spectroscopy of N2 and N2+ active species in a variable power dc
pulsed plasma used for steel nitriding”, Ind. Jour. of Pure and App.
Physics, vol. 46, pp.463-370, Feb. 2008.
[5] F. Perry, B. Stauder, G. Henrion, Ph. Pigeat, “Plasma diagnostics for
the control of reactive magnetron deposition process”, Surf. and
Coatings Tech, vol. 74-75, pp. 575-579, 1995.
[6] Irving P. Herman, “Optical diagnostics for thin film processing’, Annu.
Rev. Phys. Chem., vol. 54, pp. 277-305, Jan. 2003.
[7] T. Nakano, N. Ohnuki, S. Baba, “Pressure dependence of optical
emission from DC magnetron sputtering plasma observed with spatial
resolution’, Vacuum, vol. 59, pp. 581-585, 2000.
[8] N Britun, M Gaillard, A Ricard, Y M Kim, K S Kim and J G Han,
“Determination of the vibrational, rotational and electron temperatures
in N2 and Ar-N2 rf discharge”, J. Phys. D: Appl. Phys., vol 40, pp.
1022-1029, Feb. 2007.
____________________________________________________________________________________________________________
The Effect of Titanium Dioxide Nanorods to the

M.K. Ahmad1, M.M. N. Syafiqah1, C.F. Soon1, N. Nafarizal1, R. Sanudin1, A.B. Suriani2, A. Mohamed2, M. Shimomura3, K. Murakami3,
M.H. Mamat4 and M.F. Malek4
1Microelectronic and Nanotechnology – Shamsuddin Research Centre (MiNT-SRC)

86400 Parit Raja, Batu Pahat, Johor, Malaysia
2Nanotechnology Research Centre
Department of Physic, Faculty of Science and Mathematics

Universiti Pendidikan Sultan Idris
35900 Tanjung Malim, Perak, Malaysia
3Department of Engineering
Graduate School of Integrated Science and Technology

Shizuoka University
432-8011 Hamamatsu, Shizuoka, Japan
4Nano-ElecTronic Centre
Faculty of Electrical Engineering

UiTM, 40450 Shah Alam, Selangor, Malaysia
*Corresponding author: M.K. Ahmad

Email: akhairul@uthm.edu.my
Microelectronic and Nanotechnology – Shamsuddin Research Centre (MiNT-SRC)
Abstract — Hydrothermal process is a technique used to

synthesize nanostructured Titanium Dioxide (TiO2) I. INTRODUCTION
powder for Dye-Sensitized Solid State Solar Cell (DSSC) DSSC have been extensively researched over the past years
application. The TiO2 powder is then deposited onto glass due to their very high energy conversion efficiency and can be
substrates using Spray Pyrolysis Deposition (SPD) said as low cost of production [1]. Other than that, DSSC is
method. In this study, Fluorine-doped tin oxide (FTO) one of the best alternatives that is based on the work function
conducting glass is used as the substrate for TiO2 nanorods of semiconductor, dyes, and also electrolytes solution that
thin film. Two samples were fabricated with different time have achieved the energy conversion efficiency as high as 10-
durations, 24 hours and 12 ½ hours. There were three 11% at AM 1.5 standard solar spectrum [2]. Although solar
properties studied; structural, electrical and optical cells made from silicon are main option to be chosen, it still
properties. At the end of the research, the TiO2 thin film have competition from other materials because of high
which was fabricated for 24 hours during the manufacturing costs and problems in the electrical transport
hydrothermal preparation process gave the optimum and photo transport. Nowadays, DSSC based on nano-
results. The average gap between grains is small due to the structured TiO2 electrodes have been attracting intense
big particle size of nanostructured TiO2 on the thin film. scientific and industrial attention for quite some time because
Smaller gap between grains on the thin film produce good of their potential for low cost scale up of the corresponding
electron mobility. It indicates that it has low resistivity, synthesis procedure and applicability for diffuse light
hence high ability to conduct electricity. Therefore, the harvesting [3]. Transparent oxide thin films are widely used
TiO2 thin film which was fabricated for 24 hours during materials in new generation solar cells. A film formation of
the hydrothermal preparation is the best candidate for TiO2 thin films can be carried out by various methods such as
DSSC application. sol gel, spray pyrolysis, sputtering, pulse laser deposition and
electro deposition. Among all these techniques, spray
Keywords—TiO2; DSSC; nanorods; power conversion pyrolysis is the most promising as it is a simple, easy to
efficiency handle, cost effective technique for large area deposition and
also excellent method for the deposition of metallic oxides
____________________________________________________________________________________________________________
thin films [4]. The titanium dioxide (TiO2) has a lot of Two samples of TiO2 nanorods thin films prepared by
advantages. The structural properties and high transparency in using hydrothermal method with different parameter used.
visible range make possible application of TiO2 in Table 1 shows the parameter that affected the TiO2 nanorods
microelectronic technology [5]. Thin film of TiO2 has a large thin films for each sample during the powder preparation.
deal of welfare such as wide band gap, high refractive index,
high dielectric constant and absence of toxicity [3]. The Table 1: The parameter used for each sample during the
objective this project is to study the effect of TiO2 nanorods in powder preparation
TiO2 thin film for application as the work electrode for DSSC.
Parameter for each sample
II. METHODOLOGY
Sample HCl Titanium DI Reaction
Temp.
n-Butoxide water Time
FTO conducting glass was used as the substrate for TiO2 (°C)
(ml) (ml) (ml) (hours)
nanorods thin film and the glass substrate was cut into 2.5cm
x 1cm size. The glass substrates were immersed with Acetone, 1 120 5 120 150 24
Ethanol and Distilled Water (DI water) with ratio of 1:1:1.
Then, the substrates were cleaned using an ultrasonic cleaner 2 120 5 120 150 12½
machine for 10 minutes. Lastly, the substrates were dried in
the oven at 150°C for 15 minutes. After that, titanium n-
butoxide was mixed with different amounts of hydrochloric B. Structural Properties
acid (HCl) and DI water. The solution was stirred at the The surface morphology of the TiO2 nanopowder
temperature of 34°C with 300 rpm for 5 minutes. The have been observed using FE-SEM with different
resulting solution was then transferred into a Teflon-lined magnification. After the TiO2 nanopowder went through the
stainless steel autoclave and sealed. grinding process, the result was expected to be the surface
Then, the autoclave was placed into the oven with morphology of TiO2 nanorods powder. The TiO2 nanorods
fixed temperature of 150°C and specific reaction times. After powder was successfully obtained for sample 1 which was
completion of the reaction, the autoclave was cooled down to fabricated for 24 hours and sample 2 which was fabricated for
room temperature. The solution was then filtered and rinsed 12 ½ hours. The results are shown in Figure 2 and Figure 3.
with DI water. Lastly, the filter paper used was placed into the
oven at 150°C for a short time to obtain the TiO2 nanorods in
powder form. Grind the powder obtained with Acetic Acid to
form fine TiO2 nanorods powder. The powder was added into
a porcelain mortar and was grinded with 5.5ml of Acetic Acid
as shown in Figure 3.6. The grinded solution was then filtered
and rinsed with DI water. Lastly, the filter paper used was
placed into the oven at 150°C for a short time to obtain fine
TiO2 nanorods powder. For the deposition process, 5.5ml of
Acetic Acid was added into a porcelain mortar with 0.3g
ofTiO2 nanorods powder. The solution was then grinded for a
short time and 20ml of TiO2 Colloid solution (≈ pH6) was
added. The resulting solution was poured into a brown light
proof bottle and 5 drops of Triton X-100 were added as well. Figure 2: Structure of TiO2 nanorods for sample 1 by
Next, 30ml of Ethanol was filled into the same bottle. The using magnification x10,000
bottle was then placed into the Ultrasonic machine for 30
minutes.
Finally, the cleaned FTO glass substrates were
masked with aluminium foil. After 30 minutes, the final
solution was used to be deposited on the masked glass
substrates. The deposition process done by Spray Pyrolysis
Deposition (SPD) onto the masked glass substrates on the hot
plate set at 150°C. The TiO2 thin film were heated at 150°C
for 30 minutes and lastly were annealed in furnace at 450°C
for 1 hour time.
A. Different parameter during preparation of TiO2 nanorods Figure 3: Structure of TiO2 nanorods for sample 2 by
powder using magnification x10,000
____________________________________________________________________________________________________________
average thickness for thin film from sample 1 is 5.31µm

The surface morphology of the deposited TiO2 thin film for whereas, for thin film from sample 2, it is 5.23µm.
sample 1 and sample 2 are shown in Figure 4 and Figure 5.
The samples have been observed using FE-SEM to determine
the gap between grains on the thin film. As the result, average
gap between grains for sample 1 is smaller than the gap
between grains for sample 2. This is due to the big size of
particle for sample 1 which was fabricated for 24 hours during
the hydrothermal preparation. Smaller gap between grains on
the thin film will produce good electron mobility.
Figure 6: The average size of particles is 9.29µm
Figure 4: Average gap between grains is 48.4µm
iii. X-Ray Diffraction study

It showed that sample 1 gives higher peak of intensity
at [110] plane than sample 2 (Figure 8). The [110] peak
increased when the fabricated time duration during
hydrothermal process is increased. It should be stressed that
all peaks observed can be assigned to specific lattice planes of
Figure 5: Average gap between grains is 65.9µm rutile phase of TiO2.
ii. The average size of particles of TiO2 nanorods Sample 1

Sample 2
Generally, smaller size of particles shows greater effective 6000
R[110]
surface area for dye adsoption. By using magnification of

x10,000, the FE-SEM images proved the average size of 5000
particles for each sample tested. Figures 6 shows that the 4000
average size of particle for sample 1 is 9.29µm. For sample 2,

Height [cts]
3000
Figure 7 shows that the average size of particle is 8.51µm. As
A[103]
mentioned before, the TiO2 nanorods powder was prepared 2000 A[004]
using hydrothermal method with different parameter for each 1000

R[200]
sample. Grinding process was done to form fine TiO2
0
nanorods powder with smaller size of particles. From the
observation, sample 2 produced finer nanostructured TiO2 26 28 30 32
Pos. [°2Th.]
34 36 38 40
powder compared to sample 1. Therefore, sample 2 has larger

surface area compared to sample 1. The powder obtained from Figure 8: X-Ray Diffraction pattern of TiO2 thin film at
both samples are then used for deposition process of TiO2 different fabricated time duration during hydrothermal process
nanorods thin film using Spray Pyrolysis Deposition (SPD)
method. The thickness of TiO2 nanorods thin film after the E. The efficiency of TiO2 nanorods thin film
deposition process is measured using surface profiler. The Figure 11 shows the 1-V curves for the efficiency of
____________________________________________________________________________________________________________
TiO2 thin film for both samples which were fabricated at ACKNOWLEDGMENTS
different duration during the hydrothermal process. Table 3 The authors would like to thank the Ministry of Education
shows the solar cell performance characteristics (Voc, Isc, (MOE) Malaysia (Vot 1459), Microelectronic and
Pmax, η) to analyze the higher efficiency produced by both Nanotechnology-Shamsuddin Research Centre (MiNT-SRC)
samples. As the result, sample 1 produced higher efficiency and Universiti Tun Hussein Onn Malaysia (UTHM) for
which is 0.0350% compared to sample 2 which only produced financial support
0.0136%. During the efficiency test, the DSSC fabricated used
N719 dye. From the results, we found that the use of TiO2 REFERENCES
nanorods in DSSC leads to low dye adsorption which gives [1] Xu, H., Tao, X., Wang, D. T., Zheng, Y. Z., and
poor photoresponse. Low dye adsorption might be due to the
Chen, J. F. Enhance efficiency in dye-sensitized solar cells
smooth surface of the TiO2 nanorods itself and the dye can
only attach at the tip of the nanorods structure. based on TiO2 nanocrystal/ nanotube double-layered films.
Electrochimia Acta, (2009), pp. 1-6.
Sample 1
Sample 2 [2] Rattanavoravipa, T., Sagawa, T., and Yoshikawa. S.
0.2
Photovoltaic performance of hybrid solar cell with TiO2
0.0
nanotubes arrays. Solar Energy Materials & Solar Cells, 92
2
Current density, mA/cm
-0.2 (2008), pp. 1445-1449.

-0.4
[3] Jasim, K. E. Dye Sensitized Solar Cells-Working

-0.6
Principles, Challenges, and opportunities. Solar cell- Dye
-0.8 Sensitized devices, pp. 171-172.
-1.0
0.0 0.2 0.4 0.6
Voltage, V
0.8 1.0
[4] Manoharan, C., and Sridhar. Physical properties of
spray pyrolysis TiO2 thin films. International Journal of
Figure 9: I-V measurement to test the efficiency of TiO2 thin Recent Scientific Research, 3 (9)(2012), pp. 775 -777.
film for both samples
[5] Wojcieszak. D., Domaradzki, J., Kaczmarek, D., and
Adamiak, B. Photoelectrical properties of TOS thin films
Table 2: The solar cell performance characteristics to analyze based on TiO2 prepared by modified magnetron sputtering.
the higher efficiency produced by both samples The Seventh International Conference on Advanced
Semiconductor Devices and Microsystems, (2008), pp. 283-
Sample Pmax
(mW)
Voc
(V)
Isc
(mA)
Jsc
(mA/cm2)
FF
(%)
η (%) 286.
1 0.008 0.50 0.029 0.116 51.46 0.0350
2 0.003 0.33 0.014 0.054 65.60 0.0136
V. CONCLUSION AND SUGGESTIONS

Through this study, TiO2 nanorods powder for Dye
Solid Solar Cell (DSSC) application have been successfully
synthesized using hydrothermal method. The TiO2 nanorods
powder was well deposited onto Fluorine-doped tin oxide
(FTO) conducting glass substrates by using Spray Pyrolysis
Deposition (SPD) technique. As the result, the rutile TiO2 thin
film which was fabricated for 24 hours during the
hydrothermal preparation process gave the optimum results.
However, it shows that the energy conversion efficiency in
DSSC based on rutile TiO2 nanorods thin film is still
relatively low. This is largely due to the limited internal
surface area and the structure of the TiO2 nanorods itself
which performed low dye adsorption.
____________________________________________________________________________________________________________
NOAA Satellite Based with OpenCV Library

Weather Monitoring System Software Development
Ihsan Ibrahim#1, Gunawan Wibisono*2, Prima Dewi Purnamasari#3
#
Electrical Engineering Department, Universitas Indonesia
Faculty of Engineering, Kampus Baru UI Depok, Universitas Indonesia 16424, Indonesia
1ihsan.ibrahim@ui.ac.id
2gunawan@eng.ui.ac.id
3prima.dp@ui.ac.id
Abstract— Traditional fishermen in the middle of the sea have who use small boats, it may not be possible. They rely more on
great risk because of the absence of communication channels that the ability to read the weather changes by utilizing the
can be used to obtain information about weather in the location surrounding environment, like the movement of winds, clouds,
and weather anomalies that may change suddenly and extreme. and its temperatures. Coupled with the small dimensions of the
One solution for the problem is with a system that can monitor the
ship, the risk of danger in the face of bad weather in the middle
weather automatically. This research will focus on the software
development for the system with image processing based on of the ocean is very big. This big gap between small or
imagery from NOAA satellites and uses OpenCV library. The traditional fishermen that use small boat to sail and big vessel
software is created to provide information about the weather is exist.
conditions on the user’s location and around of its location. Image The sailors or fishermen requires a solution, which can help
processing on this software is uses characteristic of weather, them with ease in getting information weather changes that
especially cloud for determining the result. On testing, it showed occurred where they are in real time and also cost efficient. So,
that the total accuracy is at 91.25%. they can reduce risk and prevent changes to face bad weather
that could occur suddenly. Therefore, this research is focusing
Keywords - OpenCV, weather monitoring system, NOAA satellite on built a system that can accommodate the needs of such a
solution. This system automatically detects the weather
I. INTRODUCTION conditions at the location of the user. These systems utilized the
Until now, marine areas with total 70% of the surface area outcome data from weather satellites or NOAA (National
of the earth are still cannot be fully covered by Oceanic and Atmospheric Administration). From the satellite
telecommunications mobile network. It is because of the data, the system will process the image processing based on the
absence of media intermediaries in these areas. Mobile network OpenCV library. Then, the output will be displayed on a
needs that media as an important element for connecting to the computer screen using a simple GUI (Graphical User Interface)
others. Although one of the network intermediary medium is automatically. This paper is focusing on the software
the air, but still requires intermediaries such as BTSs (Base development with image processing of this solution system.
Transceiver Station) which is useful for signalling in mobile
networks. With these circumstances, the sailors, including II. OPENCV LIBRARY FOR WEATHER MONITORING
fishermen at this time can only rely on communications from OpenCV library has more than 2500 optimized algorithms,
other telecommunication networks, such as radio and satellite including a comprehensive set that includes both classic
networks. computer vision algorithms and state-of-the-art and machine
The need for information about the state and changes of learning algorithms. These algorithms can be used to detect and
weather is vital. It affects the activities of sailors, especially recognize faces, identify objects, classifying human motion in
fishermen in a vast ocean. The weather at sea has characteristics the video, track the movement of the camera and moving
that may change drastically or extreme circumstances in a short objects, extracting a 3D model of an existing object, producing
time. The extreme weather changes will affect and will be the a 3D point cloud of stereo cameras, etc., and also for weather
main obstacle for the sailors or fishermen to get a good haul or monitoring.
get to their destination. It also will be harmful when they are at In this research, to get the desired result when performing
middle at sea. image processing for weather monitoring, knowing the
For sailors or fishermen with big boat when faced bad characteristics of weather in visual is a must [1]. With the input
weather, risks will be reduced. In addition to the large of this software is an image from the conversion of NOAA
dimensions of the ship, but it also because of the large vessels satellite wave data, it has to be processed on every pixel to get
usually have sophisticated communication equipment so that it the information. Filtering one by one characteristic of weather
remains possible to know the weather conditions. As for those and cloud are the main steps for this image processing process.
Selection of pixels can be done in various ways, depending on
____________________________________________________________________________________________________________
the conditions required. There are several examples of how,

like just taking pixels that meet the conditions or terms of
existing or to discard it. This method is called thresholding
OpenCV or restriction and in this way become one of the
functions that can be immediately implemented by a direct call WXtoImage Software
from the library imgproc [2]. Antenna
However, thresholding function can only be applied to the Computer

image BW (Black and White) or Grayscale. This function can
not directly apply it on RGB images. When using RGB image RF Receiver and BNC Weather Monitoring
Connector with LNA Software
as input, it is necessary to advance the conversion of the RGB Output Display with GUI
image into grayscale image with that of the Color function is
also available in the library imgproc. With limitations into a
Fig 1. Weather monitoring system diagram
single color channel only, restrictions or segmentation can be
done directly in color images or who has three color channels Once converted by software WXtoImage, the result of image
with a specific value in each channel is inRange function [3]. needs to be processed in order to get the final result of weather
InRange function has the same function as the function of condition for the user via the LCD screen or output display.
the threshold, which is an election or segmentation area / region This process requires separate software developed specifically
you want to apply the lower limit value and lower limit value for this system.
corresponds to the name of the function itself, which means In the design and development, software for automatic
"within reach". This function has a concept by taking the values weather monitoring system will be used by a user with the
of each pixel in the array is between the lower limit value and following requirements:
upper limit value. Values that satisfy these conditions will be Users can see the weather conditions in the form of an image
changed to the maximum value, namely white concept of on the LCD screen.
segmentation distinguishes between functions inRange 2. Users can view weather information on the location and
threshold. the surrounding areas with the information display is simple
Furthermore, the function of the library imgproc moment is and easy to understand.
one useful function to get the coordinates of the midpoint or The algorithm in this paper was developed to perform image
center of gravity of the existing array. Basically, moments processing on the input data in the form of drawing image
belong to the useful functions for structural analysis and conversion results WXtoImage software. Based on some input
description of the form [4]. Moments also used to measure the data, the special specifications that exist in the image of the
mass distribution of an object, shape or contour. picture is a map which also contains the image of a cloud with
OpenCV moment later the application will do the grouping MSA (Multi Spectral Analysis) feature and also has a '+' sign
of each region or area that is detected into a separate array. The indicating the location of a station or device used to obtain the
two-dimensional array that represents the x and y coordinates. data from the satellite. MSA image has near real or visible color.
Indices [0,0] at any moment m_00 is the center of gravity of the With such characteristics of the input image as shown in Figure
region or area. Then, m_10 / m_00 and m_01 / m_00 defines 2, the algorithm created to predict the weather that occurs in a
the coordinates of the center of gravity or centroid of the area particular area consists of several stages. The weather
which is also referred to as the moment of inertia is legally conditions that could be detected are heavy rain, medium rain,
mechanics [5]. light rain, and clear where the heavy rain also could lead to the
storm that actually dangerous for sailors, especially with small
III. DESIGN AND IMPLEMENTATION OF SYSTEM SOFTWARE
boats.
Weather monitoring software is made as a part of the
automatic weather monitoring system. A set of devices of this
system is intended for high mobility use because the target
users who are local fishermen that use small boats and not
equipped with adequate facilities when they go to sea. In the
functional architecture of the system, the system is designed to
retrieve data from the NOAA satellites, process it, and show the
results to the user with easily understandable information. In
architecture, as in Figure 1, automatic weather monitoring
system consists of several devices that are useful to get input,
process, and display the output. The research focus is limited
on the weather monitoring software is shown in Figure 1 with
red colored block in a processing device.
Fig 2. Input image sample for the software
____________________________________________________________________________________________________________
The first stage is determining the coordinates of the center of be added into the index of next weather category. The required
the '+' which shows the current location of the satellite data number of array index value at any part or region is 300 for the
retrieval. Because the resulting image data is a map with a very heavy rain category, 300 for the medium rain category, and 600
broad geographic reach, is necessary in order to approximate for the light rain category. The requirement numbers for this
localization can be more specific and focused on the position of algorithm in measuring the cloud category is obtained from the
the user is located. Determination of the coordinates is done by discussion from BMKG (Badan Meterologi, Klimatologi, dan
the detection of the color of the center point of the '+' such as Geofisika) or the Indonesia government’s bureau for weather
ROI (Region of Interest). The color of the center point of the '+' and climate [7]. This algorithm used a useful variable to store
is unique and no other element on the image to match the the value of the array index number are formed, for summing
intensity of the color so as to facilitate detection. However, the index previous categories that do not meet the criteria to be
color detection algorithm is performed on this process and added in the next weather category. It is also the reason for light
further, color is detected not on channels RGB (Red Green Blue) rain category has higher requirement of index size than other
is commonly used in the image. This algorithm will detect color weather categories.
using HSV channels (Hue Saturation Value) because it can be
used for the detection of a specific color better than RGB and
time of the process is more efficient. Based on research
conducted by P. and V. Ganesan Rajini against HSV to RGB
quality, indicating that the channel is most suitable for the
detection of HSV color [6].
After getting the point coordinates of the '+', the image of the
input image will be cropped to a size of 160x160 pixels with
coordinate values as a point of making the new center. Then,
the concepts used in the next step is to detect the presence of
clouds by color. With the brightest colors indicate rain clouds
and less bright colors of the clouds indicate a light rain cloud or
cloud over it [1]. However, if only rely on the original color, it's
hard to distinguish accurately the color intensity of these clouds.
Therefore, a color image is to be changed into another colormap,
so it is easier to see and identify the intensity of different colors,
ie JET colormap representing dark colors as bright as the blue
and red colors.
Furthermore, the special function that is useful for detecting
rain on each region's imagery in order to get the final result, as
shown in Figure 3. Detection of color also plays vital role in Fig 3. Output of developed software
this function, as shown by the pseudocode in Figure 4. In this
rain detection function which is divided into four categories, //Rain Detection Function
namely heavy rain / storms, medium rain, light rain, and no rain //Heavy Rain
x = 0
or clear. Due to focus of this research is for supporting the local detect color [red – dark red];
fishermen, so that this application gives the output information get Moments of every detected area into an array;
sum array index hd(i1);
in Bahasa Indonesia, whereas heavy rain is “Hujan Deras”, If(hd(i1) => 300){
medium rain is “Hujan Sedang”, and light rain is “Hujan print “Hujan Deras”;}
x = sum of index array hd(i1);
Ringan”. The cloudy weather is not included in a category of else{
its own, because of the absence of the potential for rain, so it //Medium Rain
will be put in a clear weather category. Because of the wide Detect color [orange – red-1];
get Moments of every detected area into an array;
range value of clear weather category until light rain category, sum array index hd(i1)+x;
the categorization starts from heavy rain / storms in advance. If(hs(i1) => 300){
print “Hujan Sedang”;}
Heavy rain category is indicated by the value of red to dark red x + = sum of index array hs(i1);
color. Then, the category of medium rain with a value of the else{
//Light Rain
color orange to red-1. The numbers “-1” indicates a value below Detect color [yellow – orange-1];
the lower limit used in the previous categories. Light rain get Moments of every detected area into an array;
sum array index hr(i1) + x;
category with yellow to orange color values-1. While the value if(hr(i1) => 600){
remaining below the yellow to dark blue is said to be clear. print “Hujan Ringan”;}
This function is further categorized based on the number of x + = Jumlah Index Array hr(i1);
}
array indexes are conditioned on each category, whether it }
meets the criteria of the index number which is required in each }
region there or not. If not met, it will be processed with the next Fig 4. Pseudocode of rain detection function
category and number of indexes of the previous categories will
____________________________________________________________________________________________________________
IV. ACCURACY TESTING cloud is detected does have the potential content of light rain,
Accuracy testing is applied to this software for measuring but the amount of content that is little it cannot make the
the correctness of its information given. This test is conducted average weather in a region to be "Light rain". Examiners
with processed images by Weather Monitoring Systems consider that, the algorithm used in the manufacture of Weather
Software takes random samples taken as many as 10 pieces of Monitoring System Software Based on NOAA satellites and
images for each category of weather. Each test image was taken OpenCV library is already performed well in determining the
from each part of a whole 16 parts image. With so obtained 10 outcome of weather information on an area.
pieces of the picture for the heavy rain category, 10 pieces of
the picture for the medium rain category, and 10 pieces of the Average Accuracy Score
picture for the light rain category, as well as 10 pieces of the
picture for the clear category. This test also involves the
Clear 100%
interviewees from BMKG for determining the result testing
Weather Category
whether it showed the correct result or not [7]. Light Rain 75%
Based on Table 1, the score for each image is set to 1-3,
where 1 is not correct on two levels category or more, 2 is Medium Rain 90%
mistaken in one level category, and 3 is a correct result. It can
be seen that, in the category "Heavy Rain" and "Clear", the Heavy Rain 100%
software algorithm was successfully generating value accuracy
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
3 for each of 10 test samples provided so that the average value
Average Score
of the accuracy of the algorithm in both categories the weather
is 3. Examiners assess that, the algorithm used for weather
Fig 5. Graph of software average accuracy score
category "Heavy Rain" and "Clear" is correct. In the category
"Heavy Rain" although many covered by clouds with potential
Based on the graph in Figure 5, the weather category "Heavy
"Rain Medium" and "Light Rain" around, the software
Rain" and "Clear" managed to get 100% accuracy value
algorithm is considered appropriate to give the conclusion as
according to the success of the results corresponding to the 10
"Heavy Rain" because of its strong potential compared with
test samples given in each category. While the weather category
other categories.
"Medium Rain" gets an average score of 90% because there
TABLE I was 1 sample to get a value of 0 or is not appropriate in
TEST RESULT OF ALGORITHM ACCURACY providing the final information. With an average value of 75%
Result Test or the lowest compared to other weather categories, algorithms
Weather Accuracy
weather category "Light Rain" has five sample test results are
Category 1 2 3 4 5 6 7 8 9 10 Average less precise in providing information or misses one category of
weather. Overall, the average value of the accuracy of the
Heavy
3 3 3 3 3 3 3 3 3 3 3 algorithm in software is 91.25%.
Rain
V. CONCLUSIONS
Medium
3 3 3 3 3 1 3 3 3 3 2.8 From this research the built software can give the output of
Rain information about weather on user’s location and its
Light surrounding. The algorithm accuracy of this software is at
2 2 3 2 2 3 3 3 2 3 2.5 91.25% correct.
Rain
Clear 3 3 3 3 3 3 3 3 3 3 3 REFERENCES
[1] Japan Meteorological Agency, Analysis and Use Meteorological
Algorithm Accuracy Average 2.825
Satellite Images, Tokyo: Meteorological Satellite Center, 2002.
[2] G. B. &. A. Kaehler, Learning OpenCV, Sebastopol: O'Reilly, 2008.
[3] OpenCV.org, “Operations on Arrays,” OpenCV.org, 2015. [Online].
Similarly, the weather category "Clear", the examiner Available:
considered that, although there are other clouds which does not http://docs.opencv.org/modules/core/doc/operations_on_arrays.html?hi
have the potential for rain at all and only have the potential to ghlight=inrange#cv2.inRange. [Accessed May 20th, 2015].
be very small compared to the vast existing area. [4] OpenCV.org, “Image Moments,” OpenCV.org, 2015. [Online].
Available:
In the category of weather "Medium Rain" there is one http://docs.opencv.org/doc/tutorials/imgproc/shapedescriptors/moment
sample that scored 1, and the examiner testers should the s/moments.html?highlight=moment. [Accessed May 22th, 2015].
weather on that image was "Clear" because of the strength of [5] J. Flusser, Moments and Moment Invariants in Pattern Recognition,
existing cloud poses no potential rain, so the values given are 1 Prague: John Wiley & Sons, 2009.
[6] G. P. &. V. Rajini, “Assessment of Satellite Image Segmentation in RGB
for error category reaches 2 level. While in the weather and HSV Color Space using Image Quality Measures,” Advances in
category "Light Rain" there were 5 samples that had an error Electrical Engineering (ICAEE), 2014 International Conference on, vol.
rate of 1 level. The fifth sample was equally scored 2 of the 1, no. 14394638, pp. 1-5, 2014.
maximum value 3. Errors that occur in this category due to the [7] S. Widarko dan A. Susilawati, Interviewees, [Interview]. July 31th, 2015.
____________________________________________________________________________________________________________
Application of Microcapillary Atmospheric Pressure

Plasma Jet as a Patterned Surface Modification of
Carbon Nanotube Microarray
Tomy Abuzairi1,2, Mitsuru Okada3, Retno Wigajatri Purnamaningsih2, Nji Raden Poespawati2,
and Masaaki Nagatsu 1,3,4,*
1
2
3
Graduate School of Integrated Science and Technology, Shizuoka University, 3-5-1, Johoku, Naka-ku, Hamamatsu 432-8561,
Japan
4
Research Institute of Electronics, Shizuoka University, 3-5-1, Johoku, Naka-ku, Hamamatsu 432-8011, Japan
*
nagatsu.masaaki@shizuoka.ac.jp
Abstract— A microcapillary atmospheric pressure plasma jet motivation of the patterned surface modification of CNT
(µCAPPJ) is a promising technology with huge potential for microarray is for developing biochip sensor based on CNT.
micro-scale material surface modification. In this work, we
demonstrated the use of µCAPPJ to modify carbon nanotubes
(CNTs) surface for developing biochip sensor based on CNT. The
patterning of the CNT microarray surface was automatically
conducted by computer-controller stages with two-step plasma
treatments, i.e. pretreatment and posttreatment. The results
show that the successful of “cross” pattern of amino group
modification on CNT without any significant interference
between CNT modification. The successfully patterned
modification of amino groups on CNT microarray is valuable for
future development of CNT-based biochip sensor.
Keywords— microcapillary atmospheric pressure plasma jet;

carbon nanotubes; pattern surface modification; biochip sensor
I. INTRODUCTION
Atmospheric pressure plasma jet (APPJ) is gas discharges
that consist of free electrons, ions, radicals, and highly excited
neutral and charged species, which is generated in
atmospheric pressure (open space) rather than in low pressure
required vacuum system. Their ability to provide more active
plasma chemistry, make those sources essential for several
applications such as etching [1], ashing [2], material
deposition [3], and surface modification [4]–[6] as shown in Fig. 1 Applications of APPJ as etching, ashing, deposition, and surface
Fig. 1. modification of the substrate surface.
Furthermore, carbon nanotubes (CNTs) are nano-material
carbon allotropes with outstanding material properties for
example high surface-area to weight ratio [7], high tensile II. EXPERIMENTAL SETUP
strength [8], and strong absorption of biomaterials (e.g., The schematic of µCAPPJ is shown in Fig. 2 and the high
proteins) [9]. As a result, CNT are applied in numerous areas speed camera image of microcapillary is depicted inset Fig. 2.
for instance fuel cell [10], field-effect transistor [11], lithium- The µCAPPJ consist of a glass tube with outer diameter of 6
ion battery [12], including biosensors [13]. mm and inner diameter of 4 mm, and a capillary tip with inner
In this work, we demonstrated the use of a microcapillary diameter ~5 µm tapered by a resin micropipette tip. The glass
atmospheric pressure plasma jet (µCAPPJ) for modifying the tube was covered by two copper band electrodes, ground
micro-scale surface of CNT microarray, not only without electrode and power electrode. The microcapillary tip of
physical mask but also in the patterned modification. The
____________________________________________________________________________________________________________
atmospheric plasma jet was fabricated by using a laser lithography, RF sputtering, and thermal plasma chemical
micropipette puller (Sutter Instrument P-2000). vapor deposition [14]. Each CNT spot is comprised of a
bundle of multiwall carbon nanotube having a typical length
of about 2 µm and a diameter of about 100 nm as shown in
magnified image inset Fig. 3. The primary advantage of the
CNT materials for biochip sensor platform is the enhancement
of active surface area without sacrificing the size of the device,
which possibly will permit to detect of currently undetectable
disease markers [15], [16].
Fig. 3 FE-SEM images of the CNT biochip sensor platform with a spacing of
50 µm and a dot size of 5 µm, and a magnified image of one spot CNT is
shown inset. [5]
Fig. 2 Schematic of µCAPPJ for patterning surface modification of CNT
microarray. The right inset shows a magnified image of µCAPPJ using high
speed camera. Fig. 4(a) depicts the illustration of patterning surface
modification of CNT microarray in “cross” pattern by
µCAPPJ. The cross patterned is chosen for checking whether
To ignite the plasma in the µCAPPJ, high voltage pulses of there are interference or not between CNT dots in the plasma
±7.5 kV amplitude was employed with a multifunction treatment due to larger size surface modification of the
generator (NF WF1948) and amplifier system (AC/DC µCAPPJ. It is well documented that larger size surface
amplifier NF HVA4321). modification with minimum 5 times larger than tip size was
In order to modify the surface of CNT microarray with observed as reported by Ye et al. in monolayer graphene on Si
amino group (–NH2), two-step plasma treatments, i.e. substrate [17], Kakei et al. in photoresist on Si substrate [18],
pretreatment and posttreatment, were carried out. The detail of and Tao et al. in parylene-C film on Si substrate [19].
the surface modification of CNT by two-step plasma The bright field image of CNT microarray under
treatments has been reported by our group elsewhere [6]. fluorescent microscope is shown in Fig. 4(b). The white
Briefly, the pretreatment was performed for 0.2 second by circles in the bright field image correspond to the designated
using 700 sccm He gas flow with a -500 V substrate bias. The CNT spot modified by µCAPPJ. By using synchronization
posttreatment was conducted for 2 second by using 700/10 system of the computer-controller stages and high-voltage
sccm He/NH3 gas mixture without substrate bias. Furthermore, system, the next designated spot on the CNT microarray
for patterning the CNT microarray by µCAPPJ, the sample substrate was treated automatically with precision movement
was positioned on a computer-controlled X–Y stage (Chuo of 1 µm.
Precision QT-ADM2 and ALS-301-HM). Fig. 4(c) shows the dark field image of CNT microarray
The modified CNT microarray was finally evaluated by covered with Alexa Fluor® 488 5-SDP ester as amino group
chemical derivatization method with Alexa Fluor® 488 5- label. The green colour on CNT indicates the connection of
SDP ester as amino group label. The amino group label the CNT with amino group label that emits green fluorescence
connections to the modified CNTs were confirmed by using at 520 nm when excited by 494 nm light of fluorescent
fluorescent microscope (Leica Microsystems DMI3000B) microscope. In addition, we also prove the successful of cross
recorded with 20.8-s exposure time, 1.6 saturation, 1.0 gamma, pattern of amino group modification on CNT, without any
and 1.0× gain. significant interference between CNT. This successful
III. RESULTS AND DISCUSSION patterned modification of amino groups on CNT microarray is
valuable for future development of biochip microarray based
Fig. 3 shows typical CNT microarray (spacing of 50 µm on CNT materials.
and dot size of 5 µm) that is fabricated by using electron beam
____________________________________________________________________________________________________________
Fig. 4 (a) Illustration of patterning surface modification of CNT microarray by µCAPPJ in the cross pattern. (b) Bright field image of the CNT microarray. The
white circles correspond to the designated CNT spot modified by µCAPPJ. (c) Dark field image of the CNT microarray covered with amino group label, Alexa
Fluor® 488 5-SDP ester. (Colour figure).
[6] T. Abuzairi, M. Okada, Y. Mochizuki, N. R. Poespawati, R. W.

IV. CONCLUSIONS Purnamaningsih, and M. Nagatsu, “Maskless functionalization of a
carbon nanotube dot array biosensor using an ultrafine atmospheric
A microcapillary atmospheric pressure plasma jet was used pressure plasma jet,” Carbon, vol. 89, pp. 208–216, 2015.
to modify CNT microarray surface, not only without a [7] J. N. Wohlstadter, J. L. Wilbur, G. B. Sigal, H. A. Biebuyck, M. A.
physical mask but also in the cross pattern form. The Billadeau, L. Dong, A. B. Fischer, S. R. Gudibande, S. H. Jameison, J.
H. Kenten, J. Leginus, J. K. Leland, R. J. Massey, and S. J. Wohlstadter,
patterning of CNT microarray was automatically carried out “Carbon nanotube based biosensor,” Adv. Mater., vol. 15, no. 14, pp.
by computer-controller stages with two-step plasma treatment, 1184–1187, 2003.
pretreatment and posttreatment. The present results show that [8] J.-P. Salvetat, J.-M. Bonard, N. Thomson, A. Kulik, L. Forro, W.
the successful of cross pattern of amino group modification on Benoit, and L. Zuppiroli, “Mechanical properties of carbon nanotubes,”
Appl. Phys. A, vol. 69, no. 3, pp. 255–260, 1999.
CNT without any significant interference between CNT is [9] F. Balavoine, P. Schultz, C. Richard, V. Mallouh, T. W. Ebbesen, and C.
significant for the development of CNT based biochip sensor. Mioskowski, “Helical Crystallization of Proteins on Carbon Nanotubes:
A First Step towards the Development of New Biosensors,” Helical
ACKNOWLEDGMENT Cryst. Proteins Carbon Nanotub. First Step Dev. New Biosens., vol. 38,
no. 13/14, pp. 1912–1915, 1999.
This work was supported by Grant-in-Aid for Scientific [10] C. Wang, M. Waje, X. Wang, J. Tang, and R. Haddon, “Proton
Research (A) (No. 25246029) and Scientific Research (B) (No. exchange membrane fuel cells with carbon nanotube based electrodes,”
26289016) from the Japan Society for the Promotion of Proton Exch. Membr. Fuel Cells Carbon Nanotube Based Electrodes,
Science. The authors also acknowledge the support of vol. 4, no. 2, pp. 345–348, 2004.
[11] S. J. Tans, A. R. M. Verschueren, and C. Dekker, “Room-temperature
International Research Collaboration and Scientific transistor based on a single carbon nanotube,” Room-Temp. Transistor
Publication Grant from Ministry of Research, Technology and Based Single Carbon Nanotube, vol. 393, no. 6680, pp. 49–52, 1998.
Higher Education Indonesia. The authors would like to thank [12] C. de las Casas and W. Li, “A review of application of carbon
to Prof. Sudeep Bhattacharjee of Indian Institute of nanotubes for lithium ion battery anode material,” Rev. Appl. Carbon
Nanotub. Lithium Ion Battery Anode Mater., vol. 28, pp. 74–85, 2012.
Technology Kanpur for his help in the present study. [13] F. Lu, L. Gu, M. J. Meziani, X. Wang, P. G. Luo, L. Veca, L. Cao, and
Y.P. Sun, “Advances in Bioapplications of Carbon Nanotubes,” Adv.
REFERENCES Bioapplications Carbon Nanotub., vol. 21, no. 2, p. 139152, 2009.
[1] J. Y. Jeong, S. E. Babayan, V. J. Tu, J. Park, I. Henins, R. F. Hicks, and [14] T. Matsuda, M. Mesko, A. Ogino, and M. Nagatsu, “Synthesis of
G. S. Selwyn, “Etching materials with an atmospheric-pressure plasma vertically aligned carbon nanotubes on submicron-sized dot-catalyst
jet,” Plasma Sources Sci. Technol., vol. 7, no. 3, pp. 282–285, 1998. array using plasma CVD method,” Diam. Relat. Mater., vol. 17, no. 4–5,
[2] A. West, M. van der Schans, C. Xu, M. Cooke, and E. Wagenaars, “Fast, pp. 772–775, 2008.
downstream removal of photoresist using reactive oxygen species from [15] C. Wingren and C. A. Borrebaeck, “Progress in miniaturization of
the effluent of an atmospheric pressure plasma Jet,” Plasma Sources Sci. protein arrays - a step closer to high-density nanoarrays,” Drug Discov.
Technol., vol. 25, no. 2, p. 02LT01, 2016. Today, vol. 12, no. 19, pp. 813–819, 2007.
[3] P. Zhao, W. Zheng, J. Watanabe, Y. D. Meng, and M. Nagatsu, “Highly [16] J. W. Silzel, B. Cercek, C. Dodson, T. Tsay, and R. J. Obremski,
Conductive Cu Thin Film Deposition on Polyimide by RF Driven “Mass-sensing, multianalyte microarray immunoassay with imaging
Atmospheric Pressure Plasma Jets under Nitrogen Atmosphere,” detection,” Clin. Chem., vol. 44, no. 9, pp. 2036–2043, 1998.
Plasma Process. Polym., vol. 12, no. 5, pp. 431–438, 2015. [17] D. Ye, S. Q. Wu, Y. Yu, L. Liu, X. P. Lu, and Y. Wu, “Patterned
[4] T. Abuzairi, M. Okada, S. Bhattacharjee, and M. Nagatsu, “Surface graphene functionalization via mask-free scanning of micro-plasma jet
conductivity dependent dynamic behaviour of an ultrafine atmospheric under ambient condition,” Appl. Phys. Lett., vol. 104, no. 10, p. 103105,
pressure plasma jet for microscale surface processing,” Appl. Surf. Sci, 2014.
vol. 390, pp. 489–496, 2016. [18] R. Kakei, A. Ogino, F. Iwata, and M. Nagatsu, “Production of ultrafine
[5] T. Abuzairi, M. Okada, R. W. Purnamaningsih, N. R. Poespawati, F. atmospheric pressure plasma jet with nano-capillary,” Thin Solid Films,
Iwata, and M. Nagatsu, “Maskless localized patterning of biomolecules vol. 518, no. 13, pp. 3457–3460, 2010.
on carbon nanotube microarray functionalized by ultrafine atmospheric [19] T. Wang, B. Yang, X. Chen, X. Wang, C. Yang, and J. Liu,
pressure plasma jet using biotin-avidin system,” Appl. Phys. Lett., vol. “Nonhomogeneous surface properties of parylene-C film etched by an
109, no. 2, p. 023701, 2016. atmospheric pressure He/O2 micro-plasma jet in ambient air,” Appl.
Surf. Sci., vol. 383, pp. 261–267, 2016.
____________________________________________________________________________________________________________
The Development of Kartini Reactor Data

Acquisition System to Support Nuclear Training
Centre (NTC)
Adi Abimanyu, Syarip Syarip, Elisabeth Supriyatni, Wagirin, Djati Gunawan, Marsudi
National Nuclear Energy Agency, Centre for Accelerator Science and Technology
Babarsari Street PO BOX 6101 YKBB, Yogyakarta 55281, Indonesia
abimanyu.adi@batan.go.id
Abstract— Kartini research reactor is designed for the purpose of The control systems of the reactor is a means to regulate or
research, training and education in nuclear reactor science field. control the nuclear fission process that occurs in the reactor, in
In recent years, the role of Kartini reactor was to become a such that the reactor can be operated at a power level that is
Nuclear Training Centre (NTC). Therefore, a system of data desired, to increase, and to maintain the reactor at the level of
acquisition that are well documented in the database should be
certain power, as well as stopping or shut down the reactor
developed. The development of Kartini reactor data acquisition
system provided with a database as a means of documentation, in operation [3]. To failure protection system is implemented
such that the data is ready to be used in the NTC program. The using diversity method, where the reactor power is measured
system was developed through the stages of acquisition by two different system ie, by Wide-Range logarithmically
(communication, reading, parsing, calculation and data Power Channel (NLW-2) and linear power channel (NP 1000)
storage).Input data in this system directly derived from Kartini [4].
reactor instrumentation & control system which consisted data Research on the design and development of data acquisition
from linear power (NP-1000) and logarithmic power (NLW-2) systems Kartini reactor process parameters have been
channels, reactor period, position of control rods, reactor fuel conducted [5]. Acquisition system is a process to measure,
temperature, water tank temperature and level, reactor coolant
store, display and analyse data from a wide array of tools [6].
flow rate of primary and secondary, etc. Based on the test results,
this system has the error reading and parsing data at 0% and Each measurements requirea transducer or sensor, the device
1% error calculation. Therefore it can be concluded that the that converts a physical quantity that measured into the
developed system is feasible to be used to support the NTC electrical signal. The study describes a computer system to
program and documentation system. acquire data of parameters process on the operating reactor
with the signal processing module slots in the form of
Keywords— Kartini reactor, NTC, data acquisitions, instru- modules, namely Advantech PCL-812PG. The signals were
mentation & control processed by processing system in a data acquisition system
of Kartini reactor process parameters by installing such an
Advantech PCL-812PG on ISA BUS slot industrial computer
I. INTRODUCTION [6].In order to be used by the NTC, the data acquisition in
Nuclear Training Centre (NTC) is a training program in the computer process is sent to the NTC computer. Data on NTC
field of nuclear and applied physics particular in reactors computer is still required processing and then being stored on
utilization, for educational institutions and research personnel the database.
in the research activities in the field of ionizing radiation and The database on NTC computer can’t be directly used
reactor technology[1].The training that offered by NTC can be because the storage formatis not appropriate with the
either power reactors calibration, control rods calibration, computer program that will be developed, therefore, the
measurement of reactivity and so forth. The training was development of Kartini reactor data acquisition system is
conducted by using data taken directly from the reactor needed to support the NTC.The acquisition system is based on
operation. MySQL database using the Labview software, installed on
Kartini reactor is a research reactor designed by BATAN NTC computer.
with a power of 250 kWth for the purpose of research, training
II. KARTINI REACTOR DATA ACQUISITION SYSTEM
and education and operates under licenced on a power of 100
kWth. Kartini reactor is a type of reactor TRIGA MARK II Kartini reactor data acquisition system consists of process
with an open pool, light water coolant. Kartini reactor control and data acquisition reactor. Process control is done by
construction began in late 1974, and reached a critical regulating the configuration of the control rod position to
condition for the first time on January 25, 1979, was obtain the desired value with the safety limitation using
inaugurated on March 1, 1979 and still operated until now [2]. certain reactor period. Reactor period is defined as the time
interval required to change the reactor power by e-folding
time where e = 2.71828 [7]. Kartini reactor parameters in the
____________________________________________________________________________________________________________
form of analogous signal from three control rod positions,

reactor power and period NLW2 logarithmic channel and
reactor linear power from of NP1000, are acquired using
PCL812PG on computer control. Fig 1 shows block diagram
of data acquisition system Kartini reactor process parameters.
Fig. 3. NTC computer network
IV. THE FORMAT OF DATA COMMUNICATION ON THE NTC

NETWORK
Data transmission from control computer to the acquisition
computer conducted through the RS-232 communication with
the baud rate of 115200 bps with the format shown in Fig.4.
.
Data begins # sign as the header and still in the form of
Fig. 1. Block diagram of the data acquisition system of Kartini reactor voltage value which consists of power NP1000, power NLW2,
process parameters[6]
NLW2 period, the control rod position and time of the trip and
The control rod position analogue signal is taken from the the trip parameter.
tap voltage coming from potentiometer. Potentiometer voltage
is 5 volts and it is equivalent to 100% of control rod position
(38cm) [8] as shown in Fig. 2.
Fig. 4. The format of data being transmitted by the computer control [7]
While the physical parameters Kartini reactor such as

temperature and flow rate of the water coolant are monitored
using a neck computer and transmitted to the acquisition
computer located in the control room via RS-485
Fig. 2. The acquisition method of control rod position [9] communication with baud rateof 9600 bps.The data format
used is shown in Fig. 5. The data are sent alternately from A
III. NTC COMPUTER NETWORK to D and the value of Data is still in the form of volts and
The operator manually operated reactor by moving the initiated $ sign as the header data.
control rods to regulate the flux of neutrons through the
operation button in the control room. Control rod position
changed will lead to change neutron flux, reactor power and
period. Reactor power and period is monitored output display
of NLW2 and NP1000. In order to facilitate the operator to
monitor the value of power, period and position of the control
rods, the quantities are displayed on the computer with ISA-
BUS interface using PCL812PG. The NTC network concept is Fig. 5. The format of data being transmitted by the neck computer
shown in Fig. 3.
The process data flow derived from the NP1000, NLW2, During operation of the reactor, reactor operating
control rod position, was acquired by control computer while parameters need to be merged with the temperature of reactor
the physical quantities (temperature and flow rate) which was water coolant and coolant flow rate. The process of merging is
acquired by neck computer. These data are then sent to the conducted on a acquisition computer with the format shown in
acquisition computer for further processing. Fig.6. Data from this merger aretransmitted to NTC computer
in NTC room located in the 3rd floor of the reactor building
____________________________________________________________________________________________________________
through the RS-485 communication with baud rate of 9600

bps. These data in NTC computer are parsed, processed into
operating data information reactors, displayed and stored in a
MySQL database that is readily accessible by intranet on the
NTC client computer.
Fig. 8. Design of Reactor data acquisition GUI
A, B, C and D in Fig. 8 is data reactor tank water

Fig. 6. The format of data being transmitted by the acquisition computer temperature, fuel, coolant flow rate of primary and secondary
system, respectively. The database used consists of a table
V. RESEARCH METHOD with the name akusisi data with 13 columns, as shown in Fig.
A. Software development 9.
Reactor parameters acquisition software is created and
developed using LabView 2011 software in NTC computer,
so that the input data used haveformat as in Fig.6. This
Fig. 9. Format tabel database
software will be used to acquire the data from the reactor,
process the data into information reactor parameters and store B. Testing, reading and parsing of data
the information in a database, and display the data on a GUI
The test data reading is to determine the suitability of data
[9].The flowchart of software is shown in Fig. 7
being transmitted by the acquisition computer with the
received data / read by the NTC computer via RS-485 serial
communications in terms of similarity of data (format) and the
data length. This testing is done by monitoring the data sent
by computer acquisition using hyper terminal and then
compared with the data reading software developed on
computer acquisition. Testing of parsing the data is to find out
the parsing process data that done by the software and all data
can be parsed according to the group based on the sequence
and length of the data.
C. Testing data computation
Fig. 7. Flowchart software Testing data calculation is to determine the performed
calculations error value. This testing is done by comparing the
Data parsing process is done after the serial data reading value of the calculation result processed by NTC computer
process from the acquisition computer completed. This with the acquisition computer and see the stored value by the
process aims is to separate data which transmitted by database wether these are same as the calculated value by the
acquisition computer, into information of power, period and NTC computer.
the position of the control rods reactor and temperature and
flow rate data of water coolant. The process of parsing is done VI. RESULTS AND ANALYSIS
based on the data length of each data thatconsist of 5
A. The result of reading and parsing of data
characters for power reactors, the period and the position of
the control rods while 7 characters for the temperature and Table II shows reading and parsing data. In the process of
flow rate. reading data at 9600 bps baud rate as shown in Table II, the
After the data are successfully parsed, these data cannot transmitted and received data are exactly the same in terms of
beexecutedusing arithmetic calculations because the data are value (information) and the data length. Based on these
still in string variables. Therefore, the data must be converted reading test results error value obtained is 0%.The erroris still
into a number using the convert function in LabView. allowed/acceptable to be happened, if the baud rate values
used are different. So it must be ascertained in advance, the
Data from the calculation is displayed on the Graphical baud rate value used by computer acquisition prior to the
User Interface (GUI) and stored in the Mysql database. The process of reading the NTC computer. In testing the parsing
design reactor data acquisition GUI is shown in Fig.8. the data showed that the data can be parsed properly in
accordance with the group and the length of the data and the
____________________________________________________________________________________________________________
error value obtained at 0%. It was found that the error parsing the data length of the readings should be maintained to always
still occured if the length of the data read is changed, so that be the same at 39 characters.
TABLE I
THE RESULT OF READING AND PARSING OF DATA
Data transmit (hyper terminal) Length of data Data receive Length of data
@0.0000.9430.3600.0150.0170.013B0.00002 39 @0.0000.9430.3600.0150.0170.013B0.00002 39
1 PARSING
NP-1000 NLW-2 Period Safe Shim Reg WTR temp Fuel temp Flow primer Flow seconder
0.000 0.943 0.360 0.015 0.017 0.013 0.00000 0.00002 0.00000 0.00000
Data transmit (hyper terminal) Length of data Data receive Length of data
@0.0000.9500.3800.0140.0180.014D0.36200 39 @0.0000.9500.3800.0140.0180.014D0.36200 39
2 PARSING
NP-1000 NLW-2 Period Safe Shim Reg WTR temp Fuel temp Flow primer Flow seconder
0.000 0.950 0.380 0.014 0.018 0.014 0.00000 0.00000 0.00000 0.36200
VII. CONCLUSIONS
B. Result of testing data computation
Based on the test results it can be concluded that the
Table III shows the result of calculations performed by
software of data acquisition system of Kartini reactor is
NTC computer and databases as well as the value of the
properly in function, able to read and parses data transmitted
acquisition computer. Based on Table III, it is shown that the
by acquisition computer on baud rate of 9600 bps with error
error value of the NTC computer calculation when compared
value of 0%. Result of calculations by the NTC computer as
to the value at acquisition computer is less than 1% and the
compared to acquisition computer has a maximum error of 1%.
data stored in the database is exactly same as on the NTC
computer. This error occurs due to differences in the accuracy ACKNOWLEDGMENT
of the calculation, where the acquisition of computer using
The authors would like to thank to Director of Centre for
integers and the NTC computer using three decimal places.
Accelerator Science and Technology Yogyakarta, Head and
This can be minimized if the calculations done on the
all staffs of Reactor Division, for their support, as well as to
acquisition computer, and the transmitted data is the results of
Catur P and Aryono of Politeknik Manufaktur Negeri Bangka
the calculation so that the data shown on the NTC computer
Belitung for their participation and help in data acquisition
and the database will be matched exactly the data on the
programming.
computer acquisition. In general, the testing results shown that
the software of data acquisition system of Kartini reactor is REFERENCES
properly in function, the acquisition computer can transmit [1] G. STANESCU. NUCLEAR TRAINING CENTRE (CPSDN)
data at baud rate of 9600 bps with 0% error. Available: http://cpsdn.nipne.ro
[2] T. Suhaemi, D. D. Dj, I. K., J. S., and Setyono. (2003, December 19,
TABLE II
2015). Safety Evaluation reactor Kartini in terms of Instrumentation
THE RESULT DATA COMPUTATION
System Design. 49-60. Available: http://www.iaea.org/in-is/collec-
No Parameter Acqui - NTC Database Error tion/NCLCollectionStore/_Public/44/011/44011588.pdf
sition (%) [3] A. Istiqomah, "Simulation of Kartini Reactor Operation Using
1 Reg control rod 54 54 54 0.000 Software DCS Centum VP Based On Data Operations" Diploma IV,
Teknofisika Nuclear, STTN BATAN, Yogyakarta, 2015.
2 Shim control rod 75 75 75 0.000 [4] Prajitno. (2010, February 12, 2016). “Reactor Instrumentation and
Control Systems”.
3 Safe control rod 100 100 100 0.000 [5] Prajitno, "Design and Development of Data Acquisition Systems
4 NP-1000 power 82 82.608 82.608 0.741 Kartini Reactor Process Parameters" Yogyakarta, 2009, pp. 145-149
[6] M. C. Corporation, Data Acquisition Handbook, Third ed. vol. III.
5 NLW-2 power 82 82.327 82.327 0.399 United States of America: Measurement Computing Corporation, 2012.
[7] Y. Sardjono and W. Setiawan, "Measurement of the Reactivity Kartini
6 Period 37 37.358 37.358 0.968 Reactor Control Rods using Motorola DSP56001" Scientific Meeting
7 Fuel temperature 89 89.221 89.221 0.248 and Presentation in Basic Research in Science and Nuclear Technology,
Yogyakarta, 2003, pp. 45-51.
8 WTR temperature 84.205 84.205 84.205 0.000 [8] Prajitno, Reactor Instrumentation and Control Systems. Yogyakarta:
Pusdiklat BATAN 2010..
9 Flow primer 605 604.56 604.56 0.073 [9] C. Pebriandani and Aryono, "Realization Initiation Hybrid Reactor is
10 Flow seconder 556 556.97 556.97 0.174 Ready Connected to the Internet" D3 practical work report, Prodi
Electronic Engineering, Manufacturing Polytechnic State Bangka
Belitung, Bangka Belitung.
____________________________________________________________________________________________________________
Influence of Load Properties on the Performance of

Low-Speed Single-Phase Self-Excited Induction
Generator
Hari Santoso#1, Rini Nur Hasanah#2, I Nyoman Gde Wardana*3, Budiono Mismail#4
#
Electrical Engineering Department, Brawijaya University
Jln. MT Haryono 167 Malang 65145 Indonesia
1
hari.santoso53@yahoo.com
2
rini.hasanah@ub.ac.id
4
bud.mismail@ub.ac.id
*
Mechanical Engineering Department, Brawijaya University
Jln. MT Haryono 167 Malang 65145 Indonesia
2
wardana@ub.ac.id
Abstract— In general, an induction generator can only generate The reactances are calculated using the base frequency
voltage and power if its requirement for enough capacitive value. Under no-load condiiton, the magnetizing voltage can
reactive power is met. During an isolated operation, the reactive be found using (1),
power can be obtained using a capacitor bank, whereas in
 V 2 1/ 2

2
parallel operation it can be supplied from the power network. If
being loaded with an incandescent lamp, it it can only absorb Vm  X m I m   2  R1   X 1  I m
1 (1)
active power due to the resistive properties of the lamp. If an  I m  
energy saving lamp is chosen as load, they will also supply
reactive power to generator as most of them are of capacitive Stator reactance X1 is sufficiently low to ignore, being
properties. Each time loading is increased or decreased, it will compared with the magneitizing reactance Xm. The negative
also automatically increase or decrease the reactive power supply slip value is almost zero. The capacitor curent will be almost
to generator. On the other hand, the interaction of capacitive and the same as the magnetizing current at synchronous speed, as
inductive components in the generator windings produces expressed in (2) and (3).
oscillation which makes the current waveforms distorted. The Ic X c  Im X c (2)
research results indicate that voltage build-up in a low-power,
low-speed single-phase self-excited induction generator can be V1
achieved and the voltage can be maintained without collapsing I c  V1C  (3)
Xc
under loading condition with load types of capacitive properties.
Under steady-satte operation, based on (3) it is shown that
Vm is proportional to V1, so that
Keywords— self-excited induction generator, load properties,
V1  I m X c (4)
output performance
jX1
I. INDUCTION GENERATOR OPERATION R1 jX2
An induction generator can only generate active power + +

I1 I2
when being supplied with capacitive reactive power. In a Im
single-phase self-excited induction generator, the voltage

build-up and generation of active power are only possible with R -jXc
V1 jXm Vm R2/s
sufficient reactive power supply from the exciting capacitor
[1-3, 6-9].
Analysis of voltage, current and frequency generated in a - -
self-excited induction generator can be done using the
equivalent circuit in Fig. 1 [4]. As indicated, Vm represents the Fig. 1 An equivalent circuit of single-phase induction generator under isolated
magnetizing voltage in volts, Xm magnetizing reactance in operation [4]
ohms, Im magnetizing current in amperes, V1 generator
terminal voltage, R1 and R2 stator and rotor resistances, X1 and II. WAVE DISTORTION
X2 stator and rotor reactances, whereas s means the slip. Voltage and current harmonic distortions are commonly
found in the power system network [5]. The harmonic waves
consist of sinusoidal waveforms with frequencies being
integer multiple of the fundamental frequency. These
____________________________________________________________________________________________________________
harmonics frequencies will add to the original wave with energy saving lamps are respectively presented in Table I dan
fundamental frequency to result in a distorted non-sinusoidal II.
waveform. TABLE I
The distorted voltage waveforms can be expressed using RESULTS OF LOADING EXPERIMENT USING INCANDESCENT LAMPS
the mathematical Fourier equation,
Pin Pout
No. n (rpm) η (%)
(W) (W)
(5) 1 491 141,2 0 0
2 500 197,7 40 20,2
Vo represents the DC component of the wave, whereas V1, 3 512 235,5 75 31,8
V2, V3, …, Vn are the wave amplitude at each harmonic level. 4 522 281,4 117 41,6
When the fundamental frequency is f, then the second 5 536 327,5 156 47,6
harmonic frequency will be 2f, third harmonic frequency 3f, 6 546 361,0 198 54,8
and so on, with the nth harmonic frequency is nf. 7 566 431,2 233 54,0
By using mathematical Fourier analysis method, a 8 576 472,0 270 57,2
periodical distorted waveform can be decomposed into its 9 600 552,0 320 58,0
10 620 606,0 352 58,1
sinusoidal components with its all composing harmonic
frequencies including the fundamental one. This Fourier
analysis can explain the process on how a distorted waveform TABLE II
is produced, as shown in Fig. 2. RESULTS OF LOADING EXPERIMENT USING ENERGY SAVING LAMPS
Pin Pout
No. n (rpm) η (%)
(W) (W)
1 486 153,7 0 0
2 501 216,1 44 20,4
3 506 261,9 86 32,8
4 522 322,7 144 44,6
5 532 382,4 182 47,6
6 540 426,9 226 52,9
7 556 479,5 258 53,8
8 572 542,7 305 56,2
9 580 566,9 323 57,0
Fig. 2 The harmonic components producing distortion in a sinusoidal In Table I it is indicated that the loading with resistive loads
waveform [5] could be increased gradually up to the output power of 352
Harmonics could produce the heating of electrical watts and 58.1% of efficiency, however, the loading using
equipments like generator, the capacity derating of capacitive loads could only be increased until the value of 323
equipments, shorter economical life-time of equipments, and watts, as shown in Table II, because the load increase beyond
maybe more other undesirable losses [10]. that value made the voltage collapsed. It indicated that the
ouput power decreased. Under the same condition of output
III. EXPERIMENTS ON GENERATOR PERFORMANCES power, approximately being represented by the 9th step of
measurements in the two tables, generator resulted in higher
Laboratory experiments have been undertaken using loads
efficiency under resistive loading. It means the lower power
of two property types, i.e. resistive and capacitive. The
losses in the generator.
resistive type of load has been represented by incandescent
lamps, whereas the capacitive properties could be found on
some types of energy saving lamps. The loads were applied on
the generator and the voltage across the generator terminals
were maintained constant at 220 volts. Observation and
examination have been focused on the resulted values of
rotation speed and output power of generator, input power
given by the prime-mover, and the generated current and
voltage waveforms. The generator under consideration was
built with 12 poles, 500W of output power, 220V of nominal
voltage and used 64µF of exciting capacitors.
IV. RESULTS AND ANALYSIS Fig. 3 The resulted waveforms of voltage (in blue color) and current (in
By maintaining generator output voltage at 220V, the yellow color) in the self-excited induction generator under consideration
results of loading experiments, using incandescent lamps and
____________________________________________________________________________________________________________
Loading with energy saving lamps resulted in quite visible

distortion in the current waveform (yellow coloured curve in REFERENCES
Fig. 3). This distortion was produced by harmonics due to the [1] S. S. Murthy, B. Singh, and V. Sandeep, ”A novel and comprehensive
oscillation of winding inductance and capacitance of the performance analysis of a single-phase two-winding self-excited
induction generator,” IEEE Transactions on Energy Conversion, vol.
exciting capacitor. 27, no. 1, pp. 117-127, 2008.
[2] H. Santoso, R. N. Hasanah, I. N. G. Wardana, and B. Mismail,
V. CONCLUSIONS “Capacitor motor as low-power low-speed single-phase generator,” in
Proc. The International Conference on Electrical Engineering,
The capacitive properties of energy saving lamps help in
Computer Science and Informatics (ECCSI), pp. 290-295, 2014.
the self-excitation process to build up voltage in single-phase [3] H. Santoso, R. N. Hasanah, I. N. G. Wardana, and B. Mismail, “Self-
induction generator. exciting capacitor circuit for a low-power, low-speed single-phase
The capacitive properties of energy saving lamps interact induction generator,” Journal of Engineering and Applied Sciences,
vol.11, No.7, pp. 4860-4864, 2016.
with the inductive properties of generator windings producing
[4] B. Shakuntla, “On-set theory of self-excitation in induction generator,”
oscillation which increase harmonic distortion in the generator International Journal of Recent Trends in Engineering, Vol.2, No.5, pp.
current waveforms. 325-330, 2009.
From the point of view of customer, an energy saving lamp [5] H. Santoso, R. N. Hasanah, I. N. G. Wardana, and B. Mismail,
“Loading performances of low-power low-speed single phase induction
offers the advantage of giving similar light levels with lower
generator with energy saving lamps,” Journal of Applied Mechanics
power consumption than the traditional incandescent lamps. and Materials, vol. 785, pp. 290-294, 2015.
However, from the point of view of power producers, the [6] D. K. Palwalia and S. P. Singh, “Design and implementation of
energy saving lamp is not favoritized in terms of the resulted induction generator controller for single phase self excited induction
generator,” in Proc. The Third IEEE Conference on Industrial
generated current distortion and the required power input to
Electronics and Applications, 2008, pp. 400-404.
generator. [7] S. N. Mahato, S. P. Singh, and M. P. Sharma, “Capacitors required for
maximum power of a self excited single-phase induction generator
ACKNOWLEDGMENT using a three-phase machine,” IEEE Transactions on Energy
Conversion, vol. 23, No.2, pp. 372-381, 2008.
The authors would like to express their gratitude to the [8] B. Singh and L. B. Shilpakar, “Analysis of a novel solid-state voltage
Faculty of Engineering, Brawijaya University, for the funding regulator for a self-excited induction generator,” in IEE Proceedings
of this research publication. on Generation, Transmission and Distribution, vol. 145, No.6, pp. 647-
655, 2008.
[9] T. F. Chan, “Capacitance requirements of self-excited induction
generators,” IEEE Transactions on Energy Conversion, vol. 8, No.2,
pp. 304-311, 1993.
[10] C. Sankaran, Power Quality, Boca Raton, USA: CRC Press LLC, 2002.
____________________________________________________________________________________________________________
Bandstop Filter for Radar Application

with L Resonator
Dian Widi Astuti1, Ferdi Yansyah2, Mudrik Alaydrus3
Department of Electrical Engineering, Universitas Mercu Buana
Meruya No 1 West Jakarta, Indonesia
1
dian.widiastuti@mercubuana.ac.id
2
adiferdi.yan@gmail.com
3
mudrikalaydrus@yahoo.com
Abstract— A simple and general approach for design of norrow- been presented in recent years. Bandstop filter for X-Band
band bandstop filter is based on reactance/susceptance slop application have been presented in [1] a DGS with folded L-
parameters of the resonators. To employ a lowpass prototype for shaped arms. The resonant behaviors of the DGS used in [1]
bandstop filter design, the transition from lowpass to bandstop to introduces transmission zeroes to the filter response and
characteristics can be effected by frequency mapping. There are
consequently improves its stopband performance. In [2,3] also
two typical configuration for TEM or quasi-TEM to realize
norrow-band bandstop filter. The first is a main transmission proposed a compact bandstop filter with the resonant
line that it’s have electrically coupled to half-wavelength behaviours of the petal-shaped stub and the square-ring stub
resonator and the second is a main transmission line that it’s with four protruded arrow-shaped strips are used to introduces
have magnetically coupled to half-wavelength resonators in a transmission zeroes to the filter response and consequently
hairpin shape. In this article, we use microstrip bandstop filter improves its stopband performance. While in [4], the
with L-shaped resonators coupled to the main line both microstrip BSF was designed on both substare side by
electrically and magnetically for SART in 9.2 – 9.5 GHz. opening aperture in the ground metallization under the low-
Experimental results show that simulation results differ from impedance transmission line. Replacing of the pairs of RSLRs
measurement results even though there is a shifting of frequency.
based on Koch fractal structure, introduces transmission
We use Rogers RT/Duroid 5870 with a relative dielectric contant
of 2.33 and thickness of material is 0.508 mm. zeroes.
This paper work deals with design and development of a
microstrip filter for Radar maritm for Search and Rescue
Keywords— Bandstop filter, L-resonator, radar application, Transponder (SART) in 9.2 – 9.5 GHz. In this structure used
L-shape resonators coupled both electrically and magnetically
I. INTRODUCTION
to the main line. This reasearch used five-poles (n = 5)
Radar or Radio Detection and Ringing widely used in Chebyshev lowpass prototype with a passband ripple of 0.1
military purposes as radar used to detect, measure distances dB is chosen for desinging a microstrip bandstop filter
and create a folder objects such as aircraft, motor vehicles and because it’s easier to design and fabriction rather than [1-4].
various weather (rain). Actually radar applications can be used In addition, due to the limitations of fabricating filters are only
in the field of policing, shipping, aviation and much more. able to provide accuracy the distance between the resonator is
One type of filter by function is band-stop filter. There is 0.25 mm.
also a call to the rejected band filter, where the filter has a
function to hold the frequency in the range f1 to f2 and f1 skip II. TRANSFORMATIONS
frequencies below and above f2. The band stop filter is the
opposite of a band pass filter. A. Frequency and Element Transformations
Prototype band-stop filter for radar applications by using For frequency mapping, impedance scaling is also required
microstrip technology becomes difficult to be realized, given to accomplish the element transformation. The impedance
that most of the high-frequency special for radar applications scaling will remove the g0 = 1 normalization and adjusts the
is not to mention the realization of filters that use microstrip filter work for any value of the source impedance denoted by
subtrate because the results of insertion loss and return loss is Z0. For our formulation, it is convenient to define an
greater when compared with waveguide technology. impedance scaling factor 0 as [5]
Conventionally the microwave bandstop filter (BSF) is
Z 0 g 0 for g 0 being the resistance
implemented either by all shunt stubs or by series connected 0   (1)
high-low stepped-impedance microstrip line sections.  g 0 Y0 for g 0 being the conductanc e
Hawever, generally these are not easily available in
where Y0 = 1/Z0 is the source admittance. In principle,
microwave band due to the high impedance microstrip line applying the impedance scaling upon a filter network in such a
and the spurious pass-bands. To remove these disadvantages, way that
defected ground structures (DGSs) for microstrip lines have
____________________________________________________________________________________________________________
L   0L  ZU 
2
1
  
C  C 0 (2)  Z0  g g
0 n 1
R   0R (8)
1
xi   0 Li 
G  G 0  0 C1
2
has no effect on the response shape. Z  g0
 Z 0  U  for i  1 to n
Let g be the generic term for the lowpass prototype  Z0  g i  C FBW
elements in the element transformation to be discussed. where gi are the element value of lowpass prototype and xi are
Becaused it is independent of the frequency transformation, the the reactance slope parameter of shunt-series resonators.
following resistive-element transformation holds for any type
of filter: The relationship between the normalized reactance slope
parameter to the frequency of a microwave bandstop resonator
R   0g can be seen at
for g representing the resistance (3)
g
G for g representing the conductanc e  x  0 f0
0     (9)
 Z0  2  3dB 2 f 3dB
B. Bandstop Transformations
Frequency responses of a single L-resonator coupled to the
The frequency transformation from lowpass prototype to main line for differenst coupling spacing s are then simulated
bandstop is achieved by the frequency mapping using EM simulator based on Fig.1. In this reasearch use
 C FBW Sonnet v15.52 [6] based on Fig.1. It should be mentioned that
 (4)
0   0 
if another attenuation bandwidth other than the 3 dB bandwidth
is desirable fot extracting the normalized reactance slope
0  12 parameter, the relationship between the desired attenuation
bandwidth and the normalized reactance slope parameter can
 2  1 be derived for Eq. (9). After that, we get grafic to extract
FBW  (5) normalized reactance slope parameters against coupling
0 spacings.
where 2  1 is the bandwidth. This form of the
transformation is opposite to the bandpass transformation in
that an inductive/capasitive element g in the lowpass prototype
will transform to a parallel/series LC resonant circuit in the
bandstop filter. The elements for the LC resonators
transformed to the bandstop filter are
 1  1
C p   
 FBW  0  c  0g (6)
  FBW 
L p   c  0 g
 0 
for g representing the inductance, and
 1 0
Ls   
 FBW0  c  g (7)
  FBW  g
C s   c 
 0   0 Fig. 1 Typical simulated frequency response of a single microstip L-resonator
coupled to a main transmission line..
for g representing the capacitance. [5]
Equivalent circuits of banstop filters with shunt-series III. FILTER DESIGN
resonant branches will give: The design of this band-stop filter is used Chebyshev low-
pass filter prototype with five-pole (n = 5) such as shown in
Fig. 2.[5]
____________________________________________________________________________________________________________
Based on Fig. 3 will be obtained within the smallest

distance of the third resonator with a major axis of about 0.1
mm. While the smallest path between the resonator can be
fabricated by the workshop was 0.25 mm. Then do the
simulation iteration many times by changing the structure of
L-resonators which the horizontal lenght, lh becomes shorter
than from lh = 6.4 mm become lh = 4.3 mm and the height
become higher from lv = 5.4 mm become lv = 7.1 mm. Then
obtained a design and prototype band-stop filter, as shown in
Fig. 4.
Fig. 2. A microstrip narraow-band bandpass filter with L-resonators.
The microstrip bandstop filter uses L-shape resonators

coupled both electrically and magnetically to the main line. We
chosed a passband ripple of 0.1 dB and C = 1 have g0 = g6 =
1.0, g1 = g5 = 1.1468, g2 = g4 =1.3712, and g3 = 1.9750. The
desired band-edge frequencies are 1 = 9.2 and 2 = 9.5 GHz,
hence, the midband frequency of the stopband is f0 = 9.3488
GHz and the fractional bandwidth is FBW = 9.3488 GHz,
according to Eq. (5). Using Eq. (8) to calculate the desired
design parameters yields
x1 x
ZU  Z 0  5  22.7266
Z0 Z0
x3 x2 x
 27.1735  4  15.7786
Z0 Z0 Z0 (a)
where Z0 = 50 ohms. In this reasearch used a commercial
subtrate Rogers RT/Duroid 5870 (www.rogerscorp.com) with
a relative dielectric contant (r) of 2.33, thickness 0.508 mm
(0.020 inch) and tangent loss of 0.0005. We can use [5,7] to
get the 50 ohm main line has a width, W = 1.5 mm and for a
half of wavelenght at f0, 1/2  is 11.43 mm. We made a half of
wavelenght for L-resonators shape with lh = 6.4 mm and lv =
5.4 mm. Frequency responses of a single L-resonator coupled
to the main line for diffrent coupling spacing s are the
simulated using Sonnet v15.52. The normalized reactance
slope parameters are then extracted according to Eq. (9). The
extracted normalized reactance slope parameter are plotted in
Fig. 3.
(b)
110
Fig 4 (a) Skematic band-stop design, (b) Prototype of band-stop filter.
100
IV. RESULT AND CONCLUSIONS
90
Fig. 4b. is a photograph of fabricated microstip bandstop
80 filter. The measured and simulated performances of the filter
are illustrated in Fig. 5.
70
x/zo
60
50
40
30
20
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
s(mm)
Fig. 3 Extracted normalized reactance slope parameters against coupling

spacings.
____________________________________________________________________________________________________________
-5
reflection and transmission factor in dB
-10
-15
-20
-25
S11 measurement
S21 measurement
-30
S11 simulation
S21 simulation
-35
8 8.5 9 9.5 10 10.5
Frequency [GHz]
Fig. 5. Simulation (dashed line) dan measurement (solid line) result.
As shown in Fig. 5, for simulation has an insertion and

return losses quite a good value at 9.2 – 9.5 GHz. While for
measurement result has a frequency shiftting become low
frequency and degradation transmission and reflection factor.
REFERENCES
[1] R.Habibi, Ch. Ghobadi, M. Ojaroudi, and N Ojaroudi, "Very compact
broad band-stop filter using periodic L-Shaped stubs based on self-
cmplementary structure for X-band application," Electronic Letters,
Vol. 48, No. 23, 2012.
[2] Mohammad Pourbagher, Nasser Ojaroudi, Changiz Ghobadi, and
Javad Nourinia,"Compact Band-Stop Filter for X-Band Transceiver in
Radar Applications," ACES Journal, Vol. 30, No. 4, pp. 423-426, April
2015.
[3] N. Ojaroudi, M. Ojaroudi, and R. Habibi,"Design and implementation
of very compact band-stop filter with petal-shaped stub for radar
applications,"Microwave and Optical Technology Letters, Vol 55, No.
5, pp. 1130-1132, May 2013
[4] N. Ojaroudi, Y. Ojaroudi and S. Ojaroudi,"Novel Design of UWB
Band-Stop Filter (BSF) Based on Koch Fractal Structures," ACES
Journal, Vol. 30, No. 1, pp. 117-122, January 2015.
[5] J. -S. Hong, Microstrip Filters for RF/Microwave Applications, 2nd ed.
New Jersey: Wiley, 2011.
[6] www.sonnetsoftware.com veified on 18 June 2013
[7] Dian Widi Astuti, Juwanto and Mudrik Alaydrus. “A Bandpass Filter
Based On Square Open Loop Resonators at 2.45 GHz,” 5th ICICI-BME,
November 7-8, 2013
____________________________________________________________________________________________________________
Power System Stabilizer Optimization Based on

Modified Differential Evolution Algorithm and
Network Reduction Considering Network Losses
Rohmanita Duanaputri1, Adi Soeprijanto2, Ardyono Priyadi3, Dheny Ashari Hasan4
1,2,3
Department of Electrical Engineering, Institut Teknologi Sepuluh Nopember, Indonesia
Kampus ITS, Keputih-Sukolilo, Surabaya, 60111
1rohmanitar@gmail.com
2adisup@ee.its.ac.id
3priyadi@ee.its.ac.id
4dheny3007@gmail.com
Abstract — Power system stability in power system operations Oscillation damping on the interconnected system has a
have become a major concern, which can be achieved by major role on the security of the system operation. In this
stabilizing input and output power. Power system stability can paper, Suralaya generator’s machine, excitation system, and
be enhanced by adding Power System Stabilizer (PSS). The the PSS will be modeled as SMIB.
basic function of adjusted parameter PSS to damp the
generator rotor oscillations. This research paper proposes to In SMIB model, Req and Xeq can be searched by Rei -
adjust PSS’s parameters using Modified Differential Evolution Dimo method [2]. This method is relatively complicated and
(MDE) for improving steady state index. Damping oscillation is commonly used to model multi-machine modelling type.
based on an eigenvalue method is used as objective In this paper, the calculation method to approach the value
function in the MDE optimization algorithm. of Xeq and Req is conducted using losses analysis
Additionally, the Req and Xeq are provided for contributed by a single machine [3]. Every power system
representing the network losses. The simulation result naturally supplies electricity from generators to consumers
provided the best solution for settling time and passing through the transmission line that had losses. Based
amplitude compared with GAPSS and DEPSS in the on the Losses Concept, Req and Xeq for SMIB model can be
SMIB system. calculated with a much simpler way.
To optimization parameters of PSS can be makes
Keyword — PSS, MDE, Network Reduction considering
Artificial Intellegence (AI) likes, Genetic Algorithm (GA),
Network Losses.
Particle Swarm Optimization (PSO), Differential Evolution
I. INTRODUCTION (DE), and many other [4-6]. One of the approach to
optimally tune the parameters of PSS is Modified
The excitation current has an important part in Differential Evolution (MDE). In this paper, modified in
maintaining the stability of a power plant. In the occurrence order to incorporate information of the best solution in the
of load fluctuations, excitation controller will be functioned current population and also information of the current parent
as a controller of output generator such as voltage and power to define the new search directions by using mutation
factor by regulating the magnitudes input in order to find a operator [7]. Optimal tuning parameters is very influential in
new stability [1]. Dynamic stability is an important factor for order to stabilize the system with a faster time. However, the
the reliability of electric power systems. System instability is parameter range of equipment is very diverse and extensive.
very dangerous because it can cause a loss of Therefore, to obtain the parameter value optimally, MDE
synchronization, and then causes blackout. Electrical power optimization method could be good choice. The improvement
system on a large scale is a system that is susceptible to of engine oscillation damping can be analyzed using time-
interference as well as the Java - Bali interconnection system based or eigen-value. In this paper, multi-objective function
of 500 kV. The occurrence of a change in load demand ratio approach to improve the damping of engine oscillation is a
and the total power generated in the system can cause multi-objective eigenvalue [8-9]. Then, the real time
oscillation in the system. So as to overcome the oscillation, simulation results using SIMULINK are analyzed by
auxiliary equipment to dampen the oscillation is needed. One comparing IEEE PSS value and MDE optimized PSS.
of the tools used is PSS (Power System Stabilizer) which is
mounted on the AVR generator. The basic function of a PSS II. MODELLING SYSTEM
is to add damping to the generator rotor oscillations by
In reality the transmission power system is designed to be
controlling its excitation using auxiliary stability signals.
interconnected, which allows many machines (multi-
____________________________________________________________________________________________________________
machines) to supply load demand together. In this paper,

SMIB model is taken to analyse engine oscillation. Because,
SMIB model is observing only one machine connected to
the infinite bus through the impedance equivalent, network
reduction is needed to achieve the value of equivalent
impedance as shown in Fig. 1.
Fig 2. Linier model of single machine
Fig. 1. Single Machine Infinite Bus Model (SMIB) C. Power System Stabilizer
To improve the oscillation damping on the system,
A. Network Reduction from Multi-machine to SMIB Suralaya steam plant is added with IEEE PSS models with
considering Network Losses rotor speed changes as the input signal.
In reality, there are many kinds of network reduction
method (Rei Dimo, etc). This paper presents a network
reduction method to obtain Req and Xeq in order to observe
SMIB model using the losses concept that is much simpler.
Based on the Losses Concept, the value of Req and Xeq for
SMIB can be calculated easily. The process of network
reduction considering network losses can be explained as D. Objective Function and Constrain
Fig.3. IEEE PSS
follows [3]: The analysis of the engine oscillation damping
1) Turn off all of machine, except the observed machine. improvement can be analysed using time-based or
Set, the total loading does not exceed the machine capacity eigenvalue. In this paper, the objective function that is used
that will be observed. to observe the damping engine oscillation improvement is a
multi-objective eigenvalue [8-9].
2) Running load flow and take the result of the active
power (P), the reactive power (Q), of the active power losses (3)
(PL), the reactive power losses (QL).
PSS parameter optimization typical range is (0.001-50)
3) Calculate the current value of the observed machine
for , (4.5-5) for , (0.06-1.5) for and , (0.01-0.1)
using the equation:
for and [12]. In this paper, Modified Differential
(1) Evolution (MDE) is applied to resolve and find the optimum
value of PSS parameters. So the constraint of optimization
I is the current of the observed machine, P – jQ is the problems is:
complex power which is generated by observed machine,
Vi* is the voltage conjugate on the bus where the observed (4)
machine is located. (5)
(6)
4) Calculate the value of Req and Xeq using the losses (7)
concept with the equation:
(8)
(2) (9)
PL is active power losses, QL is reactive power losses, Zeq (10)
is equivalent impedance of network. Req and Xeq are used E. Modified Differential Evolution
to get the value of K1 to K6. K1 to K6 are a parameter of a
Differential Evolution (DE) is an optimization technique
synchronous machine used in liner models [10-11].
based on population searching. The characteristic of DE are
B. Linier Model simplicity, robustness, few control variables and fast
convergence. Generally, there are three important processes
The Linier model is chosen to simplify the analysis in DE, which are mutation, crossover, and selection. The fig.
process. Linear model used in the analysis is of Heffron – 4 is following chart is an illustration on how these steps in
Phillips [10] and De Mello - Concordia [11] to approach the DE algorithm proceed.
machine model.
____________________________________________________________________________________________________________
3) Calculate the current value I of the observed

machine using the equation (1) is 2.295 p.u
4) Calculate the value of Req and Xeq using the losses
concept with the equation (2). Req is 0.000138 p.u and Xeq
is 0.00143 p.u. Value of K1 to K6 are:
K1= 3.3088 K3= 0.1339 K5=-0.9837
K2= 2.5200 K4= 4.9094 K6= 0.0034
B. Optimization of PSS Parameters
The PSS tuning parameter simulation experiments is
performed 10 times on each optimization method. As shown
in the TABLE I and Fig. 5, MDE has a value higher than the
GA and DE. These results indicate that MDE is able to
produce optimal value better than GA and DE. The final
Fig.4. Flow chart DE algorithm
values of the optimized parameter are given in TABLE II
TABLE I
In order to optimize the parameters of PSS we use COMPARISON FITNESS VALUE
original DE by modifying mutation formula. The original
formula is Trial GAPSS DEPSS MDEPSS
(11) 1 0.4186 0.5313 0.6193
and modified into
2 0.4824 0.43269 0.62393
(12)
3 0.4572 0.5199 0.51393
where,
x = population 4 0.4472 0.5222 0.5693
= mutation factor 5 0.5376 0.5588 0.46393
R1, R2, R3 = randomly selecting factor 6 0.4178 0.4976 0.53393
= best solution in current population
7 0.5514 0.5689 0.6393
= current parent 8 0.4796 0.529 0.5593
= current population that selected 9 0.4978 0.5393 0.55393
randomly 10 0.4984 0.5439 0.6393
= factors that indicate the best influence
Max value 0.5514 0.5689 0.6393
= factors that indicate the parent solution
TABLE II
So, the result of modifying mutation is called by FINAL VALUES OF OPTIMIZED
Modified Differential Evolution (MDE) [7].
GAPSS DEPSS MDEPSS
III. SIMULATION AND ANALYSIS OF DATA
K 20 50 50
In this paper, 500 kV, 50 Hz system of Java - Bali
transmission is chosen as the study case. Suralaya steam Tw 3 4.951 5
plant is chosen as the observed machine. T1 1.4226 0.3383 0.35
A. Calculation of Network Reduction T2 0.1555 0.0701 0.075
The system of 500 kV transmissions is complicated T3 1.0386 0.8025 0.85
interconnected system. The reduction of network using the T4 0.0497 0.0263 0.03
losses concept would be much easier. Based on losses
concept network reduction, the step to get the value of Req
and Xeq is described as below:
1) Turn off all of machine, except the observed machine.
Set, the total loading does not exceed the machine capacity
that will be observed.
2) Running load flow and take the result of the active
power (P), the reactive power (Q), of the active power losses
(PL), the reactive power losses (QL).
P = 1.95 p.u PL = 0.00073 p.u
In addition, it also presented the results of real-time Fig.5. Plot of fitness value in 10 trial running
Q = 1.21 p.u QL = 0.00752 p.u
simulation using SIMULINK in Fig. 6 and 7, by comparing
the value of PSS parameters from GA, DE, and MDE.
Using the PSS optimum tuning parameter optimization
method MDE provide oscillation damping better than other
method. MDE has the fastest time to stabilize the system..
____________________________________________________________________________________________________________
IV. CONCLUSIONS
In this paper, the design of MDEPSS is applied in Single
Machine Infinite Bus (SMIB). This SMIB is the result of
Multi-machine network reduction that uses q concept
method, which is the simplest method in network reduction.
By doing this, the equivalent impedance for dynamic
stability analysis on SMIB would be found. Then, this SMIB
is installed by a PSS which its parameter is tuned with three
kind of algorithm, they are; (i) GA, (ii) DE, (iii) MDE.
After several simulations, based on the analysis of
(a) eigenvalue with multiobjective fitness function, using the
PSS optimum tuning parameter optimization method MDE
provide oscillation damping better than GAPSS, and DEPSS
method.
ACKNOWLEDGMENT
The authors acknowledgment the support of Power
System Simulation Laboratory, Institute Teknologi Sepuluh
Nopember (ITS) Surabaya, Indonesia.
REFERENCES
[1] A. Soeprijanto, Desain Kontroller untuk Kestabilan Dinamik Sistem
Tenaga Listrik. Surabaya : ITS Press, 2012.
(b) [2] F. Milano and K. Srivastava, Dynamic REI equivalents for short
circuit and transient stability analyses, in Electric Power System
Fig. 6. (a) Variation of rotor angle with optimized PSS parameters when Research, vol 79, 2009, pp. 878-887.
0.01 p.u disturbance is applied at 5s. (b) Variation of rotor angle with [3] Rusilawati, A. Soeprijanto, O. Panangsang, and D.F. Uman,
optimized PSS parameters when 0.03 p.u disturbance is applied at 10s. Determination of generator steady state stability limit using losses
concept and RBFNN, in Jurnal Teknologi, Johor Bahru: UTM Press,
2015.
[4] H. E. Mostafa, M. A. El-Sharkawy, A. A. Emary, K. Yassin, Design
and allocation of power system stabilizers using particle swarm
optimization technique for an interconnected power system, in
Electrical Power and Energy System, vol 34, 2012, pp. 57-65.
[5] Z. Sun, N.Wang, D. Srinivasan, Y. Bi, Optimal tunning of type-2 fuzzy
logic power system stabilizer based on diffetential evolution
algorithm, in Electrical Power and Energy System, vol 62, 2014, pp.
19-28.
[6] F. Mayouf A., F. Djahli, AMayouf, T. Devers, A coordinated genetic
based on type-2 fuzzy stababilizer for conventional and
superconducting generator in Electrical Power and Energy System,
vol 129, 2015, pp. 51-61.
(a) [7] E. M. Mountes, J. V. Reyes, Carlos A. C. C., Modified Differential
Evolution of Constrained Optimization, IEEE Congress on
Evolutionery Computation, Canada, 2006
[8] H Alkhatib and J Duveau, Dynamic genetic algorithms for robust
design of Multi-machine power system stabilizers, in Electrical Power
and Energy System, vol 45, 2013, pp. 242-251.
[9] N.N. Islam, M. A. Hannan, H. Shareef, A. Mohamed, M. A. Salam,
Comparative Study of Popular Objective Functions for Damping
Power System Oscillations in Multi-machine System, in Hindawi The
Scientific World Journal, 2014.
[10] WG Heffron and RA Phillips, Effect of a modern amplidyne voltage
regulator on underexcited operation of large turbine generators,
AIEE Trans. vol. 71, 1952, pp. 692-697.
[11] D Mello and C Concordia, Concepts of synchronous machine stability
(b) as affected by excitation control, IEEE Trans. 1969, pp. 316-329.
Fig. 7.(a) Variation of rotor speed with optimized PSS parameters [12] MA Abido, Robust design of power system stabilizers for Multi-
when 0.01 p.u disturbance is applied at t 5s. (b) Variation of rotor machine power systems using differential evolution, in Computational
speed with optimized PSS parameters when 0.03 p.u disturbance is Intelligence in Power Enegineering, Berlin : Springer, 2010, pp. 1-18.
applied at t 10s.
____________________________________________________________________________________________________________
Flight Control System Improvement for Quadrotor

Using Particle Swarm PID (PS-PID)
Andi Adriansyah, Anwar Minarso, Eko Ihsanto
Department of Electrical Engineering, Faculty of Engineering, Universitas Mercu Buana
Meruya Selatan, Kembangan, Jakarta 11650, Indonesia
andi@mercubuana.ac.id, anwar.minarso@gmail.com, eko.ihsanto@mercubuana.ac.id
Abstract— The rapid improvement of microprocessor, using Fuzzy Logic [5], Genetic Algorithm [6] or Ant Bee
electrical, sensors and advanced control technology make a Colony [7].
quadrotor fast expansion. Unfortunately, a quadrotor is unstable This paper addresses the problems of flight control system
and impossible to fly in full open loop system. Several using Particle Swarm Optimization (PSO). The using of PSO
methodologies have been proposed to control flight control is to tune PID parameters. The rest of this paper is organized
system for a quadrotor, such as PID Controller. However, for
as follows: First, a brief description of the quadrotor is
effective implementation of PID, parameters adjustment or
tuning procedure is required. The paper proposed a flight provided. Then, hardware design of quadrotor is described and
controller system based on a PID controller that the parameters followed by flight control design. Section 3 gives some
are tuned using Particle Swarm Optimization. Several experiment results. Finally, the paper is concluded in Section
experiments have been performed. According to these 4.
experiments the proposed system able to generate optimal and
reliable PID parameters for robust flight controller. II. METHODOLOGY
Keywords— Quadrotor, Flight Control System, PID, PSO

A. Quadrotor Model
Quadrotor movement mechanism is a resultant of propeller
I. INTRODUCTION rotation speed ( 1, 2, 3 and 4). This composition will
A quadrotor aircraft can take off and land in limited spaces, generate force on each rotor (F1, F2, F3 and F4) that affecting
even hover over a target and fly through narrow space and motion of the quadrotor body, where θ is the pitch angle, ϕ is
stay in limited-speed motion object easily. With these the roll angle and is the yaw angle in x, y and z axes,
advantages, the quadrotor has received a strong of attention in respectively. Coordinates and motion principle of quadrotor is
the last decade. However, a quadrotor is unstable and depicted in Fig. 1. Equ. (1) is a total lifting force to be
impossible to fly in full open loop system. Improved generated by the motor in order quadrotor can hover. Equ. (2)
performance expected from the new generation of quadrotor is and (3) also shows quadrotor condition to lift up or down,
possible through derivation and implementation of specific respectively [8].
control techniques incorporating limitations related to sensors
and actuators. Different methodologies have been proposed to
control flight control system for a quadrotor. The following
approaches have been used are PID control, backstepping
control, sliding mode control, linear quadratic regulator (LQR)
control, fuzzy logic (FL) control, neural network and hybrid
of them [1].
PID has become one of the selected solutions for many
practical control designs such as electronics devices, robotics,
and chemical process. There are some researches have been
performed to design flight control using PID [2-4]. PID
controller is used here in order to get adequate performances
of quadrotor such as fast response, zero steady-state error, and
minimum overshoot/undershoot.
However, for effective implementation of PID, parameters
adjustment or tuning procedure is required. The tuning
process, whereby the optimum values for the controller
Fig. 1 Quadrotor Model and Movement Principles [4]
parameters are obtained is a critical challenge. Many studies
were conducted to find the best way for tuning PID (1)
parameters in order to get satisfactory performances such as (2)
(3)
____________________________________________________________________________________________________________
B. Attitude Heading Reference System (AHRS) distance to the center of mass quadrotor. Wireless
An important component of quadrotor is an Attitude communications is one of the challenges. Several factors are
Heading Reference System (AHRS). AHRS is a 3 -axis sensor taken into consideration such as power consumption, weight,
system that provides three-dimensional position information transmission speed and reliability. Modules XBee Pro 900 HP
orientation (yaw, pitch and roll) in real time that provides data is one of the modules suitable to support these criteria. XBee
orientation. AHRS consists of a magnetometer, accelerometer Module Explorer is used to simplify the circuit and
and gyroscope on all three axes [9]. These sensors create connectivity with a microcontroller or computer.
inertial sensor system that can fully measure the attitude of Drotek 10 DOF IMU is used that contains accelerometer,
objects in 3D space. gyroscope, magnetometer and barometer. The sensor system
Quadrotor dynamic movement used state equation and contains two chips, MPU9150 that contains accelerometer,
observer equation based on quaternion. Quaternions are gyroscope, magnetometer and MS5611-01BA that contain
mathematically denoted as in Equ. (4) and (5), barometer. A DMP system is mounted in this unit. Sensor unit
where q0, q1, q2, and q3 are all real numbers. A quaternion is processed data based on IMU results which is named as
capable of describing that rotation just right [10]: AHRS. The results are angular position ( , θ, ϕ), rotational
speed, acceleration and altitude. Then AHRS data will be used
(4) in flight controller based on PID schema.
(5)
D. Flight Control Design
Finally, calculation of the attitude angle based on quaternion
is: Flight Control function is to control the quadrotor from
ground control based on AHRS data, such as maintaining the
(6) angular position, changing them or maintaining the altitude of
quadrotor. Outputs from flight controller then are distributed
to BLDC motor, respectively. Two modes of flight controller
are designed as acrobatic mode and angle mode.
In order to process the data sensor in quaternion format, In this research, Flight Control is designed using PS-PID
Digital Motion Processor (DMP) is used. DMP is a Controller. Basically, PS-PID is a PID controller augmented
technology planted on inertial sensor chip that is intended to by a tuning or learning process based on PSO. In PS-PID,
filter the data and process complex calculations quickly [11]. PSO is applied in order to search for PID parameters for a
C. Quadrotor Design particular problem and to ensure those parameter values are
optimal with respect to the design criteria. PSO is one of
Typical quadrotor system contains an UAV quadrotor, a evolutionary computation technique developed by Kennedy
ground control and a communication system. Fig. 2 shows and Eberhart in 1995 [12-13].
block diagram of the system are developed. The PID parameters constitute the optimization space,
which is then transformed into suitable position on which the
search process operates. Fig. 3 shows the concept of a PS-PID
system where PSO design and PID processing are the two
fundamental constituents.
Fig. 2 Block Diagram of Quadrotor System
The UAV system design consists the following

component: the brushless dc motor (BLDC), the propeller, the
Electronic Speed Controller (ESC) and the battery. The BLDC
motors used are Motor SunnySky X224S-16 with 2300kV and
need 11.v Volt. ESC EMAX12 A is used for speed controller Fig. 3 The concept of a PS-PID
that has PWM input signal and internal Battery Eliminator As the PSO deals with coded parameters, all parameters
Circuit (BEC) with 5V/1A rate. Six propeller HQ-PROP is that need to be tuned or learned must be encoded into a finite
used with 6 inch long and 3 inch pitch. A microcontroller length of string. The encoded strings are concatenated to form
system based on Arduino Pro mini with ATMega328 used to a complete particle. In order to learn PID parameter for yaw,
control all quadrotor system. The quadrotor is designed in X- roll and pitch control, each parameter is encoded into integer
frame type of FPV250 with diameter 25 cm and weight 109 g. codes that are based on series of Kp, Ki and Kd, respectively.
The placement of four motors on this frame has the same
____________________________________________________________________________________________________________
The coded parameters for each control system are arranged to TABLE I
FONT SIZES FOR PAPERS PID PARAMETERS
form particles of the population.
The PSO process starts with randomly generated initial Angular Position Kp Ki Kd
populations. Then, all populations of particles are evaluated Roll 0.75 0.0205 0.02343
iteratively use: Pitch 0.85 0.0125 0.02500
Yaw 0.75 0.0205 0.02343
vik 1  vik  c1 * rand (.) * ( pbest  sik )  c2 * rand (.) * ( gbest  sik ) (7) (10)
k 1 k 1
s i  s v
k
i i
(8) In order to test the performance (11) of controller, some
responses systems are shown. Other results that use PID for
where vik, vik+i , and sik , are velocity vector, modified velocity flight control tuned manually [3] are compared. Fig. 6 and Fig.
vector and positioning vector of particle i at generation k, 7 show response system for roll angular position after given
respectively. Then, pbest is the best position found by particle some disturbance.
i and gbest is the best position found by particle group. Finally,
c1 and c2 are cognitive and social coefficients, respectively.
In this work, a Sigmoid Decreasing Inertia Weight (SDIW) is
used to provide faster speed of convergence and better
accuracy of optimized value [13]. Consequently, PS-PID
would generate optimal and reliable PID parameters for robust
flight controller.

The whole of UAV quadrotor based on system design and Fig. 6 Response System for Roll Angular Position with PS-PID
GUI module are depicted in Fig. 4. A PC / Notebook
connected with XBee module and an Xbox joystick are used
input for flight control.
Fig. 7 Response System for Roll Angular Position with PID [3]
Fig. 4 A UAV Quadrotor System and GUI Module
The response system using PS-PID after given certain
disturbance has very short oscillation, fast response time in
Several experiments have been performed. Testing of
around 1 s and has minimum steady state error, as depicted in
robustness for PS-PID controller is using bench-test
Fig. 5. But, using PID that tuned manually, there is very long
equipment as shown in Fig. 5. Based on offline experiments of
oscillation, has more 7s response time and has zero steady
PS-PID, the optimum parameters of PID for flight control
state error. Fortunately, both systems have no overshoot.
system with each angular position respectively is listed in
Response systems of yaw angular position after given
Table 1.
some disturbance are shown in Fig. 8 and Fig. 9.
Fig. 8 Response System for Yaw Angular Position with PS-PID

Fig. 5 Bench Test Equipment
____________________________________________________________________________________________________________
called as Particle Swarm-PID (PS-PID). A Sigmoid

Decreasing Inertia Weight (SDIW) is used to provide faster
speed of convergence and better accuracy of optimized value
in PSO process. The UAV system design consists of a frame,
motors and propellers as actuators, IMU unit as AHRS data
source, microcontroller unit as a control module and wireless
unit as supporting communication with ground control. The
Digital Motion Processor (DMP) technology is used for data
sensor calculation. Several experiments have been performed.
A system based on manual tuned PID is used for comparison.
Fig. 9 Response System for Yaw Angular Position with PID [3]
According to these experiments it can be said that the
Fig. 8 shows that the response system using PS-PID has very proposed system able to generate optimal and reliable PID
short oscillation, fast response time in less than 1 s, has parameters for robust flight controller.
minimum steady state error and has no overshoot. In another
side Fig. 9 shows that using PID that tuned manually, there is REFERENCES
some oscillation, has response time more than 10s, has zero [1] A. Zulu, and Samuel John. “A Review Of Control Algorithms For
steady state error and has overshoot. Autonomous Quadrotors”. Open Journal Of Applied Sciences. Vol. 4:
pp. 547-556, 2014.
The last experiments are to see response systems of ptich [2] G. Szafranski, and R. Czyba. “Different Approaches of PID Control
angular position as shown in Fig. 10 and Fig. 11. UAV Type Quadrotor”, Proceedings of the International Micro Air
Vehicles conference, pp . 70-75, 2011.
[3] A. L. Salih, L., M. Moghavvemi, Haider A. F. Mohamed and Khalaf
Sallom Gaeid. “Flight PID controller design for a UAV Quadrotor”.
Scientific Research and Essays. Vol. 5(23), pp. 3660-3667, December,
2010.
[4] B. Kada, and Y. Ghazzawi. “Robust PID Controller Design for an
UAV Flight Control System”, World Congress on Engineering and
Computer Science (WCECS), Vol II, pp. 1-6, 2011.
[5] D. Zhou, Qingbo Geng. “Multi-model And Fuzzy PID Control for
Fixed-wing UAV”, 3rd International Conference on Mechatronics,
Robotics and Automation (ICMRA 2015), pp. 523-528, 2015.
[6] L. Feng, Yuxi Wang, Xiaoguang Qu. “The small UAV longitudinal
control law design based on genetic algorithms”, 2nd International
Fig. 10 Response System for Yaw Angular Position with PS-PID Conference on Electrical, Computer Engineering and Electronics
(ICECEE 2015), pp. 253-258, 2015.
[7] L. Feng, Xiaotong, Wang, Xiaoguang, Qu, “PID parameters tuning of
UAV flight control system based on artificial bee colony algorithm”,
2nd International Conference on Electrical, Computer Engineering
and Electronics (ICECEE 2015), pp. 248-252, 2015.
[8] G. Navajas, Tuta and S. Roa Prada, “Building Your Own Quadrotor: A
Mechatronics System Design Case Study”, International Conference
on Engineering Mechatronics and Automation. pp. 1-5, 2014.
[9] Q. Jiang, Zeng,Y., Liu, Q. and Jing, H, “Attitude and Heading
Reference System for Quadrating Based on MEMS Sensor”, 2012 2nd
International Conference on Instrumentation & Measurement,
Computer, Communication and Control. pp. 1090-1093, 2012.
Fig. 11 Response System for Yaw Angular Position with PID [3]
[10] E. Fresk, “Full quaternion based attitude control for a quadrotor”, 2013
European Control Conference. pp. 3864-3869, 2013.
[11] A. Adriansyah, B. Sulle And A. Minarso.. Design of AHRS for
As experiments before, the response system using PS-PID has Quadrotor Control using Digital Motion Processor. Jurnal Teknologi.
very short oscillation, fast response time in less than 3 s, has 2016. (Accepted)
minimum steady state error and has no overshoot, as depicted [12] J. Kennedy, and Eberhart, R. C., “Particle Swarm Optimization”, IEEE
in Fig. 10. In another side Fig. 10 shows that using PID that Internationl Conference on Neural Neworks, IV, pp. 1942-1948, 1995.
[13] R. C. Eberhart, and Kennedy, J., “A new optimizer using particle
tuned manually, there is also short oscillation, has response swarm theory”, Sixth International Symposium on Micro Machine and
time around than 2s, has zero steady state error and has no Human Science, pp. 39-43, 1995.
overshoot. [14] A. Adriansyah, and S.H.M. Amin, “Analytical and Empirical Study of
Particle Swarm Optimization with a Sigmoid Decreasing Inertia
IV. CONCLUSIONS Weight,” in Proceedings of 1st Regional Postgraduate Conference on
Engineering and Science (RPCES 2006), pp. 247-252, 2006.
A flight controller system for an UAV quadrotor has been
designed. The system based on a PID controller that the
parameters are tuned using Particle Swarm Optimization that
____________________________________________________________________________________________________________
Optimum Operation Stand-Alone Microgrid Using

PSO Algorithm to Considering Life Time Battery
Vicky Andria Kusuma#1, Ontoseno Penangsang*2, Rony Seto Wibowo #3
#
Electrical Engineering, Institute of Technology Sepuluh November
Jl. Raya ITS, Keputih, Sukolilo, Kota SBY, Jawa Timur, Indonesia
1
vickyf2l@gmail.com
2
zenno_379@yahoo.com
3
ronyseto@ee.its.ac.id
Abstract— Stand-Alone microgrid is an Electrical system work well when grid connected in island mode or hybrid
consisting of several Distributed Generator and not connected mode [3].
with main grid system (Grid Utility). The system basically Stand-alone microgrid usually used and applied on the
designed not connected to the main grid, because the limitations isolated island or where the main grid cannot reach. Microgrid
of grid connection or isolated island region. This study using
electrical system has two operation method, which is Stand
Photovoltaic, Wind Turbine, Diesel generators and batteries as
the energy sources. Battery has a value that serve as the state of Alone and Grid-Connected. An operation system stand-alone
charge (SOC), as an indicator of energy from battery. This microgrid used while the system is not connected main grid,
Operation Optimum research will maximize the lifetime of and an operation mode of grid connected used when main grid
battery by contrasting the cost of generation. Method Particle connected system. This stand-alone operation Distributed
Swarm Optimization (PSO) will be used to maximize the life of generator will be operating solo when supply load system.
the battery. This operation mode aiming stability of the main system
Keywords— Battery; Energy; Microgrid; Renewable; Particle frequency and voltage.
Swarm Optimization (PSO);
I. INTRODUCTION
The Microgrid is important technology as combined
renewable energy generators like Photovoltaic (PV), Wind
Turbines (WT), Diesel Generator (DG), and also Battery for
stage media [1]. The battery on Microgrid system as a reserve
and also backup energy, cause renewable energy cannot
supply load all times. An energy storage system the most
important things in microgrids, and also needs special
treatment to ensure efficiency, stability, safety, and reliability
of the whole system [2].
Battery energy storage system has a short time life
characteristic and also the cost of the battery is expensive.
Battery optimization will observe characteristic of the life
Fig. 1 Schematic diagram Stand-Alone Microgrid
time battery because of optimization extensively pretty hard
for research. Energy Management System (EMS) will be
applied as control of the energy optimization on the TABLE I
microgrid. This research will use IEEE 30 bus modified as DATA PARAMETER
the model of optimization, which covers losses of the life Lead Acid
Name PV WT Diesel
time battery cost, maintenance cost, fuel cost, for receive Battery
optimal parameter operation. Type 83 W 840 W 175 kW 2V/1000 Ah
Quantity 1000 500 2 1000
II. BACKGROUND
Capacity 83 kW 420 kW 350 kW 1000 KAh
A. Stand-Alone Microgrid
A microgrid is the system of electric power system. The
generator of microgrid around 1 kW – 50 kW using renewable
energy for example sunlight, the wind, and water flow. This
system consist of DG, Energy storage, and Load which can
____________________________________________________________________________________________________________
B. Problem Formulation Pbat-t is the battery power between t and t+∆t; Cbat is
1) PV Model: Power output of PV can be calculated by capacity of battery.
output power rate standard test condition, datasheet,
temperature, and irradiance [4] C. Constraints
Generation power should be balance with the load, Pbat is
GC
Ppv  PSTC
GSTC
 1  k  Tc  TSTC   (1) the power output of the batteries, and microgrid system should
be also following the rule of balance load demand.
PPV is power output of the PV, PSTC is the maximum Pload  Pgen  Pout  pv  Pout  wt  Pbattery (7)
module output (W), GC is the actual irradiance (W), GSTC is
solar irradiance 1000 W/m2, k is the coefficient temperature
module, TC is Cell temperature in ºC, TSTC is the references III. PARTICLE SWARM OPTIMIZATION ALGORITHM
temperature (25º). Particle Swarm Optimization (PSO) is algorithm like group
of a bird that each individual effect to another one. On this
2) WT Model: Wind turbine output can be modelling with research will finish multi-objective wherever each generator
following [4] assumed a bird or particle. While the fitness value of multi-
objective described as position with the most amount of food
 Pwt  0, vac  vci [6], applied PSO to economic dispatch problem in general can
 be defined as
 Pwt  avac  bvac  c, vci  vac  vr (2)
2
 Pwt  130, vr  vac  vco xj  [ P1 j , P 2 j ... Pij ... Pnj ]T (8)


Where Pwt is the output of the wind turbine (W), Vci is Where Pij is an output of generator i for particle j .
speed of cut-in (m/s), Vco is speed of cut-out (m/s), Vr is a Generation random data will be still considered constraints.
rating of the wind speed (m/s) and Vac is the actual wind The relative importance of these two factors can vary from
speed (m/s). one decision to another and it is reasonable to apply random
weights to each part, therefore the velocity will determine by
3) Diesel Generator Model: DG is modelled as linear of
their actual output power v v uv uv
 i (t )   i (t  1)  1.random1.( p i   i (t  1))...
uv uv
F(Pdie )  a  bPdie  cPdie 2 (3)  2 .random2 .( p g   i (t  1)) (9)
Cost function generator obtained from test heat run, value of Where φ1, φ2 are two positive numbers and random1, random2
a, b, and c from the datasheet. are two random of numbers with uniform distribution [8].
Power obtained from random process then made power
4) Battery Model: Batteries joined as one or we knew as generating initialization on the power flow process. After
battery bank, saving strategy battery lifetime is State of power flow process running power generating in the slack
Charge (SOC), must be following the formula: bus need to review, whether they constraint. When power
generating out of constraint then power generating will be
SOCmin  SOC  SOCmax (4) add or subtracted until becoming to the constraint.
Speed calculation, generating cost, and battery life loss
Pcharge  max  Pbattery  Pdischarge  max (5) cost and fitness value from multi-objective. Weighted Sum
method combines several the objective function with weights.
Equations Weighted Sum method is as follows (10), where f
SOC min is the lowest indicator for discharge battery
is the objective function k. While wk is the weights of k
usage, SOC mean an average of battery charge, SOC max is
objective function.
the maximum limitation of battery charge. Value time of SOC
t+∆t; is determined by the value of SOC at time t and also
battery power during the time period. It can be expressed [5]: fitness  w1 f1 ( P )  w2 f 2 ( P )  ...  wk f k ( P) (10)
t Result obtained in this process then used for determine

SOC( t t )  SOCt  Pbat ,t  (6) best particle for generating power combination and global
Cbat best of fitness value.
____________________________________________________________________________________________________________
Speed update and particle position to the best particle. Each

speed and particle update will do running load flow and slack
bus checked will be still constrained.
Update fitness value, best particle, and global best Repeat
update step obtained maximum iteration value. This
simulation using the initial parameter: Particle = 30, Weight =
0,4, Iteration = 100, c1 = 2, c2 = 2.
IV. SIMULATION AND ANALYSIS OF DATA

Solution optimum operation of this stand-alone microgrid
shown in three study case. The first case is completion
optimum operation when SOC in full condition (SOC=1), Fig. 4 Microgrid System when Battery Empty (SOC=0)
second case is SOC in empty condition (SOC=0). The three
case study will be shown battery function as storage and also
supply when renewable energy the generator cannot back-up Comparison value of the simulation using the multi
all of the load. From two of study case, generator will work as objective function on this study case shown in Table 2. A
backup energy supply. Figure 2 shown Load and Renewable result from two different study case vastly different, if try to
energy on the plan of the system. Figure 3 shown optimization minimize losses of the battery when status is empty, the user
of the islanded microgrid use Renewable Energy, Battery, and will receive expensive of generation cost, because diesel
Diesel engine when battery fully charged. Figure 4 shown engine works almost many times. Seen from comparison
plan microgrid system when battery status is empty, diesel better if choose to optimization of generation cost because
will help battery charge for maximizing function and life time differential of generation cost is very high, and the battery
battery. will be kept fully charge.
TABLE III
RESULT OF RUNNING MICROGRID USING PSO ALGORITHM
Microgrid Generation
Objective
Function SOC = 1 SOC = 0
kW kVAR kW kVAR
1601.346
Generation 3039.8650 3542.5264 1694.760
7
1262.631
Load 3035.4170 3535.4170 1262.6310
0
Losses 4.5688 339.8873 7.2243 433.2935
Battery
Fig. 2 Load and Renewable Energy Curve 0.58 % 0.42 %
Losses
Generation
Rp 5,271,523.03 Rp 13,064,979.20
Cost
Life Loss
Rp 7,623.30 Rp 5,499.83
Cost
V. CONCLUSION
This Research showing result of the operation optimization
Islanded Microgrid using renewable energy source, battery,
and diesel generators. PSO algorithm used to find the best
solution of the multi-objective problems, life loss cost of
battery and generation cost. Simulation shown optimization
with minimization one objective function is inversely
proportional if the generation cost is minimized, lifetime
Fig. 3 Microgrid System when Battery fully Charged (SOC=1) battery will losses 0.58% and generation cost Rp
5,271,523.03. When battery life time gets optimization will
receive 0.42% and generation cost Rp 13,064,979.20.
____________________________________________________________________________________________________________
REFERENCES
[1] Zhixin Miao, LingXu, Vahid R. Disfani, Lingling Fan. "An SOC-Based
Battery Management System for Microgrids". IEEE Transactions on
Smart Grid, Vol. 5, No. 2, March 2014
[2] Ye Yang, HuiLi, Andreas Aichhorn, Jianping Zheng, and Michael
Greenleaf. "Sizing Strategy of Distributed Battery Storage System With
High Penetration of Photovoltaic for Voltage Regulation and Peak Load
Shaving". IEEE Transactions on Smart Grid, Vol. 5, No. 2, March 2014
[3] Emiliano Dall’Anese, HaoZhu, Georgios B. Gipsoakis, Fellow.
"Distributed Optimal power flow for Smart Microgrids". IEEE
Transactions on Smart Grid, Vol. 4, No. 3, September 2013
[4] Bo Zhao, Xuesong Zhang, Jian Chen, Caisheng Wang, dan Li Guo,
“Operation Optimization of Standalone Microgrids Considering
Lifetime Characteristics of Battery energy storage system”, IEEE
Transactions On Power Systems, Vol. 4, No. 4, October 2013
[5] Haihua Zhou, Tanmoy Bhattacharya, Duong Tran, Tuck Sing Terence
Siew, Ashwin M. Khambadkone, “Composite Energy Storage System
Involving Battery and Ultracapacitor With Dynamic Energy
Management in Microgrid Applications”, IEEE Transactions On Power
Electronics, Vol. 26, No. 3, March 2011
[6] M.F. Rahman, Penangsang. O., Soeprijanto .A., “Optimisasi
Pembebanan Pembangkit (economic dispatch) Pada sistem 500Kv Jawa-
Bali Menggunakan Particle Swarm Optimization dengan
Mempertimbangkan Kapasitas Transmisi”, Jurusan Teknik Elektro FTI-
ITS, Surabaya, 2009
[7] Noel Augustine, Sindhu Suresh, Prajakta Moghe, Kashif Sheikh,
“Economic Dispatch for a Microgrid Considering Renewable Energy
Cost Functions”, IEEE. 2011
[8] Yamille del Valle, Ganesh Kumar Venayagamoorthy, Salman
Mohagheghi, Jean-Carlos Hernandez, Ronald G. Harley, “Particle
Swarm Optimization: Basic Concepts, Variants and Applications in
Power Systems”, IEEE. 2008
____________________________________________________________________________________________________________
Response of microcantilever vibration due to CO

gas in vacuum condition
Lia Aprilia1,2,3, Ratno Nuryadi4, Djoko Hartanto3, Makoto Hosoda2, Yoichiro Neo1,2, Hidenori Mimura1,2*
1
2
Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, 432 8011, Japan
3
4
Center for Materials Technology, Agency for the Assessment and Application of Technology, South Tangerang, Indonesia
*mimura.hidenori@shizuoka.ac.jp
Abstract— We present a preliminary study of cantilever response profile (resonance frequency shift and Q-factor) for each
to CO gas in vacuum condition through its frequency profile. cantilever vibration.
Here, an active microcantilever which was formed by coating ZnO
rods on the microcantilever surface as sensitive layer and an EXPERIMENTAL DETAILS
uncoated microcantilever as a reference were used. The detail
analysis of the frequency profiles shows that resonance frequency A. ZnO rods formation
and Q-factor of both cantilevers decrease due to CO gas. The ZnO rods preparation consists of seed layer coating and
resonance frequency of ZnO-coated cantilever decreases with
growth of ZnO rods. A seed solution was made from 0.2 M zinc
increasing CO gas concentration while resonance frequency of
reference one after gas exposure does not change when the gas nitrate hexahydrate and monoethanolamine (MEA) in ethylene
concentration is varied. This preliminary study opens up a further glycol. The microcantilever was coated to the seed solution by
study for detection characteristic of microcantilever in vacuum dip coating technique. The coated cantilever was annealed at
condition. 100 ºC for 1 hour. Then, ZnO rod was grown on the cantilever
Keywords: Microcantilever, ZnO, CO, resonance frequency, vacuum surface by dipping in ZnO solution at 85 ºC for 2 hours. Here,
the ZnO solution was prep ared by using 0.02 M zinc nitrate
hexahydrate and hexamethylenetetramine (HMTA) in
INTRODUCTION deionized water. Next, the ZnO coated microcantilever is dried
The advent of atomic force microscopy (AFM) in 1986 [1] at 90 ºC for 20 minutes. We applied scanning electron
triggers research reports on the study of microcantilever as a microscopy (SEM) observation to check surface morphology
sensitive tool for chemicals and bioorganisms detections of the ZnO.
because of the capability to detect ultrasmall mass. The smallest
B. Electronic circuit
mass detection using the microcantilever has been reported in
femtogram (10-15 gram) level by Sone et al. [2], attogram level The whole system of microcantilever vibration
(10-18 gram) [3] and zeptogram level (10-20 gram) by Roukes measurement is shown in Fig. 1. The system based on direct
group [4]. digital synthesizer (DDS) was constructed under control by a
For sensor application, a sensitive layer such as polymer [5], microcontroller. The microcontroller gets a command and
metal oxide sensitive layers [6], etc. which determines sensor setting frequency data from PC. This set data corresponding to
selectivity must be coated on the microcantilever surface. frequency is sent to the DDS. Then, the sinusoidal wave output
Generally, gas detection on metal oxide surface, such as zinc is amplified by an OP-AMP to gain a sufficient amplitude for
oxide (ZnO) which has a great potential in sensing application, piezoelectric device. By activating piezoelectric, the
is especially influenced by a presence of humidity [7]. microcantilever will vibrate in a certain frequency.
Therefore, the characteristic of sensor response should be
considered in various conditions, such as high humidity or
vacuum condition. In previous experiment, we have conducted
gas detection using cantilever coated by ZnO in high humidity
condition [8]. The result showed that the gas adsorption
mechanism is influenced by water vapor.
In this work, we study response of ZnO-coated cantilever
and uncoated one to carbon monoxide (CO) gas in vacuum Fig. 1. Schematic circuit system using piezoresistive microcantilever.
condition. The coated cantilever is formed by ZnO rods Here, a commercial piezoresistive microcantilever (Seiko
sensitive layer coated on microcantilever surface. We Instrument Inc.) as shown in Fig. 2 was used. A long
investigate the difference of signal response through frequency microcantilever has a length of 400 µm, a width of 50 µm,
spring constant (k) of 4 N/m [2]. The short cantilever a
____________________________________________________________________________________________________________
reference cantilever, which is crucial in order to reduce

background noise, such as thermal drift and gas turbulence [9].
In piezoresistive microcantilever, the deflection of
microcantilever due to the gas produces the resistivity change
of the piezoresistor. Such resistivity change was measured by a
Wheatstone bridge and amplified by op-amp. Using analog to
digital converter and microcontroller, the amplified wheatstone
bridge output could be monitored by PC in the form of
frequency profile.
Fig. 3. Scanning electron micrograph of ZnO which was coated on

microcantilever surface.
B. Frequency profiles of microcantilevers

Frequency profiles of coated cantilever in vacuum condition
and CO exposure are shown in Fig. 4. We monitored frequency
profiles of the microcantilever coated by zinc oxide at vacuum
level of 1.3, 10, 12.5, 24.5, 93.6, 141, and 168 Pa. The CO gas
was flowed and the frequency profile was measured at pressure
of 378, 878, and 1400 Pa. From the graph, we can see that the
Fig. 2. Scanning electron microscopy (SEM) image of piezoresistive amplitudes of waves significantly decrease due to CO
microcantilever.
exposure.
C. Carbon monoxide gas measurement
Here, each microcantilever, i.e. coated and uncoated
microcantilevers were not measured at the same time. First, we
measured the resonance frequency of the microcantilever
before applying vacuum condition (non-vacuum). The
resonance frequency of coated and uncoated cantilevers was
27.020 kHz and 36.260 kHz, respectively. Next, we conducted
the measurement in medium vacuum condition of 0 – 168 and
recorded the frequency profiles of the microcantilevers.
Finally, the carbon monoxide (CO) gas was flowed into the
chamber until the vacuum pressure increased to be 1450 Pa.
During turn on the gas, the frequency profiles of each
microcantilever were monitored at several points.
Fig. 4. Frequency profiles of the microcantilever coated by ZnO.
RESULTS AND DISCUSSIONS
Figure 5 shows frequency profiles of uncoated cantilever.
A. Characterization of zinc oxide rods We monitored the frequency profiles at vacuum of 0.564, 3.24,
Figure 3 shows the SEM image of ZnO rods coated on 10.3, 21.5, 23.5, 41.2, and 53.1 Pa. The frequency profile was
microcantilever surface. ZnO microrods combination with measured in CO exposure condition at pressure of 266, 623,
microtubes structure are found. Even the all formed rods are not 999, 1320, and 1450 Pa. This graph also shows that the
well aligned, we can see the hexagonal shape of the rods. It is amplitudes significantly decrease due to CO exposure. Here,
predicted that the crystal structures of fabricated ZnO is the resonance frequency of the microcantilever in vacuum
hexagonal wurtzite. The length and the diameter of ZnO rods condition is higher than that of one in non-vacuum condition. It
are approximately 2 µm and 250 nm, respectively. is predicted that in non-vacuum condition, the resonance
frequencies of both microcantilevers are influenced by water
vapor. The detail of resonance frequency shift is explained in
next section.
____________________________________________________________________________________________________________
Fig. 5. Frequency profiles of uncoated microcantilever Fig. 7. Resonance frequency change of the uncoated microcantilever due to
CO gas (266 – 1450 Pa).
Eventhough the both cantilevers show the resonance
C. Resonance frequency shift of microcantilevers frequency shift due to CO gas, profile of the frequency shift is
Resonance frequency of coated cantilever during quite different. The coated cantilever shows that the change of
measurement in vacuum and CO gas exposure can be seen in resonance frequency is higher at higher gas concentration
Fig. 6. The resonance frequency decreases from 27.200 kHz to (higher pressure value) whereas the frequency shift of the
27195 kHz when CO gas is inserted until pressure of 378 and uncoated cantilever is constant at various CO gas concentration
878 Pa. At pressure of 1400 Pa, the resonance frequency (various pressure value). It may exhibit the role of zinc oxide
decreases to be 27190 kHz. The higher pressure is related to as sensitive layer of the microcantilever. However, the
more CO molecules inside the chamber. Therefore, the higher sensitivity should be increased through optimization of
frequency decrease at higher pressure can be caused by more sensitive layer thickness [10].
gas molecules adsorbed on zinc oxide layer.
D. Q-factor of microcantilevers
Q-factor indicates energy losses within a oscillator. The
energy loss of a vibrating microcantilever is dominated by
attachment loss, surface loss, thermoelectric loss, and air
damping [11]. The Q-factor is defined by resonant frequency
peak (fp) divided by 3-dB bandwidth of response (f-3dB). In this
case, f-3dB is corresponded to 0.707 of the highest peak response.
The equation of Q-factor is shown as follows [12].
Q factor = fp (1)
f-3DB
From frequency profile at Fig. 4 and 5, we calculated Q-
factor of each microcantilever according Eq. 1. The calculated
Q-factor for coated microcantilever is shown at Fig. 8. We can
Fig. 6. Resonance frequency change of the coated microcantilever due to CO
see that the Q-factor value decreases from 1432 at 1.3 Pa to
gas (378 – 1400 Pa). stable value of 1046 at 93.6, 141, and 168 Pa. When CO gas
was introduced, Q-factor decreases to be 800 at 378 Pa, 632 at
Figure 7 shows resonance frequency of uncoated 878 Pa, and 604 at 1400 Pa. Higher gas concentration (higher
microcantilever in vacuum condition and CO exposure. We can pressure) provides lower in Q-factor.
see that the resonance frequency is stable at 36.345 kHz before
gas exposure. When CO gas is introduced (from pressure of 266
– 1450 Pa), the resonance frequency decreases to be 36.335
kHz. Such frequency shift is predicted due to CO gas adsorbed
on microcantilever surface.
____________________________________________________________________________________________________________
REFERENCES
[1] G. Binnig, C.F. Quate, C. Gerber, Atomic force microscope, Phys. Rev.
Lett. 56 (1986) 930.
[2] H. Sone, A. Ikeuchi, T. Izumi, H. Okano and S. Hosaka, “Femtogram mass
biosensor using self-sensing cantilever for allergy check,” Jpn. J. Appl.
Phys., vol. 45, no. 3B, pp. 2301-2304, 2006.
[3] M. Li, H. X. Tang, and M. L. Roukes, “Ultra-sensitive NEMS-based
cantilevers for sensing, scanned probe and very high-frequency
applications,” Nat. Nanotech., vol. 2, pp. 114-120, Jan. 2007.
[4] Y. T. Yang, C. Callegari, X. L. Feng, K. L. Ekinci, and M. L. Roukes,
“Zeptogram-scale nanomechanical mass sensing,” Nano Lett., vol. 6 (4),
pp. 583-586, Mar. 2006.
[5] C. Steffens, F. L. Leite, A. Manzoli, R. D. Sandovall, O. Fatibello, and P.
S. Herrmann, “Microcantilever sensors coated with a sensitive polyaniline
layer for detecting volatile organic compounds,” J. Nanosci. Nanotechnol.,
vol. 14 (9), pp. 6718-22, Sep. 2014.
[6] H. Yamagiwa, S. Sato, T. Fukawa, T. Ikehara, R. Maeda, T. Mihara, and
M. Kimura, “Detection of volatile organic compounds by weight-
Fig. 8. Q-Factor of the microcantilever coated by ZnO. detectable sensors coated with metal-organic frameworks,” Sci. Rep., vol.
4, pp. 6247, Sept. 2014.
Figure 9 shows the calculated Q-factor for uncoated [7] H. Xu, W. Fan, A. L. Rosa, R. Q. Zhang, and Th. Frauenheim, "Hydrogen
cantilever. The graph shows increasing Q-factor before and oxygen adsorption on ZnO nanowires: A first-principles study," Phys.
introducing the CO gas (from 1454 at 0.564 Pa to 1817 at 53.1 Rev. B, vol. 79, pp. 073402, Feb. 2009.
Pa). Next, CO gas is introduced to be 266 – 1450 Pa resulting [8] L. Aprilia, R. Nuryadi, D. Gustiono, D. Hartanto, B. Yuliarto, Y. Neo, H.
Mimura, “Resonance Frequency Shift due to CO Gas using ZnO-
in decreasing Q-factor to be 1253 - 757. functionalized Microcantilever,” IEEE Trans. Nanotechnol. 2016
(submitted).
[9] H. P. Lang, R. Berger, C. Andreoli, J. Brugger, M. Despont, P. Vettiger, F.
Battiston, J. P. Ramseyer, E. Meyer, T. Mezzacasa, L. Scandella, H. J.
Guntherodt, Ch. Gerber, and J. K. Gimzewski, “A chemical sensor based
on a micromechanical cantilever array for the identification of gases and
vapors,”Appl. Phys. A., vol. 66, pp. S61-S64, Mar. 1998.
[10] L. Aprilia, R. Nuryadi, D. Hartanto, “Sensitive Layer Thickness
Dependence on Microcantilever Sensor Sensitvity,” Adv. Mat. Res. vol.
789, pp. 219-224, 2013.
[11] N. Gao, D. Zhao, R. Jia, and D. Liu, “ Microcantilever actuation by laser
induced photoacoustic waves,” Sci. Rep. vol. 19935, Jan. 2016.
[12] A. Kolpekwar, R. D. S. Blanton, D. Woodilla, “ Failure modes for stiction
in surface-micromachined MEMS.” International Test Conference, pp.
551 – 556, 1998.
Fig 9. Q-Factor of the microcantilever uncoated by ZnO.
Both of cantilever signals shows that Q-factor value before

gas exposure is higher than after one. It indicates that the
quality factor in CO exposure is dominated by the viscous
losses for both cantilever. The difference of Q-factor profiles
between coated and uncoated cantilever may be due to ZnO
layer on microcantilever surface. Detail analysis of Q-factor is
under studied.
CONCLUSIONS
We have grown zinc oxide (ZnO) rods on microcantilever
surface and studied resonance frequency profiles of the
microcantilever in vacuum condition. The detail analysis of the
frequency profiles shows that resonance frequency and Q-
factor of both cantilevers decrease due to CO gas. The
resonance frequency shift of coated cantilever increases due to
higher gas concentration whereas the frequency change of the
uncoated cantilever is constant at various CO concentration.
Such preliminary study opens up a further study for detection
characteristic of microcantilever in vacuum condition.
____________________________________________________________________________________________________________
6FDQ&DUW$QGURLG%DUFRGH6FDQQHUIRU
(&RPPHUFH6WRFN,QYHQWRU\
0X]DPPLO0XG]DNLU6KDULIDK6DRQ$EG.DGLU0DKDPDG

)DFXOW\RI(OHFWULFDODQG(OHFWURQLF(QJLQHHULQJ8QLYHUVLWL7XQ+XVVHLQ2QQ0DOD\VLD
3DULW5DMD0DOD\VLD

PX]DPPLOPXG]DNLU#JPDLOFRP
VKDULID#XWKPHGXP\

NDGLU#XWKPHGXP\

$EVWUDFW² 7KH XVDJH RI $QGURLG DSSOLFDWLRQ LV LQFUHDVLQJ ZLWK
WKHLQFUHPHQWRIVPDUWSKRQHSHU\HDUVLQFHXVHUSUHIHULQWHUQHW ,, $1'52,'%$5&2'(6&$11(5)25(&200(5&(
WREHRQHRIWKHLPSRUWDQWLQIUDVWUXFWXUHLQWKHLUGDLO\OLIH7KHUH %DUFRGH LV D PDFKLQH UHDGDEOH UHSUHVHQWDWLRQ RI
DUH PDQ\ DSSOLFDWLRQV IRU $QGURLG WKDW KDV EHHQ UHOHDVHG DQG
LQIRUPDWLRQ LQ D YLVXDO IRUPDW RQ D VXUIDFH DQG LW FRQVLVW RI
VRPH RI WKHP JLYH JUHDW IHHGEDFN LQ WHUP RI IXQFWLRQDOLW\ DQG
HDV\ WR XVHG )XUWKHUPRUH LW DOVR YHU\ VXVWDLQ LQ RUGHU WR
SULQWHGSDUDOOHOOLQHV>@7KHGHYHORSPHQWRIEDUFRGHV\VWHP
LQFUHDVHWKHSURGXFWLYLW\7KHUHIRUHWKLVLVDJUHDWRSSRUWXQLW\ ZDV HVWDEOLVKHG D ORQJ WLPH DJR ,W EHJDQ ZLWK WKH GHVLJQ RI
WRFUHDWHEXVLQHVVEDVHGPRELOHDSSOLFDWLRQ7KLVSURMHFWDLPVWR WKHSXQFKFDUGV\VWHPE\:DOODFH)OLQWLQ+RZHYHUWKH
GHYHORS DQ LQWHJUDWHG $QGURLG EDUFRGH VFDQQHU ZLWK H SUREOHPV RFFXU WKDW WKH FDUG UHDGLQJ HTXLSPHQW RI WKH GD\
FRPPHUFHV\VWHP7KLVDSSOLFDWLRQDOORZVXVHUWRVFDQDEDUFRGH ZDVEXON\DQGXWWHUO\XQZLHOG\>@7KHILUVWEDUFRGHV\VWHP
DQGXSGDWHTXDQWLW\RUFUHDWHDQRUGHUE\XVLQJPRELOHFDPHUD XVLQJJORZXQGHUXOWUDYLROHWOLJKWE\%HUQDUG6LOYHULVLQ
7KLV ZLOO EH H[FHVVLYH IRU UHGXFLQJ WKH QXPEHU RI SK\VLFDO >@ +RZHYHU WKLV V\VWHP HQFRXQWHUHG SUREOHPV ZKLFK
GHYLFHV 3HUIRUPDQFHV RI GHYHORSHG $QGURLG DSSOLFDWLRQ DUH UDQJLQJ IURP LQN LQVWDELOLW\ WR SULQWLQJ FRVWV ,Q
HYDOXDWHGEDVHGRQVXUYH\E\EXVLQHVVDQGLQGXVWULDOSDUWQHUV
8QLYHUVDO 3URGXFW &RGH 83& KDV EHHQ ZLGHVSUHDG XVH RI
EDUFRGHV EHFDXVH LW GHYHORSHG E\ WKH XVHU FRPPXQLW\ DQG LW
.H\ZRUGV²$QGURLG%DUFRGH6FDQQHU(FRPPHUFH6WRFN FRQWLQXH WR GHYHORS (XURSHDQ $UWLFOH 1XPEHULQJ 6\VWHP
,QYHQWRU\ ($1WKDWZLGHO\XVHWRGD\>@0HDQZKLOHLQ :DQJ
DQG KLV FROOHJHV GLVFRYHUHG ' EDUFRGH UHDGLQJ E\ XVLQJ
, ,1752'8&7,21 FDPHUDSKRQHV>@
6PDUWSKRQH KDV EHFRPH RQH RI WKH PRVW LPSRUWDQW 7KHUHDUHSOHQW\W\SHVRIEDUFRGHIRU'VXFKDV(XURSHDQ
HOHFWURQLFGHYLFHVQRZDGD\V:LWKWHFKQRORJ\RILQWHUQHWKDV $UWLFOH 1XPEHULQJ ($1 >@ ,QWHUQDWLRQDO 6WDQGDUG
UDSLGO\ HYROYLQJ WKH IXQFWLRQDOLW\ RI WKH VPDUWSKRQH KDV %RRN 1XPEHU ,6%1&RGH >@ DQG 8QLYHUVDO 3URGXFW &RGH
EHFRPH PRUH FRQYHQLHQFH WR WKH VRFLHW\ GXH WR ORW RI 83&>@7KLVZRUNZLOOEHIRFXVLQJRQ($1EHFDXVHLW
DSSOLFDWLRQV WKDW PD\ KHOS SHRSOH GDLO\ OLIH VXFK DV VXUILQJ FRQVLVWV RI QXPEHUV RQO\ DQG LW XVHG IRU UHWDLO SURGXFW
LQWHUQHWPHVVDJLQJDQGDOVREXVLQHVVPDQDJHPHQW>@ PDUNLQJ ZRUOGZLGH )LJXUH VKRZV D W\SLFDO ($1
2QOLQH EXVLQHVV LV RQH RI WKH HOHFWURQLF FRPPHUFH %DUFRGH
HFRPPHUFHWKDWVHOOLQJWKHJRRGVRUVHUYLFHV6HYHUDOLVVXHV
PD\FDXVH GHOD\LQSXUFKDVLQJSURFHVVLQFOXGLQJVHSDUDWHRU
XQV\QFKURQLVLQJ VWRFNV TXDQWLWLHV EHWZHHQ RQOLQH DQG UHWDLO
DW WKH VDPH WLPH LW¶V DOVR FDXVH WKH GLIILFXOW\ LQ PRQLWRU DQG
XSGDWH WKH VDOHV 7KHUHIRUH PRWLYDWLRQ IRU WKLV SURMHFW LV WR
GHYHORS LQWHUIDFLQJ DSSOLFDWLRQ WKDW VXSSRUW EDUFRGH VFDQQHU
DQG PD\ V\QFKURQL]H EHWZHHQ SK\VLFDO VWRUHV UHWDLO DQG
RQOLQH IRU LQYHQWRU\ SXUSRVH ZKLFK FDQ EH PRQLWRUHG DW DQ\
WLPHHYHU\ZKHUHDQGUHDOWLPHXSGDWH
7KLV SURMHFW LQYROYH VHYHUDO HOHPHQW ZKLFK DUH (
FRPPHUFH ,QWHUQHW RI 7KLQJV ,R7 DQG DOVR GHYHORSPHQW RI 0DQXIDFWXUHU&RGH
DQDSSOLFDWLRQWKDWFRPSDWLEOHZLWK$QGURLGRSHUDWLQJV\VWHP &RXQWU\&RGH 3URGXFW&RGH
26 LQ RUGHU WR FRQQHFW WKH PDLQ ZHEVLWH RI GDWDEDVH ZLWK &KHFN6XP
EDUFRGH VFDQQHU RI SURGXFW WR UHSODFH WKH H[LVWLQJ EDUFRGH
GHYLFH>@
)LJ($1%DUFRGH>@

____________________________________________________________________________________________________________
,Q RUGHUWR UHDOLVHWKLVLGHD $QGURLGEDUFRGHVFDQQHUPD\

LQWHJUDWHGZLWKRWKHUIHDWXUHVRIHFRPPHUFHV\VWHPE\XVLQJ
GDWDEDVH FRQQHFWHG YLD 0\ 6WUXFWXUH 4XHU\ /DQJXDJH
0\64/ 2SHQ VRXUFH HFRPPHUFH LV D IUHH RI FKDUJH
SODWIRUP WKDW FDQ EH LQVWDOOLQJ LQ VHOIKRVWHG 7KLV W\SH RI H
FRPPHUFH FDQ EH GRZQORDGLQJ IRU IUHH DQG XVHU FDQ HDVLO\
DFFHVV DQG UHGHVLJQ RU PRGLI\ WKH VRXUFH FRGH ZLWK KHOS RI
FRPPXQLW\WRXQGHUVWDQGWKHIXQGDPHQWDORIWKLVV\VWHP7KLV
V\VWHPDOVRSURYLGHVZLGHYDULHW\RIDYDLODEOHH[WHQVLRQVWKDW
FDQLPSURYHWKHV\VWHPWR EHPRUHIOH[LELOLW\ EHFDXVH RIWKH
VRXUFHFRGHFDQEHPRGLILHG
:KLOH GDWD ZLOO EH VWRUH LQWR D GDWDEDVH WKDW FRQVLVW RI
WDEOH DQG VWUXFWXUH $FFRUGLQJ WR 6LOEHUFKDW] HW DO >@ WKH
GDWDEDVH LV D VHW RI GDWD FRQWDLQV LQIRUPDWLRQ RQ DQ HQWLW\
ZKLOH 3RZHOO >@ GHVFULEHG WKH GDWDEDVH DV DQ
LPSOHPHQWDWLRQ RU FUHDWLRQ RI SK\VLFDO GDWD RQ D FRPSXWHU
+RZHYHU 6XPDWKL >@ LQWHUSUHW GDWDEDVH DV D FROOHFWLRQ RI
GDWDWKDWUHODWHWRDQGKDYHDFHUWDLQPHDQLQJZKLFKLVVWRUHG
SURSHUO\DQGFDQEHDFFHVVHGLQYDULRXVZD\VDQGVHTXHQFHV
7KHUH DUH WKUHH W\SH RI GDWDEDVH WLHU ZKLFK DUH )LUVW7LHU
IRU FOLHQW ZLOO GR WKH WDVN RU VHUYLFHV DQG LW FRQWDLQ XVHU
LQWHUIDFH DQG SUHVHQWDWLRQ VHUYLFHV 6HFRQG7LHU ZLOO EH SDUW
IRU DSSOLFDWLRQ VHUYHU ZKLFK FRQWDLQV WKH SUHVHQWDWLRQ FRGH
DQGWKH64/VWDWHPHQWVIRU GDWDDFFHVVZKLOHWKH7KLUG7LHU
LV IRUGDWD VHUYHUWKDWVWRUH WKH GDWDEDVH,Q JHQHUDOLW FDQEH
FRQFOXGHG WKDW DOO GDWD ZLOO EH VWRUH HLWKHU GLUHFWO\ RU
LQGLUHFWO\ GHSHQG RQ W\SH RI ILHOG 7KXV WKLV ZRUN ZLOO EH
IRFXVLQJ RQ 0XOWLSOH7LHU $UFKLWHFWXUH GDWDEDVH ZKLFK FDQ
SURYLGH WKH PRVW EHQHILWV LQ WHUPV RI VFDODELOLW\
LQWHURSHUDELOLW\DQGIOH[LELOLW\
,,,$)5$0(:25.2)6<67(0'(9(/230(17

)LJXUH VKRZV WKH SURMHFW EORFN GLDJUDP FRQVLVW RI )LJ)ORZFKDUWRI$QGURLG%DUFRGH6FDQQHU
EDUFRGH ($1 DQG LW ZLOO VFDQQHG E\ XVLQJ 6DPVXQJ
*DOD[\ 6 PRELOH,QWHUQHWFRQQHFWLRQLVUHTXLUHGLQRUGHUWR ,95(68/76$1'',6&866,21
FRQQHFW EHWZHHQ$QGURLGDSSOLFDWLRQDQGVHUYHUGDWDEDVHYLD 6FDQ&DUW LV D EUDQG QDPHG IRU WKLV DSSOLFDWLRQ
$3,7KHZHEVLWHLVGHVLJQHGXVLQJ2SHQFDUW(FRPPHUFHLQ FRPHVZLWKWKUHHIHDWXUHVZKLFKDUHXSGDWLQJTXDQWLW\FUHDWH
RUGHU WR VHW WKH SURGXFW LQIRUPDWLRQ DQG EDUFRGH ($1 RUGHUDQGUHFHQWRUGHUV8SGDWLQJTXDQWLW\LVWKHPDLQIHDWXUH
0HDQZKLOHVHUYHUGDWDEDVHLVXVHGWRVWRUHGDWDLQIRUPDWLRQ ZKLOHFUHDWHRUGHUDQGUHFHQWRUGHUDUHDQDGGLWLRQDOIHDWXUHV
,Q WKLV ZRUN $QGURLG EDUFRGH VFDQQHU DSSOLFDWLRQ ZDV 7KLV DSSOLFDWLRQ LV UHTXLUHGLQWHUQHWFRQQHFWLRQ WR EH RSHUDWH
GHVLJQHGE\XVLQJ$QGURLG6WXGLRDVVKRZQLQIORZFKDUWDVLQ VLQFHWKHGDWDLVVWRUHGLQVHUYHUGDWDEDVH8VHUFUHGHQWLDOVFDQ
)LJXUH:KHQWKHDSSOLFDWLRQLVVWDUWHGWKHVWDUWVFUHHQZLWK EHFUHDWHE\WKHDGPLQLVWUDWRURIWKHZHEVLWHEDFNHQG)LJXUH
ORJR 6FDQ&DUW ZLOO GLVSOD\ 7KHQ WKH ORJLQ SDJH ZLWK VKRZV WKH GHYHORSHG RI LQLWLDO VWDUWXS DSSOLFDWLRQ ORJLQ
FUHGHQWLDOV ILHOG ZLOO GLVSOD\ RQ VFUHHQ $OO FUHGHQWLDOV DUH SDJHDQGPDLQPHQX
UHTXLUHG RWKHUZLVH LW ZLOO GLVSOD\ ZURQJ SDVVZRUG XVHUQDPH
RU GRPDLQ 0RUHRYHU WKLV V\VWHP LV UHTXLUHG LQWHUQHW
FRQQHFWLRQ WR SHUIRUP RWKHUZLVH WKH V\VWHP ZLOO GLVSOD\ QR
LQWHUQHWFRQQHFWLRQ

)LJ3URMHFW%ORFN'LDJUDP
)LJ,QLWLDOVWDUWXSDSSOLFDWLRQORJLQSDJHDQGPDLQPHQX

____________________________________________________________________________________________________________
8SGDWH TXDQWLW\ SDJH LV YHU\ FUXFLDO LQ RUGHU WR VKRZ WKH
FXUUHQW TXDQWLWLHV DQG LWDOVR FDQ EH XVH WRDGG RU GHGXFWWKH 7KLV V\VWHP VXFFHVVIXO FRQQHFWHG DQG FRPPXQLFDWHG ZLWK
TXDQWLW\ IURP WKH GDWDEDVH %\ XVLQJ WKH EDUFRGH VFDQQHU LW ZHEVLWH EDFNHQG ZHEVLWH IURQWHQG DQG GDWDEDVH )URQWHQG
VFDQ ($1 &RGH IRUPDW WKDW DOUHDG\ EHHQ VWRUHG LQ WKH ZDV GHVLJQHG XVLQJ PLQLPDOLVW GXH WR PDUNHW GHPDQG IRU (
GDWDEDVHWRUHWULHYHDQGHGLW WKH TXDQWLW\ YDOXH 7KH VWHSE\ FRPPHUFH )LJXUH VKRZ WKH IURQWHQG SDJH RI ZHEVLWH WKDW
VWHSSURFHVVIORZLVVKRZQDVLQ)LJXUH FRQVLVWV RI VOLGHVKRZ ZRUGLQJ DQG LFRQV :KLOH )LJXUH
7KH SHUIRUPDQFH RI $QGURLG DSSOLFDWLRQ KDV EHHQ WHVWHG VKRZVWKHFDWHJRU\VHFWLRQIRUSURGXFWWKDWKDVEHHQVHOHFWHG
XVLQJ 6DPVXQJ 6 ZLWK $QGURLG /ROOLSRS 26 DQG DV FDWHJRU\ µ%DUX¶ 7KH PLQLPDOLVW ZHEVLWH LV WKH EHVW DV LW
1H[XV ZLWK $QGURLG 0DUVKPDOORZ 26 $OO KHOSVYLVLWRUWRXQGHUVWDQGWKHFRQFHSWRIWKHZHESDJH
SHUIRUPDQFH WHVW ZDV GRQH E\ UXQQLQJ WKH DSSOLFDWLRQ 7KHEDFNHQGZHEVLWHLVXVHGSXUSRVHO\IRUDGPLQLVWUDWRUWR
RSHUDWLRQ IRU DSSUR[LPDWHO\ WKUHH KRXUV 7KH WHVW DUH FRQVLVW PDQDJHWKHV\VWHP7KLVEDFNHQGFRQVLVWRIXVHUDGPLQLVWUDWRU
RI EDWWHU\ FRQVXPSWLRQ 5$0520 XVDJH GDWD XVDJH DQG DFFRXQW VDOHV VWDWLVWLFV WUDFN FXVWRPHU LQIRUPDWLRQ SURGXFW
FRQQHFWLYLW\ EHWZHHQ $QGURLG DSSOLFDWLRQ DQG VHUYHU 7KH LQIRUPDWLRQ FXVWRPL]H WKH VWUXFWXUH RI WKH ZHE DQG PDQDJH
UHVXOW RI WKLV WHVW VKRZHG WKDW WKH SHUIRUPDQFHV RI WKLV SD\PHQWDVLQ)LJXUH
DSSOLFDWLRQ LV LQ JRRG FRQGLWLRQ ZLWK JUHDW EDWWHU\
FRQVXPSWLRQ DQG PLQLPDO 5$0 DQG 520 XVDJH 7DEOH
VKRZVWKHSHUIRUPDQFHFRPSDULVRQEHWZHHQWKLVWZRGHYLFHV

)LJ)URQWHQG:HEVLWH

)LJ6WHSE\6WHSRI2UGHU3URFHVV

7DEOH&RPSDULVRQ*DOD[\6DQG1H[XVIRUDSSOLFDWLRQSHUIRUPDQFH

6DPVXQJ
1H[XV
*DOD[\6
%DWWHU\
%DWWHU\
5$0
520
5$0
520
'DWD
'DWD
0%
0%
0%
0%
.%
.%

)LJ&DWHJRU\6HFWLRQ

____________________________________________________________________________________________________________
7KH GDWDEDVH ZLOO EH YLVXDOL]H E\ XVLQJ SKS0\$GPLQ

7KLV XVHULQWHUIDFHZLOOGLVSOD\ WKHWDEOH RI GDWDEDVHWKDWKDV
EHHQFUHDWHGDWZHEVLWHEDFNHQG)LJXUHVKRZWKHUHJLVWHUHG 9 &21&/86,21
FRQWHQWIRUGDWDEDVHWDEOHRISURGXFW $V D FRQFOXVLRQ WKLV 6FDQ&DUW LQWHJUDWHG $QGURLG
$QRWKHULPSRUWDQWIDFWRULQRUGHUWRLPSURYHWKHTXDOLW\RI EDUFRGH VFDQQHU ZLWK (FRPPHUFH V\VWHP DQG GDWDEDVH ZDV
SURGXFW RU EXVLQHVVHV LV D XVHU DQG FXVWRPHU IHHGEDFNV VXFFHVVIXOO\WHVWHGDQGYHULILHGE\HQWUHSUHQHXUVIRXQGHUDQG
7KHUHIRUH WKUHH PDLQ SRLQWV RI IHHGEDFN KDYH EHHQ DVVHVVHG LQGXVWU\ZKLFKDUH5LQDVDOOHKERXWLTXHIRXQGHUDQGWKHLUFR
ZKLFKDUHGHVLJQDQGXVHULQWHUIDFHV\VWHPIXQFWLRQDOLW\DQG IRXQGHU %HOODP$PPDUD +LMULDKRPH DQG &ORXG+RX]H
VDWLVIDFWLRQ 7KH UHVSRQGHQW IHHGEDFNV DUH VXPPDULVH DV LQ IRXQGHU0DQDJHUIURP.RSHUDVL8QLYHUVLW\7XQ+XVVHLQ2QQ
7DEOH %DVHG RQ WKH IHHGEDFN DQDO\VLV WKH GHYHORSHG 0DOD\VLD 87+0 DQG 0HODND ,&7 +ROGLQJ 6GQ %KG
V\VWHPFRQVLGHUPHHWWKHXVHUDQGFXVWRPHUUHTXLUHPHQWZLWK 0,&7+ ZLWK SRVLWLYH UHVXOWV LQ WHUP RI IXQFWLRQDOLW\
SRVLWLYHUHFRPPHQGDWLRQIRUIHDWXUHVLPSURYHPHQW IHHGEDFNDQGLQWHJUDWLRQ
%DVHGRQVXUYH\IHHGEDFNDQGFRPPHQWZHEHOLHYHGWKLV
V\VWHP DUH XVHIXO IRU 6PDOO0HGLXP (QWHUSULVH 60( WR
HQKDQFH WKHLU SURGXFWLYLW\ DQG XSGDWH WKHLU VWRFN LQYHQWRU\
HDVLO\ +RSHIXOO\ WKLV V\VWHP LV DEOH WR EH XSJUDGH DQG
FRPPHUFLDOL]HLQIXWXUH
9,$&.12:/('*(0(17
7KH ILQDQFLDO VXSSRUW UHFHLYHG IURP WKH )XQGDPHQWDO
5HVHDUFK *UDQW 6FKHPH )5*6 9RW 2IILFH IRU
5HVHDUFK ,QQRYDWLRQ &RPPHUFLDOL]DWLRQ DQG &RQVXOWDQF\
0DQDJHPHQW 25,&& 8QLYHUVLW\ 7XQ +XVVHLQ 2QQ
0DOD\VLDLVJUDWHIXOO\DFNQRZOHGJHG
)LJ%DFNHQG3DJH

5()(5(1&(6
>@ +3LHWHUVHDQG062OLYLHU³$QGURLG%RWQHWVRQWKH5LVH7UHQGVDQG
&KDUDFWHULVWLFV´,QIRUPDWLRQ 6HFXULW\IRU6RXWK $IULFD ,66$SS

>@ ,JRU 3LKLU 9DOHQWLQD 3LKLU DQG 6WMHSDQ 9LGD ³,PSURYHPHQW RI
:DUHKRXVH 2SHUDWLRQV WKURXJK ,PSOHPHQWDWLRQ RI 0RELOH %DUFRGH
6\VWHPV $LPHG DW $GYDQFLQJ 6DOHV 3URFHVV´ ,QIRUPDWLRQ 7HFKQRORJ\
,QWHUIDFHV,7,SS
>@ .:DQJ < =RX DQG + :DQJ ³' EDU FRGH UHDGLQJ RQ FDPHUD
SKRQHV´ ,QWHUQDWLRQDO -RXUQDO RI ,PDJH DQG *UDSKLFV YRO QR
SS
>@ 6 7RQ\ ³%DU FRGHV 6ZHHS 7KH ZRUOG´ ,QYHQWLRQ 7HFKQRORJ\
0DJD]LQHYROQR
>@ ' 6KDUPD 9 6KDUPD % 6KULYDVWDYD 5 .XPDU DQG 3 6KDUPD
³5HJXODWRU\ $VSHFW 2I %DUFRGH 7HFKQRORJ\´ $VLDQ -RXUQDO 2I
3KDUPDFHXWLFDO6FLHQFHV$QG5HVHDUFKYROQR6HS
>@ : .HQW ³8QGHUVWDQGLQJ DQG 2EWDLQLQJ D 8QLYHUVDO 3URGXFW &RGH´
&HQWHUIRU$JULEXVLQHVVDQG(FRQRPLF'HYHORSPHQW
>@ *RZWKDP01.DQFKDQD³*HQHUDWLQJ($16WDQGDUG%DUFRGHV´
)LJ'DWDEDVH7DEOHYLHZ ,QWHUQDWLRQDO-RXUQDORI6FLHQFHDQG5HVHDUFK
>@ 3KLOLS %UDGOH\ ³%RRN QXPEHULQJ WKH LPSRUWDQFH RI WKH ,6%1´ 7KH
7DEOH8VHUDQG&XVWRPHU)HHGEDFNLQ3HUFHQWDJH ,QGH[HUYRO$SULO
>@ 6LOEHUVFKDW].RUWK6XGDUVKDQ³'DWDEDVH 6\VWHP &RQFHSWV´ )RXUWK
$ERYH %HORZ (GLWLRQ7KH0F*UDZ+LOO,QF
&ULWHULD *RRG $YHUDJH $YHUDJH $YHUDJH 3RRU >@ 3RZHOO*DYLQ³%HJLQQLQJ'DWDEDVH'HVLJQ´,QWHUQDWLRQDO-RXUQDORI
&RPSXWHU6FLHQFH ,QIRUPDWLRQ7HFKQRORJ\,-&6,7YROQR2FW
,QVWDOODWLRQ

3URFHVV >@ 6 6XPDWKLDQG 6 (VDNNLUDMDQ ³)XQGDPHQWDOV RI5HODWLRQDO'DWDEDVH
8VHU)ULHQGO\ 0DQDJHPHQW 6\VWHPV´ 3XEOLVKHU 6SULQJHU9HUODJ %HUOLQ +HLGHOEHUJ
'HVLJQ

)HDWXUHV
6\VWHP

3HUIRUPDQFH
6\VWHP

SURGXFWLYLW\

____________________________________________________________________________________________________________
Comparative Study on Metal Resistor and MOSFET-

Based THz Bolometers
Hiroshi Inokawa#1, Ajay Tiwari#, Takeo Ueta#, Hiroaki Satoh#, Catur Apriono*, Eko Tjipto Rahardjo*,
Norihisa Hiromoto§
#
Research Institute of Electronics, Shizuoka University
3-5-1 Johoku, Naka-ku, Hamamatsu 432-8011, Japan
1
inokawa.hiroshi@shizuoka.ac.jp
*
Department of Electrical Engineering, Faculty of Engineering, Universitas Indonesia
Kampus Baru UI Depok, Depok 16424, Indonesia
§
Graduate School of Science and Technology, Shizuoka University
3-5-1 Johoku, Naka-ku, Hamamatsu 432-8011, Japan
Abstract— This report compares the characteristics of terahertz heater is ~10 µm. The heater and thermistor should be fitted to
(THz) antenna-coupled bolometers with two different tempera- this length. Apart from this antenna-coupled bolometer, we
ture sensing scheme, i.e. metal-resistor-based thermistor and also fabricated bolometers with various heater/thermistor
SOI-MOSFET-based one. The former is simple, and its
lengths to verify the scaling trend of their performance.
performance is predictable as is described by the scaling trend.
The later shows higher responsivity due to the amplification The length dependences of the electrical responsivity (Rv)
function of the MOSFET, and features better manufacturability and the cutoff frequency (fc) of integrated thermistor and
by standard integrated circuit process. heater reveal that the former gets degraded and the latter
improved as the length is reduced if the width and thickness of
Keywords—Terahertz (THz), Bolometer, temperature coefficient
the heater-thermistor are kept the same. To predict the
of resistance (TCR), silicon-on-insulator (SOI), metal-oxide-
semiconductor field-effect transistor (MOSFET) Antenna (~/2)
I. INTRODUCTION
Heater
A bolometer detects the temperature increase caused by Thermistor Thermal
electromagnetic wave radiation, and is widely used in infrared isolation
(a)
focal plane arrays operating at room temperature [1],[2]. For
the THz radiation with longer wavelength, the size of the Dipole Stacked
absorber to receive the radiation becomes too large to be antenna heater/
structurally supported and thermally isolated. Therefore, an thermistor
antenna-coupled bolometer as schematically shown in Fig.
1(a) [3] is preferable, in which the radiation is first received
by an antenna, and then converted to heat by an isolated
resistive load. In this report, we compare the characteristics of
bolometers with two different temperature sensing schemes,
i.e. metal resistor-based thermistor [Fig. 1(b)] and silicon-on- Cavity
insulator (SOI) metal-oxide-semiconductor field-effect tran- in Si Thermistor leads
(b)
sistor (MOSFET)-based one [Fig. 1(c)], and clarify their
features. Dipole Source Suspended
antenna MOSFET
II. METAL RESISTOR-BASED BOLOMETER
Titanium (Ti) is selected as a resistive temperature sensor
(thermistor) and also as a heater that receives current from the
antenna to produce an increase in temperature, because of its
low thermal conductivity and high resistivity. The width of the Gate/
heater
Ti thermistor is made as narrow as 0.1 µm to attain high
responsivity. The detailed fabrication steps are reported Cavity
in Si Drain
elsewhere [4]. The fabricated antenna-coupled Ti bolometer (c)
for 1-THz range is shown is Fig. 1(b). As is understood from
Fig.1(a) Schematic structure of the antenna-coupled bolometer, (b) metal
the figure, the available length for integrated thermistor and resistor type, and (c) SOI MOSFET type, both for 1-THz detection.
____________________________________________________________________________________________________________
TABLE I
SCALING TREND FOR BOLOMETER UNDER CONSTANT-VOLTAGE CONDITION
Parameter Factor Remarks

Heater Length k
Thermisitor Thick., Width 1
Appl. Volt. (V) 1 Const. volt.
Curr. (I), Input power (Pin) 1/k
Elec. (Re ), Therm. (Rt) resist. k
Heat cap. (Ct) k
(a)
Temp. rise (T) 1  PinRt
Output volt. (Vout) 1  IRe T
Responsivity (RV) k  Vout/Pin
2
Cutoff freq. (fc) 1/k  1/(RtCt)
responsivity and response speed, scaling trend is explicitly

described in Table 1. The factor k is used to represent the
change in the physical dimensions of heater and thermistor,
and the resultant change in parameters such as voltage, current,
electrical and thermal resistances, and thermal capacitance is
considered under the constant-voltage condition. The overall (b)
effect of these parameters is reflected to the Rv and fc. The fc is
increased by the length reduction in inverse proportion to k2, 104 Slope ~ - 2 Experimental 104
Responsivity(V/W)
but the Rv is reduced in proportion to k. To verify the scaling Rv fc

trend experimentally, the integrated thermistor and heater with
different lengths were fabricated, and the length dependence Slope ~ 1
fc (Hz)
Expected fc
of the Rv and fc of the integrated thermistor and heater is 103 103
shown in Fig. 2. As is obvious from the figure and predicted Expected Rv
by the scaling trend, the Rv decreases with decreasing length.
At 11 µm, which we have to choose to realize the antenna-
102 102
coupled bolometer for 1 THz as mentioned previously, the
expected Rv becomes ~110 V/W. 10-5 10-4
Another issue in the down scaling of the thermistor is the (c) Length (m)
narrow-width effect as shown in Fig. 3. If the width is reduced
Fig. 2(a) Input-output characteristics of metal resistor-based bolometer, (b)
from 5 to 0.1 m, the resistivity increases by factor 3.7 and frequency response of the same, and (c) length dependence of the responsivity
the TCR decreases by factor 4.0. Additional material research RV from (a) and cutoff frequency fc from (b). Thickness and width of
is necessary to understand and alleviate this effect. thermistor are 46 nm and 0.1 m, respectively, and those of heater are 96 nm
and 1.24 m, respectively.
III. SOI MOSFET-BASED BOLOMETER
Fig. 1(c) shows the fabricated structure of the device. The
MOSFET is thermally isolated from the silicon (Si) substrate
by the cavity formed inside, and supported by source/drain
leads of doped single-crystalline Si and gate leads of doped
polycrystalline Si. The gate electrode and the leads also
function as the heater (resistive load for the antenna). Gate
length (L), thicknesses of gate oxide, SOI, and buried oxide
(BOX) are 1 m, 9 nm, 100 nm, and 400 nm, respectively.
The channel width (W) and lead length have variations of 5
and 10 m.
Fig. 4 shows the circuit diagram for the evaluation of
electrical responsivity. The MOSFET is heated by applying
balanced AC voltages to both ends of the gate (heater), and
the change in the threshold voltage caused by the temperature
Fig.3 Width dependence of the Ti resistivity and temperature coefficient of
rise is reflected to the drain current Io and output voltage Vout. resistance (TCR) for 50-nm thick film at 300 K.
____________________________________________________________________________________________________________
MOSFET Vdd
with gate Io
& heater RL
AM Vout
signal
Vin fm 2fm
~
fm fcar
Vg
Fig.4 Test circuit to evaluate the electrical responsivity. Instead of flowing

current from the antenna, balanced AC signals from a transformer are fed to
the gate (heater). Amplitude-modulated (AM) signal with carrier frequency
fcar=5 MHz and various modulation frequencies fm is supplied.
20 Fig.6 Frequency response of the MOSFET bolometers.

L/W/Lead Responsivity
1/ 5/ 5 fm 250 Hz
Output volt. (mV)
15 1/ 5/10
1/10/10 1512 IV. CONCLUSIONS
(µm) Characteristics of the terahertz (THz) antenna-coupled
10 bolometers with two different temperature sensing scheme, i.e.
metal-resistor-based thermistor and SOI-MOSFET-based one,
643 are compared. The former is simple, and its performance is
5
predictable as is described by the scaling trend, but the issue
439 (V/W) of the narrow-width effect in TCR is noted.
0 The latter could successfully attain electrical responsivity
0 2 4 6 8 10
Input power (µW) four times larger than that of the Ti one with similar
dimensions. Newly developed test circuit also enabled to find
Fig.5 Input-output characteristics of the MOSFET bolometers at fm=250 Hz the thermal response frequency more than 10 kHz. Since SOI
and Io=10 A for RL=100 k. The slope is electrical responsivity. MOSFET can be fabricated by standard integrated circuit
processing without the use of exotic thermistor materials, the
In this setup, amplitude-modulated (AM) signal with high results are promising for THz bolometer arrays with an
carrier frequency fcar=5 MHz is used to widen the measurable integrated read-out circuit.
range of the thermal response frequency. 1st or 2nd harmonic
(fm or 2fm) signal at Vout is detected by a lock-in amplifier. For ACKNOWLEDGMENT
the 1st harmonic output, the effective input power is given by This work was supported by JSPS KAKENHI Grant
Pin_fm=2mVcar2/Rh, (1) Number 15H03990, and the Cooperative Research Project of
Research Institute of Electronics, Shizuoka University.
where m, Vcar and Rh are modulation depth, root-mean-square Authors are indebted to the Solid State Division of
(RMS) amplitude of the carrier signal and the heater re- Hamamatsu Photonics K.K. for providing SOI MOSFETs.
sistance, respectively. For the 2nd harmonic output, the effec-
tive input power becomes REFERENCES
Pin_2fm=(mVcar) /(2Rh). 2
(2) [1] P. W. Kruse, Uncooled Thermal Imaging, Bellingham, WA: SPIE
Press, 2001.
Fig. 5 shows the linear relationship between input power to [2] M. V. S. Ramakrishna, G. Karunasiri, P. Neuzil, U. Sridhar, W. J. Zeng,
"Highly sensitive infrared temperature sensor using self-heating
the heater and the 1st harmonic output voltage detected by the compensated microbolometers," Sensors and Actuators A, no. 79, no. 2,
lock-in amplifier. The slope, corresponding to the electrical pp. 122-127, Feb. 2000.
responsivity (Rv), is larger for wider W and longer leads, and [3] T. Shimizu, H. Moritsu, Y. Yasuoka, and K. Gamo, "Fabrication of
rises to a maximum value of 1512 V/W. For the fair Antenna-Coupled Microbolometers," Jpn. J. Appl. Phys., vol. 34, no.
12A, pp. 6352-6357, Dec. 1995.
comparison with Ti bolometer at L=11 m, the smallest one [4] A. Tiwari, H. Satoh, M. Aoki, M. Takeda, N. Hiromoto, and Hiroshi
(W=5 m and Lead=5 m) should be selected. Still, the Inokawa, "Analysis of Microbolometer Characteristics for Antenna-
MOSFET has factor 4 advantage over the Ti-based bolometer. Coupled THz Detectors," Asian J. Chem., Vol. 25, pp. S358-S360,
Fig. 6 shows the frequency response of the MOSFET 2013.
[5] A. Tiwari, H. Satoh, M. Aoki, M. Takeda, N. Hiromoto, and H.
bolometers. It can be seen that the shorter lead and narrower Inokawa, "THz Antenna-Coupled Microbolometer with 0.1-μm-wide
W result in higher cutoff frequency (fc). Due to the improved Titanium Thermistor," Int. J. ChemTech Res., vol. 7, no. 2, pp. 1019-
test circuit in Fig. 4, fc of more than 10 kHz is observed. This 1026, Feb. 2015.
fc is also superior to that of Ti bolometer. Since there is a [6] A. Tiwari, H. Satoh, M. Aoki, M. Takeda, N. Hiromoto, and H.
Inokawa, "Fabrication and analytical modeling of integrated heater and
tradeoff between Rv and fc, Rv could further be improved, if thermistor for antenna-coupled bolometers," Sensors and Actuators A,
slower response is acceptable. no. 222, pp. 160-166, Feb. 2015.

____________________________________________________________________________________________________________
Silicon Single Boron Transistor

Yukinori Ono#1, Masahiro Hori#2, Akira Fujiwara*3
#
Research Institute of Electronics, Shizuoka University
3-5-1, Johoku,Naka-ku,
Hamamatsu 432-8011, JAPAN
1
ono.yukinori@shizuoka.ac.jp
2
hori.masahiro@shizuoka.ac.jp
*
NTT Basic Research Laboratories
3-1 Morinosato Wakamiya, Atsugi, Kanagawa, 243-0198, JAPAN
3
fujiwara.akira@lab.ntt.co.jp
Abstract— We will make a brief review of our research on the

detection of single dopant (boron) atom in silicon transistors. The
conductance modulation due to individual boron acceptors in
silicon nano-transistors is detected at 26 K. In addition, the single
charge transistor operation is demonstrated at 6 K using the
transistor containing just a single boron atom, i.e., the single
boron transistor. We will also discuss the possibility of
controlling the localized vibration mode of single boron atoms,
based on our single-boron transistors.
Keywords— single dopant, boron, silicon, transistor Fig. 1: Top- and cross-sectional views of the nano MOSFET fabricated
on a (100) SOI substrate. The channel doping was carried out to the
I. INTRODUCTION entire Si layer by ion implantation, which was followed by thermal
activation in N2 at 900˚C. The SOI is 32-nm thick and the lower-gate
Manipulation of single charges using dopant atoms in
channel is 40-nm long and 40-nm wide.
silicon has recently attracted much attention from the
viewpoint of quantum information technology and the so- doped source/drain, making it possible to investigate the effect
called beyond CMOS technology [1,2]. Electron transport via of the low-concentration channel dopants without the
single and minute number of dopants have thus been annoyance of dopant diffusion from the heavily doped
intensively investigated by several groups [3−13]. source/drain. Furthermore, the channel and the source/drain
Furthermore, fabrication techniques to create devices where have the same conduction type (p-type). Thus, with a
the position of each dopant atom in the active device area is sufficiently low temperature, the conductance is expected to
controlled are being developed taking both top-down and be influenced by the trapping/detrapping of holes by/from the
bottom-up approaches [14,15]. acceptor level, which is located right over the valence band
We have so far reported the detection of single [5,7,8] and edge.
coupled [6] boron acceptors in a transistor, and also the single
electron transfer by a single arsenic atom [12]. Here, we will III. MEASUREMENT RESULTS
make a brief review of our work [5] on low-temperature Conductance (G) was measured as a function of the lower
charge transport in doped MOSFETs, which is a significant gate voltage (VLG) using the back (substrate) gate voltage (VBG)
step towards the realization of single-dopant technology. We as a parameter, with the upper gate voltage kept constant (VUG
will also touch on the possibility of controlling the localized= -5 V). Figure 2 shows two examples of the results, measured
vibration mode of single boron atoms, based on our single- at 26 K, for MOSFETs with the doping concentration NA = 2 x
boron transistors. 1016 cm-3. This concentration corresponds to the mean boron
number in the nano-channel region of around one. Only one
II. DEVICE STRACTURE FOR SINGLE-DOPANT DETECTION series of modulation humps is observed in the subthreshold
Nano MOSFETs, fabricated on a silicon-on-insulator (SOI) region, which is the signature of the single-hole capture. In
substrate, comprise a boron-doped Si wire, a p+ source/drain, fact, no such humps were observed in undoped MOSFETs.
and a two-layered (lower and upper) poly-Si gate (Fig. 1). The The statistics of the appearance of the hump structure with
lower gate controls the electrical potential of a small wire several samples indicated that the number of the associated
region underneath it, while the upper gate, with a negative accepters is a few, or even just one [5], consistent with the
voltage, creates a hole accumulation layer acting as leads for above-mentioned mean boron number estimated from NA. We
the lower-gate nano MOSFET. Owing to this two-layered gate, thus ascribe these humps to the tunneling via a single boron
the channel under the lower gate is separated from the heavily acceptor. It should also be noticed that the pattern, or the VBG

____________________________________________________________________________________________________________
10-3 10-3 exponent 0.04

NA = 0.2 x 1017 cm-3 N = 0.2 x 1017 cm-3
10-4 10-4 A -14
device #LST-1 VBG = device #LST-2 -13 0.03
10-5 10-5 VBG = -12
VD (V)
-10 V -8 V -11
10-6 10-6 0.02
-10
G (S)
10-7 10-7 -6
0.01
10-8 10-8
10-5
10-9 10-9 # LST-1 0.00
10-6 T = 28 K to 6 K
10-10 VBG 10-10 VBG = 10-7 in 2-K step -0.01
10-11 =8V 10-11 10 V VBG =
10-8
G (S)
10-12 10-12 - 4 V -0.02
-1.6 -1.2 -0.8 -0.4 0.0 -1.2 -0.8 -0.4 0.0 0.4 10-9
VLG (V) VLG (V) 10-10 -0.03
10-11
Fig. 2: Examples of the conductance characteristics of the nano -0.04
MOSFETs with NA = 2 x 1016 cm-3, measured at 26 K. The VUG was 10-12 -0.75 -0.70 -0.65 -0.60 -0.55 -0.50
-0.7 -0.6 -0.5
fixed at -5 V and VBG was changed in 1-V steps. The estimated VLG (V)
VLG (V)
number of boron acceptors underneath the lower gate is estimated to
be around one from the NA and channel volume. Fig. 4: Temperature dependence of the hump (left) and Coulomb
diamond measured at 6 K for device # LST-1 in Fig. 2. VBG and
dependence of the hump conductance level was found to be VUG were fixed at -4 and -5 V, respectively. VD denotes the drain
different from one sample to another. This difference provides voltage.
information on the depth of the associated acceptors [6].
When the NA becomes higher, the modulation structure
IV. INTERPLAY WITH SINGLE PHONON CONTROL
becomes more complicated. Figure 3 shows two examples of
the results for MOSFETs with NA = 9 x 1016 cm-3. This The current single dopant technology employs the dopant
concentration corresponds to the mean boron number of atom as a substitute of the artificial atom or quantum dot, and
around five. A mixture of several series of humps is observed relies on either electronic, magnetic, or optical propery of
and the modulation pattern is different from one device to various kinds of dopant species. One of unique properties that
another due to a random doping. the dopant quantum dot possesses, but has not yet been
A sample with the lower NA was further investigated by explored, is the atomic vibration. For example, the dopant
lowering the temperature to 6 K. The humps become more with a lighter mass than that of the host material, e.g., boron in
distinct with decreasing temperature (Fig. 4, left). The silicon, has a local vibration mode [16], which is in effect the
differential conductance data (Fig. 4, right) reveals that there localized phonon. The local vibration mode of boron atoms in
is only one Coulomb diamond, suggesting that the acceptor silicon can couple with valence band holes and was observed
site can capture only one hole. This is a strong implication that using the inelastic electron tunnelling spectroscopy [17,18].
what we have observed is single-charge-transistor operation We anticipate that the localized vibration mode of the
with a single boron acceptor. In other words, we have single boron atom could be excited by using the single hole
successfully detected a single boron atom in nano MOSFET tunnelling in the present single boron transistors (Fig. 5). The
channels by the transport measurements. energy of the boron local vibration mode (78 meV) is close to
those of the zone-center optical phonons (65 meV) [17,18],
10-3 10-3 and the intervalley phonons, such as g-LO (63 meV) [19],
N = 0.9 x 1017 cm-3 N = 0.9 x 1017 cm-3 which are the cause of the heat bottleneck in silicon
10-4 A 10-4 A
10-5
device # L-1
VBG = 10-5
device # L-2
VBG = MOSFETs [20]. Therefore, control of the local vibration mode
10-6 -8 V 10-6 -10 V excitation may give us a possible route to control the
generation and annihilation of these travelling phonons,
G (S)
10-7 10-7
10-8 10-8
leading to a new concept for the heat control based on the
single dopant technology.
10-9 10-9
10-10 10-10 hole vibrational state
VBG VBG
10-11 10-11 =8V
= 10 V
10-12 10-12
-1.6 -1.2 -0.8 -0.4 0.0 -1.6 -1.2 -0.8 -0.4 0.0
VLG (V) electronic state
VLG (V)
boron atom
Fig. 3: Examples of the conductance characteristics of nano
MOSFETs with a low NA = 9 x 1016 cm-3, measured at 26 K. The Ground state Localized phonon excited state
VUG was fixed at -5 V and VBG was changed in 1-V steps. The
estimated number of boron acceptors underneath the lower gate is Fig. 5: Excitation of the local vibration mode of boron atom by the
estimated to be around 5 from the NA and channel volume. energy transfer from the valence band holes.

____________________________________________________________________________________________________________
CONCLUSIONS [7] Y. Ono, M. A. H. Khalafalla, K. Nishiguchi, K. Takashina, A. Fujiwara,

S. Horiguchi, H. Inokawa, Y. Takahashi, “Charge transport in boron-
We have made a brief review of our research on the doped nano MOSFETs: Towards single-dopant electronics,” Appl. Surf.
detection of single boron atom in silicon nano-transistors. The Sci., vol. 254, pp. 6252–6256, (2008).
conductance modulation due to individual boron acceptors in [8] M. A. H. Khalafalla, Y. Ono, K. Nishiguchi, A. Fujiwara, “Horizontal
position analysis of single acceptors in Si nanoscale field-effect
the transistors is detected at 26 K, and the single charge transistors,” Appl. Phys. Lett., vol. 94, pp. 223501_1-3, (2009).
transistor operation is demonstrated at 6 K. The present single [9] M. Pierre, R. Wacquez, X. Jehl, M. Sanquer, M. Vinet, O. Cueto,
boron transistor is expected to provide the way for controlling “Single-Donor Ionization Energies in a Nanoscale CMOS Channel,”
the single localized phonons of the boron atom, opening up a Nat. Nanotechnol., vol. 5, pp. 133-137, (2009).
[10] M. Fuechsle, J. A. Miwa, S. Mahapatra, H. Ryu, S. Lee, O. Warschkow,
new field of single phonon control based on the single dopant L. C. L. Hollenberg, G. Klimeck, M. Y. Simmons, “A single-Atom
technology. Transistor,” Nat. Nanotechnol., vol. 10, pp. 1038_1-5, (2012).
[11] M. Tabe, D. Moraru, M. Ligowski, M. Anwar, R. Jablonski, Y. Ono, T.
ACKNOWLEDGMENT Mizuno, “Single-Electron Transport through Single Dopants in a
Dopant-Rich Environment,” Phys. Rev. Lett., vol. 105, pp. 016803_1-4,
This work was supported by the Japan Society for the (2010).
Promotion of Science (JSPS) KAKENHI Grant Nos. [12] G. Lansbergen, Y. Ono, A. Fujiwara, “Donor-Based Single Electron
16H06087, 15K13970, 16H02339 and 26289105. We thank M. Pumps with Tunable Donor Binding Energy.” Nano Lett., vol. 12, pp.
Kalafalla for his supports of low-temperature measurements. 763-768, (2012).
[13] D. Moraru, A. Samanta, L T. Anh, T. Mizuno, H. Mizuta, M. Tabe,
“Transport Spectroscopy of Coupled Donors in Silicon Nano-
REFERENCES Transistors,” Sci. Rep., vol. 4, pp. 6219_1-6, (2014).
[1] F. A. Zwanenburg, A. S. Dzurak, A. Morello, M. Y. Simmons, L. C. L. [14] T. Shinada, S. Okamoto, T. Kobayashi, I. Ohdomari, “Enhancing
Hollenberg, G. Klimeck, S. Rogge, S. N. Coppersmith, M. A. Eriksson, Semiconductor Device Performance using Ordered Dopant Arrays,”
“Silicon Quantum Electronics,” Rev. Mod. Phys., vol. 85, pp. 961-1019, Nature vol. 437, pp. 1128-1131, (2005).
(2013). [15] J. C. Ho, R. Yerushalmi, Z. A. Jacobson, Z. Fan, R. L. Alley, A. Javey,
[2] X. Jehll, Y.-M. Niquet, M. Sanque, “Single Donor Electronics and “Controlled Nanoscale Doping of Semiconductors via Molecular
Quantum Functionalities with Advanced CMOS Technology,” J. Phys.: Monolayers.” Nat. Mater., vol. 7, pp. 62-67, (2008).
Condens. Matter, vol. 28, 103001_1- 18, (2016). [16] M. D. McCluskeya, “Local vibrational modes of impurities in
[3] H. Sellier, G. P. Lansbergen, J. Caro, S. Rogge, N. Collaert, I. Ferain, semiconductors,” J. Appl. Phys., vol. 87, pp. 3593-3617, (2000).
M. Jurczak, S. Biesemans, “Transport Spectroscopy of a Single Dopant [17] E. L. Wolf, “Evidence of hole-optical-phonon interaction in degenerate
in a Gated Silicon Nanowire,” Phys. Rev. Lett., vol. 97, pp. 206805_1- silicon in tunneking measurements,” Phys. Rev. Lett., vol. 20, pp. 204-
4, (2006). 207, (1968).
[4] L. Calvet, R. Wheeler, M. Reed, “Observation of the Linear Stark [18] D. E. Cullen, E. L. Wolf, W. D. Compton, “Tunneling Spectroscopy in
Effect in a Single Acceptor in Si,” Phys. Rev. Lett., vol. 98, pp. Degenerate p-Type Silicon,” Phys. Rev. B, vol. 2, pp. 3157-3169,
96805_1-4, (2007). (1970).
[5] Y. Ono, K. Nishiguchi, A. Fujiwara, H. Yamaguchi, Y. Takahashi, [19] S. G. Pavlov, R. Kh. Zhukavin, V. N. Shastin, H.-W. Hubers, “The
“Conductance Modulation by Individual Acceptors in Si Nanoscale Physical Principles of Terahertz Silicon Lasers based on Intracenter
Field-Effect Transistors,” Appl. Phys. Lett., vol. 90, pp. 102106_1-3, Transitions,” Phys. Status Solidi B, vol. 250, pp. 9-36, (2013).
(2007). [20] E. Pop, “Energy Dissipation and Transport in Nanoscale Devices,”
[6] M. A. H. Khalafalla, Y. Ono, K. Nishiguchi, A. Fujiwara, Nano Res., vol. 3, pp. 147-169, (2010).
“Identification of Single and Coupled Acceptors in Silicon Nano-Field-
Effect Transistors,” Appl. Phys. Lett., vol. 91, pp. 263513_1-3, (2007).

____________________________________________________________________________________________________________
Improvement in Film Quality of Hexagonal Boron

Nitride Grown by Chemical Vapor Deposition
K. Hara#1, N. Umehara#1, A. Masuda#2, T. Shimizu#2, T. Kouno#2 and H. Kominami#2
#1
Graduate School of Science and Technology, Shizuoka University
3-5-1 Johoku, Naka-ku, Hamamatsu, 432-8011 Japan
hara.kazuhiko@shizuoka.ac.jp
#2
Graduate School of Integrated Science and Technology, Shizuoka University
3-5-1 Johoku, Naka-ku, Hamamatsu, 432-8561 Japan
Abstract— The hexagonal BN (h-BN) films were grown by and so on [3-5]. Thus far, high quality h-BN bulk crystals
chemical vapor deposition using BCl3 and NH3 as sources on c- have been grown by the high pressure and high temperature
plane sapphire substrate. The growth at 20 kPa led to the technique [1]. At present, however, their size is limited to
significant improvement of crystalline quality compared to those several mm in diameter. Therefore, high-quality single-crystal
grown at an atmospheric pressure. The dependence on the
h-BN thin films on a large-area substrate are required for such
growth temperature (Tg) for the low-pressure growth indicated
that the crystalline quality is most improved for the sample electronics applications.
grown at 1200 C, in which the uniform in-plane orientation was To this effect, h-BN thin films have been grown by several
confirmed. This condition enhanced the lateral growth, resulting techniques such as metalorganic chemical vapor deposition
in the formation of grains with flat top surfaces. Reducing the (MOCVD) and molecular beam epitaxy [6-10]. Our study has
growth pressure to 10 kPa improves the luminescence property, intended to develop the manufacturing method of high quality
resulting in the first observation of the intrinsic exciton emission h-BN thin films by selecting the CVD with BCl3 and NH3
at room temperature from the h-BN films grown on sapphire gases for precursors, which is currently used for the
substrates. production of commercially-available pyrolytic BN. So far,
we have reported that the low pressure growth at 20 kPa leads
Keywords— Hexagonal boron nitride, chemical vapor to the significant improvement of crystalline quality
deposition, thin film, cathodoluninescence compared to those grown at an atmospheric pressure [11].
However, the quality of the grown films is required to be
I. INTRODUCTION further improved in terms of atomically flat surface and
Hexagonal boron nitride (h-BN) has a graphite-like crystal efficient band edge luminescence. In this study, we have first
structure consisting of sp2-hybridized atomic sheets of boron investigate the film growth mechanism during the CVD
and nitrogen. In addition to high thermal and chemical process in detail to improve the crystalline quality of the films.
stability, h-BN shows the characteristic luminescence in the This investigation has led us to achieve the growth of h-BN
ultraviolet spectral region including the free exciton emission films showing the free exciton emission around 215 nm at
at 215 nm and the impurity related emissions such as that room temperature.
around 320 nm owing to its wide bandgap energy of about 6.0
eV [1, 2]. From such features, h-BN has been expected to be II. EXPERIMENTAL PROCEDURES
applied for deep ultraviolet light emitters, substrates for The CVD apparatus used in this study consists of a BN
graphene devices, releasing layers for nitride semiconductors ceramics reactor tube (i.d. = 40 mm) and a high temperature
tubular furnace [11]. The source gases were BCl3 diluted to
0.03 % in N2 and 99.9997 % - pure NH3. BCl3 gas was
supplied through a coaxial gas nozzle with a sheath flow of
N2 outside the BCl3 flow. A c-plane sapphire substrate was
placed 30 mm away from the nozzle. Prior to growth, the
substrate was subject to a heat-treatment in a flow including
NH3 at growth temperature (Tg) for 10 min. The growth was
B
carried out for 2 h at reactor pressures of 5 - 20 kPa. The
N
typical supply rates of BCl3 was 0.2 sccm, whereas those of
NH3 was varied from 100 to 600 sccm. Tg was varied from
B-N: 0.145 nm C-C: 0.142 nm 1000 to 1500 °C. The growth rate of the samples was about
Inter layer: 0.334 nm Inter layer: 0.335nm 0.6 µm/h.
Fig. 1. Crystal structure of h-BN. That of graphite is also shown for
comparison.

____________________________________________________________________________________________________________
The structural and luminescence properties of the samples h-BN

were evaluated by X-ray diffraction (XRD) and
cathodoluminescence (CL) measurements, respectively. The -Al2O3
CL spectra were taken at room temperature and 20 K. Surface
morphology was observed using a scanning electron (1) Formation of nuclear
microscope (SEM).

A. Structural Property (2) Growth of grains
The XRD measurement indicated that the films grown at
20 kPa were oriented towards c-axis for Tg from 1000 to 1500
C. However, a disordered phase of BN tended to be formed
with increasing Tg from 1300 C, which is evidenced by the
tail structure observed in the lower angle region of the (002)
diffraction peak. On the other hand, the diffraction peak (3) Coalescence of grains
became broader and weaker with decreasing Tg from 1100 C. Fig. 3 Proposed film growth mechanism of h-BN
These result reveals that the crystalline quality was most during the CVD process.
improved at Tg of 1200 C for the growth at 20 kPa. It should
be noted that this growth temperature is much lower than have flat surface and coalesce each other for the sample
1500 C for the case of atmospheric pressure growth. grown at about 1200 C, suggesting that the lateral growth
The most significant improvement in crystalline quality for occurred more dominantly due to the enhanced surface
the film grown at low pressures is the in-plane orientation of migration of precursors under this condition. From the results
crystal. In contrast to the result obtained for the sample grown of XRD measurement and surface observation, we have found
at the atmospheric pressure, the pronounced diffraction peaks that the crystallographic property of the films is dramatically
were observed every 60 degree in the in-plane XRD patterns improved by reducing the growth pressure from the
of the sample grown at 1200 °C, which reflects the six-fold atmospheric pressure to 20 kPa. This condition is effective to
symmetry of the crystal structure of h-BN. The comparison suppress an unfavorable reaction between BCl3 and NH3 in
with those observed for the sapphire substrate confirms that the gas phase before reaching the substrate surface, which
the in-plane epitaxial relationship is {100}h-BN // {110}sapphire., makes the reaction for the film formation more sensitive to
which agrees with that reported for the very thin h-BN layer the atomic alignment of substrate surface.
grown by MOCVD [5]. It can be seen in Fig. 2 that the grains To investigate the growth mechanism of h-BN film on a c-
plane sapphire, we prepared several samples grown for
different time from 6 to 120 min at 1150 °C. The SEM
observation of the film surfaces indicates that at the early
stage of the growth (t = 6 min), small nuclei which has flat
top surfaces were found on the substrate surface. Such a
morphology indicates that these nuclei are oriented with the
c-axis normal to the substrate surface. With increasing growth
time, from 30 to 60 min, the size of the c-oriented nuclei were
increased and the surface consisted of hexagonal grains with
diameters of about 500 nm, which reflects the crystal
2 m symmetry of h-BN. It should be noted that the island like
grains grew lateral direction and began to coalesce each other
Randomly oriented grains
in this stage. These coalescence of grains was not completed
even at t = 120 min. On the other hand, as shown in the
bottom image in Fig. 2, we noticed that randomly-oriented
grains often exist between grains.
Based on the observation mentioned above, we propose a
growth model of h-BN crystals shown in Fig. 4. At the early
stage of growth, crystalline nuclei are formed. Most of them
are oriented towards the c-axis. But some are misoriented.
Then, the grains grow dominantly towards the lateral
500 nm direction, resulting in flat top surfaces. The crystal growth
preferentially toward a-direction can be explained by the
Fig. 2 Surface morphology of the films grown at 1150 C. strongly anisotropic nature of h-BN crystal. Finally, the

____________________________________________________________________________________________________________
215 nm ACKNOWLEDGMENT
10 kV RT Part of this work was carried out under the Cooperative
CL Intensity (a.u.)
Research Project Program of the Research Institute of
CL Intensity (a.u.)
Electrical Communication, Tohoku University.
REFERENCES
200 210 220 230
W avelength (nm) [1] K. Watanabe, T. Taniguchi, and H. Kanda, Nature Mater. 3, 404
(2004).
[2] K. Hara, X. Liu, M. Yamauchi, Y. Kawanishi, H. Kominami, and Y.
Nakanishi, phys. stat. sol. (c) 8, 2509 (2011).
[3] K. Watanabe, T. Taniguchi, T. Niiyama, K. Miya, and M. Taniguchi,
Nature Photon. 3, 591 (2009).
[4] C. R. Dean, A. F. Young, I. Meric, C. Lee, L. Wang, S. Sorgenfrei, K.
200 300 400 500 600 Watanabe, T. Taniguchi, P. Kim, K. L. Shepard, and J. Hone, Nature
Wavelength (nm) Nanotech. 5, 722 (2010).
[5] Y. Kobayashi, K. Kumakura, T. Akasaka, and T. Makimoto, Nature
Fig. 4 Room temperature CL spectra of the samples 484, 223 (2012).
grown at 10 kPa. [6] C. L. Tsai, Y. Kobayashi, T. Akasaka and M. Kasu, J. Cryst. Growth
331, 3054 (2009).
[7] S. Majety, J. Li, X. K. Cao, R. Dahal, B. N. Pantha, J. Y. Lin, and H.
grains coalesce each other, which forms larger grains. X. Jiang, Appl. Phys. Lett. 100, 061121 (2012).
However, randomly-oriented grains may disturb the perfect [8] L. Song, L. Ci, H. Lu, P. B. Sorokin, C. Jin, J. Ni, A. G. Kvashnin, D.
coalescence. The crystal growth model clarifies that the G. Kvashnin, J. Lou, B. I. Yakobson, and P. M. Ajayan, Nano Lett. 10,
control of nucleus formation on the sapphire substrate at the 3209 (2010).
[9] O. Thuda, K. Watanabe, and T. Taniguchi, Jpn. J. Appl. Phys. 46, 13
early stage is essential to further improvement of h-BN films. (2007).
[10] M. Chubarov, H. Pedersen, H. Hogberg, J. Jensen, and A. Henry,
B. Luminescence Property Cryst. Growth and Design 12, 3215 (2012).
The CL spectra of the samples grown at 20 kPa were [11] N. Umehara, A. Masuda, T. Shimizu, I. Kuwahara, T. Kouno H.
Kominami and K. Hara, Jpn. J. Appl. Phys. 55, 05FD09 (2016).
dominated by the impurity related emission bands around 318 [12] K. Watanabe and T. Taniguchi, Int. J. Appl. Ceram. Technol., 8, 977
or 350 nm, which probably originates from O or C impurities (2011).
in h-BN. Even the sample grown at 1200 C, of which
crystalline quality was most improved, did not show clear
band edge emissions. On the other hands, a pronounced
emission was observed around 215 nm from those grown at 5
or 10 kPa although the broad emissions around 350 nm were
still dominant. This emission can be ascribed to the
recombination of intrinsic excitons of h-BN if it is compared
with that observed at 215 nm for the h-BN bulk crystal [12].
It is noteworthy that the intrinsic exciton emission was
observed at room temperature for the first time from the h-BN
film grown on a c-plane sapphire substrate. The dependence
of CL spectra on the accelerating voltage of an electron beam
indicates the improvement of film quality with the increase of
its thickness.
IV. SUMMARY
The h-BN films were grown by low pressure CVD using
BCl3 and NH3 as sources on c-plane sapphire substrates. The
growth at 20 kPa led to the significant improvement of
crystalline quality compared to those grown at an atmospheric
pressure. We have also found that reducing the growth
pressure to 5 kPa improves the luminescence property,
resulting in the first observation of the free exciton emission
at room temperature from the h-BN films grown on sapphire
substrates.

____________________________________________________________________________________________________________
Minimising the Number of Tower of Base

Transceiver Station by Considering the Coverage
Area using Fuzzy Clustering Means and Particle
Swarm Optimization
Faqih Rofii#1, Diky Siswanto#2, Fachrudin Hunaini#3, Ahmad Rijal Kafy#4
#1,2,3,4
Departement of Electrical Engineering, Universitas Widyagama Malang
Jl. Taman Borobudur Indah 03 Malang, Indonesia
1faqih@widyagama.ac.id
2dsiswanto@widyagama.ac.id
3fachrudin@widagama.ac.id
4ar_kafi@widyagama.ac.id
Abstract— The increasing number and the distribution is already very crowded, very irregular placement, and on the
expansion of mobile subscribers need the cellular network to other hand telecommunications providers continue to develop
expand its service coverage, especially in urban areas. It in the region are already full.
encourages cellular operators to continually extend its cellular To arrange the number of existing telecommunications
coverage area by constructing new towers of base transceiver
towers and anticipate the growing amount of mobile telecom
station (BTS) in certain areas. BTS development that continues
to grow in the limited urban areas causing environmental clutter. towers and to realize the tower together at the point of use
So it requires rearranging the presence of the tower. This paper collocation [3], we need a study to recalculate the number of
proposes a novel method to minimise the number of existing base stations that already exist and maintain its coverage area,
cellular towers that able to cover an urban area of Malang city. using optimization techniques. One of the problems in the
The Fuzzy C-Means and Particle Swarm Optimization methods development of mobile telecommunications network is the
are used to optimise the problem. Briefly, the proposed method is selection of proper placement of BTS locations, so as to reach
to make a GSM standard-based cellular network planning to customers optimally with adequate traffic services. To
obtain the ideal number of cells, and then to optimise the BTS’s overcome this problem, some optimization techniques
composition for finding the best tower with the widest regional
placement of base stations has been carried out, ie mobile
coverage. In this study, there are two types of cellular providers
based on the frequency spectrum used. The result showed that network optimization with genetic approaches [4] and the
there are 48 of 154 existing towers selected. The result also application of genetic algorithms for optimization of radio
revealed that the new BTS’s composition covers 75.39% of the network coverage [5].
total area for the frequency bandwidth of 10 MHz, while this new
composition covers 64.49% of the total area for the frequency II. FUZZY CLUSTERING
bandwidth of 7.5 MHz.. Clustering generally used for processing large amounts of
data, which is the process of classifying objects based on
Keywords— Tower, BTS, coverage area, FCM, PSO information obtained from the data describing the relationship
between objects in principle to maximize the similarity
I. INTRODUCTION between members of the class and to minimize the similarity
Mobile telecommunications users are always increasing, between classes / clusters. The goal is finding a qualified
stimulating the mobile telecommunications operators increase cluster within a reasonable time. Clustering is a pattern
the area of coverage by building towers Base Transceiver classification into a group (cluster). Clustering problem, much
Station (BTS) in each place. Tens of thousands of towers that discussed by researchers in many contexts and disciplines for
have been built are increasingly making the existing the analysis and exploration of data [6]. There are several
environmental clutter. BTS tower construction without clustering algorithms of data, one of which is the Fuzzy C-
considering various factors can lead to tower forest, disrupting Means (FCM). FCM clustering is a technique of data, where
flight traffic and other negative effects [1], [2]. The the existence of each data point in a cluster is determined by
government has imposed a ministerial regulation and technical the degree of membership [7].
guidance on structuring telecommunications through the use The basic concept of FCM, the first is to determine the
of telecommunication towers together. The problem is the center of the cluster, which will mark the average location for
amount of cellular telecommunications tower in an urban area each cluster. In the initial condition, the center of the cluster is
still not accurate. Each data point has a degree of membership

____________________________________________________________________________________________________________
for each cluster. By fixing the center of the cluster and the V. RESULT
degree of membership of each data point is repeated, it will be Clustering existing tower is to get the center point of the
seen that the center of the cluster will move to the right cluster of the most prevalent areas of the city of Malang.
location. This loop is based on minimization of the objective Tower clustering process, conducted by the point coordinates,
function that describes the distance from data supplied to while the number of clusters is determined by the number of
center cluster membership degree weighted by the data points. BTS base planning results, the aim is establishing the particles
The output of the FCM is a row of the center cluster and some in PSO operations. The results of FCM is for the tower
degree of membership for each data point. operator A, is 28 clusters, while for the operator B and C is 24
clusters, where the number of towers each cluster is different,
III. PARTICLE SWARM OPTIMIZATION
according to the density of the tower in each cluster were
The initial idea of the 'particle swarm' is basically intended selected, as shown in figure 3 and 4.
to produce a computational intelligence through a simple
-7.9
analogy of social interaction, not just a purely individual
intelligence capabilities. This idea was later developed by -7.92
Kenedy and simulated become the method of particle swarm
optimization (PSO). In PSO, a number of simple entities -7.94
(particles) are placed on the search space of some problem or

-7.96
function, and each evaluate the objective function at its
latitude
current location. Each particle further define its movement -7.98
through the search space by combining some aspects of the
history of the current and the best location with one or more -8
members of the group at random. Each individual of the herd

-8.02
vector particles are prepared by a three-dimensional D, where
D is the dimension of the search space, namely: the current -8.04
112.56 112.58 112.6 112.62 112.64 112.66 112.68 112.7
position (xi), the previous best position and speed. PSO longitude
algorithm step sequence, are as developed by Kenedy
Eberhart [8]. PSO is widely used for various purposes of Fig. 2 Mapping FCM clustering results for the tower operator A
optimization, in the field of control [9], the design of mobile
networks [10] and on wireless sensor networks [11]. -7.9
-7.92
IV. OPTIMIZATION OF NUMBER OF BTS TOWER
Optimization is applied to the Malang city region with an -7.94
area of 110.06 km², assuming the entire area of the city of
Malang is urban morphology characterized by using -7.96
latitude
propagation models Cost 231 [12]. There are three operators,

-7.98
namely operator A, B and C with operating frequency
bandwidth are 10, 7.5 and 7.5 Mhz respectively, and the -8
number of the existing BTS are 55, 52 and 47 sites
respectively. At this planning assumed a 30-meter high tower, -8.02
so by using a link budget calculation [12], radius BTS are 0,88

-8.04
and 0,96 KM. 112.56 112.58 112.6 112.62 112.64 112.66 112.68 112.7
longitude
Fig. 3 Mapping FCM clustering results for the tower operator B and C
After using FCM clustering tower, then choose a tower

each cluster to form the composition of the distribution of
base stations that can generate the largest area coverage using
PSO. The objective function of this optimization is to get
maximum coverage area of BTS selected, expressed in :
L TotalBTS − L ∩
Coverage area = …………..……….…..…. (1)
L City
Where L TotalBTS is the sum of the area of each BTS elected,
L ∩ is the area of the slices between BTS and LCity is an area
of a city (region).
Fig. 1 Map of the distribution of the existing BTS tower in the city of Malang

____________________________________________________________________________________________________________
N convergent iteration 37th where the coverage area reaches

∑ ∑L
1 M
…………………………….…….….... (2)
L∩ = ∩ij 64.49%, of the total area planned.
2 i =1
j=1
Fitness function of this optimization is to find a minimum

area of the slice between base stations, is expressed :
 0 if i = j
 0 if rTi + rTj < D ij
 (3)

L∩ ij = 
  D ij   D 2 
 2ri r jcos   − D ij  ri r j  ij  others
  r1 + r j    4 
     
L∩ij is the area of the slice between BTSi to BTSj, is the

radius of BTSj, and is distance BTSi to BTSj. In this work,
using the sectoral antenna in 1200 to form the coverage area of
a circle. Correction cell radius is expressed: r = d x 27.. …..(4) Fig. 4 Optimization of BTS and Coverage area using FCM-PSO for 24 BTS
Where r is the radius correction cells, whereas d is the
radius of a cell planning results.
The optimization steps are as follows:
1. Initialization problems
The number of particles (N), lower bounary (Lb), Upper
boundary (Ub), cognitive learning (c1), Social learning
(c2), Constriction Factor (K) and Initial Velocity (Vo)
2. Generate particles through the generation of random
numbers (0.1) with dimensions (N, Ub). Each particle is
generated, is a representation of the identity/BTS number.
3. Sort each particle resulting from step 2, starting from the
smallest to largest value. The result of particle sorting is
the composition of the BTS identity
4. Check the matrix of BTS tower composition, result of
sorting particle above to the cluster members. In order to
know each member of the BTS matrix included in what Fig. 5 Optimization of BTS and Coverage area using FCM-PSO for 28 BTS
cluster
5. If there is a cluster value equal to the value of the previous BTS tower elected from optimization results for 24 and 28
cluster, changing the cluster value to 0 (zero) base stations, then plotted on a digital map shown in Figure
6. Make the cluster value is not 0 (zero) as the composition 5.6 and 7 below.
of the selected BTS.
7. Get fitness function, by calculating the total area wedge
between BTS for each particle.
8. Compare the results of the evaluation of the fitness
function of each particle to get the best particles (Pbesti),
ie particles with minimum wedge between BTS
9. Update the particle's position is based on the speed
10. Get an objective function, coverage area of the
composition of the selected BTS, for example for 24 BTS
is elected, then the coverage is
L TotalBTS = ∑ L i =112.6 KM
M 2
i =1
∑ ∑
1 M N
L∩ = L ∩ ij = 29.627 KM2
2 i =1 j=1 Fig. 6 Plot of optimization result for 24 BTS (Operator B )
L TotalBTS − L ∩
Coverage area = x100% =75.39% Having obtained the BTS was selected as a result of
L City
optimization, for operators A, B and C, as in Table 1, then the
results are plotted simultaneously on a digital map, so it can
Fig. 5 and 6 show the results of the fitness value of FCM-
be visually seen in Figure 8. In Figure it can be seen that in
PSO iterations, such as the coverage area to 24 and 28 BTS.
areas that are not covered is an area with few customers.
On optimization for 24 BTS, fitness reaches the value of
converging iteration 37th where the coverage area reaches
75.39%, whereas in 28 BTS, fitness reaching a value of

____________________________________________________________________________________________________________
VI. CONCLUSION
The applicability of the FCM in this study could simplify
the selection process to be used as an individual BTS/
particles used for process optimization PSO, this technique is
useful to accelerate the achievement of convergence on PSO
iteration, because it reduces the size of the slices between the
BTS. Merger techniques FCM and PSO proven can be used to
minimize the number of base stations, of 154 BTS tower in
the city of Malang, being only 58 base stations, the coverage
area reaches 64.49% for operators A and 75.39 for the
operator B and C of the total planning area, where an area not
covered by the BTS tower is an area with few customers.
Optimization results were obtained 4 BTSs, which can tower
sharing because it is in the same location.
Fig. 7 Plot of optimization result for 28 BTS (Operator A) ACKNOWLEDGMENT
My thanks to LPPM Universitas Widyagama Malang, and
the Ministry of Research, Technology and Higher Education
of the Republic of Indonesia that have supported this research
activity.
REFERENCES
[1] W. A. Prijono, “Penataan Menara BTS (Cell Planning),” J. EECCIS,
vol.4, no. 1, p. pp–50, 2012.
[2] A. Anwar and S. Idaiani, “Gambaran Gangguan Cemas Masyarakat Di
Sekitar Menara Base Tranceiver Station/Bts Di Bandung Dan
Jakarta,” Bul. Penelit. Kesehat., vol. 42, no. 4 Des, pp. 267–272, 2014.
[3] A. Qustoniah and others, “Analisis Colocation Site Sebagai Acuan
Dalam Penataan Menara ℡Ekomunikasi Seluler Di Wilayah Kota
Malang,” WIDYA Tek., vol. 22, no. 1, 2014.
[4] P. Calégari, P. Kuonen, F. Guidec, and D. Wagner, “A genetic
approach to radio network optimization for mobile systems,” in
VTC’97, 1997, vol. 2, pp. 755–759.
[5] C. O. Alenoghena, J. O. Emagbetere, and F. O. Edeko, “Application
Fig. 8 Plot of optimization result for 24 dan 28 BTS (Operator A,B dan C) of Genetic Algorithm in Radio Network Coverage Optimization–A
Review,” Int. J. Comput. Appl., vol. 66, no. 12, 2013.
Table 1 below is the result of selection for the 24 and 28 [6] A. K. Jain, M. N. Murty, and P. J. Flynn, “Data clustering: a review,”
BTS towers. In the table, can be found 4 the same base station ACM Comput. Surv. CSUR, vol. 31, no. 3, pp. 264–323, 1999.
[7] J. C. Bezdek, R. Ehrlich, and W. Full, “FCM: The fuzzy c-means
(cell-colored), which indicates that the BTS tower located on clustering algorithm,” Comput. Geosci., vol. 10, no. 2, pp. 191–203,
the same location, so that it can be combined (tower sharing). 1984.
[8] M. Clerc, Particle swarm optimization, vol. 93. John Wiley & Sons,
2010.
TABLE I [9] F. HUNAINI, I. ROBANDI, and N. SUTANTRA, “Optimization of
24 AND 28 SELECTED BTS automatic steering control on vehicle with steer by wire system using
No. of No. of BTS No. of BTS No. of No. of BTS No. of BTS particle swarm optimization,” Turk J Electr Eng Comput Sci, pp. 1–35,
for 28 for 24 for28 for 24 2013.
Selected selected Selected selected [10] A. A. El-Saleh, M. Ismail, R. Viknesh, C. C. Mark, and M. L. Chan,
Cluster BTS BTS Cluster BTS BTS “Particle swarm optimization for mobile network design,” IEICE
1 151 152 15 138 99 Electron. Express, vol. 6, no. 17, pp. 1219–1225, 2009.
2 67 3 16 13 74 [11] R. V. Kulkarni and G. K. Venayagamoorthy, “Particle swarm
3 116 6 17 129 17 optimization in wireless-sensor networks: A brief survey,” Syst. Man
4 6 30 18 112 44
Cybern. Part C Appl. Rev. IEEE Trans. On, vol. 41, no. 2, pp. 262–
267, 2011.
5 102 26 19 72 117
[12] J. B. Andersen, T. S. Rappaport, and S. Yoshida, “Propagation
6 142 75 20 28 132 measurements and models for wireless communications channels,”
7 85 19 21 48 94 Commun. Mag. IEEE, vol. 33, no. 1, pp. 42–49, 1995.
8 80 76 22 134 122 [13] P. K. Sharma and R. K. Singh, “Cell coverage area and link budget
9 108 141 23 59 31 calculations in GSM system,” Int. J. Mod. Eng. Res. IJMER Vol, vol.
10 136 148 24 26 9 2, pp. 170–176, 2012.
11 10 4 25 66
12 127 143 26 126
13 3 48 27 153
14 46 71 28 64

____________________________________________________________________________________________________________
Implemention of Lyapunov Function to PID

Voltage Control for Single Phase Rectifier using
Low Cost Microcontroller
Shamsul Aizam Zulkifli, Muhammad Azri Mustafa
Department of Electrical Power Engineering, Faculty of Electrical and Electronic Engineering

Universiti Tun Hussein Onn Malaysia, Batu Pahat, Johor, Malaysia
s.aizam@yahoo.com
Abstract— This project is focused on development of single phase continuous cycling (CC) method and Process Reaction Curve
rectifier with voltage control application using a Proportional (PRC) method, while the PRC method is specially proposed
Integral Derivative (PID) controller. The PID controller uses a for controlled processes with time delay [3]. As for the
Lyapunov function in order to find the best gain values for the Lyapunov function, several important equations have been
PID control gains for good accuracy and improved output
listed below [7],
respond. The uses of the Lyapunov involves with mathematical
equations that have been model in MATLAB-Simulink in order
to be applied with the low cost microcontroller where are the (1)
Raspberry Pi and Arduino. At the end, the simulation and where (t, x) is the solution to the system differential
hardware tests setup have been conducted and its show and equation, defining the trajectory starting at x at time t = 0.
prove that the Lyapunov function gives good result in terms of The general equation of any plant transfer function with
the rectifier output voltage
the PID control as given below,
Keywords— Lyapunov function, Raspberry Pi, PID controller,
Arduino, Matlab-Simulink
(2)
I. INTRODUCTION Hence, the open looped transfer function of control system
Most of the electrical applications are being monitored yields
" #$ " #" #%
and controlled by a single controller. The common controller !
% . (3)
is the Proportional Derivative Integral (PID) control where it &$ " &" &%
calculates an "error" value from the difference between a
measured process variable and a desired point [1] before the Meanwhile, the error differential equation is derived as,
accurate output can be generated. At the same time, this ' ( ) ' ( ' * ') (
controller has some disadvantage such as the gain values are '+ ( ,. (4)
determined by try and error technique. Currently this method
has been improved by using the mathematical equation to The Lyapunov parameters of the equations are defined as
ensure the value for each gain can be determined ' - ' '+ .
automatically. However, the optimal value of PID control ' . ') .
gains can be achieved by multiple ways such as by (5)
implementing the mathematical equation such as the Ordinary At the meantime, the optimal PID control can minimize
Differential Equation (ODE). The ODE is a differential the performance index of the given control weight matrix and
equation that contains functions or one independent variable initial states [8],
and its derivative may be respect to more than one
/
independent variable [2]. This concept have path a way to / / /
(6)
optimize the value of the gains for the optimal value of PID
controller that can contribute for better output at the plant. 0$ 1 02 1&% 1#" 03
. , (7)
One of the examples of ODE is the concept of Lyapunov #$ #% #%
Functionn [3]-[4] because many ODE admit an energy This project is focused on the development of single
function that is non-increasing along solution. As a result, the phase rectifier with voltage control using the PID controller.
Lyapunov function equations can be used to improve the PID The PID controller gains will be optimized after the
gains selection. As known, traditional PID control needs to be implementation of Lyapunov function. The block diagram on
tune empirically by the method proposed by Zeigler Nichols how the Lyapunov function being implemented to the PID in
(Z-N) and Cohen-Coon [5]-[6]. The Z–N method includes the single phase rectifier is shown in Fig. 1.

____________________________________________________________________________________________________________
DOUBLE OUTPUT
RECTIFIER
AC SUPPLY
VOLTAGE
~ RECTIFIER FILTER
DIVIDER
ANALOG TO
PID CONTROLLER DIGITAL (ADC) Fig. 2 Hardware
ardware setup
WITH LAYPUNOV CONVERTER AND
GATE DRIVER FUNCTION ( DIGITAL TO
USING ANALOG (DAC)
RASPBERRY PI) CONVRTER USING
ARDUINO UNO
Fig.1 Block diagram of the project
The Raspberry Pi (RPi) and Arduino are been used

because its behave as low cost micro-controller
controller and able
communicate with MATLAB Simulink [8]-[11]. Fig. 3 PID control strategies develop in MATLAB-Simulink
Nevertheless, these microcontroller are eased
eas to connect to
MATLAB and it has friendly user interface.
interface At the meantime, TABLE 1: PID GAIN PARAMETER
PARAM
RPi also gives flexibility in the choice of programming
PID setting Kp Ki Kd
languages and installation that could be used for designing the
PID controller. This control strategy is able to been tested on a Without Lyapunov 0.009 0.9 0
simple control format by only using the mathematical
mathematica blocks With Lyapunov 0.008999981 0.899999982 0
function available in the simulink with limit of time.
time
III. RECTIFIER TESTING WITH
WI PID VOLTAGE
II. IMPLEMENTATION THE LYAPUNOV
PUNOV ON THE
T SINGLE CONTROL
PHASE RECTIFIER WITH VOLTAGE CONTROL
C A. Closed Loop Simulation
For the first test,, it has been conducted in simulation
Fig. 2 shows the overall setup for the hardware test in
environment shown in Fig.44 where the closed loop has
this project. The hardware setup consists of a RPi as a
undergone a voltage feedback system in order to control the
microcontroller that produces the PWM signal for rectification
output from the rectifier. For this simulation test, the AC
operation. The Arduino is used for converting ing the Analogue to
input is been set to 20Vac while the target rectifier output
Digital signal (ADC) for 3 bits value and also to be the digital
voltage control is set at 15Vrms inside
in the PID controller.
input to the RPi. This hardware has been set-up
set in the Makmal
Elektronik Kuasa, FKEE, Universiti Tun Hussein Onn
Malaysia.
Meanwhile, for the controller design,design it has been
model in MATLAB-Simulink link using the mathematical
equations given (6) and (7) and shown in Fig. Fig 3. This block
diagram will be downloaded to the RPi by converting c this
block into the C+ language. This language is generated
generat by the
MATLAB with the specific library function and not been
discussed in this paper.
The value of the PID gains for the each PID gains is
given in Tab.1. Here, two sets of value have been given which
are the gains that based on try and error and the second is
based on Lyapunov equations given in equations
(4),(5),(6),(7).. The rectifier plant is not been discussed in this
paper due to simple transfer function that basedba on the single
phase rectifier without considering the average small sma signal
converter pattern.
Fig. 4 Single phase rectifier in MATLAB blocks
b

____________________________________________________________________________________________________________
Fig. 5 Close loop rectifier output voltage

oltage
Fig. 7 Complete RPI read and write with ADC and DAC blocks
Fig. 6 Duty cycle respond on voltage control

c
Figs. 5 and 6 show the output of the simulation test.
Fig. 8 Rectification output with hardware test
From Fig. 5, the rectifier output shows the respond on full
rectification signal before a filter has been used which give Fig. 8 shows the output of the rectifier before the filtering
filter
about 18Vm on average. This result is achieved due to the process. The input source is supplied for 20Vac. It can be
voltage control has able to respond the target reference. At the seen that, the Vp value is about 15V .
meantime, Fig. 6 shows the duty cycle generating from f the
controller where it constant at 0.7 throughout the simulation.
simulation
B. Hardware Closed Loop Testing

Two types of Raspberry Pi GPIO blocks
block have been
used in hardware closed loop testing which are GPIO write
and GPIO read. GPIO write is connected to the PWM gate
driver meanwhile GPIO read is connected to Arduino digital
output block as to be a input sensor to the PID controller in
order to respond the rectifier output.
The closed loop part involves the additional circuit of
voltage divider and ADC converter from the Arduino due to
limitation of the RPi that cannot received the analogue input
from the hardware. In order to see the effectives of the
rectifier the voltage reference has been set to 15Vrms and
tested using try and error PID values and new gains PID with
Laypunov function.
Fig. 9 Rectifier voltage with try and error PID values

____________________________________________________________________________________________________________
Fig. 9 shows the output of rectifier when it uses only [3]N. P. Bhatia and G. P. Szeg¨ o, Stability Theory of Dynamical Systems,
Springer Verlag, Berlin, Heidelberg, New York, 1970
the try and error PID values. This rectifier output has been set
[4]A. M. Liapunov, Stability of motion, With a contribution by V. A. Pliss and
to 15Vref for the control parameter. At the meantime, the an introduction by V. P. Basov. Translated from the Russian by Flavian
output from the rectifier has an average voltage of 13V which Abramovici and Michael Shimshoni. Mathematics in Science and
is lower than the expected value from the target reference. Engineering, Vol. 30, Academic Press, New York, 1966.
[5]J. G. Ziegler, N.B. Nichols, Optimum settings for automatic controllers,
This indicates that the PID control and the microcontroller are
Trans. ASME, vol. 64, pp. 759-768, pp. 433-444, 1942.
able to work together. [6]G. H. Cohen and G. A. Coon, Theoretical Consideration of Related
Control, Trans. ASME, 75, pp.827-834, 1953.
[7]Xian Hong Li1, Hai Bin Yu1, Min Zhe Yuan, " Design of an Optimal PID
Controller Based on Lyapunov Approach ", IEEE 2009 978-1-4244-4994-
1/09/
[8]Zulkifli, S.A., Hussin, M. N. and Saad, A S, "MATLAB-Arduino as a low
cost microcontroller for 3 phase inverter", 2014 IEEE Student
Conference on Research and Development (SCOReD), 2014 , Page(s): 1
–5
[9] Zulkifli, S.A., Ab Ghani, M.K.A., Deraman, F., Berahim, N., Abdullah,
A.H. and Md Tomari, M.R., “Investigation on various Voltage
Controllers for 3 phase rectifier using Arduino as a Low Cost
Microcontroller”, Journal of Appiled Mechanics and Material, Vol 785,
July 2015
[10]Zulkifli, S.A., Yusof, S.M., Hussian, A.H., Arifin, A.I.M,
Ismail@Marzuki, M.S.N., WanChek, W.A.K., and Tazudin, F.R.M., “
Matlab-Simulink Controller Design for Arduino Target on AC Motor
Control Application”, Inter. Journal. Eng. Pow. Engineering Research
July 2015
[11] S.A. Zulkifli, I. Mamat, M.N.H Mokhtarudin, S.A. Shah, "Investigation
on Raspberry Pi Microcontroller as Digital Processing Controller in
Back to Back Converter", 9th International Conference on Robotic,
Vision, Signal Processing and Power Applications, Springer
Fig. 10 Rectifier voltage with PID Laypunov Functions
Science+Business Media Singapore, Chp: 8, will be appeared in 2017.
Fig. 10 shows the same output but the ripple scale and the
average voltage have been increased to 13.8V when using the
optimal PID value that been calculated using Laypunov
functions parameters. It can be seen that, this small
improvement has made the rectifier output voltage is more
compare to previous case.
IV. CONCLUSION
As a conclusion, this project has showed that, by using a

low cost microcontroller such as the RPi and Arduino, its can
be used as a digital signal processing due to its advantages
which are easy to operate and able to communicate with the
MATLAB-Simulink. Based on the results obtained in the
simulation and in experiment, its show that the Lyapunov
function is able to behave as the PID controller and able to
increase the rectifier output to follow the specific given
reference voltage compare to try and error PID gain selection.
ACKNOWLEDGEMENT
Authors would like to thanks RAGS (R037), Makmal
Kuasa Elektronik, Power and Renewable Energy Team
(PaREnT), FKEE, UTHM to undertake this research activity.
REFERENCES
[1]Kanai, M. M, Nderu J. N., and Hinga, P.K.,Modeling and Analysis of AC-
DC Converter PID Controller Optimized with Pattern Search Algorithm,
Eger. J. Sci. Technol. Nairobi, 2011 11: 104-118
[2]Gerald Teschl, Ordinary Differential Equations and Dynamical Systems,
American Mathematical Society (AMS), Vienna, Austria, April 2012.

____________________________________________________________________________________________________________
The Effect of Titanium Dioxide Nanorods to the

M.K. Ahmad1, N. S. Mohamed Mustakim1, C.F. Soon1, N. Nayan1, R. Sanudin1, S. Abu Bakar2, A. Mohamed2, S. Masaru3, M. Kenji3, M.H.
Mamat4 and M.F. Malek4
1Microelectronic and Nanotechnology – Shamsuddin Research Centre (MiNT-SRC)

2Nanotechnology Research Centre
Department of Physic, Faculty of Science and Mathematics

Universiti Pendidikan Sultan Idris
35900 Tanjung Malim, Perak, Malaysia
3Department of Engineering
Graduate School of Integrated Science and Technology

Shizuoka University
432-8011 Hamamatsu, Shizuoka, Japan
4Nano-ElecTronic Centre
Faculty of Electrical Engineering

UiTM, 40450 Shah Alam, Selangor, Malaysia
*Corresponding author: M.K. Ahmad

Email: akhairul@uthm.edu.my
Microelectronic and Nanotechnology – Shamsuddin Research Centre (MiNT-SRC)
Abstract — Hydrothermal process is a technique used to

synthesize nanostructured Titanium Dioxide (TiO2) I. INTRODUCTION
powder for Dye-Sensitized Solid State Solar Cell (DSSC) DSSC have been extensively researched over the past years
application. The TiO2 powder is then deposited onto glass due to their very high energy conversion efficiency and can be
substrates using Spray Pyrolysis Deposition (SPD) said as low cost of production [1]. Other than that, DSSC is
method. In this study, Fluorine-doped tin oxide (FTO) one of the best alternatives that is based on the work function
conducting glass is used as the substrate for TiO2 nanorods of semiconductor, dyes, and also electrolytes solution that
thin film. Two samples were fabricated with different time have achieved the energy conversion efficiency as high as 10-
durations, 24 hours and 12 ½ hours. There were three 11% at AM 1.5 standard solar spectrum [2]. Although solar
properties studied; structural, electrical and optical cells made from silicon are main option to be chosen, it still
properties. At the end of the research, the TiO2 thin film have competition from other materials because of high
which was fabricated for 24 hours during the manufacturing costs and problems in the electrical transport
hydrothermal preparation process gave the optimum and photo transport. Nowadays, DSSC based on nano-
results. The average gap between grains is small due to the structured TiO2 electrodes have been attracting intense
big particle size of nanostructured TiO2 on the thin film. scientific and industrial attention for quite some time because
Smaller gap between grains on the thin film produce good of their potential for low cost scale up of the corresponding
electron mobility. It indicates that it has low resistivity, synthesis procedure and applicability for diffuse light
hence high ability to conduct electricity. Therefore, the harvesting [3]. Transparent oxide thin films are widely used
TiO2 thin film which was fabricated for 24 hours during materials in new generation solar cells. A film formation of
the hydrothermal preparation is the best candidate for TiO2 thin films can be carried out by various methods such as
DSSC application. sol gel, spray pyrolysis, sputtering, pulse laser deposition and
electro deposition. Among all these techniques, spray
Keywords—TiO2; DSSC; nanorods; power conversion pyrolysis is the most promising as it is a simple, easy to
efficiency handle, cost effective technique for large area deposition and
also excellent method for the deposition of metallic oxides

____________________________________________________________________________________________________________
thin films [4]. The titanium dioxide (TiO2) has a lot of Two samples of TiO2 nanorods thin films prepared by
advantages. The structural properties and high transparency in using hydrothermal method with different parameter used.
visible range make possible application of TiO2 in Table 1 shows the parameter that affected the TiO2 nanorods
microelectronic technology [5]. Thin film of TiO2 has a large thin films for each sample during the powder preparation.
deal of welfare such as wide band gap, high refractive index,
high dielectric constant and absence of toxicity [3]. The Table 1: The parameter used for each sample during the
objective this project is to study the effect of TiO2 nanorods in powder preparation
TiO2 thin film for application as the work electrode for DSSC.
Parameter for each sample
II. METHODOLOGY
Sample HCl Titanium DI Reaction
Temp.
n-Butoxide water Time
FTO conducting glass was used as the substrate for TiO2 (°C)
(ml) (ml) (ml) (hours)
nanorods thin film and the glass substrate was cut into 2.5cm
x 1cm size. The glass substrates were immersed with Acetone, 1 120 5 120 150 24
Ethanol and Distilled Water (DI water) with ratio of 1:1:1.
Then, the substrates were cleaned using an ultrasonic cleaner 2 120 5 120 150 12½
machine for 10 minutes. Lastly, the substrates were dried in
the oven at 150°C for 15 minutes. After that, titanium n-
butoxide was mixed with different amounts of hydrochloric B. Structural Properties
acid (HCl) and DI water. The solution was stirred at the The surface morphology of the TiO2 nanopowder
temperature of 34°C with 300 rpm for 5 minutes. The have been observed using FE-SEM with different
resulting solution was then transferred into a Teflon-lined magnification. After the TiO2 nanopowder went through the
stainless steel autoclave and sealed. grinding process, the result was expected to be the surface
Then, the autoclave was placed into the oven with morphology of TiO2 nanorods powder. The TiO2 nanorods
fixed temperature of 150°C and specific reaction times. After powder was successfully obtained for sample 1 which was
completion of the reaction, the autoclave was cooled down to fabricated for 24 hours and sample 2 which was fabricated for
room temperature. The solution was then filtered and rinsed 12 ½ hours. The results are shown in Figure 2 and Figure 3.
with DI water. Lastly, the filter paper used was placed into the
oven at 150°C for a short time to obtain the TiO2 nanorods in
powder form. Grind the powder obtained with Acetic Acid to
form fine TiO2 nanorods powder. The powder was added into
a porcelain mortar and was grinded with 5.5ml of Acetic Acid
as shown in Figure 3.6. The grinded solution was then filtered
and rinsed with DI water. Lastly, the filter paper used was
placed into the oven at 150°C for a short time to obtain fine
TiO2 nanorods powder. For the deposition process, 5.5ml of
Acetic Acid was added into a porcelain mortar with 0.3g
ofTiO2 nanorods powder. The solution was then grinded for a
short time and 20ml of TiO2 Colloid solution (≈ pH6) was
added. The resulting solution was poured into a brown light
proof bottle and 5 drops of Triton X-100 were added as well. Figure 2: Structure of TiO2 nanorods for sample 1 by
Next, 30ml of Ethanol was filled into the same bottle. The using magnification x10,000
bottle was then placed into the Ultrasonic machine for 30
minutes.
Finally, the cleaned FTO glass substrates were
masked with aluminium foil. After 30 minutes, the final
solution was used to be deposited on the masked glass
substrates. The deposition process done by Spray Pyrolysis
Deposition (SPD) onto the masked glass substrates on the hot
plate set at 150°C. The TiO2 thin film were heated at 150°C
for 30 minutes and lastly were annealed in furnace at 450°C
for 1 hour time.
A. Different parameter during preparation of TiO2 nanorods Figure 3: Structure of TiO2 nanorods for sample 2 by
powder using magnification x10,000

____________________________________________________________________________________________________________
average thickness for thin film from sample 1 is 5.31µm

The surface morphology of the deposited TiO2 thin film for whereas, for thin film from sample 2, it is 5.23µm.
sample 1 and sample 2 are shown in Figure 4 and Figure 5.
The samples have been observed using FE-SEM to determine
the gap between grains on the thin film. As the result, average
gap between grains for sample 1 is smaller than the gap
between grains for sample 2. This is due to the big size of
particle for sample 1 which was fabricated for 24 hours during
the hydrothermal preparation. Smaller gap between grains on
the thin film will produce good electron mobility.
Figure 4: Average gap between grains is 48.4µm
iii. X-Ray Diffraction study

It showed that sample 1 gives higher peak of intensity
at [110] plane than sample 2 (Figure 8). The [110] peak
increased when the fabricated time duration during
hydrothermal process is increased. It should be stressed that
all peaks observed can be assigned to specific lattice planes of
Figure 5: Average gap between grains is 65.9µm rutile phase of TiO2.
ii. The average size of particles of TiO2 nanorods Sample 1

Sample 2
Generally, smaller size of particles shows greater effective 6000
R[110]
surface area for dye adsoption. By using magnification of

x10,000, the FE-SEM images proved the average size of 5000
particles for each sample tested. Figures 6 shows that the 4000
average size of particle for sample 1 is 9.29µm. For sample 2,

Height [cts]
3000
Figure 7 shows that the average size of particle is 8.51µm. As
A[103]
mentioned before, the TiO2 nanorods powder was prepared 2000 A[004]
using hydrothermal method with different parameter for each 1000

R[200]
sample. Grinding process was done to form fine TiO2
0
nanorods powder with smaller size of particles. From the
observation, sample 2 produced finer nanostructured TiO2 26 28 30 32
Pos. [°2Th.]
34 36 38 40
powder compared to sample 1. Therefore, sample 2 has larger

surface area compared to sample 1. The powder obtained from Figure 8: X-Ray Diffraction pattern of TiO2 thin film at
both samples are then used for deposition process of TiO2 different fabricated time duration during hydrothermal process
nanorods thin film using Spray Pyrolysis Deposition (SPD)
method. The thickness of TiO2 nanorods thin film after the E. The efficiency of TiO2 nanorods thin film
deposition process is measured using surface profiler. The Figure 11 shows the 1-V curves for the efficiency of

____________________________________________________________________________________________________________
TiO2 thin film for both samples which were fabricated at ACKNOWLEDGMENTS
different duration during the hydrothermal process. Table 3 The authors would like to thank the Ministry of Education
shows the solar cell performance characteristics (Voc, Isc, (MOE) Malaysia (Vot 1459), Microelectronic and
Pmax, η) to analyze the higher efficiency produced by both Nanotechnology-Shamsuddin Research Centre (MiNT-SRC)
samples. As the result, sample 1 produced higher efficiency and Universiti Tun Hussein Onn Malaysia (UTHM) for
which is 0.0350% compared to sample 2 which only produced financial support
0.0136%. During the efficiency test, the DSSC fabricated used
N719 dye. From the results, we found that the use of TiO2 REFERENCES
nanorods in DSSC leads to low dye adsorption which gives [1] Xu, H., Tao, X., Wang, D. T., Zheng, Y. Z., and
poor photoresponse. Low dye adsorption might be due to the
Chen, J. F. Enhance efficiency in dye-sensitized solar cells
smooth surface of the TiO2 nanorods itself and the dye can
only attach at the tip of the nanorods structure. based on TiO2 nanocrystal/ nanotube double-layered films.
Electrochimia Acta, (2009), pp. 1-6.
Sample 1
Sample 2 [2] Rattanavoravipa, T., Sagawa, T., and Yoshikawa. S.
0.2
Photovoltaic performance of hybrid solar cell with TiO2
0.0
nanotubes arrays. Solar Energy Materials & Solar Cells, 92
2
Current density, mA/cm
-0.2 (2008), pp. 1445-1449.

-0.4
[3] Jasim, K. E. Dye Sensitized Solar Cells-Working

-0.6
Principles, Challenges, and opportunities. Solar cell- Dye
-0.8 Sensitized devices, pp. 171-172.
-1.0
0.0 0.2 0.4 0.6
Voltage, V
0.8 1.0
[4] Manoharan, C., and Sridhar. Physical properties of
spray pyrolysis TiO2 thin films. International Journal of
Figure 9: I-V measurement to test the efficiency of TiO2 thin Recent Scientific Research, 3 (9)(2012), pp. 775 -777.
film for both samples
[5] Wojcieszak. D., Domaradzki, J., Kaczmarek, D., and
Adamiak, B. Photoelectrical properties of TOS thin films
Table 2: The solar cell performance characteristics to analyze based on TiO2 prepared by modified magnetron sputtering.
the higher efficiency produced by both samples The Seventh International Conference on Advanced
Semiconductor Devices and Microsystems, (2008), pp. 283-
Sample Pmax
(mW)
Voc
(V)
Isc
(mA)
Jsc
(mA/cm2)
FF
(%)
η (%) 286.
1 0.008 0.50 0.029 0.116 51.46 0.0350
2 0.003 0.33 0.014 0.054 65.60 0.0136
V. CONCLUSION AND SUGGESTIONS

Through this study, TiO2 nanorods powder for Dye
Solid Solar Cell (DSSC) application have been successfully
synthesized using hydrothermal method. The TiO2 nanorods
powder was well deposited onto Fluorine-doped tin oxide
(FTO) conducting glass substrates by using Spray Pyrolysis
Deposition (SPD) technique. As the result, the rutile TiO2 thin
film which was fabricated for 24 hours during the
hydrothermal preparation process gave the optimum results.
However, it shows that the energy conversion efficiency in
DSSC based on rutile TiO2 nanorods thin film is still
relatively low. This is largely due to the limited internal
surface area and the structure of the TiO2 nanorods itself
which performed low dye adsorption.

____________________________________________________________________________________________________________
Air Pollution Monitoring Integrated on Smart

Home System Using ZigBee Wireless Sensor
Networks
Octarina Nur Samijayani, Linear Addien, and Hadiansyah Ismawan

Electrical Engineering Department, Faculty of Science and Technology
University Al Azhar Indonesia
Jakarta, Indonesia
Octarina.nur@uai.ac.id
security, energy saver for several appliances, and air pollution

Abstract— In densely populated city, air pollution should monitoring. In this paper the smart home system is developed
become a concern because it can gradually be harmful to the as the prototype using three Sensor Nodes and one
human body. The implementation of Wireless Sensor Networks Coordinator Node as the sink node. Sink node is connected to
(WSN) for air quality monitoring can be further integrated with
the Internet of Things (IoT) technology. This paper discusses the the computer server as the monitoring and controlling center.
implementation of WSN to monitor the level of CO and CO2 in
air, which is integrated on Smart Home system. Air pollution II. AIR POLLUTION MONITORING
level detected by Sensor MQ-7 (CO) and MG-811 (CO2) in Many studies developed the air pollution monitoring
several Sensor Nodes will be sent to Coordinator Node. Smart
Home system have some basic features including home system using variety of technology. In 2015, air monitoring
appliances and home security monitoring, then it improved with system was deployed using WSN by Ahmed Broubima et.al.
air pollution monitoring. It is also able to control the home WSN are considered as a suitable technology for this purpose
appliances remotely. Gas sensor module has been calibrated and due to their substantial benefits including low cost and
tested with the error rate of about 38.95% and 2.39% for CO autonomy [3]. Ref [4] presents a system which provides a low
and CO2 reading, respectively. Data transmission from three cost data acquisition of CO and smoke pollutants that can be
Sensor Nodes to Coordinator Node was tested using transparent
mode and API frame mode, with maximum indoor range is read online for air pollution monitoring for environmental
about 20 meter between nodes. Data of air pollution are sent and awareness. The air pollution monitoring application was
displayed on serial monitor on PC in real time. build bases on internet, which the information can be
accessed anywhere and in real time.
Keywords: WSN, ZigBee, Air Pollution, Smart Home. Study done by Fuertes et.al which developed a distributed
system as IoT for a new low cost air pollution wireless
monitoring in real time. Agile methodologies such as Scrum
I. INTRODUCTION and Extreme Programming were used in order to ensure
software quality. The electronic device is equipped with three
A ir pollution is one crucial factors affecting life and health

of human [1]. To provide better and safer environment
for human being, animals and plants we need to monitor the
sensors, which determines carbon monoxide (CO) as well as
carbon dioxide (CO2) concentrations and powder density [5].
Another project deploy WSN based pollution monitoring
pollution which is created around us. This project propose the system in metropolitan cities using Bluetooth to connect the
integrated system of smart home, which has the ability to controller with client and the client connects with the server
monitor the air pollution including the CO and CO2. The via web services [6]. Wireless sensors are used to calculate
system implement Wireless Sensor Networks (WSN) using the percentage of harmful gases present in the air and also it
Zigbee standard. forecasts to avoid future pollution and send the warning
WSN is a distributed set of sensors deployed to work message to the particular polluted area.
together for collective sensing and possible data processing. Recent application of air pollution monitoring system was
WSN can be used to monitor environmental behavior and implemented also in school environment. A real time ambient
structural integrity in a variety of application fields, thus air quality monitoring wireless sensor network was developed
becoming an integral part of the consumer electronics of for schools in smart cities by H. Ali et.al in 2015 [7]. In order
smart buildings in smart cities [2]. The proposed system is to help schools in smart cities to assess whether ambient air
supporting the Internet of Things (IoT) application, of which exposure is unhealthy for students, a low cost solar powered
a physical object is embedded with electronics, software, air quality monitoring system is used based on ZigBee
sensors and wireless connectivity to allow monitoring air wireless network system technology. Many recent studies
pollution on real-time [3]. bring a potential of the WSN system since it is low cost and
The proposed smart home system aims to provide the able to cooperate many sensors to gather information in real
monitoring and controlling system, including the room time.

____________________________________________________________________________________________________________
III. SYSTEM DESIGN WSN Smart Home prototype is containing three Sensor
Nodes and one Coordinator Nodes. Sensor nodes are placed
A proposed Smart Home is based on WSN principle using in three different rooms, containing multi sensors; PIR
Zigbee standard which working on 2.4GHz, with maximum sensor, CO and CO2 sensor, and switch appliances. The
data rate is 250 kbps. The transceiver used in this system is communication between Sensor Nodes and Coordinator
Xbee series 2 basic and pro series, which can transmit power Node is working in two ways. From SN to CN for monitoring
about 2mW and 63mW respectively. and from CN to SN for controlling function. The processor is
Smart home is designed to work as the monitoring and using Arduino UNO Rev 3 and powered by 5 V from battery
controlling system. As a monitoring system, Smart Home or adaptor. The routing and sensing processing is done on
monitor the presence of human, home appliances and air software Arduino IDE and XCTU to configure the API frame
pollution, while as the controlling system, Smart Home is based communication. General specification of Smart Home
able to control the home appliances remotely. In supporting system is specified on Table 1.
the home safety, the presence of human in a room is detected
using PIR sensor. Some home devices are connected to the Table 1. Specification of Smart Home Monitoring System
Sensor Node to monitor the condition of devices. Air Microcontroller ATmega328P
pollution is detected using CO and CO2 sensor, MQ-7 and Digital I/O Pins 14 (of which 6 provide PWM
output)
MG-811, respectively. The controlling function allow the
PWM Digital I/O Pins 6
house owner to activate or turning off the appliances
Analog Input Pins 6
remotely. This feature may support the energy saving when
Flash Memory 32 KB (ATmega328P) of
the owner are away from the house. which 0.5 KB used by boot
Generally, the WSN system component are divided into loader
Sensor Nodes (SN) and Coordinator Nodes (CN). Sensor SRAM/EEPROM 2 KB (ATmega328P)/1 KB
Nodes has some input data from several sensor. One Sensor (ATmega328P)
Node consist of a set of sensor, microcontroller included the Transceiver Xbee Series 2 (basic and Pro
ADC part, Zigbee transceiver, and power supply. Component series)
of Sensor Node are depicted in Fig1. Transmit Power (2 mW and 63 mW)
Input Sensor CO2 Sensor, CO Sensor, PIR
Sensor
Actuator 2 appliances per-Sensor Node
IV. RESULTS AND DISCUSSION
A. CO and CO2 Measurement Result

Figure 3 and 4 shows the measurement of CO and CO2
indicating pollution level in air in indoor. DT sense with MG-
811 Sensor is used to detect CO2 level and MQ-7 Sensor is
used to detect CO level. These result presents the
measurement result received in Coordinator Node and
displayed on PC in monitoring center. CO and CO2 reading
Fig. 1. Sensor Node Component in Coordinator node is then compared with the reference in
order to evaluate the system performance. DT- SENSE sensor
Sensor Node will send the sensor data wirelessly to module has been calibrated with reference ppm meter, and the
Coordinator Node. The wireless communication is built using experiment shows that CO sensor working with average error
Zigbee protocol using Xbee Series 2 component. Coordinator of about 38.95% and CO2 sensor with 2.39% of average error
Node is connected to the internet using XBee gateway or in temperature of 270C.
directly connected to a PC as monitoring center. The
communication between Sensor Node and Coordinator Node CO2 Experiment Result
are provided in Fig.2.
CO2 Sensor CO2 Meter
1200
1000
CO2 LEVEL (PPM)
800
600
400
200
0
1 2 3 4 5 6 7 8 9 10
EXPERIMENT NUMBER
Fig. 2. Sensor Node and Coordinator Node Component Fig. 3. Experiment result of CO2 sensor testing

____________________________________________________________________________________________________________
CO Experiment Result
CO Sensor CO Meter
10
CO LEVEL (PPM)
0
1 2 3 4 5 6 7 8 9 10
EXPERIMENT NUMBER
Fig. 4. Experiment result of CO sensor testing
The concentration of CO2 in indoor will increase if the

number of people is increased. CO2 measurement is done in
one room with 4 m2 area, it produces 600-900 ppm of CO2.
The concentration of CO2 which reach 5000 ppm can
interfere and may endanger to human health, moreover signs Fig. 5. Display of Data Sensor Transmission in XCTU and Serial Monitor
of intoxication will be produced by a 30-minute exposure at
50,000 ppm, it will immediately dangerous to life [8]. The
Another important feature of this smart home monitoring
measurement result in figure 4 indicates that the CO2 in
system is the range between node since this communication
specified room is still in normal level. This condition is
is built in wireless condition. Zigbee communication testing
reported in display monitoring system.
result with various range is presented in Table 3. The
Measurement result in Figure 4 indicating that CO level in
measurement is done for range 5 m to 40 m in indoor. The
one room is about 1-2 ppm. The average level of CO in indoor
power level is relatively decreased with the increasing of
is about 0.5 – 5 ppm and it has been stated that a 1-hour
range, and it will lead to packet loss. The data sensor
exposure to 1,000 to 1,200 ppm would cause unpleasant but
transmission is lost and no connection when range is reach 30
no dangerous symptoms, and 1,500 to 2,000 ppm might be a
m above. Fig 6 shows the graph of power decreasing when
dangerous concentration after 1 hour [9]. This condition is
the range is increased.
reported in display monitoring system, otherwise when CO
and CO2 detected as air pollution is reach the endanger
Table 3. Communication testing with variation Range
threshold level, the system will give a warning indicator.
Range Power Packet lost Result Time
between (dBm) Response
B. Zigbee Communication Experiment Result Xbee (m) (mean)
5 -76 0 Ok 1s
The wireless communication can be built with various
10 -75 0 Ok 1s
option of API Frame type and supported with frame generator
on XCTU enable the communication of API Frame based. 15 -75 0 Ok 1s
The flexibility of frame which had wide range data slot makes 20 -74 1 Ok 1s
communication of API frame based easier to be used. The 25 -79 1 Ok 1.5s
system is working properly to deliver multi input sensing data
30 -78 8 Ok 2s
with maximum pin which can be used as input or output are
8 pins, include 4 pins as analog input. Digital input/output 35 -78 7 - -
requires 2 byte data slot for 8 pin available, while analog 40 Xbee-81
Range Testing
2 in indoor
- -
requires 2 bytes data slot for each pin used. Memory used for Power (dBm)
maximizing the function of pin in frame type (0x92) is 30 -70
bytes within 100 bytes, with Xbee transfer rate can reach 250
-72
kbps.
Power (dBm)
In order to test the communication between Sensor Node -74
and Coordinator Node, the packet data containing the CO and -76
CO2 data is displayed on XCTU. Figure 5 shows the Display -78
of Data Sensor transmission in XCTU. The packet contain is -80
extracted and displayed in XCTU and Arduino Serial
-82
monitor. 5 10 15 20 25 30 35 40
Power (dBm) -76 -75 -75 -74 -79 -78 -78 -81
Range between Xbee
Fig. 6. Experiment of Xbee Range Testing

____________________________________________________________________________________________________________
Based on the range testing communication presented in

Table 3 and Fig 6, the power level in range of 5, 10, 15 and
20 meters is maintained in the level around 74 – 76 dBm. This
range still provide good communication between nodes.
However, for range above 25 meter, the power level is
decreased into 78-81 dBm, and also there is packet data loss.
Drastically when the range reach 35 meters, the
communication between nodes is failed to be connected.
V. CONCLUSION
Air pollution monitoring application integrated with the

Smart Home system is proposed using the Wireless Sensor
Networks working on Zigbee standard. The percentage error
of CO and CO2 data sensor reading is about 38.95% and
2.39%, and communication range maximum is about 30
meter. This proposed system provide more function for smart
building concept, and it is working for monitoring and
controlling as well. The WSN system bring more beneficial
features as it is low cost and easy to be configured using API
type of wireless communication which allow multi input data
sensor.
ACKNOWLEDGMENT
We wishes to acknowledge the Directorate General of
Higher Education (DIKTI) of Indonesia and
Telecommunication Laboratory of Electrical Engineering
under Faculty of Science and Technology in University Al
Azhar Indonesia.
REFERENCES
[1] Souhir Bedoui, Sami Gomri, Hekmet Samet, “Abdennaceur Kachouri:
1. Air pollution monitoring system using LabVIEW”, Systems,
Signals & Devices (SSD), 2015 12th International Multi-Conference
on, 1 – 6, March 2015.
[2] By Madhupreetha L. Rajaram, Elias Kougianos, Saraju P. Mohanty,
and Uma Choppal, “Wireless Sensor Network Simulation Frameworks:
A Tutorial Review; MATLAB/Simulink bests the rest”, IEEE
Consumer Electronics Magazine, vol.5, 63 – 69, April 2016.
[3] Ahmed Boubrima, Frederic Matigot, Walid Bechkit, Herve Rivano,
“Optimal Deployment of Wireless Sensor Networks for Air Pollution
Monitoring”, 2015 24th International Conference on Computer
Communication and Networks (ICCCN), 1-7, Aug. 2015.
[4] Abdul Hadi Nograles H.1 , Agbay, Christopher Paolo D.2 , Flores, Ian
Steven L. 2, Linsangan, Manuel Jr. A. 2 , Salonga, John Bethany C.
“Low Cost Internet Based Wireless Sensor Network for Air Pollution
Monitoring using Zigbee Module” Digital Information and
Communication Technology and it's Applications (DICTAP), 2014
Fourth International Conference on, 310 – 314, May 2014.
[5] Walter Fuertes et.al, “Distributed System as Internet of Things for a
New Low-Cost, Air Pollution Wireless Monitoring on Real Time”,
2015 IEEE/ACM 19th International Symposium on Distributed
Simulation and Real Time Applications (DS-RT), 58 – 67, Oct. 2015.
[6] Shwetal Raipure, Pune, Deepak Mehetre, “Wireless sensor network
based pollution monitoring system in metropolitan cities”, 2015
International Conference on Communications and Signal Processing
(ICCSP), 1835 – 1838, April 2015.
[7] H. Ali, Callaghan, J. K. Soe, Steven. R. Weller,” A real-time ambient
air quality monitoring wireless sensor network for schools in smart
cities”, Smart Cities Conference (ISC2), 2015 IEEE First International,
1-6, Oct. 2015.
[8] Carbon dioxide: IDLH Documentation", Centers for Disease Control
and Prevention. The National Institute for Occupational Safety and
Health, Updated December 4, 2014, Retrieved February 2015.
http://www.cdc.gov/niosh/idlh/124389.html.
[9] Carbon monoxide: IDLH Documentation", Centers for Disease Control
and Prevention. The National Institute for Occupational Safety and
Health, Updated December 4, 2014, Retrieved February 2015.
(NIOSH)http://www.cdc.gov/niosh/idlh/630080.html.

____________________________________________________________________________________________________________
Dual-Port SRAM Design with Low Power Technique
Nabihah Ahmad and Nurdyana Mat Jainudin

Nano Simulation Group (NanoSIM), Department of Electronic Engineering, Faculty of Electrical and Electronic Engineering,
Universiti Tun Hussein Onn Malaysia. Malaysia.
nabihah@uthm.edu.my
Abstract— Static Random Access Memory (SRAM) filled (Dual-Vt) and adding the stacking transistors in the cell. This
more than half of the silicon area on System on Chip (SoC), and technique able to reduce the leakage power consumption.
consumed the major portion of the power and energy budget. Power consumption is improved about 89%, however its delay
Therefore, it needs to satisfy a low-power consumption increases about 40%. Singh et al. [7] modified basic eight
requirement without scarifying the performance. This paper transistors (8T) cell by applying pass transistors technique.
aimed to compare a conventional DP SRAM and DP SRAM with Power consumption improved about 20% and the write delay
low power techniques which are power gating, full stack and about 8%. Report in [8] modified the 1-RW cell and focused
pass transistor techniques implemented in 130-nm CMOS on the layout compaction. Write delay is about 17% improved
technology. The simulation and comparison has been carried out as well compare to the conventional design.
by using Cadence Virtuoso. Based on the comparison result, a
full-custom Dual-Port SRAM (DP-SRAM) of 2-Read/Write (2- Another technique is by using MTCMOS where leakage
RW) type with Multi-threshold Complementary Metal Oxide current could be minimized and a large scale (macro) of
Semiconductor (MTCMOS) power gating technique have been SRAM could be divided into multiple sub-macros to improve
designed. This technique used to reduce the overall power the read and write stability and ability and also reducing the
consumption and delay of the SRAM architecture and optimized
the SRAM performances. This design achieved a power
dynamic power. There are three types of Vt which are low,
dissipation of 12.60µW and propagation delay is 73.52ps. The normal and high-Vt. Low-Vt (LVt) will give a high
result demonstrated that the power consumption of DP SRAM performance in the circuit design because the propagation
with MTCMOS technique achieved 24.86% smaller than the delay could be reduced in the critical path and can be used
conventional memory cell, while the propagation delay improved whenever high performance is required. However, it could
about 67.35%. causes the leakage current to be increased in the sub-
threshold. High-Vt (HVt) can be used to reduce leakage
Keywords— DP-SRAM, MTCMOS, power gating, low currents and the power consumption in the shortest path [3].
power
Design in [9] and [10] modified the 1-RW cell. In [9], the
power is reduced about 9%, yet the read and write delays
I. INTRODUCTION improved about 80%. Zhu et al. [10] designed the 128x64-bit
Memory subsystem in System on Chip (SoC) plays a major memory architecture of SRAM with its power consumption
part in the silicon are which 70% of the SoC silicon area are and delay is reduced but only with a small amount. Mohan et
occupied with the static Random Access Memory (SRAM) as al. [11] introduced the MTCMOS power gating. It modified
it provides a link with CPU [2]. So it is crucial to design a low the conventional 9T and makes the number of transistor to
power and high speed design. SRAM is volatile where the data increase became 15T. Hence, by using this technique the
will be stored as long as supply is applied. In addition, it is power is improved about 66%. In contrast, write delay
able to work at an acceptable speed compare to other type of increases after the modification. Besides, the supply voltage
memory. Unlike DRAM, SRAM cells do not need to be and CMOS technologies also play an important roles in
refreshed. Thus, they are available for reading and writing data affecting the power consumptions and delay [12]. Selection of
at most all of the time. Besides, they also consumed much low CMOS technology also able to enhance the design
power and very compatible with CMOS technology. performance, as it has been stated that the leakage current will
The existing SRAM design in [3] reported the increase for the technologies below 100-nm, thus, affecting
maximization by using the MTCMOS technique that plays the overall power consumption as well [15].
with different threshold voltage (Vt). In research article [4] and II. DP- SRAM ARCHITECTURE
[5], both improved the conventional design of DP-SRAM with
2-RW memory cell type. Due to the low power and high A full custom asynchronous DP-SRAM has been designed
stability issues in designing the SRAM, there are a various and implemented in 130-nm CMOS technology. Fig. 1 shows
techniques such as full stack [6], pass transistors [7,8], and the memory architecture of SRAM with several sub-circuits
MTCMOS [9-11]. In [6], it modified the conventional Single- needed to operate the SRAM.
Port SRAM (SP-SRAM) by applying dual threshold voltage

____________________________________________________________________________________________________________
the virtual VDD (supply voltage) and GND (ground). The LVt,
Pre-Charge which is the conventional 2-RW cell is utilized, permit cell to
operate at low power supply values with sufficient speed [14].
While during standby mode (inactive), the SLEEP signal is
Address Decoder
activated by supplying the signals with HIGH signal (1V etc.).
RAM Array
The HVt is turned OFF and causes the virtual VDD and GND
Address input
(consists of SRAM
cells) to float, hence limiting the leakage current. This allowing the
LVt transistors to work normally, at low leakage current. Fig.
Data in 3 shows the MTCMOS with power gating technique.
Data out
Sense Amplifiers and
Write Drivers
Fig. 1. SRAM memory architecture.
A. Conventional 2-RW DP SRAM

DP SRAM cells is used in order to achieve a higher SRAM
density. As each port must be able to access the cell
independent of each other the basic access lines; bit line, bit
line bar and word line need to be duplicated. There are two Fig. 3. DP SRAM with MTCMOS power gating technique.
types of memory cells are used for the DP SRAM, the 1-
Read/Write (1-RW) memory cell and 2-RW memory cell C. Full Stack Technique
[13]. Both of the memory cells have eight transistors (8T) in
common. 2-RW can operated as 1-RW and its structure is Stack effect is the condition when two or more stacked
similar with the conventional SP SRAM which consists of six transistors are turned off together, thus able to reduce the
transistors (6T). As compared with 1-RW, the 2-RW cell leakage power consumption [6]. Fig. 4 shows the combination
more access-flexibility of data. This 2-RW or known as of the DP SRAM with the full stack technique. Two PMOSs
multiple port type of DP SRAM comprises two cross-coupled were stacked on the top of the pull-up transistors, M1 and M2
inverters from M1 until M4 that are used to store the data, the of the conventional DP SRAM cell and the other two NMOSs
rest M5 until M8 are used to access the stored data in the cell. were stacked at the bottom of pull-down network, M3 and
It has two bit lines where either one of the bit line can be used M4. These PMOSs and NMOSs are named as the PMOS
to carry the data and stored in the memory cell whenever the stacking transistors and NMOS stacking transistors
corresponding word lines are activated. Thus, both of the respectively. The reverse bias will be induced between the
ports in this 2-RW type are able to do the read and write stacked transistors and the memory cell and the two cross-
operation at the same time. Fig. 2 shows the circuit diagram coupled inverters, thus resulting a lower sub-threshold
of the conventional 2-RW DP SRAM. leakage current. However, this would increase the delay and
VDD affect the circuit performances due to the large number of
transistors.
VDD
BL2 BL2
BL1 BL1
WWL2
WWL2
WWL1 WWL1
GND
BL2 BL2
Fig. 2. Circuit diagram of conventional DP SRAM cell.
BL1 BL1
WWL2
WWL2
B. MTCMOS Power Gating Technique
WWL1 WWL1
Two sleep transistors are applied between the supply
voltage and the LVt gate, and between the ground line and the
HVt device is known as power gating. The sleep transistors
will reduce the subthreshold leakage during sleep mode. There GND
are two modes in this technique. During active mode, the
SLEEP signal is deactivated by supplying the sleep transistors Fig. 4. DP SRAM with full stack technique.
with LOW signal (0V). The HVt, which are the sleep
transistors is turned ON and causes the LVt to be connected to

____________________________________________________________________________________________________________
D. Pass Transistor Technique TABLE I

COMPARISON OF DIFFERENT TECHNIQUES APPLIED TO THE SRAM
The gate of PMOS pass transistor is connected with the DESIGN
ground while the gate of NMOS pass transistor is connected Total
Rise Fall
with voltage supply. This keeps the PMOS and NMOS low Techniques
Power
Time Time
Propagation
and high, respectively, all the time. The concept of this pass (µW)
(ps) (ps)
Delay
(ps)
transistor technique is to break any short circuitry in the cell.
50.69 68.88 40.28 109.16
The circuit of this technique is shown in Fig. 5. Conventional
Full Stack 41.18 53.20 34.69 74.02
VDD
Pass
25.20 44.82 29.20 87.89
Transistor
GND
MTCMOS 12.60 46.03 27.49 73.52
BL2 BL2
B. DP-SRAM with MTCMOS Power Gating Technique
BL1 BL1
The proper and suitable signal must be given to the sleep
transistors in order for the memory cell to work properly.
WWL2 From Fig. 6. during both word lines labelled WL1 and WL2
WWL2
are LOW, output Q observed from the cross-coupled inverter
WWL1 WWL1 is LOW. When WL1 HIGH and WL2 LOW, bit line BL1 is
VDD selected at output Q. When WL2 HIGH and WL1 LOW, BL2
is selected at output Q. Finally when both word lines are
GND HIGH, the output Q is the AND operation from both BL1 and
BL2. During the active mode, the sleep transistors are ON and
Fig. 5. DP SRAM with pass transistor technique.
causes the memory cell to be supplied with the virtual VDD
and GND, thus allowing the circuit to run at low power
supply values with sufficient speed. While during the inactive
III. RESULTS AND DISCUSSION mode, the memory cell is working normally. However, the
output Q generated during this mode is slightly smaller than
A. Comparison of DP SRAM designs during the active mode. This is because when the sleep
The conventional DP SRAM and different low power transistors are OFF, the memory cell is forces to pass through
techniques of DP SRAM are designed and simulated in some resistance value of the transistors in order to be
Cadence Tools using 130nm CMOS technology with 1V connected with the actual VDD and GND. This, as much will
supply. The simulation results obtained from the conventional affecting the output voltage and hence, limiting the leakage
DP SRAM and low power techniques observed in this project current.
is tabulated in Table 1. Based on the table, the DP SRAM Both WL1 and WL1 =1 WL1 =0 Both WL1 and
with MTCMOS power gating technique has the least power WL2 = 0 WL2 = 0 WL2 = 1 WL2 = 1
consumed in the cell and the shorter propagation delay

compared to the others. By applying the MTCMOS power
gating technique, there was significant improvement to both
total power consumption and propagation delay in the cell BL1 pulse
when being compared with other low power design

techniques. It achieved 24.86% smaller than the conventional BL2 pulse
memory cell, while the propagation delay improved about
67.35%.
Output Q
No output at Q Q follow BL1 Q follow BL2 Q from AND

( Hold state) pulse pulse operation of BL1
and BL2
Fig. 6. Output waveform of DP SRAM with MTCMOS power gating

technique.
The layout of the design in shown in Fig. 7. The area of the

layout is about 9µm x 17µm.

____________________________________________________________________________________________________________
17um in Proc. IEEE Conference on Information and Communication

Technologies, 2013, pp. 1243-1246.
WL1 SLEEP [6] B. Rajlaxmi, T. Pramod Kumar, R. Obaleppa, Dasar, S. Narmada
SLEEPB WL2
VDD and H. P. Rajani, “Design and Implementation of Low Leakage
Power SRAM System Using Full Stack Asymmetric SRAM,”
International Journal of Advanced Research in Electrical, Electronics
and Instrumentation Engineering, Vol. 2, Issue 7, pp. 3083-3091,
July 2013.
[7] A. A. Singh, A. Choubey and R. K. Maddheshiya, “Low Power
SRAM Cell with Improved Response,” International Journal of
Research in Engineering and Technology, Vol. 3, Issue 6, pp. 483-
488, Jun-2014.
BL2 [8] S. A. Bota, B. Alorda, G. Torrens and J. Segura, “Pass-Transistor
9um
BL1
BL2B PMOS Based 8T SRAM Cell For Layout Compaction,” in Proc.
BL1B IEEE CDE, 2011.
[9] A Mitra, “Design and Analysis of an 8T Read Decoupled Dual Port
SRAM Cell for Low Power High Speed Applications,” International
Journal of Electrical, Computer, Energetic, Electronic and
Communication Engineering, Vol.8, No.4, pp. 697-700, 2014.
[10] H. Zhu and V. Kursun, “Novel Triple-Threshold-Voltage Eight-
Transistor SRAM Circuit with Enhanced Overall Electrical Quality,”
IEEE, 2012.
GND [11] S. Mohan Das, K. S. Kiran Kumar and A. Madhulata, “A Novel Low
Power Energy Efficient SRAM Cell with Reduced Power
Consumption using MTCMOS Technique,” IOSR Journal of VLSI
and Signal Processing (IOSR-JVSP), vol. 5, issue 1, ver. 1, pp. 42-
Fig. 7. Layout of DP SRAM with MTCMOS power gating technique 49, Jan-Feb. 2015.
[12] S. Sapna, A. Neha, S. Meenakshi and G. Neha, “Performance
Evaluation Of Different Sram Cell Structures At Different
Technologies,” International Journal of VLSI design &
IV. CONCLUSION Communication Systems (VLSICS), Vol.3, No.1, pp. 97-109,
February 2012.
The DP-SRAM has the ability to read and write data [13] A. Mohammad Mahdi, I. R. Mamun, A. Md. Syedul Amin and F.H.
simultaneously. This project compared the DP-SRAM of 2- Hashim, “Design of an 8-cell Dual Port SRAM in 0.18-µm CMOS
RW type with applying MTCMOS power gating technique Technology,” Research Journal of Applied Sciences, Engineering
and Technology, pp. 2565-2568, Mar. 2013.
with conventional DP SRAM and other low power [14] K. Sung-Mo, L. Yusuf and K. Chulwoo, “Semiconductor
techniques. DP SRAM with MTCMOS power gating Memories,” in CMOS Digital Integrated Circuits Analysis and
technique has the least power consumed in the cell and the Design, 4th ed., New York: McGraw-Hill, 2015, pp. 402-450.
[15] M. Manghisoni, L. Ratti, V. Re, V. Speziali, and G. Traversi,
shorter propagation delay compared to the others. Therefore, “Resolution Limits in 130 nm and 90 nm CMOS Technologies for
this proved that the MTCMOS power gating technique able to Analog Front-End Applications”, IEEE Transactions On Nuclear
improve the overall performance in DP SRAM architecture. Science, Vol. 54, No. 3, pp. 531-537 , June 2007.
ACKNOWLEDGMENT
This work was supported by Universiti Tun Hussein Onn
Malaysia.
REFERENCES
[1] T.S. Rajesh Kumar, “On-Chip Memory Architecture Exploration of
Embedded System on Chip,” Ph.D. dissertation, Dept.
Supercomputer Education and Research Centre, Indian Institute of
Science, Bangalore, September 2008.
[2] M. Rajeshwari, K. Geetanjali and S.K.Shirakol, “Dual port SRAM,”
in Proc. IJCET- NCWSE, 2013, pp. 348-352.
[3] W. Bo, Z. Jun and T. Kim Tony, “Maximization of SRAM Energy
Efficiency Utilizing MTCMOS Technology,” IEEE 4th Asia
Symposium on Quality Electronic Design, pp. 35-40, 2012.
[4] Y. Ishii, H. Fujiwara, K. Nii, H. Chigasaki, O. Kuromiya, T. Saiki,
A. Miyanishi and Y. KiharaA, “28-nm Dual-Port SRAM Macro
With Active Bitline Equalizing Circuitry Against Write Disturb
Issue,” in IEEE Symposium on VLSI Circuits/Technical Dig. Tech.
Papers, pp. 99-100, 2010.
[5] Balakrishna Kankanala, Avireni Srinivasulu and Sarada Musala, “7-
T Single End and 8-T Differential Dual-Port SRAM Memory Cells,”

____________________________________________________________________________________________________________
Design and Implementation of M2M Technology in

Smart Street Light System Node with RS 485
Communication and Hybrid Power Source
Gunawan Wibisono#1, Antonius Putranto#2
#
Electrical Engineering, University of Indonesia
Kampus Baru UI, Depok, Indonesia
1gunawan@eng.ui.ac.id 2antonius.listyo@ui.ac.id
Abstract— M2M technology is implemented on a smart street

lights system based on RS 485 serial communication with bus Nowadays, street lighting become an important facilities in
network configuration and using a hybrid power source. This a city. Street lighting has several function such as provide
system use Arduino Nano 328p to control 18 Watt LED lights light at night, improve security and also prevent criminality.
which has the power source of 12 VDC. The system is also Almost all luminaires that are used in street is conventional
equipped with sensors that detect and capture parameters
luminaires which is responsible for the consumption of 40%
around street lights. RS 485 serial communication is used as a
communication system between the coordinator and node lights. of city budget [1]. The conventional luminaires is street
From the test results can be seen that the system node smart lighting system that are not able to do communication, control
street lights can work as designed algorithms (resources and management features such as the ability to measure the
switching algorithms and damage system detection algorithms). information that is useful from the system, energy
On data communication testing for transmitting data from node consumption savings or tracing system failure [2]. Therefore
to coordinator obtained data reception success rate of 97.14%. the technology M2M proposed to create a smart way light
Then from efficiency results obtained total power usage of smart system that is more efficient in the use of energy and able to
street lights per year amounted to 65.87 KWh. So smart fulfill the function of monitoring, control and data storage.
streetlight system node can make more efficient in the process of
Research on [3] and [4] made the system M2M smart street
operation, monitoring and maintenance
lamp using microcontroller ATMega32 as device to control
the street light . Zigbee wireless communication 2.4 GHz is
Keywords— M2M, smart street light system, Arduino Nano used as wireless transmission between the coordinator of light
328p, RS-485, serial communication, bus network configuration, and the light node. There are some disadvantages of this
hybrid power resource system in which susceptibility to high temperatures on the
I. INTRODUCTION side of the node, susceptibility through Zigbee data delivery
failure and the maximum distance from each node [4].
In this research, author design clustering system on smart
street lighting where each cluster consist of group of street actuator (relay on-off and relay power source) and RS 485
lights (nodes) that will be controlled by the coordinator. communication. Sensors will capture the lamp information
Coordinator will perform on-off control, monitoring and such as movement detection, failure detection and capture
failure detection if one of node has failure. This system was power consumption. Actuator will perform on-off switching,
proposed to perform efficiency in the process of the dimming and power source switching based on sensor
controlling and monitoring. Each node has a movement detection.
detection sensor (PIR), voltage sensor and current sensor and Node communication system uses RS485 serial cable to
use hybrid power source where the main source of energy perform communication beetween node and coordinator using
from sunlight from the solar cells and the back up source from bus topology. Coordinator will be the master control and
PLN (power plant). RS485 serial cable communication with nodes will be slaves. Each node will responsible to receive
bus topology is used between nodes and light coordinator that order from coordinator to on/off the lights and to transmit
are more reliable and good noise immunity from information data. The information data is consist of lamp
environmental conditions. This research design and build node energy consumption (current and voltage), light status (on, off,
embedded system based on smart street light clustering system. or broken), status of battery, power source status and failure
system status.
II. SMART STREET LIGHT SYSTEM NODE In order to save energy consumption, this system use LED
The node system is design to automatically manage the lights and on-grid technology that uses solar cells as main
light consumption based on the environment around the lamp. power resources. The system was designed to be able
Fig. 1 shows, system is controlled by microcontroller and dimming the lights according environment around the light
equipped with sensors (light, presence, current and voltage), node if there is a movement from object or human. The

____________________________________________________________________________________________________________
system automatically switch the power sources if the battery Start
cappacity is below the setpoint and use PLN source. The

system also has failure detection if the lamp or the battery is
System
initialize
broken. Serial Receive

B
Mode (Rx)
Capacity Battery
N Damage Y Failure Status :
> setpoint battery ? Battery Failure
Y N
Power Source : Power
Battery Source : PLN
Y
Serial Receive Check On/off order
Fig. 1 Smart street light node system chart N

Light Light
status = 0 status = 1
The components use for the design are as follows :
a) LED lights 12 VDC 18 watt as the main light
resource. Light status check Off
b) Microcontroller Arduino Nano 328p is used as the
device controller on the node embedded system. On
c) Passive infra-red (PIR) is presence sensor thath ses to
detec human presence around the lights node. Movement Dimmer On
d) Voltage sensor is used to measure the value of the detection ? Y
voltage on the battery and lights. Voltage sensor also N
detect if the battery is broken. Dimmer Off
e) Current sensor is used to measure the current flow in

the lights. Current sensor also detect if the lamp is Capture power
consumption
broken (no current flow in the lights).
f) IC MAX 485 is used to perform thewired RS 485
communication between light node and lights Y
coordinator. Lamp failure ? ( I = 0 and Failure status :
g) Solar panel and battery are used to convert sunlight Lamp status =1 ) Lamp failure
into electrical energy. N

Failure status :
In Fig. 2 Shows the communication architecture in

Normal
clustering system. One coordinator is responsible to manage

severeal nodes in one cluster. Coordinator gives order to
Y Y
several node is cluster to operate on/off lights and ask for each Serial receive Update Status Sent
mode (Tx)
check Order
node to sent their cureent status every periodical time. The N
N
nodes is responsible to manage the lights locally and response Sent data to
to coordinator order. If failure is detected, node will sent the coordinator
failure notification to coordinator and coordinator will sent to

the webserver and redirecting to the engineer via website or Y
short message service (SMS). Is system on ? B
N
End
Fig. 3 Flowchart system
III. DESIGN AND IMPLEMENTATION

In order to implements the system, the node expansion
board is design in altium designer software to make a compact
PCB that embed the components such as arduino nano 328p,
IC MAX 485 and relay system. Sensors will be connect
through the socket in order to easy installation and
replacement. Fig. 4 show the design layout system and Fig. 5
shows the embedeed system.
Fig. 2 Clustering sytem architecture

____________________________________________________________________________________________________________
Title must be in 24 pt Regular font. Author name must be

in 11 pt Regular font. Author affiliation must be in 10 pt Italic.
Email address must be in 9 pt Courier Regular font.
B. The testing of Work Function of Lights Node System
The testing of lights node system is accomplished to see the
performance of the sensor in regulating the operation of the
lights. The system is tested for 23 hours long from 15.58 to
15.36 in order to test the algorithm and measure the power of
system.
TABLE I
DATA RESULT FOR EACH NODE ON TRANSITION OF LIGHTING CONDITIONS
N Parameters
o State Hour I V Power Stat Batt.
d (A) (Volt) Sources us Cap.
e (%)
N Morni 05:52 1.26 11.39 Battery On 58.49
Fig. 4 Layout system in altium designer o ng 05:56 0 0 Battery Off 60.11
d
Night 18:00 0 0 Battery Off 62.8
1 18:04 1.24 10.98 Battery On 61.72
o ng 05:56 0 0 Battery Off 75.91
d
e Night 18:00 0 0 Battery Off 81.63
2 18:04 1.32 9.92 Battery On 80.36
o ng 05:56 0 0 Battery Off 81.08
d
e Night 18:00 0 0 Battery Off 85.97
3 18:04 1.14 11.53 Battery On 84.36
N Morni 05:52 1 11.51 PLN On -
o ng 05:56 0 0 PLN Off -
d Night 18:00 0 0 PLN Off -
e 18:04 0.99 11.51 PLN On -
4
Table 1 shows that system in each node working properly
when the coordinator sent order to on or off in the night and
day transition. It is also shown thath in the afternoon, battery
Fig. 5 The fabrication of embedded system
capacity is always higher than in the morning state, except for
IV. TEST AND ANALYSIS the node 4 because because it test without using the battery,
which indicate that the solar panels managed to charged the
A. The testing of Resources Switching Algorithm battery. Then the capacity will decrease when it used for
Resources switching algorithm is designed to make the lighting. From the curent and voltage measurement result, itis
prioritizing system that use of resources from the battery indicate that the value of the current and voltage are varues at
(which is derived from solar panels), but it can also the moment light on bright or dim. The value of the lights
automatically switch to electricity grid resources when battery current and voltage are shown in Table 2.
capacity is below the lower threshold. The mechanism of TABLE II
resources selection is done by utilizing the analog signal CURRENT AND VOLTAGE VALUES ON DIFFERENT LIGHT OPERATION
reader features of the microcontroller. Analog reader is used
Current (A) Voltage (V) Power (W)
to compare the input voltage with an internal reference
Light Bright Dim Bright Dim Bright Dim
voltage of the microcontroller. The selection of system Mode
resources run based on battery capacity, if the battery capacity Average 1.42 1.16 12.00 11.14 17.08 13.00
is less than 50% (12 volts) then the system will automatically
switch the power source to the electricity grid, but if the C. The testing of Data Communication Algorithm
battery capacity back to 50 % or more then system resources Data communication between nodes and coordinator is
will switch back to the battery. performed by applying the bus network configuration. In

____________________________________________________________________________________________________________
testing, use 4 nodes as end device and one node as a into two stages, namely the design of hardware and software.
coordinator node that connected using wired RS 485 serial On testing data communications for 23 hours by using a bus
communication. The data communication results from nodes network configuration, obtained a success rate of 97.14% for
to coordinator is shown in Table 3. data communication between nodes and coordinator. Data
TABLE III
reception failure caused due to two factors, namely due the
CURRENT AND VOLTAGE VALUES ON DIFFERENT LIGHT OPERATION faiure in the process of sending data by RS485 or because of
the failure in the process of parsing data by coordinator. Then
Node Data Valid Data The success rate of from the calculation of the system efficiency obtained the total
uploaded to Data error data received by
power usage per year of 65,87 KWh/lights. So the smart street
coordinator coordinator (%)
Node 1 318 303 15 95.28 % lighting system use energy much more efficiently than other
314 312 2 99.36 %
existing street lighting system, which is about 1/15 of the
Node 2
Node 3 314 305 9 97.13 % energy use of conventional street lights powered by electricity
Node 4 313 303 10 96.60 % grid.
Average 97.14 %
REFERENCES
[1] Mahoor, Mohsen dkk. " A Smart Street Light Control System for
Based on test result, it shown from tabel 3 that the average Optimization of Energy Consumption and Lamp Life," 22nd Iranian
results with a success rate of data received by coordinator is Conference on Electrical Engineering (ICEE), pp. 1290 – 1294, 2014.
97,14%. Failure reception of data can be caused due to two [2] Braga, Henrique dkk. “Economic Analysis of a Controllable Device
factors, namely due the faiure in the process of sending data with Smart Grid Features Applied to LED Street Lighting System,”
by RS485 or because of the failure in the process of parsing 2015 IEEE 24th International Symposium on Industrial Electronics
(ISIE), pp. 1184 – 1189, 2015.
data by coordinator. [3] Yoziarde, L. “Rancang Bangun dan Implenetasi M2M Teknologi pada
Lampu Jalan dengan Konfigurasi Jaringan Star,” Tesis S2, Universitas
V. CONCLUSIONS Indonesia, 2015. (In Bahasa)
[4] Bayhaki, A. “Rancang Bangun dan Implementasi Sistem Lampu Jalan
The smart street light node systems based on smart-grid and Pintar dengan Konfigurasi Jaringan Ad-Hoc, ” Tesis S2, Universitas
wired sensor network using RS485 with bus network Indonesia, 2015. (In Bahasa)
configuration has been designed and implemented. The [5] Holler, J. From Machine-to-Machine to the Internet of Things:
system is designed to be able to perform the function of make Introduction to a New Age of Intelligence, 1st Edition. Oxford :
Elsevier. 2014.
it more effective and efficient in the process of operation, [6] Axelson, Jan (2007), “Serial Port Complete: COM Port, USB Virtual
control and maintenance. The design of the system is divided COM Port, and Ports for Embedded Systems Second Edition”.

____________________________________________________________________________________________________________
Log Periodic Microstrip Antenna

For Digital Video Broadcasting Application
Indra Surjati#1, Yuli Kurnia Ningsih#2, Syah Alam#3

#
Graduate Programe of Electrical Engineering, Faculty of Industrial Technology Trisakti University
West Jakarta, Indonesia
1
indra@trisakti.ac.id
2
yuli.kn1731@gmail.com
3
alam_bizkit@yahoo.com
Abstract— This paper proposed a new design of log digital terrestrial radio frequency bands is between 478-694
periodic microstrip antenna using peripheral slits for Digital MHz [3].
Video Broadcasting application in Indonesia. Dimensions of the There are several studies that have been done to produce a
proposed antenna is 400 mm x 150 mm with 4 patches in digital television receiver antennas with certain expected
different frequencies which are connected with log periodic characteristics. As has been done by [4], using a slot technique
technique using FR4 Epoxy substrate. The results shown that that produces a dimension of (204.8 x 160.6) mm with
the antenna is capable to work at the frequency range of 482 omnidirectional radiation pattern and VSWR ≤ 2. Another
MHz - 740 MHz with a VSWR ≤ 2 and return loss ≤ - 10 dB and
study by [5] using 5 pieces of element patch log periodic
bandwith of the proposed antenna is 257,8 MHz (482 MHz - 740
MHz) can be achieved from this research. produce dimension of (169,4 x 113,8) mm and in [6] to
produce a 2 x 2 array with dimension of (515 x 300) mm.
Keywords— Digital VideoBroadcasting, Peripheral Slits, Log The technique to be able to produce a form of compact
Periodic, Rectangular Microstrip Antenna antenna and small dimension is by using peripheral slits.
Using this technique microstrip patch antenna size can be
reduced up to 33% of its original size. Therefore peripheral
INTRODUCTION slits technique is one way to optimize the size of the
dimensions of the microstrip antenna [7]. Peripheral slits are
Digital television broadcasting is a technology that can not
one method of size miniaturization techniques microstrip
be avoided by any country in the world. The development of
antenna which works by making some parts (slits) on the side
digital television broadcasting technology becomes a global
of the patch antenna.
demand which nearly each country has and in the process
Based on the results of the previous studies it can be infered
towards the transition from analogue to digital broadcasting
that the peripheral slit method can reduce the antenna size and
systems. The advantages implementation of digital television
the log periodic technique can increase gain of the antenna.
broadcasting include reception of picture and sound sharper
Therefore this paper proposed a new design of log periodic
and better and the allocation of radio frequencies more
microstrip antennausing peripheral slits for Digital Television
efficiently [1].
Broadcasting application in Indonesia.
Digital television broadcasting standard has also been
undergoing in development of Digital Video Broadcasting -
Terrestrial (DVB-T) to Digital Video Broadcasting - Second
Generation Terrestrial (DVB-T2). The Indonesian DESIGN ANTENNA
Government with its Minister of Communications and
The proposed antenna is realized on one layer substrate with
Information Technology has state regulation No. 05 / PER /
relative permitivity (εr)= 4.3, substrate of thicknes (h) = 1.6
M.KOMINFO / 2/2012 on standard fixed and not paid Digital
mm and loss tangen (tan δ) = 0.0265. The dimension of the
Terrestrial Television Broadcasting (free-to-air), known as
rectangular patch antenna is given by
DVB-T2 digital terrestrial television broadcasting free-to-air
standar in Indonesia [2]. Regarding to the Regulation of
=
Minister of Communication and Information Technology
( )

Number 23 / PER / M.KOMINFO / 11/2011 on the master (1)
plan of radio frequencies for television broadcasting in the

____________________________________________________________________________________________________________
=
( ) ( )
(2)
= √
(3)
= (4)
The proposed antenna is working at various frequencies for

Digital Television Broadcast application for DKI Jakarta area
(Zone 4) as seen in Table I.
TABLE I
THEDVB FREQUENCIES FOR DKI JAKARTA AREA
No Patch Channel Frequency

Figure 1. Design of The Log Periodic Antenna
1 RCTI, Global TV, MNC TV 24 498 MHz
2 Metro TV 32 562 MHz
3 TV One, ANTV and Sport One 34 578 MHz
4 Berita Satu TV (BSTV) 36 594 MHz
RESULT AND DISCUSSION
5 Transcorp (Trans TV, Trans 7) 40 626 MHz
and Kompas TV After several iteration, the bandwith can increased by
6 SCTV, Indosiar, and O Channel 44 658 MHz adjusting W and L of each patch antenna. The first iteration is
7 RTV 48 690 MHz with W1= 77,4 mm, L1=83,5 mm , W2 = 70 mm, L2 = 80
mm, W3 = 65, L3 = 73 mm and W4 =58 mm, L4 = 65 mm.
Theresult is shown on Figure 2.
The group of working frequencies of the four patches is
seen in Table II below. The four patches are combined using
log periodic technique and Figure 1 shows the design of the
proposed antenna.
TABLE II
THE FREQUENCY OF EACH PATCH ANTENNA
No Patch Frequency
1 Patch 1 580 MHz
2 Patch 2 630 MHz
3 Patch 3 670 MHz
4 Patch 4 720 MHz
From Figure 1 it can be clearly seen that log periodic

antenna has 4 patches with different size. Patch 1 has
dimension W1 = 77,4 mm , L1 = 89,5 mm , Y1 = 27,8 mm and
Ls1 = 24 mm. Patch 2 has dimension W2 = 77 mm , L2 = 80
mm , Y2 = 28 mm and Ls2 = 24 mm. Patch 3 has dimension
W3= 65 mm , L3 = 73 mm , Y3 = 23 mm and Ls3 = 14 mm and
Figure 2. Simulation result from the first iteration
patch 4 has dimension W4= 58 mm , L4 = 65 mm , Y4 = 23
mm and Ls4 = 14 mm. The dimension of microstrip line of log
periodic antenna is ZL = 332,6 mm.
The second iteration is with W1= 77,4 mm, L1 = 83,5 mm,
W2 = 73 mm, L2 = 76 mm, W3 = 65, L3 = 73 mm and W4
=58 mm, L4 = 65 mm. The result is shown on Figure 3.

____________________________________________________________________________________________________________
TABLE III
ITERATIONS OF ANTENNA
Iterations Parameters (mm)

W1 W2 W3 W4 L1 L2 L3 L4
1 77,4 70 65 58 83,5 80 73 65
2 77,4 73 65 58 83,5 73 73 65
3 77,4 77 65 58 89,5 80 73 65
TABLE IV
SIMULATION RESULT
Iteration Return Loss (dB)

C.24 C.32 C.34 C.36 C.40 C.46 C.48
1 -11,6 -25,9 -17,8 -28,5 -11,7 -9,3 -13,1
2 -6,3 -15,7 -14,6 -14,3 -21,7 -12,7 -12,7
3 -16,0 -32,6 -24,6 -15,6 -11,9 -10,3 -10,2
Figure 3. Simulation result from the second iteration

Furthermore, it can be seen from Figure 5 that log periodic
microstrip antenna can increase the gain from 4,963 dB to
The third iteration is with W1= 77,4 mm, L1 = 89,5 mm, 5,398 dB.
W2 = 77 mm, L2 = 80 mm, W3 = 65, L3 = 73 mm and W4
=58 mm , L4 = 65 mm. The result is shown on Figure 4.
Figure 5. Gain of the proposed antenna
Figure 4. Simulation result from the third iteration CONCLUSIONS

A new design of log periodic microstrip antenna is
The overall iteration results can be summarize into Table
eventually well proposed. The best value of return loss ≤ - 10
III and Table IV as follows. From the twoTables, it can be
dB can be obtained by adjusting the dimesion of W and L of
seen that the best result of return loss value ≤ - 10 dB (482,7
each patch antenna. The impedance bandwith of the antenna is
MHz – 740,5) can be obtained through the arrangement W1=
77,4 mm, L1 = 89,5 mm, W2 = 77 mm, L2 = 80 mm, W3 = 65, 247,8 MHz (482,7 MHz – 740,5 MHz) that suitable for
L3 = 73 mm and W4 =58 mm , L4 = 65 mm. Digital Video Broadcasting in Indonesia. Beside that the log
periodic technique succeed to increase the gain of the antenna
from 4,963 dB to 5,398 dB.

____________________________________________________________________________________________________________
REFERENCES
[1]. Seminar of the Ministry of Communications and

Information Directorate General of Post and
Informatics. "Indonesia Goes Digital Communication
Information", Jakarta: Administrator, p. 1-15, 2012.
[2]. Regulation No. 23 / PER / M.KOMINFO / 11/2011
on Standards for Digital Terrestrial Television
Broadcasting Revenue Fixed Not Paid (Free-To-Air).
p. 1-3.
[3]. Regulation No. 23 / PER / M.KOMINFO / 11/2011
on Master Plan (Masterplan) Radio Frequency For
Purposes of Terrestrial Digital Broadcast TV On the
Radio Frequency Band 478-694 MHz. p. 1-10.
[4]. A Adrian Gulfyan Putranto and Aloysius
AdyaPramudita. "Designing Microstrip Slot Antenna
Receiver System For Digital Television". RiTekTra.
Jakarta, 2013
[5]. Sri AnggaraeniKadiran, "Log Periodic Patch
Techniques To Widen Bandwidth Microstrip
Antenna TV Receiver", ORBIT, Vol.9, No. 2, July
2013
[6]. Andrew Mulia, "Design and Realization Based
omnidirectional UHF Television Antenna Microstrip
For Applied in Tasikmalaya Region",
TELEKONTRAN, Vol.1, No.2, October 2013
[7]. IndraSurjati et al, "Antenna Peripheral Slits
Berbentuk Cincin Persegi Denagn Pencatuan
Electromagnetic Coupled” Seminar Nasional
Microwave, Antennas dan Propagasi (SMAP), 2013,
Department of Electrical Engineering FT Universitas
Indonesia, October 2013

____________________________________________________________________________________________________________
A Paddy Soil Monitoring System Based on

Wireless Sensor Networks
E. H. Putra, M. Susantok, N. F. Khoiruddin, M. Dani
Department of Electrical Engineering, Politeknik Caltex Riau
Jalan Umban Sari No 1 Pekanbaru, Indonesia
emansa@pcr.ac.id
Abstract— Land remains a vital factor in farming for almost all implementation of the Wireless Sensor Networks (WSNs) that
agricultural products in the form of plants that require fertile land use a multi-sensory system. A monitoring system based on the
as a place to grow, the supply of nutrients, and water. In order to WSNs has been used in agriculture to monitor weather and frost
make the rice plant grow well, the soil should be ideally suited to in Washington State. The system uses the 900 MHz FHSS
the needs of the rice crop. A paddy soil monitoring system based
radio/logger [2]. The WSN technology has also been
on the Wireless Sensor Networks (WSNs) is created to monitor soil
fertility and provide references for fertilizing paddy rice land incorporated to be used to measure temperature, light and soil
according to the conditions that exist. The paddy soil monitoring moisture in the greenhouse [3], monitor the quality of water in
system consists of four nodes. Each node has three types of sensors, rivers and lakes [4], and irrigation management [5]. The WSN
namely temperature sensors, soil moisture sensors, and soil acidity technology development has also been possible to transmit
(PH) sensors. All sensors are integrated with a microcontroller multimedia data from the image sensor and videos over
ATMEGA 8535 and CVAVR programming language. The wireless networks. The technology is known as Wireless
transmission of data is sent over the XBee interface. Then, they Multimedia Sensor Networks (WMSNs) [6, 7]. In this research,
are sent to a monitoring server. This monitoring system has a paddy soil monitoring system based on the WSNs created to
successfully displayed the measurement results from the sensors
monitor soil fertility and provide references for fertilizing
and displayed the fertilization recommendations based on the
measured paddy soil conditions. paddy rice land according to the conditions that exist.
The rest of this paper is organized as follows. Section II
explains the Wireless Sensor Networks (WSNs) System
Keywords— paddy soil, wireless sensor networks, fertilization consisting of temperature sensors, soil moisture sensors, and
soil acidity (PH) sensors. Section III describes the Paddy Soil
I. INTRODUCTION
Monitoring System. Section IV explains the simulation and
Rice plants require the supply of nutrients with balanced results. Finally, the conclusion is presented in Section V.
proportion that can be absorbed from the soil. The fertility rate
of rainfed areas is lower than the irrigated land. Besides that, II. WIRELESS SENSOR NETWORK SYSTEM
soil moisture levels also limit nutrient uptake by plants. The Wireless Sensor Network (WSN) system constitutes a
Therefore, to reach the expected results, the supply of the measurement process, computing, and communications that
nutrients (fertilizer) for the rainfed areas has to have a high provide administrative capabilities to a device, observation, and
number and high varieties. Furthermore, the time of the any treatment for any events and phenomena that occur in the
fertilization also needs special attention as when the fertilizing environment using wireless technology. This system is much
land is in a dry condition. It does not have to wait until the soil more efficient than the use of cables. This system has a function
becomes moist. In general, the fertilizer is needed for growth for various types of applications. The WSN technology
and rice production in the rainfed areas such as nitrogen (urea), provides the foundation to conduct experiments on the
phosphate, and potash fertilizer. Soil that is rich in nutrients and environment [8].
contains enough organic, greatly affects the growth and
development of the rice plants to grow optimally. However, it A. Temperature Sensors
is not easy to create an ideal condition of the soil for rice plants LM35 temperature sensor is a transducer that has a function
[1]. to convert the temperature scale into electrical quantities in the
The paddy soil monitoring system aims to help farmers to form of voltage. It has high accuracy and ease of design
get information about PH measurement, humidity, and compared to other temperature sensors. It also has a low output
temperature of the paddy soil fields. Through this system, impedance and high linearity so that it can be easily connected
farmers can monitor and calculate the fertilizer needed for their to a series of special control and does not require a further
farms to achieve maximum yields. If the process of the adjustment. The temperature can be measured by the IC is quite
fertilization is done too excessively, the production process of wide, between -55 oC to 150 oC. The voltage required by the
the rice plants will also be affected. LM35 to work is at 5 volts. Every 1ºC rise in temperature means
In some literature, there are several studies related to the

____________________________________________________________________________________________________________
that the output voltage is divided by 10 mV as shown in the D. Xbee Module

following equation [8]:
XBee module is a module that uses ZigBee protocol. The
VLM35 = T (ºC) x 10 mV, T (ºC) = (1) Zigbee uses Personal Area Network IEEE 802.15.4 standard.
The XBee works at a frequency of 2.4 GHz and has low power
B. Soil Acidity (PH) Sensors
consumption so that it is suitable for measuring and monitoring
For the acidity of the soil using a sensor amplifier (non- an environment which is not covered by the residential
inverting) with strengthening as much as 11 times is shown in electricity.
Fig. 1. In this circuit there are LM358 IC component and
resistor R1 (10K) as Rf and the resistor R2 (1K) as Ri. To III. THE PADDY SOIL MONITORING SYSTEM
calculate and of PH sensor circuit, use the following
formula: A. System Design
= 1+ × (2) The paddy soil monitoring system consists of 4 nodes (node
A, B, C, and D) as shown in Fig. 3.
= (3)
Node Node
Temperature sensor C D Temperature sensor
Soil moisture sensor Soil moisture sensor
PH sensor PH sensor
Power source Power source

Microcontroller Microcontroller
XBEE S2 XBEE S2
Node Node
Temperature sensor A B Temperature sensor
Fig. 1 PH sensor circuit Soil moisture sensor Soil moisture sensor
PH sensor PH sensor
C. Soil Moisture Sensors Power source Power source

Microcontroller Microcontroller
The soil moisture sensor is designed to be capable of
measuring soil moisture level of the rice paddies. This sensor
uses two probes to pass current through the ground and then it
reads the resistance to get the moisture level. More water makes XBEE S2 XBEE S2
the soil moisture sensor more easily conduct electricity (less

resistance), while dry soil conducts electricity poorly (more
resistance). Fig. 2 shows a series of humidity sensors used in
paddy soil condition monitoring systems.
Server
Fig. 3 The design of the paddy soil monitoring system
Each node has 3 sensors consisting of a temperature sensor

that measures ambient temperature in the paddy fields,
humidity sensor measures the humidity of the paddy soil, and
PH sensors that measure the soil acidity. Each node uses an
ATMega8535 microcontroller and is programmed with
CVAVR programming languages to regulate its work. The
Fig. 2 Soil moisture sensor circuit wireless network communication modules use the XBee Series

____________________________________________________________________________________________________________
2 in each node. The results of the measurement of the nodes C TABLE IIIII
TESTING OF MOISTURE SENSOR ON NODE A
and D will be sent to the node A and B firstly. Then node A
dam B will transmit the data from all measurement results to a Measurement (Dry)
monitoring server. Depth Soil Grove Vout of Vout of %
(cm) Moist. Moist. Sensor Soil Grove Error
B. Results Sensor (%) Moisture Moist.
The LM35 sensor testing aims to test the accuracy of the data (%) (V) (V)
generated by the LM35 sensor output. Then, the result is 1 14 12 0,52 0,4 16,6%
compared with the value of the temperature on the thermometer. 2 21 25 0,8 0,84 16 %
Table I shows the LM35 sensor testing on Node A. 3 28 29 1,1 1 3,4 %
Moist
TABLE I 1 32 36 1,2 1,23 11,1%
THE LM35 SENSOR TESTING ON NODE A
2 45 49 1,6 1,73 8,1 %
NO N LM35 Measurement Thermometer % 3 67 70 2,4 2,5 4,2 %
O Output (º C) (ºC) Error Wet
D (Volt) 1 73 77 2,6 2,72 5,1 %
1 E 0,298 29 30 3,4 % 2 89 92 3,3 3,4 3,2 %
3 100 100 3,57 3,5 0%
2 0,299 30 30 0%
A
3 0,308 31 31 0% Fig. 4 shows the measurement results of the 4 sensor nodes
4 0,310 32 31 3,1 % on the paddy soil monitoring system. Fig. 5 shows a
fertilization recommendation based on the measured paddy soil
5 0,328 33 32 3% conditions on sensor node 2 when the PH measurement is 6.
The recommendation is that the use of agricultural lime is given
The result of the measurement on Vin of the PH sensor as a fertilizer to balance the PH/acidity of the soil. The dose
circuit, at 52 mV, obtained at the value of the PH meter is 7. depends on 0.75 tons/ha.
Then, the measurement result on the Vout of the op-amp output,
0.56 V, is used to calculate the value of the Vout (theory) by
the equation (2), and the value of the Vout (theory) is 0.572 V.
To calculate the value uses the equation (3). The value of
the (measurement) is 10.76 V and the value of the
(theory) is 11 V, as shown in Table II for the measurement of
the PH sensor on the sensor node A.
TABLE III
TESTING OF PH SENSOR ON NODE A
Measurement Theory %
PH Error
Meter Vin Vout Av Vout Av
(mV) (V) (v)
7 52 0,56 10,76 0,572 11 2,1 % Fig. 4 The results of the 4 sensor nodes measurement on the paddy soil
monitoring system
6 58 0,64 11,03 0,638 11 0,27 %
5 115 1,24 10,78 1,265 11 2%
4 160 1,75 10,93 1,76 11 0,63 %
3.5 195 2,13 10,92 2,145 11 0,72 %
Table 3 shows the results of the testing of the soil moisture

sensor at node A. It can be seen that the deeper the probe is
inserted into the ground, the higher the value of the is, and
the humidity value (measurement) is also higher. As the more
the probe is plugged, the water content in the soil also increases.
More water makes the soil moisture sensor more easily conduct
electricity (less resistance), while dry soil conducts electricity Fig. 5 A fertilization recommendation based on the measured paddy soil
poorly (more resistance). conditions on sensor node 2 when the PH measurement is 6

____________________________________________________________________________________________________________
Fig. 6 shows a fertilization recommendation based on the 802.11 e EDCA and H. 264/SVC." 2015 International Conference on
Science in Information Technology (ICSITech). IEEE, 2015.
measured paddy soil conditions on sensor node 2 when the PH
[8] Putra, Emansa Hasri, Muhammad Yanuar Hariyawan, and Arif
measurement is 7. The recommendation is that this condition is Gunawan. "Wireless Sensor Network for Forest Fire
ideal so that nutrients and essential compounds can be absorbed Detection." TELKOMNIKA (Telecommunication Computing
by the rice plant. Electronics and Control) 11.3 (2013): 563-574.
Fig. 6 A fertilization recommendation based on the measured paddy soil

conditions on sensor node 2 when the PH measurement is 7
IV. CONCLUSIONS
The paddy soil monitoring system consists of four nodes.
Each node has three types of sensors, i.e. temperature sensors,
soil moisture sensors, and soil acidity (PH) sensors. This
monitoring system has successfully displayed the measurement
results from the existing sensors and the fertilization
recommendations based on the measured paddy soil conditions.
ACKNOWLEDGMENT
This work is fully supported by the Ministry of Research,
Technology and Higher Education of the Republic of Indonesia
under the grant of Hibah Bersaing Dikti. The authors would
like to thank to Kopertis Wilayah X and Politeknik Caltex Riau
for their support.
REFERENCES
[1] Dobermann, A., PC Sta Cruz, and K. G. Cassman. "Fertilizer inputs,
nutrient balance, and soil nutrient-supplying power in intensive,
irrigated rice systems. I. Potassium uptake and K balance." Nutrient
Cycling in Agroecosystems 46.1 (1996): 1-10.
[2] Pierce, F. J., and T. V. Elliott. "Regional and on-farm wireless sensor
networks for agricultural systems in Eastern Washington." Computers
and electronics in agriculture 61.1 (2008): 32-43.
[3] Liu, Hui, Zhijun Meng, and Shuanghu Cui. "A wireless sensor network
prototype for environmental monitoring in greenhouses." 2007
International Conference on Wireless Communications, Networking and
Mobile Computing. IEEE, 2007.
[4] Nasirudin, Muhammad Azwan, Ummi Nurulhaiza Za'bah, and O. Sidek.
"Fresh water real-time monitoring system based on Wireless Sensor
Network and GSM." Open Systems (ICOS), 2011 IEEE Conference on.
IEEE, 2011.
[5] Kim, Yunseop, Robert G. Evans, and William M. Iversen. "Remote
sensing and control of an irrigation system using a distributed wireless
sensor network." IEEE Transactions on Instrumentation and
Measurement 57.7 (2008): 1379-1387.
[6] Zacharias, Sven, and Thomas Newe. Technologies and Architectures for
Multimedia-Support in Wireless Sensor Network. INTECH Open Access
Publisher, 2010.
[7] Putra, Emansa Hasri, Risanuri Hidayat, and I. Wayan Mustika. "Cross-
layer design of wireless multimedia sensor network based on IEEE

____________________________________________________________________________________________________________
Greedy Algorithm and FIS Sugeno for

Determining Evacuation Routes in Eruption of
Merapi Mountain in Indonesia
Abdul Aziz1, Ananto Setyadi2, Esti Suryani3

Informatics Department Sebelas Maret University
Surakarta, Indonesia
1
aaziz@staff.uns.ac.id
2
ananto.setyadi@gmail.com
3
suryapalapa@yahoo.com
Abstract— Indonesia is a country that has many active crucial when a disaster occurs. Residents who mobilized
volcanoes. One of the most active volcanoes in Indonesia is should be escorted to safety placed through the fastest route.
Mount of Merapi. The eruption of Merapi in 2010 caused many The evacuation process to be done as best as possible and as
victims. Sleman is a region with a lot of some victims. One of the quickly as possible in moving the residents.
important things to reduce the number of victims when the
eruption occur is the evacuation process which one of the factors Research [4] [5] search the shortest route with multi
is the route of evacuation. These research aims are determining parameters using the fuzzy method to determine its weight.
evacuation routes by using a Greedy algorithm and FIS Sugeno Also, there are other studies [6][7] which solve Travelling
in the eruption of Merapi. FIS Sugeno is used to manipulate the Salesman Problem (TSP) and greedy algorithms.
distance parameters, and some refugees became weights that are
used in the looking for the evacuation routes. Findings the Based on researches that have been done before, this
routes using Greedy algorithm because it took the greatest value research will do a searching of disaster evacuation routes
in each stride, so it fits in the determination of the route of with multi-parameter utilizing a greedy algorithm and method
evacuation relating to human lives. Looking for the routes of fuzzy. Evacuation roads which will be passed are
starting from the post of refugee then passes the entire village considering three parameters i.e. distance, some refugees, and
which will be evacuated and returned to the refugee evacuation road barriers with ignoring the number and capacity of the
barrack. The results of this research are the evacuation route on vehicle. The length parameters are the distances between one
Merapi's eruption can be determined with the Greedy algorithm of the village to the other village, a meanwhile setting of road
and FIS Sugeno. Testing was done by comparing the results of barriers are whether could or not a connector road between
Greedy routes with the routes of Exhaustive algorithm results. the village to be crossed.
The result of testing is from 25 times of trial by the highest
tolerance is 5%, generates 88% routes from Greedy Algorithm This research aims to determine the route of eruption
is revealed properly. disaster of Mount of Merapi. The greedy algorithm that used
to search for the route is appropriate to the case which search
Keywords--- Greedy Algorithm, FIS Sugeno, Exhaustive of the route requires evacuation routes which fast handling
algorithm, Evacuation Route due to concerns the souls of many people.
I. INTRODUCTION II. BASIC THEORY

Indonesia is an archipelago that had formed from a chain A. Disaster Prone Regions
of volcanoes that are included in a single cluster of circles of
fire or often called the ring of fire [1]. One of the most active The disaster-prone regions are those areas for a certain
volcanoes in Indonesia is Mount of Merapi.Mount Merapi period that is not capable of reducing the harms of danger [8].
erupted in 2010; these eruptions resulted in much damage and The disaster-prone region zoning in Sleman Regency in
loss as well as loss of life [2]. Sleman is a Regency with the Indonesia which is based on on a map of Merapi disaster-
most number of casualties. There are some victims died as prone region is divided into five sectors. Division of disaster-
many as 194 inhabitants, of whom 163 residents in Sleman prone areas in the zone of Sleman Regency based on the flow
Regency, while 498 people hospitalized, and as much as of the river that originates on Mount of Merapi [9].
370,028 inhabitants evacuated in Java and Yogyakarta [3].
B. Graph
When the eruption happens, one of the significant The graph is a collection of points (vertices) connected to
prevention of disaster is optimizing the evacuation of victims. each other through the edge [10]. A graph consists of two
The selection of the route in conducting the evacuation is sets, name of the set is V, and the set E. V is the set of

____________________________________________________________________________________________________________
vertices which are limited and not empty. E is the set of arcs With αi is the degree of membership of the implications
which connect a pair of vertex [11]. Graph notation: G (V, function rule i and zi is the value of output at rules i.
E), meaning that V is the set of point or vertexV from graph
G i.e. V = {v1, v2, v3, ...,vn} and E is the set of side or edge F. Greedy Algorithm
E from graph G, i.e., E = {e1, e2, e3, ..., em} [12]. The graph A greedy algorithm is an algorithm which solves the
can be distinguished in some types, such as a weighted graph problem steps by steps in which at each step choosing the
and complete graph [13]. local optimum with the hope that the next step leads to the
global the optimum solution [22]. The solution was found in
C. Shortest Path Problem the algorithm of Greedy could be not the global optimum
The shortest path problem (shortest path problem) is a solution yet, but local optimum solution only.
network of travel where someone wants to determine the Here is a pseudo code of Greedy algorithm:
shortest route between the two cities based on alternative Input: a set of candidates
routes that available, where only one destination [14]. This its Output: Set Solutions
issue using the graph for modeling of problems experienced Process:
by making it easier to finish it. The problem is how to visit 1. Check the adjacent vertex with vertex currently just
the vertex on the graph from early vertex to last vertex with call it vertex a.
minimum weight. The graph G is called connected if every 2. Check the connected edge that has not been visited
vertex of the graph has an edge that is connecting both (marked with a 0).
vertexes, or in other words, the graph is connected if any two 3. Select the edge that is available with the largest
vertex, i.e., vi and vj in a graph there exists at least one edge weights, take the distance,e (i).
[15]. 4. Calculate the distance d (i) = d + e (i) (i).
5. Tick the vertices visited by number 1.
D. Fuzzy Logic 6. Make the selected vertex as a.
Fuzzy in the lexicon meaning is fuzzy or vague. 7. Repeat step 1.
According to [16], a fuzzy is value can be true or false at the
same time. How big the values are truth and faults depending
on degrees of membership. In the fuzzy logic, membership III. METHODOLOGY
degrees have a value of 0 to 1. The value is provided by the
membership function is called the degree of membership A. Collecting Secondary Data
[17]. The degree of membership of a set of fuzzy was At this stage the data collected is required i.e. map
obtained from membership functions. disaster site that includes the location of the village, the
evacuation site, and road traffic with its distance.Location of
Map disaster is obtained from a map issued by the Sleman
E. FIS Sugeno Regency BPBD. The distance between the village is obtained
One of the applications of fuzzy logic that have grown by using digital map that is accessed at www.mapcarta.com.
very wide now it is fuzzy inference systems (FIS) which are a The data numbers of refugee and street barriers are assumed
mapping of the formulation process of the input to the output and dynamically, so can describe the circumstances by the
by using fuzzy logic [18]. Fuzzy inference system serves to reality at the time of eruption disasters.
take decisions through a particular process using the rules of
inference based on fuzzy logic. There are three methods in B. Graph Modelling
the system fuzzy that often used inference,i.e.,Tsukamoto, Based on map data and the distance between the villages
mamdani, and sugeno [19]. which was already obtained is made a labeling graph and
Sugeno method was introduced by Takagi, Sugeno, and weighted.Weights on the edge show range with units of
Kangin 1985. Sugeno method is using a linear output meters while the label on vertex shows the number of
membership function or the shape of constanta[20]. Stages in refugees. Examples of graph modeling results can be seen in
the Sugeno method is as follows: Figure 1 where shows a weighted graph with does it weigh
1. Fuzzification, at this point, the input variable is stated distance. Green vertex is a vertex of evacuation site,
and the white vertex is a village that will be evacuated.
divided into one or more set of fuzzy by using
membership functions.
2. Application of implicating function, on Sugeno
method, the function that is used is Min.
3. Defuzzification, Sugeno method using Weight
Average Method to do defuzzification (Z *) as
follows equation 1[21].
(1)
Fig. 1 Graph Modeling

____________________________________________________________________________________________________________
C. Normalization Graph combination of distance parameter and the number of

The graph that is formed will be checked whether the refugees. A rule that is formed is shown in Table 1.
shape of a complete graph or not, if not then the graph will be
transformed into a complete graph by adding a dummy edge TABLE 1. RULES OF FIS SUGENO
on a vertex which is not connected. The steps of adding No. Rules
dummy edge is as follows:
1 If distance NEAR and A LOT OF refugees z = 100
For each vertex in the graph is doing:
1. Checkvertex that is not adjacent with a call the 2 If distance NEAR and MEDIUM refugees z = 75
vertex b. 3 If distance NEAR and LITTLE refugees z = 50
2. Find the shortest distance between a greedy
4 If distance MEDIUM and A LOT OF refugees z = 75
algorithm with b, call it d (i).
3. Add a dummy edge that connects with d (i). 5 If distance MEDIUM and MEDIUM refugees z = 50
4. Repeat step 1.
6 If distance MEDIUM and LITTLE refugees z = 25
7 If distance FAR and A LOT OF refugees z = 50
D. Rules of FIS Sugeno
The first step on the stage of the FIS Sugeno is to 8 If distance FAR and MEDIUM refugees z= 25
determine the degree of membership on the distance 9 If distance FAR and LITTLE refugees z = 10
parameters and the number of refugees. Limit values for each
condition for each parameter are as follows:
Distance Each rule will be evaluated using the Min implications
Near: 0 - 2000 meters function. Equation 2 is a function Min implications on one
Medium: 2000 – 5000 meters rule:
Far: more than 5000 meters
(2)
number of refugees
little : 0 - 50 people
Medium : 50 – 150 people information :
Amount: more than 150 people α= minimum value between the degree of membership of
distance with the number of refugees
Figure 2 shows the curve of the shoulder for the distance µ= degrees of membership
and Figure 3 shows the curve of the shoulder to the number x= value of distance parameter
of refugees with the function of its membership. y=parameter value of the number of refugees
The last step in FIS Sugeno is defuzzification. The
process of defuzzification is done by Formulas of Weight
near medium far Average as follow equation 3.
(3)
IV. RESULT
distance
Fig. 2 Distance Curve A. Simulation and testing
Simulation and testing are done by comparing the result
little medium A lot of greedy algorithms by the route that is generated from the
of
exhaustive algorithm. Exhaustive algorithms will give all
possibilities of routes with a total distance and it is sorted
from smallest to largest distance. A greedy algorithm is
optimal levels can be seen from the rankings of the total
distance route of greedy on the list that is produced by
exhaustive algorithms. The steps in testing steps are as
Refugee number follows:
Fig. 3 Curve of The Number of Refugees 1. Determine the drag path
2. Normalization of graph
After getting degrees of distance parameter and the 3. Determine the number of refugees
number of the refugees then made rules which are a 4. FIS Sugeno

____________________________________________________________________________________________________________
5. Searching route of Greedy Algorithm vehicle capacity and time, using other methods for instance
6. Searching route exhaustive Algorithm methods in the process of adding dummy edge.
The experiment was conducted as many as five times in
each sector. Table 2 shows the results of all route finding REFERENCES
experiments using the greedy algorithm along with a
description of routes feasibilities that are generated after its is [1] Blair, L., & Blair, L. (2010). Ring of Fire: An Indonesia
applied top rank tolerance of 5%. Odyssey.Singapore :Bantam Books.
Table 2. Experiment results of the search algorithm are [2] BNPB. (2010). Ketangguhan Bangsa Dalam Menghadapi
the Greedy route. Bencana.Jakarta :Badan Nasional Penanggulangan Bencana.
[3] Suhendi, A. (2010). Korban Tewas Letusan Gunung Merapi Menjadi
TABLE 2. RESULT OF COMPARING GREEDY AND EXHAUSTIVE 194. [on-line] :
http://www.tribunnews.com/regional/2010/11/11/korban-tewas-
Greedy letusan-gunung-merapi-menjadi-194.
Rank
Rank [4] Hernandes, F., Lamata, M. T., Verdegay, J. L., &Yamakami, A.
Experime Route Comparin Descripti
Sector Toler (2007). The shortest path problem on networks with fuzzy
nt Total g on parameters.ELSEVIER, 1561-1570.
ance
Exhaustiv
e [5] Ichsan, M. H., Yudaningtyas, E., & Muslim, M. A. (2012). Solusi
Optimal Pencarian Jalur Tercepat dengan Algoritma Hybrid Fuzzy-
1 389 19 45 unfeasible
Dijkstra. EECCIS, 155-160.
2 645 32 10 feasible
A 3 447 22 33 unfeasible [6] Lukman, A., Rubinah, A. R., & Nurhayati. (2011). Penyelesaian
4 626 31 26 feasible Travelling Salesman Problem dengan AlgoritmaGreedy.STIMED Nusa
Palapa Makassar.
1 700 35 14 feasible [7] Cahyawijaya, S. (2010). Penyelesaian TSP Simetris dengan Algoritma
2 648 32 1 feasible Greedy Dua Arah. Institut Teknologi Bandung.
B 3 696 34 1 feasible [8] Anonim, (2007). UU No. 24 Tahun 2007.
4 879 43 13 feasible [9] BPBD. (2013). Jalur dan Ruang Evakuasi Bencana Erupsi. Sleman.
5 620 31 9 feasible
1 374 18 4 feasible [10] Rosen, K. H. (2007). Discrete Mathematics and its Applications. New
York :McGraw-Hill.
2 440 22 1 feasible
C 3 417 20 1 feasible [11] Zakaria, T. M. (2005). Konsep dan Implementasi. Bandung
4 425 21 2 feasible :Informatika.
5 521 26 15 feasible [12] Fletcher, P., Hoyle, H., & Patty, C. W. (1991). Foundations of Discrete
1 709 35 36 feasible Mathematics. Boston :PWS-KENT Publishing Company.
2 744 37 20 unfeasible [13] Munir, R. (2004). Matematika Diskrit.Bandung :Departemen
D 3 436 21 1 feasible Informatika.
[14] Wallis, R. (1992). Applications of Discrete mathematics. New York :
5 379 18 16 feasible McGraw-Hill, Inc.
1 478 23 9 feasible
2 429 21 20 feasible [15] Harary, F. (1969). Graph Theory. Canada : Addison-Wesley.
E
3 705 35 8 feasible [16] Setiadji. (2009). Himpunan & Logika Samar serta Aplikasinya. Edisi
4 587 29 15 feasible Pertama. Yogyakarta :Graha Ilmu, Yogyakarta.
[17] Klir, G. J., & Yuan, B. (1995). Fuzzy Sets And Fuzzy Logic. Theory
Testing of 25 experiments obtain the results 88% routes And Applications. New Jersey :Prentice Hall.
that are generated by the greedy algorithm are revealed [18] Ross, T. J. (2004). Fuzzy Logic With Engineering Applications.London
feasible. Ranking of the route that is generated from greedy :John Wiley & Sons, Ltd.
algorithm has many variations; it is strongly influenced by [19] Jang, J. R., & Gulley, N. (1995). Fuzzy Logic Toolbox. Math Works
the values of the distance parameters and refugees, so that is Inc.
generated are not always optimal but not always bad. [20] Wang, L. X. (1997). A Course in Fuzzy Systems and Control. New
Jersey :Prentice Hall International.
V. CONCLUSIONS AND FUTURE WORKS
[21] Takagi, T., &Sugeno, M. (1985). Fuzzy Identification of System and
From the research that has been done, some things can be Its Applications to Modeling and Control. IEEE TRANSACTIONS ON
summarized up that Evacuation Route on Eruption of Mount SYSTEMS, MAN, AND CYBERNETICS, 116-132.
of Merapi can be determined by a combination of Greedy [22] Pasino, K. M., &Yurkovich, S. (1997). Fuzzy Control. Canada
Algorithm and FIS Sugeno with 88% of the route that was :Addison-Wesley.
generated is declared feasible. The route that is generated
from the Greedy algorithm is strongly influenced by the
distance parameter and the number of refugees.
In next works, Suggestions are submitted for further
research development are added the other parameters such as

____________________________________________________________________________________________________________
Comparison Analysis of Phase-Shifted and Level-

Shifted Carrier-Based PWM on the Five-Level
Single-Phase Cascaded H-Bridge (CHB) Multilevel
Inverter
I B F Citarsa#1, I N W Satiawan#2, I K Wiryajati#3
#1,2,3
Electrical Engineering Department, Universitas Mataram
Jalan Majapahit Nomor 62, Mataram, Nusa Tenggara Barat, Indonesia
1
ferycitarsa@unram.ac.id
3
kjatiwirya@yahoo.co.uk
2
nwahyus@yahoo.com
Abstract— A Five-level Single-phase Cascaded H-Bridge One of the multilevel inverter topology that is popular in
multilevel inverter is designed in this research. The inverter is use is the Cascaded H-Bridge multilevel inverter. Cascaded
controlled using Level-shifted Carrier-based PWM and Phase- H-Bridge multilevel inverter is a kind of bridge multilevel
Shifted Carrier-based PWM. The PWM signals to the inverter inverters that are arranged in increment positions. The major
are applied using Arduino Mega 2560. The results from the use
advantages of this type of inverter are: it produces good
of those Carrier-based PWM methods are compared to
investigate which switching method produces the best profile for voltage waveform, small harmonic distortion, and it can
the Five-level Single-phase Cascaded H-Bridge multilevel handle large power utility.
inverter. The results show that the nearest waveform to A Five-level Single-phase Cascaded H-Bridge multilevel
sinusoidal waveform is produced by the Phase-Shifted Carrier- inverter is designed in this research. The inverter is designed
based PWM. The Phase-shifted Carrier-based PWM also using Carrier-Based method for the pulse signal generator
produce the lowest low order harmonics and the lowest THD modulation techniques. Carrier-Based method is mainly
value of 16.40 %. Hence, from all of those results, the best profile divided into two categories: Level-Shifted [5-6] and Phase-
is produced by the Phase-Shifted Carrier-based PWM for this Shifted methods [6]. The Level-Shifted methods meanwhile
particular inverter topology.
can be divided into three schemes: In-Phase Disposition (IPD),
Phase Opposition Disposition (POD) and Alternate Phase
Keywords— Carrier-based PWM, Level-shifted, Phase-shifted, Opposition Disposition (APOD) modulations [7].
Five-level Single-phase Cascaded H-Bridge multilevel inverter In this study, the Phase-shifted Carrier-based PWM and
the three schemes of Level-shifted Carrier-based PWM are
I. INTRODUCTION applied in the Five-level Single-phase Cascaded H-Bridge
There are three types of multi level inverters: flying multilevel inverter using Arduino Mega 2560 as the PWM
capacitor, diode clamped and cascaded h-bridge multi level signals controller. The results from those Carrier-based PWM
inverter [1]. Among those three inverters, the cascaded h- methods are compared to investigate which switching method
bridge multilevel inverter has been developed for utility produces the best profile for this particular inverter topology.
applications including voltage regulation, VAr compensation,
utility interface of renewable energy, and harmonic filtering in II. METHODOLOGY
power systems [1]. Three switching frequency classes for
multilevel modulations are: fundamental, high, and hybrid A. Cascaded H-Bridge (CHB) multilevel inverter
switching frequency. Fundamental Switching Frequency This type of multilevel inverter uses only capacitors and
methods produce switches commutations at fundamental switches. The combination of a pair of capacitors and switches
frequency output and can be aimed to cancel some particular is called the Bridge (H-Bridge), where the input DC voltage
low frequency harmonics [2-3]. High Switching Frequency for each bridge (H-bridge) cell is separated. One of the
methods are meant to switch at very high frequency, about 10 advantages of this multilevel inverter is that it requires fewer
to 20 kHz [4]. In this class there are two methods: Carrier- components than diode-clamped and flying-capacitor
based or Sinusoidal methods, and Space Vector methods. multilevel inverters.
Mixed Switching Frequency methods are those in which
switches commutate at different frequencies, and are
particularly suited for hybrid inverters where different cells
can easily commutate at different frequencies [1].

____________________________________________________________________________________________________________
PWM, all the carrier triangles have the same frequency and
peak to peak amplitude as shown in Fig. 2. The phase shift
between two adjacent carrier waves, cr can be counted with:
cr = 360°/(m – 1) (2)
Where m is the number of inverter voltage levels.
Fig. 1 Five-level Single-phase Cascaded H-bridge (CHB) multilevel inverter

[8]
As shown in Fig. 1 and Table I, in Five-level Single-phase
Cascaded H-Bridge (CHB) multilevel inverter, when the Fig. 2 Phase-shifted Carrier-based PWM [8]
switches S11, S21, S12, and S22 are ON, the output voltage of
the cells (H-Bridge cells) H1 and H2 are vH1 = vH2 = E, and the
inverter phase voltage resultant is Van = VH1 + VH2 = 2E.
The inverter terminal voltage A is connected to the inverter
neutral N. Similarly, if the switches S31, S41, S32, and S42
are ON, then Van = -2E. (a) In-Phase Disposition (IPD)
TABLE I
SWITCHING STATES OF THE FIVE-LEVEL SINGLE-PHASE
CASCADED H-BRIDGE (CHB) MULTILEVEL INVERTER
Output State vH1 vH2
voltage S11 S31 S12 S32
VAN
2E 1 0 1 0 E E (b) Alternative Phase Opposite Disposition (APOD)
1 0 1 1 E 0
1 0 0 0 E 0
E
1 1 1 0 0 E
0 0 1 0 0 E
0 0 0 0 0 0
0 0 1 1 0 0
(c) Phase Opposite Disposition (POD)
1 1 0 0 0 0 Fig. 3 Level-shifted Carrier-based PWM [8]
0
1 1 1 1 0 0 2) Level-shifted Carrier-based PWM
1 0 0 1 E -E
Similar to the Phase-shifted Carrier-based PWM, the
0 1 1 0 -E E
multilevel inverter using this PWM technique with m voltage
0 1 1 1 -E 0
0 1 0 0 -E 0 levels requires m – 1 carrier triangles, all with the same
-E frequency and amplitude. These m-1 triangular carriers are
1 1 0 1 0 -E
0 0 0 1 0 -E vertically arranged so that they occupy adjoining ribbon.
-2E 0 1 0 1 -E -E Frequency modulation index, mf is given by:
The number of voltage levels, m in Cascaded H-Bridge (CHB) mf = fcr / fm (3)
multilevel inverter is always odd. It can be calculated by: Where fm is the sine (reference) wave frequency, and fcr is the
m = (2H + 1) (1) triangle (carrier) wave frequency.
Where H is the number of cells in Cascaded H-Bridge (CHB) This mf is equal to the Phase-shifted modulation scheme,
multilevel inverter while the amplitude modulation index, ma is defined by:

B. Carrier-based Pulse Width Modulation (PWM) Vm

ma  
for 0  ma  1 (4)
In Carrier-based PWM, there are two pulse signal Vcr m  1
modulation techniques using multi carrier PWM technique, Where Vm is the sine (reference) wave magnitude, and Vcr is
which are multi carrier modulation by shifting the phase the triangle (carrier) wave magnitude.
(Phase-shifted) and shifting the level (Level -shifted). In Fig. 3 are shown three schemes for the Level-shifted
1) Phase -shifted Carrier-based PWM Carrier-based PWM, including:
In general, the multilevel inverter with m voltage levels a. In-Phase Disposition (IPD), where all the carrier triangles
requires m – 1 carrier triangle. In Phase-shifted Carrier-based are in-phase;

____________________________________________________________________________________________________________
b. Alternative Phase Opposite Disposition (APOD), where The circuit in Fig. 5 and Fig. 6 (a) uses 12 Volt DC battery
all the carrier triangles are in the alternative phase power supply and voltage regulator IC LM7805 to turn on the
opposite disposition; Arduino Mega 2560 system minimum requirement voltage of
c. Phase Opposite Disposition (POD), where all the carrier 5 volt. The C language with Arduino IDE software is used to
triangles above the zero reference are in phase but those make the desired coding program in accordance with their
below the zero reference is opposite to the phase. respective inverter PWM modulation techniques. An
oscilloscope is used to display the results of each PWM
III. SIMULATIONS AND RESULTS program on the inverter. Results show that outputs of the
Arduino Mega 2560 and the Five-level Single-phase Cascaded
A. Simulation circuit
H-Bridge (CHB) multilevel inverter designed (shown in Fig. 6
Fig. 4 shows the overall simulation circuit using (b)) can match the simulation results in MATLAB/Simulink.
MATLAB/Simulink software. It includes an oscillator, a This indicates that the Arduino Mega 2560 and the Five-level
comparator, and two cells of bridge type inverter. Single-phase Cascaded H-Bridge (CHB) multilevel inverter
designed is valid to be used in this research.
C. Comparative analysis of various PWM on the Five-level
Single-phase Cascaded H-Bridge (CHB) multilevel
inverter
Fig. 4 Five-level Single-phase Cascaded H-Bridge (CHB) multilevel inverter

circuit modeled using MATLAB/Simulink Fig. 7 Voltage waveform using Phase-shifted Carrier-based PWM
The circuit working principle is that a reference voltage
(vm) is compared with four triangular signals or carrier signals
(vcr) to produce four kinds of pulse signals. The four pulse
signals are the input for electronic switches at the upper
position of each inverter cell. The electronic switches at the
lower position of each inverter cell get the inverse of the four
pulse signals for the input. It aims to set the on and off states
for each switch to avoid short circuit.
B. Validating the Arduino Mega 2560 and the Five-level Fig. 8 Harmonic spectrum using Phase-shifted Carrier-based PWM
Single- phase Cascaded H-Bridge (CHB) multilevel inverter
Fig. 5 Measuring circuit for the output voltage of the Arduino Mega 2560 and
the Five-level Single-phase Cascaded H-Bridge (CHB) multilevel inverter Fig. 9 Voltage waveform using IPD Level-shifted Carrier-based PWM
(a) (b)
Fig. 6 The measuring circuit and the inverter designed Fig. 10 Harmonic spectrum using IPD Level-shifted Carrier-based PWM

____________________________________________________________________________________________________________
Phase-shifted Carrier-based PWM. The harmonic spectrum

graphs in Fig. 7 until Fig. 14 indicates that the Phase-shifted
Carrier-based PWM produces the lowest low order harmonics
content compared with the three schemes of the Level-shifted
Carrier-based PWM. Measured values of the inverter in Table
II show that the lowest THD value of 16.40 % is produced by
the Phase-shifted Carrier-based PWM. From all of above
results, it can be concluded that the best profile is produced by
the Phase-shifted Carrier-based PWM for the Five-level
Fig. 11 Voltage waveform using POD Level-shifted Carrier-based PWM
Single-phase Cascaded H-Bridge multilevel inverter.
IV. CONCLUSIONS
In this study, the Phase-shifted Carrier-based PWM and
the three schemes of Level-shifted Carrier-based PWM are
applied in the Five-level Single-phase Cascaded H-Bridge
multilevel inverter. The results show that the nearest
waveform to sinusoidal waveform is produced by the Phase-
shifted Carrier-based PWM. The Phase-shifted Carrier-based
Fig. 12 Harmonic spectrum using POD Level-shifted Carrier-based PWM PWM also produce the lowest low order harmonics and the
lowest THD value of 16.40 %. Hence, from all of those results,
it can be concluded that the best profile is produced by the
Phase-shifted Carrier-based PWM for the Five-level Single-
phase Cascaded H-Bridge multilevel inverter.
ACKNOWLEDGMENT
The authors would like to thank the Ministry of Research
Technology and Higher Education of Republic of Indonesia
Fig. 13 Voltage waveform using APOD Level-shifted Carrier-based PWM for the financial support in this research and the Electrical
Engineering Department of University of Mataram for
providing the laboratory equipment and other facilities.
REFERENCES
[1] M. Derakhshanfar, Analysis of different inverter topologies Master of
Science Thesis, Division of Electric Power Engineering Department of
Energy and Environment, Chalmers University of Technology, 2010.
[2] C. Gomathi, S.V. Navyanagath, and S. Veerakumar, ―Comparison of
PWM methods for multilevel inverter,‖ International Journal of
Advanced Research in Electrical,Electronics and Instrumentation
Fig. 14 Harmonic spectrum using APOD Level-shifted Carrier-based PWM Engineering, vol.2, pp.6106-6114, 2013.
From the voltage waveform and harmonic spectrum [3] B. Ismail, S. I. S. Hassan, R. C. Ismail, A. R. Haron, and A. Azmi,
produced by various PWM on the Five-level Single-phase ―Selective harmonic elimination of five-level cascaded inverter using
particle swarm optimization‖, International Journal of Engineering
Cascaded H-Bridge (CHB) multilevel inverter shown in Fig. 7 and Technology (IJET), vol. 5, pp.5220-5232, 2014.
until Fig. 14, the measured values can be summarized in Table [4] N. V. Nho, and M. J. Youn, ―Comprehensive study on space vector
II. PWM and carrier based PWM correlation in multilevel inverters,‖ IEE
TABLE II Proceedings Electric Power Applications, vol.153, pp.149-158, 2006.
MEASURED VALUES OF THE INVERTER VOLTAGE WAVEFORMS [5] J. G. Monicka, and V. Jamuna, ―Optimal switching strategy of level
AND HARMONIC SPECTRUMS shifted carrier based PWM technique for asymmetric multilevel
Measured value inverter,‖ Proceeding of International Conference on Interdisciplinary
Carrier-based Research In Engineering and Technology (ICIDRET), pp.146-150,
Vrms Vpp Frequency THD 2014.
PWM [6] A. Venkatakrishna, R. Somanatham, and M. S. Reddy, ―Phase shifted
(V) (V) (Hz) (%) and level shifted PWM based cascaded multilevel inverter fed
induction motor drive,‖ International Journal of Current Engineering
Phase-Shifted 8.22 25.20 47.79 16.40
and Technology, vol.4, pp. 350-354, 2014.
IPD Level-Shifted 8.48 25.80 51.36 23.60 [7] E. Sambath, S. P. Natarajan, and C. R. Balamurugan, ―Performance
evaluation of multi carrier based PWM techniques for single phase five
POD Level-Shifted 8.38 26.40 50.25 22.80 level H-bridge type, ―FCMLI IOSR Journal of Engineering (IOSRJEN),
vol.2, pp.82-90, 2012.
APOD Level-Shifted 8.43 27.20 51.20 24.30 [8] B. Wu, High Power Converter and AC Drives, The Institute of
Electrical and Electronics Engineers, Inc., 2005.
Voltage waveforms in Fig. 7 until Fig. 14 show that the
nearest waveform to sinusoidal waveform is produced by the

____________________________________________________________________________________________________________
Multi-Area Smart Monitoring of Electrical

Quantities Based on Mini Single Board Computer
BCM 2835
Dikpride Despa#1, Mardiana#2, Gigih Forda Nama#3 , M. Komarudin#4, Michael Zontche Bernard*5
1
Department of Electrical Engineering, 2,3,4 Department of Informatics Engineering
#
University of Lampung, Lampung-Indonesia
despa@eng.unila.ac.id, mardiana@eng.unila.ac.id, gigih@eng.unila.ac.id, komar@eng.unila.ac.id
5
Department of Electrical and Computer Engineering
*
University of Alberta - Edmonton, Alberta, Canada
mzbernard@ualberta.ca
Abstract— Multi-Area Smart Monitoring Systems are are not properly monitored, they can result in electrical
applications and devices that can monitor the amount of outages and even damage to the equipment.
electricity at various locations in the distribution panel. This To alleviate the above-mentioned problems, Unila built a
paper presents a scheme for multi-area smart monitoring of system that can provide real-time information in cases of both
electrical quantities trends in real-time, to overcome the
limitations of techniques that are applicable only to one panel of
normal and abnormal events. With such a system, it is
a three-phase distribution system. The implementation of the expected that if an abnormal condition is monitored, the
proposed scheme increases the possibility of taking preventive operator can take proper action early, so that problematic
action for protection in case of abnormal conditions. A Single conditions such as sudden blackout and the risk of equipment
Board Computer BCM2835 SoC, ARM11767JZF-S 700 MHz damage can be avoided.
processor is utilized to serve as an interface for obtaining data of In fact, Unila has many buildings that need to be monitored
the electrical quantities. It collects the data and saves them to a in order to take preventive action for protection, in the case of
database system. All data utilized in this study are obtained from abnormal conditions. Hence, this research is developed for
the LAB-JTE and the ICT-unila. Both locations are on the multi-area systems, in an effort to enable the possibilities of
campus of the UNILA. These data have been recorded and sent
to a computer server through the Local Area Network
monitoring the electrical quantities at multiple locations
connection and monitored in real time. The monitoring of data panels in a three-phase power distribution network that are
which was conducted from the 29th May to the 31st May 2016 located in different buildings at the same time. Multi-Area
produced significant results, showing that the voltage, current smart monitoring of electrical quantities is advantageous in
and load connected to the monitored phase, relatively has that, it overcomes the limitations associated with techniques
asymmetric trends. The results also demonstrated that at noon, that are applicable only to one panel of a three-phase
the electrical energy consumption trends at the Unila is in distribution system.
accordance with the working hours there. Finally, by comparing In this study of multi-area smart monitoring, the Mini
the results obtained from the monitoring of the electrical Single Board Computer BCM 2835, ARM11767JZF-S 700
quantities trends at the LAB-JTE and the ICT-Unila, it is shown
that the conditions of the electrical quantities at the ICT-Unila
MHz processor (Raspberry Pi model B) are the main
has better measurement results. components utilized. Raspberry Pi serves as an interface for
the acquisition of the electrical quantities data and saves them
Keywords— Multi-area panel distribution, real-time monitoring,
Electrical Quantities, BCM2835, Unila
to the database system. The electrical quantities data which
are measured by sensors via two distribution panels is sent
through the Internet and stored on a computer server that can
I. INTRODUCTION be accessed by the manager. The trend information of
electrical quantities at multiple locations can then be
University of Lampung (Unila) has always been monitored simultaneously in real-time. Furthermore, the
implementing some of its developmental plans by working expected results of monitoring may be used as a reference for
tirelessly to increase its quality and services in various ways. stakeholders, in helping them to make decisions with regards
Some of its key efforts include the development of to maintenance and planning for future models.
information technology, improvement of infrastructure as well The rest of this paper is organized as follows: Section 2
as laboratory equipment, which have an influenced in deals with related work, including limitations of the works
increasing the number of electrical energy consumption. previously carried out in this area of research. This section
Generally, electrical equipment should be able to work also provides the contribution of the current work. In section 3,
normally within their pre-specified limits or standard values. the research methodology is presented along with the block
But in reality, abnormal conditions may occur within the diagram system. The two hardware prototypes utilized in this
electrical system for a long period of time. If these conditions

____________________________________________________________________________________________________________
research are also presented in the section. Section 4 focuses on (GPIO) pin. Data such as current, voltage, power, energy
results and discussions where the prototypes testing have been consumption and power factor are recorded in real-time via
implemented and graphically illustrated to validate the database servers: Linux and MySQL servers. These data are
effectiveness of the proposed monitoring scheme. The paper is further processed by a web server and appear in the form of a
concluded in section 5 where significant directions for future historical statistical chart that can be accessed online through
work are also provided. the Internet.
II. RELATED WORK III. RESEARCH METHODOLOGY

A. Previous Research A. Blok Diagram System
The application of real-time monitoring of electrical
quantities and stability of power systems have been
considered by several studies [1]-[4]. In addition, previous
studies regarded as significant references in this research
considered the use of Microcontroller, Raspberry Pi, Arduino
and web application for the measurement and monitoring
system [1],[4]-[11]. [1] and [4] are also considered key
references of this work. These references are mainly
associated with the design and prototype of monitoring
electrical quantities such as voltage, current, power, the power
Fig. 2 Block Diagram System
factor (cos ϕ), and energy consumption application using
BCM2835, and they focused on one of the panels a three The block diagram shown in Fig. 2 is described to facilitate
phase distribution system that exist at Unila. and a better understanding of the system.
In [4],[5] the measured data are stored in real-time on a
database server such as Linux and MySQL server. Data B. Hardware Design
obtained are further processed by the web server so that it In this study, two hardware prototypes with the same
appears in the form of a historical statistical chart that can be design are made and positioned in two of the three-phase
accessed online via the Internet. However, [4] and [5] are only distribution panels at different locations. The design of the
applicable to one panel of a three phase distribution system in hardware is structured by electronic components as illustrated
Unila. The favorable properties of applying the monitoring in Fig. 3 below.
techniques of the above-mentions references which include
Arduino Converter DC to DC
their applicability to only one panel of a three phase
distribution system are diminished by the need for the BCM2835 Signal
Conditioning
monitoring of multi-area system in real time, which is an Trafo Circuit
objective achieved in this research. Current
Sensor
Rasbery
Pi
B. Hardware Design Adaptor
Voltage Sensor
Fig. 3 Hardware Design
C. Programme Design
Python programming [12] is used to create several function
and routine that run on Raspberry Pi. These include sensors’
data program reader, a calculation program for obtaining the
electrical quantities and recording the electrical quantities data,
a web interface display, and a database system receiver
program.
Current and voltage values are measured directly, while
other electrical quantities are determined by the use of the
equations below [13]:
E = (P.t) / 1000 (kWh) ……(1)
P = E / t……….................... (2)
Fig. 1 Hardware Topology Design for Smart Monitoring System P = VI cos ϕ…...………........(3)
Pf = cos ϕ =P / (VI).…...........(4)
Fig. 1 illustrates a design of the hardware topology for where P is the Power in Watts, E is the Energy consumption
smart monitoring devices connected to Arduino and BCM in KWh, t is the time in hours, V is the voltage in Volts, I is
2835 (Raspberry Pi) through a general-purpose input/output the current in Ampere and Pf is the power factor.

____________________________________________________________________________________________________________
IV. RESULT AND DISCUSSION ICT-Unila
A. Prototype Testing
The designed prototype has been tested with variations of
load.
TABLE I : Fig. 4(b) Current Monitoring
TEST RESULTS
No Various Test Results
of Load Volt (v) Current (I) Pf (cos ¢) LAB-JTE
Proto Volt Proto Ampere Proto Cos ¢
type meter type meter type meter
1 Load 1 221,10 220 2,2 2,2 0,99 0,99
2 Load 2 221,95 222 5,28 5,27 0,99 0,99
3 Load 3 220,75 221 0,44 0,59 0,99 0,99 ICT-Unila
Table.1 shows the data of the measurement test results

using the hardware prototype and compared the results to
those of a standard measuring device, namely the multi-meter. Fig. 4(c) Power Monitoring
It can be seen from the table that the prototype and multi-
meter have a very small difference in results. This indicates LAB-JTE
that the tool is significantly reliable.
B. Results of System Testing

The testing of the system was done at two different
locations: the 3-Phase distribution panels of LAB-JTE ICT-Unila
building and ICT-Unila building.
The results of the monitoring system in both buildings are
stored in a database system: MySQL server. The final step is
displaying the data from the existing database into a web
interface. In this test, the database was split into five tables. Fig. 4(d) Power Factor Monitoring
They are voltage, current, power, cos ϕ and energy
consumption. The trends of measurement results are displayed LAB-JTE
in the statistical chart of the web interface. The monitoring
charts results for each measurement described are shown in
Fig. 4 (a, b, c, d, and e).
LAB-JTE ICT-Unila
Fig. 4(e) Energy Consumption

ICT-Unila
Fig. 4 Results obtained from Monitoring from 29th May to the 31st May 2016
As seen in Fig.4, the graphs represent the results obtained

Fig. 4(a) Voltage Monitoring
for multi-area monitoring, from the 29th May to the 31st May
2016. The monitoring results and the trends of the electrical
quantities shown in Fig. 4 (a) –(e) for LAB-JTE and ICT-
LAB-JTE Unila are discussed as follows:
a. Voltage Monitoring Results
The voltage rating is about 200 V to 240 V. It can be seen
from the graph of Fig.4 (a) that, for each phase shown, the
voltage value at the two locations are not balance. The voltage
drop occurs during the day (working hours) with the lowest
value for the three-phase system observed in both LAB -JTE
and ICT Unila. This illustrates the imbalance load mounted on
each phase and electric energy consumption patterns are not
evenly distributed, especially during a certain time of the day.

____________________________________________________________________________________________________________
energy consumption in Unila is higher during the day,

b. Current Monitoring especially during working hours, as compared to the evening
The data shows that the monitored current in each phase at or night time. Furthermore, it has been shown that the
two locations is not in balance condition. From the graph of conditions of the electrical quantities within the ICT-Unila is
Fig. 4 (b), it is observed that the current increases during better as compared to conditions of the electrical quantities
working hours (8:00 to 16:00) with 35A being the highest within the LAB-JTE building.
value at ICT-Unila building. This is because there are more
B. Future Works
activities at Unila during working hours. At night, the
In order to make the utilized system become better, this
activities at Unila are reduced significantly. During this time,
application should be integrated with a notification system.
the university is basically used for lighting and air
The development of a mobile application for this project is
conditioning only.
also highly recommended. Furthermore, the project should be
c. Power Monitoring
extended to several locations, for experimentations with more
According to Fig. 4 (c), the power consumption tends to be
than two areas. Finally, the data results from this research
higher during the working hours. If we note the power at each
should be used as a reference for planning, estimation as well
phase, then the electrical energy consumption in ICT-Unila
as the improvement of the electrical system of Unila in the
building is greater than that of LAB-JTE. This is
future.
understandable because of the many activities carried on
during the day at the ICT-Unila building. REFERENCES
d. Power factor (cos ¢) Monitoring [1] D. Despa, A. Kurniawan, M. Komarudin, Mardiana, G. F. Nama,
The power factor monitoring charts shows where the value "Smart monitoring of electrical quantities based on single board
of the power factor was obtained from the calculation using computer BCM2835", 2nd International Conference on Information
voltage and current measurement data. This is illustrated in Technology, Computer, and Electrical Engineering, pp 315-320,
Indonesia, 2015.
Fig. 4 (d), where the power factor at the ICT building ranged [2] Qing Liu, Despa. D, Y. Mitani “ Application of Phasor and Node
from 0.7 to 0.99, while in the LAB-JTE building it is around Voltage Measurements to Monitoring Power Flow Stability
0.3 to 1. This means that the ICT-Unila building has a better “ International Journal Electrical Engineering and Informatics, Printed
power factor as compared to LAB-JTE. One of the factors that ISSN 2085-6830/online, Volume 4. Number 3 Pp 276-288, July 2012.
[3] D. Despa, Y. Mitani, M. Watanabe, “PMU Based Monitoring and
can influence the value of the power factor is a reactive load. Estimation of Inter-area Power Oscillation for Singapore-Malaysia
Hence, if the value of the power factor is unstable, a possible Interconnection Power System”, Journal of Energy and Power
cause is due to the changes in the reactive power of the load. Engineering, USA, 2012.
e. Energy Consumption Monitoring [4] Dikpride Despa, Mardiana, Gigih F N, “ Smart Monitoring Data
In general, since the load connected to each phase is not Centre base on Mini Single Board Computer BCM 2835″, The 1st
International Conference on Science, Technology, and Interdisciplinary
balanced, the energy consumption of each phase will behave Research (IC-STAR), Indonesia,2015
in the same manner. The total energy consumption at ICT- [5] G. F. Nama, M Komarudin, Hanang. P, Mardiana, H. D. Septama,
Unila building is greater than the total energy consumption at "Electricity, Temperature, and Network Utilization Monitoring at
LAB-JTE. This scenario is illustrated in Fig. 4 (e) which Lampung University Data Centre Using Low Cost Low Power Single
Board Mini Computer", Regional Conference On Computer
explains that there are more activities involving the use of Information Engineering, Indonesia, pp. 184-189, 2014.
electrical appliances within the ICT-Unila building as [6] E. Baker, “Open source data logger for low-cost environmental
compared to that of Lab-JTE. monitoring”, Biodiversity Data Journal, 2014.
[7] G. Aceto, A. Botta, D. Donato, P. Marchetta, A. Pescape, G. Ventre,
“ Open Source Platforms for Internet Monitoring and Measurement”,
V. CONCLUSIONS AND FUTURE WORK Eighth International Conference on Signal Image Technology and
A. Conclusions Internet Based Systems (SITIS), G. 2012.
This paper carried out monitoring of the electrical [8] T. Michael Silver, “Monitoring Network and Service Availability with
Open-Source Software”, Information Technology & Libraries. Mar
quantities trend at LAB-JTE and compares the result to that of 2010, Vol. 29 Issue 1, 2010.
ICT-Unila. The results of the monitoring may be utilized as a [9] Riyanto, R.O Wiyagi, “Sistem Monitoring Suhu Ruang Server Berbasis
reference for stakeholders, in helping them make decisions Web”, Junal Ilmiah Elite Elektro, 2(1), pp. 50 – 54, 2011.
with respect to maintenance and planning for future models. [10] G. F. Nama, M. Komarudin, H. D. Septama, "Performance Analysis of
ArubaTM Wireless Local Area Network Lampung University", 2015
The results obtained from the monitoring of electrical International Conference on Science in Information Technology
quantities at two different locations show that there are (ICSITech), pp 41-46, Yogyakarta, Indonesia, 2015.
unbalance conditions during working hours. It is also [11] BCM2835 ARM Peripheral, http://www.raspberrypi.org/wp-
observed that the results in the trends of electrical quantities content/uploads/2012/02/BCM2835-ARM-Peripherals.pdf. Accessed:
24 July 2016.
that were monitored in ICT-Unila are better as compared to [12] B. Rhodes, J. Goerzen, 2010. Foundation of Python Network
those of LAB-JTE. Programming. 2nd Ed. New York: Apress.
On the other hand, the electrical quantities for each phase [13] Fowler, Richard J, “Electricity Principles & Applications, McGraw-
that have been monitored in Unila are not in a balance Hill”, 2008.
condition. This indicates that the load distribution for each
phase is unbalanced and needs to be fixed. The electrical

____________________________________________________________________________________________________________
Schottky Diodes Comparison for Differentially-

Driven Rectifier
Fitri Yuli Zulkifli#1, Amirsyah Rayhan Mubarak#2, Basari#3, Eko Tjipto Rahardjo#4
#
Department of Electrical Engineering, University of Indonesia
Kampus UI – Depok, West Java 16424, Indonesia
1
yuli@eng.ui.ac.id
2
amirsyah.rayhan@gmail.com
3
basyarie@eng.ui.ac.id
4
eko@eng.ui.ac.id
Abstract— This paper discusses the comparison of various types characteristics, such as its unique signal detection, sensitivity,
of schottky diode for Differentially-Driven Rectifier as the most and its application. Therefore, it is interesting to observe the
important component in a rectifier system. In order to harvest influence of the schottky diodes to DDR.
RF energy at frequency 2.4 GHz ISM, with low power input In this paper, a comparison of several schottky diodes is
range for WLAN application, this rectifier’s diode performance
implemented for DDR. The circuit can be used as a rectifier
has to be considered. Comparison between these diodes
implemented in schematic from ADS simulation is presented. unit of any rectenna operating at 2.45 GHz frequency band.
The diodes should show its good performances, its output and its The simulation is observed from low input power to medium
RF-to-DC efficiency. The results show that HSMS-2850 is the power (-20 dBm to +20 dBm). The outline of this paper at
best result to be implemented in the proposed circuit because of first, the basic principle of rectifier is discussed. In later
its good output and good efficiency at low input power level. section, the design methodology of the RF energy circuit and
Keywords— Differentially-Driven Rectifier, Schottky Diode, the schottky diode is presented. Finally the simulation results
Energy harvesting are shown to describe the characteristics of the circuit.
I. INTRODUCTION II. BASIC PRINCIPLE OF RECTIFIER

The topic of energy harvesting becomes one of the most Simply defined, rectifier is an electronic device that
interesting researches in recent years. Energy harvesting converts the alternating current (AC) to direct current (DC). It
device converting ambient energy into electrical energy have takes current that flows alternately in both directions.
attracted much interest to be utilized. This research wants to Microwave rectifiers have different topologies, depending
get a system that can be useful for applications around the on the position and number of HF diodes shown in Fig.1. The
environment; one of the benefits is to power the wearable series optimized for low power levels (below 1 mW or 0
electronics (cell phones, radio communication devices, etc). A dBm), and the shunt for medium incident power levels
possible source of energy comes from ubiquitous radio (between 1 mW and 100 mW or 0 to 20 dBm respectively).
transmitters. Every single day, radio frequency, such as from They have proved a good sensibility for very low incident
BTS tower and Wi-Fi are always radiating and much of the power levels. For higher power levels, bridge type rectifiers
energy are stored wirelessly. Especially in urban areas, many and rectenna associations have proved to offer better
public places are now declared as hotspot zones operating at performances, most of all due to their higher power handling
2.45 GHz resonant frequency. This potential energy can be capabilities and this rectifier topologies is used for high
harvested because the energy is radiated 24 hours a day. incident power levels (higher than 100 mW or 20 dBm) [1].
The device used to harvest RF energy is called ‘Rectenna’. Also there was rectifier topology called as a voltage doubler.
Rectenna is basically a rectifying antenna; which is used to Thus, the voltage on output capacitor is roughly two times the
harvest RF signal, and then it converts the RF signal to DC peak voltage of the RF source.
output. Most of the works are based on designing the suitable
antenna for receiving RF signal and increasing the efficiency
of the rectifier.
Many studies have discussed how to develop and to
improve the functionality of the system, especially in the
rectifier side. Various configurations from different basic
types are proposed [1], to increasing, not only the efficiency,
but also the output voltage. Differentially-Driven Rectifier [2]
with its design is proposed to fulfil those goals. The circuit,
especially the diode, mainly determines the RF-to-DC
(a)
conversion efficiency. Schottky diodes are produced with its

____________________________________________________________________________________________________________
0.15–0.45 volts, with lower voltage drop can provide higher

switching speed & better system efficiency [3].
(b)
(a)
(c)
(d)
Fig. 1 Rectenna circuit topologies: (a) Single series-mounted diode; (b) (b)
Single shunt-mounted diode; (c) Diode bridge; and (d) Voltage Doubler diode. Fig. 2 Rectifier circuit topologies, (a) Two single-driven rectifiers, (b)
Optimized differentially-driven rectifier [2]
A. Differentially-Driven Rectifier
This design comes from an idea, not only to gain the There are various kinds of schottky diodes produced by
output voltage, but also to get optimum efficiency. Based on many manufactures, and have their own characteristics itself.
[2], the rectifier consists of two-shunt-rectifier topologies They are manufactured with different kinds of operations,
which then is integrated into a close rectifier with two ports as features, and also package code, whether it is single, series,
the input. As a result, this RF-to-DC configuration can unconnected pair, etc shown in fig.4. Virtually all of today's
optimally increase the output voltage up to 1,5 times higher, mass produced electronics hardware is manufactured using
with a high efficiency conversion. Fig. 2 shows the optimized surface mount technology, SMT. The associated surface
DDR from two-shunt-rectifier to be a simplified DDR with mount devices, SMDs provide many advantages over their
optimization value for each component. leaded predecessors in terms of manufacturability and often
Based on this topology, this work is done by simulating performance [3].
these circuit using software Advanced Design System (ADS)
in Microstrip Line. The simulation consists of how the III. SIMULATION OF SCHOTTKY DIODES ON DIFFERENTIALLY
performances from this rectifier and DC output voltage with DRIVEN RECTIFIER DDR
various range of input power. As the rectifier circuit is fixed Fig. 4 shows the Differentially-Driven Rectifier
with the specifications, the candidates of schottky diode are configuration in ADS simulation. This rectifier is designed to
applied based on the different availability types from ADS work at WLAN frequency 2,4 GHz with different
Library. specification in detail. The simulation is conducted using
substrate Taconic with εr = 2,2, and thickness of 1,52 mm.
B. Schottky Diode Also, the DDR rectifier has been set with proper length and
The Schottky diode is a semiconductor diode with a low width in schematic. Schottky diode D1 then will be substituted
forward voltage drop and a very fast switching action. A as the test subject of which schottky diode can give good
Schottky diode is a special kind of diode with a very low performances for output DC Voltage and the circuit efficiency.
forward voltage drop. When current flows through a diode, Table II shows the detail of width and length of each
therefore there is a small voltage drop across the diode microstip lines.
terminals. This diode’s voltage drop is between approximately

____________________________________________________________________________________________________________
diode will work optimum with this rectifier configuration. The

considerations to be taken are having a high DC Output
Voltage at low input power, and having high conversion
MSUB efficiency. The results in detail can be seen in Table IV.
Εr = 2,2
H = 1,52 mm
Tan d = 0,0009
Pin = -20 to +20 dBm
Fig. 4 Circuit schematics of the DDR Simulation with parameters, with Cd1
= 36 pf; Cd2 = 150 pf; Cd3 = 2.4 pf; Rd = 2kΩ.
TABLE II
THE PARAMETERS OF WIDTH AND LENGTH OF DDR RECTIFIER
TL W (mm) L (mm) TL W (mm) L (mm)

1 1.1 7 9 3 11
2 1.1 6.3 10 3 12
3 1.1 9.5 11 1.1 12
4 1.4 12 12 3.5 5.7 (a)
5 1.4 11.5 13 6 10
6 1.1 12.3 14 6 10.5
7 1.1 9.5 15 3.5 5.7
8 1.1 12
The performance comes from S-Parameter simulation (S11

and S21, Input Impedance) which the targets are the return loss
(S11) below the level of -10dB or the insertion loss (S21) is
above -3dB, then the input impedance is matched at 50Ω. The
DC Voltage comes from Transient simulation in ADS. The
results of the simulation can be seen in Fig. 5.
Schottky diodes are being tested from different kinds of
available diodes from ADS library. First, from S-Parameter
simulation, results in detail is shown in Table III.
(b)
TABLE III
THE S-PARAMETER RESULTS OF DDR RECTIFIER SIMULATION
Schottky S-Parameter
Diode Input Impedance
S11 (dB) S21 (dB)
(Z0) at 2.45 GHz (Ω)
HSMS-2850 -32,275 -0,099 49,4
HSMS-2860 -33,611 -0,058 49,8
HSMS-2820 -24,26 -0,108 50,2
BAT-5400 -17,98 -0,179 51,05
HSMS-8101 -20,99 0,169 51
The DC Voltage output from range of input power Pin in

dBm. With the load of 2kΩ resistor as a dummy, the Vout is
achieved. The conversion efficiency can be calculated by
∆
²
100% (1)
where ∆V is the result between output voltage with input
voltage in square and Vin is the input voltage. All these (c)
equation in square because the power is equivalent with the Fig. 5 The S-Parameter simulation of DDR Rectifier: (a) S11; (b) S21; (c)
square of voltage. The transient simulation is used to achieve Input Impedance from HSMS-2850
comparison between Vout with Vin converted from Pin. With
the characteristics of each schottky diode, we can see which

____________________________________________________________________________________________________________
TABLE IV
THE COMPARISON OUTPUT VOLTAGE AND EFFICIENCY OF DDR BASED ON
VARIOUS SCHOTTKY DIODES
Transient Vout – Pin

Schottky Diode Pin Vin Vout Efficiency
(dBm) (V) (V) (%)
-20 0,060 0,067 19,970
-15 0,104 0,125 30,457
-10 0,188 0,239 38,278
-5 0,329 0,469 50,963
HSMS-2850 0 0,585 0,923 59,881
Fig. 6 The Transient Vout – Vin Simulation Results of DDR Rectifier at Pin
5 1,013 1,811 68,712 = 10dBm from HSMS-2850
10 1,769 3,586 75,656 From those two parameters simulation, we have shown that
15 3,310 5,523 64,091
all of these diode can operate into this topology, as the S-
20 5,964 8,397 49,562
Parameter works inside the frequncy of WLAN (2.4 – 2.485
-20 0,059 0,063 12,295
-15 0,107 0,113 10,183
GHz) , pointed at resonant frequency of 2.45 GHz. Then, from
-10 0,190 0,204 13,025 the transient simulation for DC output voltage, we have seen
-5 0,337 0,390 25,159 which kinds of the diode’s characteristic exhibitsthe level
HSMS-2860 0 0,591 0,803 45,792 output voltage. As the best, the HSMS-2850 gives a good
5 1,033 1,622 59,402 performance and better efficiency from the other schottky
10 1,798 3,285 70,042 diode in order to have a diode that operate good at low power
15 3,154 6,624 77,330 input. With input power of -5 to 0 dBm, this diode can convert
20 5,882 10,039 65,670 the AC signal power to DC power up to 0.92 V, instead of the
-20 0,061 0,064 8,340 input voltage from 0.58 V.
-15 0,109 0,113 7,205 However, the oher diodes doesn’t mean that they have poor
-10 0,191 0,201 9,958 output voltage. Considering HSMS-2860, the output voltage
-5 0,345 0,358 7,425 and efficiency have been resulted as good as HSMS-2850, but
HSMS-2820 0 0,607 0,642 10,662 merely the input power needed is too high to get the optimum
5 1,088 1,169 13,428 work. This conditions are similar for HSMS-2820 and the
10 1,929 2,223 24,697
other diodes.
15 3,420 4,252 35,306
20 5,982 8,105 45,522 IV. CONCLUSIONS
-20 0,061 0,081 43,28
-15 0,107 0,145 45,54 The comparison of schottky diodes for a Differentially-
-10 0,186 0,257 47,62 Driven Rectifier configuration which works at WLAN
-5 0,343 0,459 44,15 Frequency 2.4 GHz has been discussed. All of the Schottky
BAT-5400 0 0,615 0,824 44,29 Diode’s candidates simulated can conduct in the DDR
5 1,04 1,47 49,95 configuration. However, the HSMS-2850 is the best choice for
10 1,733 2,68 58,18 the configuration as it shows good results at output voltage
15 3,46 4,88 49,72 with low input power and also high efficiency compared to the
20 6,06 8,89 53,53 other Schottky Diodes. The simulation results that with the
-20 0,061 0,075 33,85 HSMS-2850 Schottky Diode, the DDR can give the output
-15 0,107 0,134 36,24 voltage up to 3.58 V instead of input voltage of 1.77 V with
-10 0,189 0,237 36,40 the efficiency of DDR is 75.65% at 10 dBm input power.
-5 0,343 0,423 34,25
HSMS-8101 0 0,608 0,759 35,83 ACKNOWLEDGEMENT
5 1,09 1,4 39,38
This research was partially supported by “Penelitian
10 1,93 2,66 47,36
Unggulan Perguruan Tinggi (PUPT)”, Indonesian Directorate
15 3,43 0,96 52,18
20 6,14 9,16 55,07
General of Higher Education (DIKTI).
Fig. 6 shows one of the transient simulation results. As we REFERENCES
can see the voltage output increases up to 1.5 times from input
[1] B. Allard, V. Marian, C. Vollaire. “Strategy for Microwave Energy
voltage where the output voltage has reached 3.58 V with Harvesting from Ambient Field or a Feeding Source”. IEEE
input power 1.77 V with HSMS-2850 schottky diode. The Transactions on Power Electronics. 2012.
fluctuation in V is caused by the persistence of the conductor [2] Sun, H. “An Enhanced Rectenna Using Differentially-Fed Rectifier For
Wireless Power Transfer”. IEEE Antennas and Wireless Propagation
component values, as seen in Fig. 6, giving a gap about 0.2 to
Letters, IEEE (Volume:PP , Issue: 99). 2014.
0.5 V as a result. The fluctuation increases as the input power [3] Radio-Electronics: Schottky Barrier Diode. Available:
increases. http://www.radioelectronics.com/info/data/semicond/schottky_diode/

____________________________________________________________________________________________________________
A Strategy to Reduce Stray Capacitance on

Electrical Capacitance Tomography Hardware
Arba’i Yusuf#1,2, Dodi Sudiana#1, Agus Santoso Tamsir#1, Harry Sudibyo S#1
#1
Department of Electrical Engineering, Universitas Indonesia
Depok, Indonesia
1
arbai@c-techlabs.com, dodi.sudiana@ui.ac.id; agus.tamsir@gmail.com;
harisudi@eng.ui.ac.id
#2
C-Tech Labs, Edwar Technology Co
Tangerang, Indonesia
Abstract— The research relates to tomography technique based component of PCB, and adds the analog switch between signal
on electrical mass properties measurement. This study proposed generator and multiplexer. With this new design, it is expected
a strategy to reduce stray capacitance on electrical capacitance to reduce stray capacitance significantly.
tomography (ECT) hardware. Stray capacitance is a parasitic
signal present from other sources that has initially been there and II. DESIGN
strongly influences measurement accuracy, sensitivity, and
resolution. One of strategies to reduce of stray capacitance is in The principle of electrical capacitance tomography is
the design of electronic circuit at the front end. The front end injection of a square wave or sine wave signal into the
circuit consists of switch configuration and charge amplifier. This electrode, which will cause fringing electric field inside sensor
proposed system is then compared with other design to investigate area as shown in Figure 1. Then, the signal is detected by the
the result. Experiment shows the effect of stray capacitance is detector to be further processed and reconstructed into an
fairly small. image using soft computing algorithm.
Keywords— electrical capacitance tomography, stray capacitance,
front end circuit, noise signal.
I. INTRODUCTION
Electrical capacitance tomography (ECT) is one of imaging
technique based on permittivity distribution of object inside the
sensor. The permittivity can be determined by measuring
capacitance from electrode pair of the sensor, which
capacitance changes is in femtofarad (10-15) ranging from 0.3
to 80 fF [1], [2]. The capacitance measurement can be
influenced by noises originated from coaxial cable, screen
enclosure, CMOS switch for selecting electrode pair [3],
electromagnetic interference (EMI) from circuit design,
component layout, and grounding effect [4], [5]. The noise as
described above is named stray capacitance.
Stray capacitance is a parasitic present from other sources
that has initially been there and strongly influences Figure 1. Principle of electrical capacitance tomography (ECT)
measurement accuracy, sensitivity, and resolution. Many ways The switch configuration formed as the letter “T”; hence, it
to reduce stray capacitance as described in [6], that the stray is called T-switch as shown in Figure 2. With this configuration,
capacitance can be reduced by using T-switch configuration each electrode can be connected to excitation source or
and applying a charge amplifier on each channel. Another detection circuit depending on the given control signal. In this
researcher in [7] described that the stray capacitance can be design, Complementary Metal Oxide Semiconductor
reduced by using T-switch configuration and applying radial Integrated Circuit (CMOS IC) DG470 manufactured from
and earthed screen in sensor design. Vishay Siliconix are used. The switch type in this chip is SPDT
The robust design of electronic circuit of the ECT system is (single pole dual tap), which has one common connection pin
very necessary. Key achievement of capacitance measurement and two connection pins i.e. NO (normally open) and NC
is at the front end system, which consists of switch (normally closed). Some specifications include supply voltage
configuration and charge amplifier design as described in this 44 volt, analog signal range ±15 volt, low on-resistance 3.6
paper. The design applies T-switch configuration, places
charge amplifier on each channel, newly layouts electronic

____________________________________________________________________________________________________________
ohm, off-capacitance of 85 pF, on-capacitance of 125 pF, from detection electrode into voltage signal for further
capacitance-to-ground of 37 pF [8]. processing.
Figure 2. T-Switch configuration for each electrode sensor
Each channel of ECT system contains T-Switch Figure 4. Charge amplifier circuit
configuration as depicted in Figure 2, hence for eight channels
of ECT has eight T-Switch configurations and placed on single The object inside sensor that has complex impedance is
Printed Circuit Board (PCB). This PCB also contains a charge considered equivalent to capacitor and resistor in parallel in
amplifier circuit. In the previous design, the DDS signal measurement circuit and can be expressed as:
generator and buffer for excitation source was placed on = + (1)
another single PCB, which will generates a continuous
sinusoidal signal at each channel. The excitation signal of each = (2)
channel goes through directly into the PCB of T-Switch and
charge amplifier circuit as shown in Figure 3(a). Therefore, the −1
charge amplifier circuit is not immune due to interferences by = (3)
the excitation signal, which will increase a stray capacitance.
Thus, the modification of electronic circuit is necessary as Where, is complex impedance of the object, is
shown in Figure 3(b). The modification of ECT circuit system resistance of the object, is capacitance of the object, all of
is done by adding the analog switch between excitation signal which are measured inside sensor.
PCB and T-Switch PCB. With the new circuit, other unused
excitation signal will not flow directly to the T-Switch PCB.
III. EXPERIMENTAL SETUP
The charge amplifier circuit is stray immune, therefore it can
decrease the stray capacitance. Figure 5 shows a PCB of signal generator as excitation
source used in the ECT system. The signal generator has buffer
for each channel and controlled by a microcontroller to set
which channel pair needs to be activated. However, other
buffers of the signal generator still generate sinusoidal signal
which can interfere the charge amplifier circuit and increase
the stray capacitance. Therefore, modification hardware of
signal generator is necessary as depicted in Figure 5(b) by
adding analog switches.
Figure 3. A Part of Circuit Diagram of ECT System; (a) before modification;

(b) after modification
The basic principle of measurement capacitance on ECT is

Figure 5. Signal generator as excitation source; (a) before modification; (b)
using charge amplifier circuit as shown in Figure 4, which after modification by adding analog switches
consists of Operational Amplifier (OpAmp) and feedback
components (resistor and capacitor ). and are Experimental setup of ECT system, 8 channel capacitance
sine wave signal input and output respectively, is object sensor, and cylinder rods are shown in Figure 6. In the
impedance will be measured, and are parasitic following experiment, DDS signal generator generates
impedance from coaxial cable and sensor screen. The charge excitation signals of 500 kHz, 21.6 Vpp passed through to
amplifier circuit has a function to convert the current signal buffer signal for each channel. The channel pair is selected by
microcontroller, allowing excitation signal flows to excitation
electrode, producing an output signal on detection electrode.

____________________________________________________________________________________________________________
Digital signal processing (DSP) is then used to calculate The experiment was conducted using a cylinder pipe made
impedance and capacitance. from polyvinyl chloride (PVC) with 6 cm of diameter filled
with solid salt (εr=15). To investigate the effect of stray
capacitance with respect to ECT system’s performance, image
reconstruction was carried out as shown in Figure 7. The
reconstruction data pattern into image utilizes Iterative Linear
Back Projection (ILBP) algorithm (not discussed in this paper).
Thorough explanation about algorithm’s implementation can
be found in [9] [10]. For image reconstruction, capacitance
measurement is normalized using equation:
−
= (4)
Where is normalized capacitance, is measured

Figure 6. Photograph of experimental setup; (a) sensor and data acquisition capacitance, is capacitance at empty calibration. Empty
system; (b) Cylinder rods used in experiment
calibration is when the sensor filled with low permittivity
An ECT cylinder sensor with 8 channel of electrodes was material, such as dry rice (εr=3.5).
used for the experiment. The material of the vessel is plastic
(εr=2.4) with 13.5 cm diameter and 1 mm of thin. The
dimension of the copper sheet of the electrode is 4.5x7.2 cm;
the gap between electrodes is 9 mm, the earth screen at top and
bottom side of the sensor is 4 mm.
IV. RESULTS AND DISCUSSION

The experimental result of the ECT system was conducted,
firstly to find voltage output correlation of charge amplifier and
stray capacitance. Secondly, the result of ECT system before
adding switches and after adding switches were compared, and
the result is shown in Table 1. The ECT hardware before
modification has stray capacitance as high as 1.49 pF; whereas
after modification it has lower stray capacitance of 0.24 pF.
Stray capacitance affects the measurement such as resolution,
sensitivity, and measurement range. If the stray capacitance is
high enough, measurement range tends to be high, hence the Figure 7. Stray capacitance with respect to image reconstruction; (a) Cylinder
ECT system cannot measure a small capacitance. rod object; (b) before adding switches; (c) after adding switches
Consequently, the system is less sensitive. The result of image reconstruction shows almost no
difference between them. The shape is same, though, there is
Table 1. Stray capacitance of ECT system little difference: Figure 7(b) yields a thicker image than Figure
7(c). It indicates that stray capacitance is bigger. According to
Before adding After adding
the experiment, the effect of stray capacitance has less
switches switches
influence on image reconstruction because the capacitance
Voltage output from
charge circuit (V) 0.68 0.11 measurement value is normalized using (4) before being
processed. However, stray capacitance has influence on
Stray capacitance
(pF) 1.49 0.24 measurement range and sensitivity as described above.
Measurement range
(pF) 1.49 – 6.8 0.24 – 3
V. CONCLUSION
Sensitivity (V pF-1) 0.22 0.38
The front end circuit on ECT hardware plays important role
to get a good performance system. It is necessary to consider
The maximum measurement value of ECT hardware in PCB design and components layout, where excitation signal
depends on the maximum reading of analog-digital-conversion should not pass through PCB T-switches which can cause stray
(ADC) which is used, usually 1.5 Vpp. After adding switches, capacitance. The modification design at front end circuit by
the stray capacitance is decreased, so that the signal gain can adding switches can decrease stray capacitance significantly
be increased in order to produce a voltage output closer to the about 0.24 pF comparing to that without modification. With
maximum reading of ADC. Accordingly, reducing the stray lower stray capacitance, the ECT hardware can measure small
capacitance in the ECT hardware is very necessary. capacitance at femtofarad level.

____________________________________________________________________________________________________________
ACKNOWLEDGMENT
The Author would like to thank all persons who kindly
contributed to this research. This work was supported by a
grant from Hibah PITTA DRPM Universitas Indonesia; grant
number: 2076/UN2.R12/HKP.05.00/2016.
REFERENCES
[1] Yang, W. Q., Stott, A. L., & Beck, M. S. (1994). High frequency and
high resolution capacitance measuring circuit for process tomography.
IEE Proceedings-Circuits, Devices and Systems, 141(3), 215-219.
[2] Yusuf, A., Muttakin, I., Rudin, A., Widada, W., & Taruno, W. P. (2014,
October). Single signal conditioning multi electrode for ECVT data
acquisition system. In TENCON 2014-2014 IEEE Region 10
Conference (pp. 1-6). IEEE.
[3] Yang, W. Q. (1996). Hardware design of electrical capacitance
tomography systems. Measurement Science and Technology, 7(3), 225.
[4] Wei, Z. (2012). The Electromagnetic Interference Model Analysis of
the Power Switching Devices. Indonesian Journal of Electrical
Engineering and Computer Science, 11(1), 167-172.
[5] Qiang, C. H. E. N. G., Ning, C. H. E. N. G., & Zhen-Shuang, L. (2013).
Grounding Effect on Common Mode Interference of Underground
Inverter. Indonesian Journal of Electrical Engineering and Computer
Science, 11(9), 5189-5194.
[6] Yusuf, A., Sudibyo, H., Sudiana, D., Tamsir, A. S., Muttakin, I.,
Widada, W., & Taruno, W. P. (2016). Switch Configuration Effect on
Stray Capacitance in Electrical Capacitance Volume Tomography
Hardware. TELKOMNIKA (Telecommunication Computing
Electronics and Control), 14(2).
[7] Yang, W. Q., Stott, A. L., & Gamio, J. C. (2003). Analysis of the effect
of stray capacitance on an ac-based capacitance tomography transducer.
IEEE Transactions on Instrumentation and Measurement, 52(5), 1674-
1681.
[8] DG469/470 Data Sheet Book. 2007, Vishay Siliconix. Document
Number: 71470 S-72541-Rev. C.
[9] Warsito, W., Marashdeh, Q., & Fan, L. (2007). Electrical capacitance
volume tomography. IEEE Sensors Journal, 7(3/4), 525.
[10] Herdian, H., Muttakin, I., Saputra, A., Yusuf, A., Widada, W. and
Taruno, W.P., (2015, November). Hardware implementation of linear
back-projection algorithm for capacitance tomography. In 2015 4th
International Conference on Instrumentation, Communications,
Information Technology, and Biomedical Engineering (ICICI-BME)
(pp. 124-129). IEEE.

____________________________________________________________________________________________________________
Injection Duration Control in a Four Strokes Engine

based on Throttle Position
Muhammad Aziz Muslim#1, Goegoes Dwi Nusantoro#2, Dwi Fadila Kurniawan#3, Ganda Lesmana#4
#
Jl. MT Haryono 167, Malang, Indonesia
1
muh_aziz@ub.ac.id
2goegoesdn@ub.ac.id
3
iwan_fadilla@yahoo.com
4
ganda16lesmana@gmail.com
Abstract— Fuel injection duration is one of the important The rest of this manuscript is organized as follows. The
parameters for the efficient combustion process in a four strokes proposed research method will be described in session 2,
engine. In this study, the injection duration based on throttle contains research apparatus and proposed control schema.
position that reflects the desired engine speed setting is being Followed by session 3 explaining resulting experiments and
considered. A fuzzy logic controller that has the ability to use the
their discussions. Session 4 closes the paper with a conclusion
language of human reasoning is employed for this purpose. The
results showed the use of fuzzy logic control can meet the and further research plans.
specifications of the desired response with an error rate of
injection duration and engine rotation speed is less than 2 %. II. RESEARCH METHODS
A. Research Apparatus
Keywords— injection duration, four strokes engine, throttle Combustion engine is one of the motor piston with the
position, fuzzy logic control combustion process in the cylinder (internal combustion).
Combustion engine in this study is a small gasoline engine
I. INTRODUCTION
motorcycles with four working steps. The four steps include
Fuel injection system is a system that is responsible for the suction, compression, power and exhaust step. Those steps
managing the time and duration of fuel injection to machines. require two turns of the crankshaft per one cycle. Figure 1 (a)
Technological developments of fuel injection in combustion - (d) shows the sequence of the four processes at motor the
process has undergone very rapid progress. Recent injection motor cycle.
system mostly equipped with electronic devices so the amount
of fuel sprayed into the engine become more precise. An
Engine Control Unit (ECU), which is an electronic device,
responsible for controling the ignition timing, frequency and
pulse width to the fuel injectors, and controls the amount of
fuel injected [1]. An ECU becomes a very important part for a
vehicle in this modern era.
The contents of an ECU is always kept secret by
manufacturers of motor vehicles. Data contained in the ECU
is data that has be set by the manufacturer in accordance with
the standard requirements of each type of vehicle, hence it is
relatively difficult to modify the standard ECU system. For Fig 1. Steps sequence in a four strokes engine:
example, to modify the injection system, one needs to design a (a) suction (b) compression (c) power and (d)
control system which is able to adjust the time and frequency exhaust
of injection to the condition of engine speed [2]. From Figure 1, it can also be seen that one needs some
Proposed study concern with modification of the injection sensors to understand condition in the combustion chamber in
system in a 4 strokes motorcycle. For that purpose, an order to decide proper timing for injection and ignition and
integrated control system among engine speed, throttle also proper duration of injection. Some required sensors are as
position and duration of fuel injection are being considered. follows:
Throttle position reflects the magnitude of desired engine Throttle Position Sensor (TPS)
speed, which will be taken into account in the scheme of Throttle is part of the injection machine that regulates the
Fuzzy Logic Control (FLC). Expected resulting engine speed entry of air into the combustion engine. The function of TPS
would be in accordance with the desired setpoint. include monitoring whether throttle position is partially open,
fully open or closed. These sensors are usually located on the

____________________________________________________________________________________________________________
butterfly spindle, so that it can directly monitor the position of

the throttle. Engine Control Unit (ECU) can control the
throttle position.
Crankshaft Position Sensor (CPS)
Crankshaft Position Sensor is an electronic device used in
internal combustion engines to monitor the position or the
rotational speed of the crankshaft. This information is used by
the engine management system to control the timing of the
ignition system and other engine parameters.
In this study, only one engine parameter to be controlled,
which is the duration of injection. Hence, there only one
actuator will be used, namely injector. Injector spraying fuel
(gasoline) to the engine chamber for mixing with air. To ease
mixed fuel-air production, the fuel is atomized into steam.
Injector controlled electromagnetically by ECU through a
negative signal. The pulse width of the signal from ECU will
determine the amount of steamed fuel. The longer the pulse, Fig 3. Fuzzy Sets of Input (Error and Refference)
the more steamed gasoline will be injected.
B. Design of Fuzzy Logic Controller

Figure 2 is a speed control scheme proposed in this study.
Throttle position sensor converts the throttle position to the
desired engine speed (in rpm). On the other hand, the actual
engine speed is detected by the crankshaft position sensor.
The difference between the reference speed and the actual
speed of the engine is calculated and refer as the error signal.
The error signal and the reference value considered as the Fig 4. Output (Injektor) Fuzzy Sets
input for the fuzzy logic controller. From the results of fuzzy
inference process, a control signal will be generated telling the
injector to spray fuel into the combustion chamber at a III. EXPERIMENTAL RESULTS AND ANALYSIS
specific duration .
The first step to designing a fuzzy logic controller is to Experiments conducted on a 4 strokes motorcycle engine.
define the fuzzy set of input and output. For simplicity, we This bike has a 108cc engine capacity and maximum power of
employed input fuzzy set (error and reference) with 3 8.52 ps / 8,500 rpm. Fuzzy logic control algorithm is planted
membership function and output fuzzy set (signal injector) on an ATmega32 board microcontroller. The output of the
also with three membership functions . Figure 3 and Figure 4 microcontroller is a signal that drives the injector. Injector will
illustrates the fuzzy set of input and output . be active (squirt fuel) when given a LOW signal, and vice
versa will not be active when the driver signals is HIGH.
Injection duration is set by adjusting the duration of LOW and
HIGH signal. Figure 5 shows the relationship between the
Throttle Throttle Reff. Fuzzy Speed (rpm)

4 Strokes
Position Logic Injector
Position Speed Engine
Sensor + Controller
(rpm
-
Crankshaft
Position
Sensor
Fig 2. Proposed control scheme

____________________________________________________________________________________________________________
desired speed with the throttle position . From the figure it can
be seen that the desired speed is proportional to throttle
position.
(a)
Fig 5. Relation of desired speed and throttle position
(b)
Fig 7. Engine speed (a) throttle position 0% (b) throttle

position 5%
Injection Duration (ms)
5
)s 4.5 IV. CONCLUSIONS
4
(m
n 3.5
o
tia 3
r 2.5
u
This study has been successfully designed an injection
D
n 2 Ideal duration control system using fuzzy logic control. The
io
cte 1.5 Experiment experimental results showed a good performance with an error
j 1
In
0.5 rate of less than 2 % compared to ideal conditions. There are
0
still many factors other than the duration of the injection
should be controlled by an ECU. Further research needs to be
Engine Speed (rpm)
done to establish an intelligent ECU, capable of working at a
Fig 6. Experimental Result of Injection Duration various operating conditions.
ACKNOWLEDGMENT
Furthermore, the task of the fuzzy logic controller is This research is funded by Directorate of Research and
driving injection duration corresponds to the desired engine Public Service, Directorate General of Research and
speed. Fig 6 shows the results of injection duration control by Development, Ministry of Research, Technology and Higher
fuzzy logic controller. The result is very close to ideal Education under contract number
conditions, with an error rate of 1.4 %. The results of the 033/SP2H/LT/DRPM/II/2016, February 17th, 2016
engine speed control for the throttle position of 0 % (speed set REFERENCES
point of 1500 rpm) and throttle position of 5 % (speed set
[1] T. C. Weil, H. Selamat, and A. J. Alimin, “Modeling and Control of an
point of 2000 rpm) are shown in Fig 7. These results are also Engine Fuel Injection System,” Int. J. Simul. Syst. Sci. Technol., vol. 11,
very close to ideal conditions, with an error rate of no. 5, pp. 48–60, Sep. 2010.
approximately 2 %. [2] S. H. Lee, R. J. Howlett, and S. D. Walters, “Engine fuel injection
control using fuzzy logic,” in Total Vehicle Technology: Finding the
Radical, Implementing the Practical (3rd International Conference),
2005, vol. 10, p. 287.

____________________________________________________________________________________________________________
Conception of an Autonomous Driving of Unmanned

Vehicles using Stereo Vision
Youness Lahdili #1, Afandi Ahmad *2 Abbes Amira
*
Bioelectronic & Instrumentation Res. Lab. (MiNT-SRC), Computer Science and Engineering Department,
#
Embedded Computing System (EmbCos), FKEE, College of Engineering, P. O Box 2713, Qatar University
Universiti Tun Hussein Onn Malaysia abbes.amira@qu.edu.qa
P. O Box 101, 86400 Batu Pahat, Johor, Malaysia
afandia@uthm.edu.my
Abstract— In the last five years, the emergence of unmanned

vehicles as a new class of robots. These range from unmanned II. PROCESS DESCRIPTIONS
aerial vehicles (UAVs), micro aerial vehicles (MAVs) commonly
called drones, to remotely operated underwater vehicles (ROVs), A. Overview
through inflatable dirigibles balloons (airships), driverless cars With stereo vision we refer to all cases where the same
and all other amateur and hobbyists gadgets in between, like the scene is observed by two cameras at different viewing
popularized quadcopters. They are now more affordable than positions. Hence, each camera observes a different projection
anytime before, and are slowly but surely populating our roads,
of the scene, which allows us to perform inference on the
waters and skies, to the point they will become ubiquitous
scene’s geometry. The obvious example for this mechanism is
elements in the traffic and logistics. But till today, these vehicles
are only being guided remotely by on-ground human operators the human visual system. Our eyes are laterally displaced,
who are prone to error and time-conscious. In this study, the which is why we observe a slightly different view of the
application of unmanned vehicles will be taken to a new level, by current scene with each. This allows our brain to infer the
sparing them the need for a human operator, and making them depth of the scene in view, which is commonly referred to as
fully autonomous. This will be possible by harnessing the power stereopsis. Although it has for long been believed that we are
of two computer vision methods that are essential parts in only able to sense the scene depth for distances up to few
photogrammetry technology: stereo vision depth and structure meters, Palmisano et al. [1] recently showed that stereo vision
from motion (SfM). Our contribution will allow the unmanned
can support our depth perception abilities even for larger
vehicle to be auto-aware of the dangers and obstructions that will
cross its way without any human intervention. distances. Using two cameras, it is possible to mimic the
human ability of depth perception through stereo vision. An
introduction to this field has been provided by Klette [2].
Keywords— Stereo Vision, Features Detection, UAVs, Depth perception is possible for arbitrary camera
Photogrammetry configurations, if the cameras share a sufficiently large
I. INTRODUCTION common field of view. We assume that we have two idealized
pinhole-type cameras C1 and C2 with projection centers O1
The proposed solution for this problematic of navigation and O2, as depicted in Figure 1. The distance between both
autonomy resides in the design of an intelligent trajectory projection centers is the baseline distance b. Both cameras
estimation system. This system will predict and trace a safe observe the same point p, which is projected as p1 in the
itinerary, also known as a planned path, clear from all sorts of image plane belonging to camera C1. We are now interested
obstacles. The vehicle will then only have to follow the in finding the point p2, which is the projection of the same
coordinates of this itinerary until reaching its given destination. point p on the image plane of camera C2. In literature, this
This itinerary must be continuously auto-adapting itself with task is known as the stereo correspondence problem, and its
the real-time change of environment, traffic movement, and solution through matching p1 to possible points in the image
the occurrence of unexpected obstacles, especially those plane of C2 is called stereo matching.
moving. So our aim is to make stereo vision depth and SfM p
coexist on the same System-on-Chip (SoC), and to apply them
in real autonomous navigation situation by testing them in the Image Epipolar Plane
Image
field. In that way, one method can mitigate the error that the Plane 1 Plane 2
other could have propagated, in the manner of hybrid systems. Epipolar Epipolar
Line 1 Line 2
This implies building a field programmable gate array (FPGA) p
prototype and mounting it on a binocular mid-range drone for p1 2
Baseline
validation and deployment. MATLAB has been selected as O2
O1
our abstraction tool, and Vivado HLS will be the middleware Epipole
between the hardware description language (HDL) and FPGA. s the epipolar geometry
Fig. 1 Example for

____________________________________________________________________________________________________________
In order to implement stereo vision depth awareness on The shown sparse example is precisely what we intend to
unmanned vehicles, we have first to solve this stereo matching apply in this research, which only finds disparity labels for a
problem, which goes back to the question of how to make the set of selected corner features. The color that is displayed for
FPGA able to tell that two points on two images taken of the these features corresponds to the magnitude of the found
same scene belongs to the same scene feature. disparity, with blue hues representing small and red hues
To achieve this result, we have to go through three (3) main representing large disparity values. The method used for the
stages. We will elaborate on each of them in the following dense example is the gradient-based belief propagation
sections, while pointing to the limitations observed and how algorithm that was employed by Schauwecker and Klette [14]
we intend to tackle them in our proposed research and Schauwecker et al. [15]. The results of this algorithm are
dense disparity maps that assign a disparity label to all pixels
B. Image Rectification in the left input image.
The common approach to stereo vision includes a Although sparse methods provide much less information
preliminary image rectification step, during which distortions than common dense approaches, this information can be
are corrected. The resulting image after rectification should sufficient for a set of applications, including UAV trajectory
match the image received from an ideal pinhole camera. To be estimation and obstacle avoidance such as proposed here in
able to perform such a correction, we first require an accurate our research.
model of the image distortions. The distortion model that is
most frequently used for this task today, is the one introduced D. Feature Detection
by Brown [3]. Using Brown’s distortion model, we are able to In computer vision, a feature detector is an algorithm that
calculate the undistorted image location (ũ, ṽ) that selects a set of image points from a given input image. These
corresponds to the image location (u, v) in the distorted image. points are chosen according to detector-specific saliency
criteria. A good feature detector is expected to always select
the same points when presented with images from the same
scene. This should also be the case if the viewing position is
changed, the camera is rotated or the illumination conditions
are varied. How well a feature detector is able to redetect the
same points is measured as repeatability, for which different
Existing implementations of the discussed algorithms can be definitions have been postulated by Schmid et al. [16]; and
found in the OpenCV library (Itseez, [4]) or the MATLAB Gauglitz et al. [17].
camera calibration toolbox (Bouguet, [5]), and that is how we Feature detectors are often used in conjunction with feature
plan to resolve this question of image rectification. descriptors. These methods aim at providing a robust
identification of the detected image features, which facilitates
C. Sparse Vision Method their recognition in case that they are re-observed. In our case,
Despite the groundbreaking work by [6]–[11], there is a we are mainly interested in feature detection and less in
gap regarding the speed performance of their systems. Our feature description. A discussion of many existing methods in
examinations of their work revealed that they have employed both fields can be found in the extensive survey published by
dense stereo matching methods which considers search of Tuytelaars and Mikolajczyk [18]. Furthermore, a thorough
matching points in the entire input stereo images, thus evaluation of several of these methods was published by
increasing the computational load of their systems. One way Gauglitz et al. [17].
to greatly speed-up stereo matching is to not process all pixel Various existing feature detectors extract image corners.
locations of the input images. While the commonly used dense Corners serve well as image features as they can be easily
approaches find a disparity label for almost all pixels in the identified and their position can generally be located with
reference image -usually the left image- sparse methods like good accuracy. Furthermore, image corners can still be
in [12] and [13], only process a small set of salient image identified as such if the image is rotated, or the scale or scene
features. An example for the results received with a sparse illumination are changed. Hence, a reliable corner detector can
compared to a dense stereo matching method can be found in provide features with high repeatability.
Figures 2 (a) and 2 (b). One less recent but still popular method for corner
detection is the Harris detector (Harris and Stephens, 1988). A
computationally less expensive method for detecting image
corners is the Smallest Univalue Segment Assimilating
Nucleus (SUSAN) detector that was proposed by Smith and
Brady [19].
A more advanced method that is similar to the SUSAN
detector is Features from Accelerated Segment Test (FAST).
One of the most influential methods in this category is the
Fig. 2 (a) Sparse stereo matching results received with the presented method
Scale Invariant Feature Transform (SIFT) by Lowe [20]. For
and (b) dense results received from a belief propagation based algorithm.
Color scale corresponds to the disparity in pixels [13]

____________________________________________________________________________________________________________
this method, two Gaussian convolutions with different values

Greyscale Cameras
for σ are computed for the input image.
A more time-efficient blob detector that was inspired by Baseline = 11cm
SIFT, is Speeded-Up Robust Features (SURF) by Bay et al.
[21]. Instead of using a DoG for detecting feature locations,
Bay et al. rely on the determinant of the Hessian matrix,
which is known from the Hessian-Laplace detector
(Mikolajczyk and Schmid [22]). Both SIFT and SURF exhibit
USB
a very high repeatability, as it has been shown by Gauglitz et Port
al. [17]. However, what Gauglitz et al. also have demonstrated
is that both methods require significant computation time.
PIXHAWK Cheetah
In this research we are going to address this gap as well, by
designing a slightly modified architecture of FAST corner
detection algorithm. Microprocessor Board
III. MODELLING THE DESIGN Serial IMU

FPGA SoC
These three key elements of our trajectory estimation Link
system will be supplemented by other filters, snippets and PID Controller Pose
SfM modules namely the Simultaneous Localization and Attitude
Mapping (SLAM), as depicted in Figure 3. low-level flight
control software
Low-Level Process
I2C
Feature Detection Stereo Matching Bus
Propellers of the MAV
High-Level Process
Pose Estimation
Local SLAM EKF Sensor Fusion Quadrotor

(4) Motors
Controller
Fig. 3 Processing pipeline of the proposed FPGA-implementation
The overall architecture will be synthesized on

reconfigurable hardware, consisting of field programmable
gate arrays (FPGAs) [23], [24]. These platforms promise to be Fig. 4 System implementation of the processing at the MAV physical level
adequate system building block in the building of
sophisticated devices at affordable cost. They offer heavy
IV. CONCLUSION
parallelism capabilities, considerable gate counts, and comes
in low-power packages [25]–[28]. Much progress has been achieved in the computer vision
Based on the existing work limitations, this project is community since the interest in sparse stereo matching
concerned with an efficient implementation of trajectory declined. In particular, new feature detectors have since been
estimation for autonomous navigation of unmanned vehicles, published, which can be used for the construction of new
with a special interest on aerial ones. sparse stereo methods, and then applied in MAV navigation.
Figure 4 shows the anatomy of the projected overall system But given the ever increasing need for highly efficient stereo
to be implemented. It is our aspired target to realize such matching, it seems logical to choose a sparse stereo method.
architecture and put it into application in different fields of life, The lack of current sparse stereo algorithms that fulfill these
like in aerial imaging, shipping parcels, search & performance requirements, while also delivering accurate
reconnaissance mission and many more. Moreover, this matching results, can became the motivation to design a novel
project will avail us of a locally-built solution that will not be sparse stereo matching system.
bound to foreign royalties or at risk of patent infringements
claims.

____________________________________________________________________________________________________________
ACKNOWLEDGMENT [13] K. Schauwecker and A. Zell, "On-board dual-stereo-vision for the

navigation of an autonomous MAV," Journal of Intelligent & Robotic
The authors would like to express their appreciation for the Systems, vol. 74, no. 1-2, pp. 1–16, Oct. 2013.
support of the VASYD Lab in FKEE UTHM, and the [14] K. Schauwecker and R. Klette, "A comparative study of two vertical
MathWorks Inc. for approving the free usage of the road Modelling techniques," in Computer Vision – ACCV 2010
Workshops. Springer Science + Business Media, 2011, pp. 174–183.
computational software of theirs. [15] K. Schauwecker, S. Morales, S. Hermann, and R. Klette, "A
comparative study of stereo-matching algorithms for road-modeling in
the presence of windscreen wipers," IEEE, 2009, pp. 12–7.
REFERENCES [16] C. Schmid, R. Mohr, and C. Bauckhage, International Journal of
[1] S. Palmisano, B. Gillam, D. G. Govan, R. S. Allison, and J. M. Harris, Computer Vision, vol. 37, no. 2, pp. 151–172, 2000.
"Stereoscopic perception of real depths at large distances," Journal of [17] S. Gauglitz, T. Höllerer, and M. Turk, "Evaluation of interest point
Vision, vol. 10, no. 6, pp. 19–19, Jun. 2010. detectors and feature Descriptors for visual tracking," International
[2] R. Klette, Concise Computer Vision, 2014th ed. London: Springer. Journal of Computer Vision, vol. 94, no. 3, pp. 335–360, Mar. 2011.
[3] D. C. Brown, "Decentering distortion of lenses," Photometric [18] T. Tuytelaars and K. Mikolajczyk, "Local invariant feature detectors: A
Engineering, vol. 32, no. 3, pp. 444–462, 1966. survey," Foundations and Trends® in Computer Graphics and Vision,
[4] Itseez, "OpenCV," 2015. [Online]. Available: http://opencv.org. vol. 3, no. 3, pp. 177–280, 2007.
Accessed: Apr. 2, 2016. [19] S. M. Smith and J. M. Brady, International Journal of Computer
[5] J. Y. Bouguet, "Camera Calibration Toolbox for MATLAB," 2013. Vision, vol. 23, no. 1, pp. 45–78, 1997.
[Online]. Available: http://vision.caltech.edu/. Accessed: Mar. 3, 2016. [20] D. G. Lowe, "Object recognition from local scale-invariant features,"
[6] M. Achtelik, T. Zhang, K. Kuhnlenz, and M. Buss, "Visual tracking vol. 2, IEEE, 1999, pp. 1150–2.
and control of a quadcopter using a stereo camera system and inertial [21] Bay, Herbert, Tinne Tuytelaars, and Luc Van Gool. ―SURF: Speeded
sensors," IEEE, 2012, pp. 2863–2869. up Robust Features.‖ Lecture Notes in Computer Science. N.p.:
[7] D. Pebrianti, F. Kendoul, S. Azrad, W. Wang, And K. Nonami, Springer Science + Business Media, 2006. 404–417. Web.
"Autonomous hovering and landing of a Quad-rotor micro aerial [22] Mikolajczyk, K., and C. Schmid. Indexing based on scale invariant
vehicle by means of on ground stereo vision system," Journal of interest points. IEEE, 2001. Web. 4 Apr. 2016.
System Design and Dynamics, vol. 4, no. 2, pp. 269–284, 2010. [23] P. Dang, ―VLSI architecture for real-time image and video processing
[8] L. R. García Carrillo, A. E. Dzul López, R. Lozano, and C. Pégard, systems," Journal of Real-Time Image Processing, vol. 1, pp. 57–62,
"Combining stereo vision and inertial navigation system for a Quad- 2006.
Rotor UAV," Journal of Intelligent & Robotic Systems, vol. 65, no. 1-4, [24] T. Todman, G. Constantinides, S. Wilton, O. Mencer, W. Luk, and P.
pp. 373–387, Aug. 2011. Cheung, ―Reconfigurable computing: architectures and design
[9] T. Tomic et al., "Toward a fully autonomous UAV: Research platform methods," Computers and Digital Techniques, IEE Proceedings, Vol.
for indoor and outdoor urban search and rescue," IEEE Robotics & 152, No. 2, pp. 193–207, 2005.
Automation Magazine, vol. 19, no. 3, pp. 46–56, Sep. 2012. [25] A. Ahmad, B. Krill, A. Amira, and H. Rabah, ―Efficient architectures
[10] A. Harmat, I. Sharf, and M. Trentini, "Parallel tracking and mapping for 3D HWT using dynamic partial reconfiguration," Journal of
with multiple cameras on an unmanned aerial vehicle," in Intelligent Systems Architecture, Vol. 56, No. 8, pp. 305–316, 2010.
Robotics and Applications. Springer Science + Business Media, 2012, [26] A. Ahmad, B. Krill, A. Amira, and H. Rabah, ―3D Haar wavelet
pp. 421–432. transform with dynamic partial reconfiguration for 3D medical image
[11] M. Nieuwenhuisen, D. Droeschel, J. Schneider, D. Holz, T. Läbe, and compression," in Biomedical Circuits and Systems Conference, 2009.
S. Behnke, "Multimodal obstacle detection and collision avoidance for BioCAS 2009. IEEE, 2009, pp. 137–140.
micro aerial vehicles," IEEE, pp. 12–7. [27] A. Ahmad and A. Amira, ―Efficient reconfigurable architectures for 3D
[12] S. Shen, Y. Mulgaonkar, N. Michael, and V. Kumar, "Vision-based medical image compression," in Field-Programmable Technology,
state estimation for autonomous rotorcraft MAVs in complex 2009. FPT 2009. International Conference on, 2009, pp. 472–474.
environments," IEEE, 2010, pp. 1758–1764. [28] B. Krill, A. Ahmad, A. Amira, and H. Rabah, ―An efficient FPGA-
based dynamic partial reconfiguration design flow and environment for
image and signal processing IP cores," Signal Processing: Image
Communication, Vol. 25, No. 5, pp. 377–387, 2010.

____________________________________________________________________________________________________________
A Numerical Study on Twist-Tappered Blade for

Low-Speed Horizontal Wind Turbine
Joke Pratilastiarso*, Setyo Nugroho, Teguh Hady Ariwibowo, Akhmad Khilmy
Dept. of Energy Generation System
Electronic Engineering Polytechnic Institute of Surabaya
Surabaya, Indonesia
*
E-mail: joke@pens.ac.id
Abstract— As a country with a relatively low and fluctuated wind

speed, Indonesia has its own wind turbine design characteristic.
Therefore, the presence of angular response of the pitch against
wind speed is required to maximize energy production, along
with an appropriate turbine design with wind conditions. This
study analyses a laboratory scale Clark-y twist-tapered-blade
wind turbine. The turbine is simulated in Blade Element
Momentum Method on wind velocity 3.5 to 7 m/s. Pitch Angle
variations are 2o, 4o, 6o, 8o. The result shows that there are some
differences on the performance on each variation of the pitch
against the Tip Speed Ratio (TSR) value. The Simulation
illustrates each pitch angle has maximum Cp at different TSR.
The pitch angle 2⁰ generates max Cp 0.5 at TSR 5. The pitch
angles 8⁰ results max Cp 0.37 at TSR 3. the pitch angle 4⁰
produces the greatest max Cp 0.43 at TSR 3.5. This circumstance
actuated at the pitch angle approaches the angle of attack 5⁰. this Fig 1. Pitch angle settings on wind turbines
angle has the greatest glide ratio.
Figure 1 describes a definition of pitch angle, twist angle, and
tapered blade
Keywords— blade element momentum; pitch angle; horizontal
axis wind turbine; clark y airfoil; low speed II. WIND TURBINE
I. INTRODUCTION A. Coefficient of Power

In Indonesia, the availability of wind energy is almost 50% A wind turbine has a limit that is unable to convert
of the area of Indonesia, with normal wind speeds between 3.5 energy effectively. This is Betz limit, Coeffisien Of
to 8 m/s [1]. However, the implementation of wind energy Performance (Cp), which value is 0.59. It is the ability of wind
into electrical energy only reached 10% of the potential turbine to absorb wind energy. Cp is ratio of wind turbine
energy [1]. In Indonesia, the wind speed is low and also power to wind power ratio.
fluctuative. Therefore, it takes a conditioned blade pitch angle,
B. Airfoil
in order to adjust its behavior of wind turbines upon the wind
speed. There are some airfoil segments that can generate lift
Reference [2] shows that a setting of a blade pitch angle is force and rotation upon wind turbine. Clark – Y’s airfoil is
able to maximize the performance of a horizontal axis wind selected to achieve high glide ratio (Cl/Cd). This performance
turbine at a certain wind speed. The blade design, however, is is gained in 5˚ angle of attack . Figure 2 describes the
for a high wind speeds, and has two blades [2]. Reference [3] definition of lift force, drag force, and angle of attack
shows because of using the pitch angle settings, a turbine
wheel can be set to a constant speed.
In practical way, we found out that the blade design with
taper and twist angle has good performance at low speed.
Angle of twist and tapper is useful to balance on blade’s force
distribution. The angle of attack at the blade’s root would be
greater than at the blade’s tip. In designing the twist angle and
tapper the method is Schmitz distribution[4]..Therefore in this
paper, numerical study of response to use pitch’s twist-
tappered-blade has been performed to get maximum energy. Fig. 2. Airfoil Clark-Y

____________________________________________________________________________________________________________
C. Pitch Angle
Pitch angle could be set to adapt condition to turbine so it
could maximize performances which depend on airfoil
characteristics.
Fig. 4 Velocity triangle on the blade
The study correlates effects of changes in local pitch angle.

This correlation can be written as follows
θ = θp + β
Axial induction factor (a) and tangential induction factor
(a') is aspect tangential velocity and the absolute velocity loss.
Both of these values are capably calculated by practicing some
Fig. 3 Characteristic curve Cp - TSR WECS
iteration of the following steps.
Ref [3] reveals that pitch angle go up, wind turbine
performance tend to decrease. High pitch angle has good Step (1) Initialize a and a ',
works on low Tip Speed Ratio (TSR) as shown on figure 3.
Choose 0 as initial value.
D. Twist-tappered Step (2) Calculate the flow angle φ
Schmitz distribution is implemented as an idea to
Step (3) Calculate the local angle of attack
construct twist and tapper on the wind turbine. The reference
is desirable TSR and highest glide ratio from the Clark – Y Step (4) Read the value of Cl (α) and Cd (α) on the graph
airfoil[5]
Step (5) Calculate Cn and Ct
( ) ( ( )) (1)  (2)
( )
 (3)
αD is an angle of attack in the optimum glide ratio (Cl / Cd).
The blade is divided into 15 segments in order to obtain the Step (6) Calculate a and a ' by using the equation:
distribution of chord and twist angles.
 a
III. BLADE ELEMENT MOMENTUM
The Blade Element Momentum (BEM) method was  a’
invented by Glauert [5]. It is able to calculate the steady load
on a blade. The method also count trust and power in some
variation of wind speed, rotational speed, and pitch angle. This Solidity can be found by the following equation
technique combines with the Stream Tube Momentum theory. ( )
This theory which is applied in one dimensional geometry  ( )
fragments a N’s annular segmental discretization [6]. Step (7) If a and a 'have changed over a certain tolerance,
There is velocity triangle which affects force generation on proceed to step 2, or if it has not changed much, the iteration
blade segment in BEM. The relative velocity (Vrel) is a has completed. Convergence criteria of iteration is 5 %.
combination of wind’s absolute velocity of the wind (Vo) and Relative velocity at the blade segment influences lift and
rotor’s tangential velocity (ωr). Inflow angle ( ) , a drag force.
combination of the angle of attack (α), and the pitch angle (θ),
is an angle between relative and tangential velocity as shown Lift and drag force per unit length can be evaluated by the
in figure 4. The angle can be computed by applying the following equation
equation below
L = ρVr2cCl (4)
( )
( )
; = –θ D = ρVr cCd 2
(5)
Lift and drag force are projected into the normal force (PN)
and tangential (PT) in the following equation

____________________________________________________________________________________________________________
PN= L cos ϕ + D sin ϕ (6) P=ω (11)

PT = Lsin ϕ –D cos ϕ (7)
Tangential force is required to determine torque blade.
IV. RESULT AND DISCUSSION
There are several airfoil segment that generate lift force
then rotate wind turbine. Clark – Y airfoil is used in this
research as high glide ratio (Cl/Cd) on 5˚ angle of attack[7] .
Fig. 7 illustrates response of pitch angle to power on
variation of wind velocity. Cut-in speed happens at 3.5 m. In
this condition all variations don’t produced power. Increased
wind speed is accompanied by a quadratic increase in power.
Fig. 5 Local force projection on the blade

Because PN and PT should be in per unit length, the normal
force (dT) and torque (DM) in the rotor are:
dT = B PN dr (8)
dM = r B PT dr (9)
Torque distribution per segment (dr) blade splitted into
several segments so torque value each blade elements can be Fig 7. the effect of pitch angle on the power of wind velocity
evaluated to generate power. variation
The increase of pitch angle, however, followed by decrease of

power. At the wind speeds of 7.5 m/s, the lowest power is
pitch angle 8o. The second lowest power occurs at pitch angle
6o. Pitch angle 2 o and 4 o have coincidence point because the
angle of attack on each blade segment is almost 2o.
Fig. 6. Force distribution on the blade

Fig. 6 shows a model of enumeration segment at BEM
method which is calculated by the following formulation.
Torque dM for a small part of the blade length dr is:

( ) Fig. 8. the effect of pitch angle on the coefficient of power of
The contribution Mi,i +1 on the total torque of the linear axis wind velocity variation
tangential between ri and ri+1 are able to :
Fig. 8 shows the correlation between the coefficients of power
to the pitch angle variation. The dramatic improvement occurs
from 3.5 to 4 m/s because the blades experience the cut-in
( ) ( ) speed and it suddenly generates power at 4 m/s. On wind
speed 4.5 to 7.5 m/s there is change in coefficient of power
Total torque is the sum of all Mi,i+1 along blade then it is
when wind speed variation is not significant. This
multiplied by the number of blades:
phenomenon is caused by absorbed and released energy of
∑ (10) turbine is similar.
The theoretical power is calculated by this formula

____________________________________________________________________________________________________________
V. CONCLUSIONS
Wind turbine simulation has been done by using Blade
Element Momentum method based on variations in pitch
angle in the range of wind speed 3.5 to 7 m/s. Pitch angle can
affect the performance of wind turbines in rotation and torque
turbine. The Simulation of Blade Clark Y generates maximum
performance on the pitch angle 2⁰ which produces TSR 5 and
Cp 0.5. the pitch angles 8⁰ produce Cp 0.37 at TSR 3 and the
pitch angle 4⁰ produce the greatest Cp 0.43 at TSR 3.5. This
phenomenon is caused by setting the pitch angle approaches
the angle of attack 5⁰. This angle of attack has the greatest
value of the glide ratio.
Fig. 9 the influence of the pitch angle on cp based on ACKNOWLEDGMENT

variations of TSR This research was supported by Decree number 0581 / E3 /
BEM Simulation input base on wind speed and TSR on 2016 Directorate of Research and Community Service Ministry
Reynolds number 105843. Fig. 9 shows the relationship of Research, Technology and Higher Education Republic of
between Cp with TSR. The fig 9 shows that the value of Cp Indonesia.
change regularly at angle 2⁰ and 4⁰ because the BEM
calculations use the same Reynolds number as the input data REFERENCES
base of airfoil. Reynolds number is influenced by the speed of [1] Soeripno Martosaputro, Wind Hybrid Power Generation Market
the wind so that the use of TSR in accordance with the Development Initiatives, Proceeding of APCRES 2013, Bali, 2013
[2] Guda Sreedhar Reddy. Modeling And Power Management Of A Hybrid
Reynolds number. Windmicroturbine Power Generation System.2005. Montana State
TSR changes may affect the inflow angle in the triangle University.
velocity. This is solved by changing the angle of attack of the [3] Ahmed, M. H; Wael, A. F. & Osama, A. M; (2011).Modelling and
airfoil at the point of maximum for the best value of glide control of Direct driven PMSG for ultra large wind turbines, World
Academy of Science, Engineering and Technology 59
ratio (Cl / Cd). This glide ratio value indicates airfoil has a [4] Søren Gundtoft . Wind Turbines. 2009. University College of Aarhus,
greater lift than drag so that the airfoil can produce high Aarhus: Denmark.
torque. The BEM simulation Reviews largest Cp Showed the [5] H. GLauert, ―Airplane Propellers,‖ Springer Berlin Heidelb., pp. 169–
TSR on 4 and 5 on the pitch angle 4⁰ and 2⁰. This pitch value 360, 1935.
[6] M. O. L. Hansen, Aerodynamics of Wind Turbines, Second., no. 11.
approaches angle of attack 5⁰ in almost all of segments of the London: Earthscan, 2008.
airfoil. [7] Bozeman, Montana T. Prashanth, S. SampathKumar, P. Harshitha
Reddy, K.Chandra Shekar, Subhash Deo Hiwase.‖Aerodynamic CFD
Study of Clark Y Airfo‖.IJERT.Vol. 3 Issue 2.February – 2014;India
`

____________________________________________________________________________________________________________
Label-Free Cellular Structure Imaging with High

Resolution Using an Electron-Beam Excitation-
Assisted Optical Microscope
Yoshimasa Kawata1,3,*, Masahiro Fukuta,2 Yuriko Masuda,2 and Wataru Inami,1,3
1
Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu 432-8561, Japan,
2
Graduate School of Science and Technology, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu 432-8561,
Japan
3
Adaptable and Seamless Technology transfer Program through Target driven R&D, Japan Science and Technology
Agency, Japan
Abstract—We present label-free and high spatial-resolution Imaging with super-resolution microscopes is based on
imaging for specific cellular structures using an electron-beam fluorescent labeling of cellular structures. Fluorescent
excitation-assisted optical microscope (EXA microscope). Images labelling permits high-contrast imaging of specific structures.
of the actin filament and mitochondria of stained HeLa cells, In addition, biomolechlar interactions with other structures
obtained by fluorescence and EXA microscopy, were compared
can be observed by multicolor imaging. However, organic
to identify cellular structures. Based on these results, we
demonstrated the feasibility of identifying label-free cellular fluorescence dyes are toxic to cells, and the uniformity of the
structures at a spatial resolution of 82 nm. Using numerical stain process always introduces the possibility of artifacts. In
analysis, we calculated the imaging depth region and determined addition, the fluorescent intensity of labeling bleaches the
the spot size of a cathodoluminescent (CL) light source to be 83 images under excitation light irradiation. It is difficult, as well,
nm at the membrane surface. to observe long-duration activities or the alteration of
biological cells.
Keywords— Bio imaging, optical microscopy, electron There many advantages of label-free technique, for
microscopy example, non-preparation before observation, reduction of
damage or deformation of biological cells due to fluorescence
I. INTRODUCTION labeling and no photobleaching of fluorescent dye. From these
High-resolution imaging of cellular structures, such as actin advantages, the label-free imaging techniques contribute long-
filaments, mitochondria, and nuclei, is a key tools for term dynamic observation of biological cells. In addition,
analyzing cellular activities, including movement, organelle label-free imaging technique is suited to observation for stem
transportation, and energy generation [1]. Although imaging cells since the labeling of stem cells is undersirable. The label-
of cell structures by phase-contrast microscopes and free imaging techniques are required in biomedical field and
fluorescent microscopes revealed the fundamental drug discovery.
mechanisms of cellular movement and organelle We have developed an electron-beam excitation-assisted
transportation [2], many questions about the details of cellular optical (EXA) microscope, which can observe label-free cells
functions or activities remain. Since the spatial resolution of with high spatial resolution [9-11]. The cathodoluminescent
conventional optical microscopes is bounded by the (CL) light source is excited in the luminescent thin film by the
diffraction limit, it is difficult to observe a cell’s detailed irradiation of an electron beam. The EXA microscope
structures. To reveal new mechanisms of cellular function, observes the spacimen beyond the diffraction limit by raster
improved optical microscopes are needed for achieving scanning the CL light source, where the CL spot is a few tens
greater spatial resolution. nm of nanometers in diameter.
In recent years, super-resolution fluorescence microscopes, Using the EXA microscope, we demonstrated label-free
used in stimulated emission depletion microscopy [3], biological cell imaging with 82 nm spatial resolution. We
structured illumination microscopy [4], photoactivated identified cellular structures specifically actin filaments,
localization microscopy [5], and stochastic optical mitochondria, and nuclei by comparing the fluorescent images
reconstruction microscopy [6], have demonstrated nano- and EXA images for the labeled cellular structures. We also
structure observations of less than 100 nm in lateral spatial present calculations and estimates of the theoretical spatial
resolution. These super-resolution microscopes have proven resolution, where the depth of the observed region was
capable of imaging the detail strcuctures and dynamics of estimated from the intensity distribution analysis of the CL
cellular structures, such as receptors, microtubules, and light source.
lysosomes[7,8].
II. CONFIGURATION OF THE EXA MICROSCOPE

____________________________________________________________________________________________________________
Figure 1(a) shows the schematic of the EXA microscope. Zn2SiO4, the intensity variation of this ZnO is reduced from
The EXA microscope is constructed from an electron 24.0% to 11.3%. The intensity variation is two times higher
microscope, a culture dish, and an optical microscope. The than that of Zn2SiO4.
inverted scanning electron microscope (APCO Ltd., MINI-
EOC) is used to generate the nanometric light source in the
EXA microscope. The electron beam from a field-emission-
type electron gun irradiates the luminescent film at the culture
dish. The CL intensity excited in the luminescent thin film is
detected by the photomultiplier tube (Hamamatsu Photonics K.
K., H10721-20). The observation image is obtained by raster
scanning of the electron beam. In the EXA microscope,
vacuum and atmosphere are separated by the substrate and
fluorescent thin film at the culture dish [9-11]. Atmospheric
pressure is maintained in the observation location of the
specimens so that biological cells can be observed in a living
state in the culture solution.
Figure 1(b) shows the detail of the square region in Fig.
1(a). In this study, the membrane substrate and fluorescent
film materials are silicon nitride (Si3N4) and zinc oxide (ZnO),
respectively. Si3N4 with a thickness of 30 nm to 50 nm can
serve as a boundary between the vacuum and atmosphere
because the Si3N4 has a high mechanical strength of 1GPa at
room temperature. A ZnO luminescent film is deposited to the
back side of the Si3N4 because ZnO film causes toxicity to
biological cell if the ZnO contacts with biological cell. The
biological cells are cultured directly on the Si3N4. The cells
are observed in CL light from the ZnO transmitted through of
the Si3N4 membrane. We formed the as-deposited ZnO film
by radio-frequency magnetron sputtering that utilized a ZnO
target with an oxygen-and-argon mixed reactive plasma. After
the sputtering process, the ZnO thin film was annealed at 800
ºC in the presence of N2. The EXA microscope can observe
the specimen with high spatial resolution because the CL spot
size at the Si3N4 surface remains a few tens of nanometers in
diameter. The image contrast is produced by the scattering or
absorption by the specimens illuminated by the evanescent
light at the Si3N4 surface.
Figure 1(c) shows the CL spectrum of ZnO in this study.
The CL single-emission peak of the ZnO appears at 380 nm.
This CL emission arises from the band gap of the ZnO. As Fig.
1(c) illustrates, the ZnO luminescent thin film presents a
monochromatic peak emission.
In this study, ZnO luminescent film was fabricated by
annealing at 800 ˚C for 15 minutes in N2. In previous report in
Ref. [9-11], we utilized Zn2SiO4 as the luminescent film by
annealing ZnO at 1000 ˚C for 60 minutes in N2. Although the
Zn2SiO4 emits bright CL emission, CL from Zn2SiO4 includes
intensity variability of 24.0% as shown in Fig. 2(a). This Fig. 1 (a) Schematic of the EXA microscope. (b) Enlarged image of the
intensity variability is caused by inhomogeneous crystalline highlighted square region in Fig. 1(a). (c) CL spectrum of the ZnO
luminescent thin film.
growth in the annealing process. To suppress the intensity
variability of CL emission, we change the annealing
temperature and annealing time from 1000 ˚C to 800 ˚C, and
60 minutes to 15 minutes, respectively. From these
adjustments, we suppressed the intensity variability of CL.
Figure 2(b) shows the CL image of ZnO by annealing at 800
˚C for 15 minutes in N2. The CL from ZnO film includes
11.3% root mean square variation. In comparison with

____________________________________________________________________________________________________________
Fig. 2 Comparison of CL image. (a) CL image of the Zn2SiO4. RMS

variability of CL intensity includes 24.0%. (b) CL image of the ZnO. RMS
variability of CL intensity includes 11.3%.
III. LABEL-FREE CELL IMAGING USING THE EXA MICROSCOPE

WITH HIGH SPATIAL RESOLUTION
Figure 3(a) shows the result of high spatial-resolution

imaging of the HeLa cell. The acquisition time to obtaine Fig.
3(a) was 200 seconds. The acceleration voltage of electorn
beam was 5 kV and current density was 1 nA. The inset shows
the enlarged image of square region in Fig. 3(a).
Figure 3(b) shows the intensity profile in the vicinity of the
intracellular granules, highlighted by arrows in Fig. 3(a). The
full width at half maximum (FWHM) of the intracellular
granules is 82 nm and the signal-to-noise ratio (SNR) is 10.5.
The SNR must exceed 5 to sustain a calim that the imaged
object is the structure in question. In this study, the achieved
SNR of 10.5 is more than two times higher than the minimum
neeeded. The EXA microscope observed HeLa cells with
spatial resolution of 82 nm and high SNR due to the reduction
of the CL intensity variation.
Fig. 3 (a) High resolution EXA microscopic imaging of the cellular granules
IV. CELL STRUCTURE IMAGING USING FLUORESCENCE AND THE (b) Intensity profile of the cellular granules highlighted by arrows in Fig. 2(a).
EXA MICROSCOPE The full width at half maximum is 82 nm and signal-to-noise ratio is 10.5.
In order to identify the label-free cellular structures in the Figure 4(b) shows the observation result, obtained with the
EXA microscope image, we observed the stained cells both EXA microscope, of the same area as shown in Fig. 4(a). The
with fluorescence and the EXA microscope to determine the acceleration voltage was 4.8 kV and the irradiation current
points of correspondence between the imaging results. was 1 nA. The acquire time to obtain Fig. 4(b) was 200
Figure 4(a) shows the observation results, obtained with a secconds. In Fig. 4(b), the EXA image was obtained only with
fluorescence microscope (IX-71, Olympus America Inc.), of the CL from the ZnO transmited through the 390 nm band-
fixed HeLa cells with stained actin filaments and path filter (FF01-390/40-25, IDEX Corpration). The cellular
mitochondria. HeLa cells were fixed by 1% glutaraldehyde. stuructrures in Fig. 4(b) can be identified by comparing them
The HeLa cells were placed in a phosphate-buffered saline with the corresponding structures in Fig. 6(a). In Figs. 4(a)
solution. The actin filament and mitochondria were stained by and 4(b), the corresponding actin filaments and mitochondria
AD48-81 (ATTO-488, ATTO-TEC GmbH) and Mitotracker are indicated by arrows and triangles. The actin filaments and
Orange CMTMRos (M7510, Thermo Fisher Scientific Inc.), mitochondria are brighter than the background CL because the
respectively. Stained actin filament and mitochondria were evanescent light is converted to propagation light by scattering
observed with fluorescence filter cubes of U-MNIBA3 and U- with the actin filaments and mitochondria near the Si3N4
MWIG3 (Olympus America Inc.), respectively. surface. As Fig. 4(b) displays, the EXA microscope
successfully imaged the actin filaments and mitochondria by
means of the CL from the ZnO; these cellular structures were
obserebed as bright regions when compared to the background
CL intensity.

____________________________________________________________________________________________________________
spatial resolution. We also calculated and estimated the

intensity distribution of the CL both inside and outside of the
ZnO and Si3N4 and estimated the CL spot size and observation
depth of the EXA microscope in this study. From the results
of this analysis, we determined the CL spot size to be 83 nm at
the Si3N4 surface. The CL spot size corresponded with the
spatial resolution of the EXA microscope. The observation
depth was 50 nm. According to the simulation results, the
observation results of the cellular structures were constructed
in the region 50 nm from the Si3N4 surface.
In addition, we identified cellular structures by comparing
the EXA microscope images with the labeled observation
results. The actin filaments and mitochondria of the HeLa
cells were identified in the observation image of the EXA
microscope. In the EXA image, the cellular structures were
observed as brighter regions as compared with the background
CL intensity. By comparing the EXA microscopic observed
results with images of label-free HeLa cells, we confirmed
that it was possible to indentify the cellular structures, such as
actin filaments, mitochondria, filopodia, nuclei, and cellular
granules.
REFERENCES
[1] D. C. Chan, “Mitochondria: Dynamic Organelles in Disease, Aging,
and Development,” Cell 125(7), 1241–1252 (2006).
[2] D. J. Stephens and V. J. Allan, “Light Microscopy Techniques for Live
Cell Imaging,” Science 300(5616), 82–86 (2003).
[3] S. W. Hell and J. Wichmann, “Breaking the diffraction resolution limit
by stimulated emission: stimulated-emission-depletion fluorescence
microscopy,” Opt. Lett. 19(11), 780–782 (1994).
[4] M. G. L. Gustafsson, “Surpassing the lateral resolution limit by a factor
of two using structured illumination microscopy,” J. Microsc. 198(2),
82–87 (2000).
[5] E. Betzig, G. H. Patterson, R. Sougrat, O. W. Lindwasser, S. Olenych,
J. S. Bonifacino, M. W. Davidson, J. L-Schwartz, and H. F. Hess,
“Imaging Intracellular Fluorescent Proteins at Nanometer Resolution,”
Science, 313(5793), 1642–1645 (2006).
[6] M. J. Rust, M. Bates, and X. Zhuang, “Sub-diffraction-limit imaging
by stochastic optical reconstruction microscopy (STORM),” Nat.
Methods 3(10), 793–795 (2006).
[7] C. A. Wurm, D. Neumann, M. A. Lauterbach, B. Harke, A. Egner, S.
W. Hell, and S. Jakobs, “Nanoscale distribution of mitochondrial
import receptor Tom20 is adjusted to cellular conditions and exhibits
an inner-cellular gradient,” Proc. Natl. Acad. Sci. USA 108(33),
13546–13551 (2011).
[8] M. Heilemann, S. van de Linde, M. Schüttpelz, R. Kasper, B. Seefeldt,
A. Mukherjee, P. Tinnefeld, and M. Sauer, “Subdiffraction-Resolution
Fluorescence Imaging with Conventional Fluorescent Probes,”
Angewandte Chemie International Edition 47(33), 6172–6176 (2008).
[9] W. Inami, K. Nakajima, A. Miyakawa, and Y. Kawata, “Electron beam
Fig. 4 (a) (a) Fluorescent microscopic image of the HeLa cells with stained excitation assisted optical microscope with ultra-high resolution,” Opt.
actin filament and mitochondria. (b) EXA microscope image for the HeLa Express 18(12), 12897–12902 (2010).
cells. The observation region is the same as Fig. 6(a). The corresponding actin [10] M. Fukuta, S. Kanamori, T. Furukawa, Y. Nawa, W. Inami, S. Lin, Y.
filaments and mitochondria are indicated by arrows and triangles. Kawata, and S. Terakawa, “Dynamic nano-imaging of label-free living
cells using electron beam excitation-assisted optical microscope,” Sci.
Rep. 5, 16068 (2015).
V. CONCLUSION [11] W. Inami, M. Fukuta, Y. Masuda, Y. Nawa, A. Ono, S. Lin, Y. Kawata,
and S. Terakawa, “A plastic scintillator film for an electron beam-
excitation assisted optical microscope,” Opt. Rev. 22(2), 354–358
In this study, we observed the label-free cellular structures (2015).
of HeLa cells at 82 nm spatial resolution by using the EXA
microscope. We succeessed in reducing the CL intensity
variation from 24.0% to 11.3%. This high spatial resolution
imaging was achieved by suppressing CL intensity variation.
As a result, cellular granules were observed at the same 82 nm

____________________________________________________________________________________________________________
Non Invasive Sensors for Bearing Damage

Identification Using Sound Signal Frequency
Analysis
Muhammad Aswin#1
#
Department of Electrical Engineering, Brawijaya University
Jl. MT Haryono 167 Malang- Indonesia
1
maswin@ub.ac.id
Abstract— Rotating bearing always generate sound. The sound harmonics, the berings will be tested on varying the
characteristics depend on bearing conditions. Usually, damaged frequency axis.
bearing sound have more complicated frequencies than normal Spectrum analysis will be done to investigate the damage
bearing. The source of the bearing sounds are the vibration of level of the bearings. The damage level can be found at
damages bearing. Researchers have indentified the damaged
around highest harmonics..
condition by analyze the bearing vibration frequencies directly.
This method have practical major problem when applied to the II. BEARING FREQUENCIES
operating bearings. Vibration sensors difficult to be placed. This
paper presents new method to determine damage bearing levels Minebea Co. Ltd. on the website (www.eminabea.com)
using the sound sensors. Two different damage level bearings provides vibration equation caused by inner race rotation in
were rotated in the same frequency separately. The generated frequency fr. With Dw is ball diameter, Dpw pitch distance,
sounds were recorded using smartphone microphone. The the contact angle αo, and the number of balls Z, then the
frequency content of each bearing sound than analyzed to frequency vibrations are:
identify the damages.
(1)
Keywords— Non Invasive sensor, bearing damage, frequency
analysis.
fb = fa (2)
I. INTRODUCTION
Bearing is an important part in the mechanical system. The
main function is to minimize the friction between moving (3)
components, especially under rotational motion. But, along
with the application, the bearings performance usually
fd = Zfa (4)
decreases. Corrosion, wear, overloading, unbalanced
installation, or contamination also can fasten the bearing
degradation. The bearing damage is often caused huge losses, fdi = Z(fr – fa) (5)
or even of the overall machine failure. Therefore, damage
early warning is very necessary. All units are in Hz :
Detecting and identifing damage bearing is very important
to prevent failure spreading of the machinery systems. In this fa = vibration frequency caused by ball revolution
research, as a basic analysis, parameters of signals from fb = vibration frequency caused by rotation
normal bearing are used as a reference. The damaged bearing fc = vibration frequency caused by ball spin
parameters are compared with the normal parameters to get fd = vibration frequency on outer race
the level of damage. For this purpose, 2 bearings with fdi = vibration frequency on inner race
different damage level have been carefully chosen.
The first parameter will be searched is dominant
frequency harmonics (have significant value) from the III. DAMAGED BEARING FREQUENCIES
bearings sound signal using autocorrelation analysis. The vibration frequency will become more varied in
Dominant harmonic frequency can be looked as peaks in the existence of one ore more defects, such as cavity or bulge on
index correlation value. Period of the signal can be calculated the rotating bearing parts. Defects on inner race will generate
by computing time shift of the peak values found. Once the some vibration frequencies in axial (fet) and radial (fer)
dominant harmonic frequency have been found, the direction.
infestigation focus on the area around these frequencies. To
find more certain result, the appearance of a dominant fet = nZ(fr – fa) (6)

____________________________________________________________________________________________________________
vibrate in rotation frequency and its harmonics in Zone 1.

fer = fet ± fr (7) Sometimes, some high frequency vibrations appear in Zone 4.
Normal condition shown on top and underneath (NORMAL
Vibration frequencies coused by defects on the outer ring and STAGE I captions) in Figure 1. There are no significant
can be calculated using equation : damage in first level. STAGE II damage level happen when
natural frequencies of the bearing appear in the Zone 3. These
ff = nZfa (8) vibrations are caused by defect on the bearing ball or race.
In STAGE III, fundamental frequencies of the rotated
When one or more balls of the bearing have defect, bearing (BPFO = the outer race frequency, BPFI = the inner
vibrations will occur in some frequencies in the direction of circle frequency, BSF = ball spin frequency, and CF = cage
axial and radial. frequency) will begin to appear in zone 2. Vibration
harmonics associated with the frequency shaft rotation depend
fgt = 2nfc (9) on the amount and distribution of the defects. The amplitude
of the very high frequency vibration in zone 4 increase also. A
fgr = fgt ± fa (10) step before totally breakdown, the bearing reach STAGE IV.
Harmonics in Zone 2 will increase. Meanwhile, very high
frequenciy vibrations amplitude will decrease meanwhile.
IV. DAMAGED BEARING LEVELS
Rotated bearings are naturally vibrate with frequencies V. EXPERIMENT SETUP

that are highly dependent on the conditions. Figure.1 was
taken from STI Vibration Monitoring Inc. (www.stiweb.com ) Figure 2. shows the bearing test installation. An aixs is
that shows 4 frequency zones of the operated bearing. placed at the top with two supporter bearings. Removeable
tested bearing is put in one of shaft end. In the other end, a
pulley was mounted that connected to an electric motor via a
v-belt. The rotational speed of the axis was set by changing
the voltage supply using a voltage transformer. Rotation rpm
is the measured parameter. Sound signal is recorded using
smartphone microphone about 5 cm at the axial direction of
tested bearing.
Figure 2. Bearing Test Installation.
Two 6205 2RS type bearings are chosen carefully. The

first bearing is brand new, the other is damaged one. Both
have 9 balls inside and 00 contact angle. Inner race has 25mm
diameter, outer race has 52mm, and 15mm thickness. Figure 3.
show the 2 tested bearings.
Figure 1. Frequency zones of the operating bearing.
Figure 1. shows damage level that are indicated by

various vibration frequecies in the zones. Normal bearing only

____________________________________________________________________________________________________________
spectra curve. Comparison ranges from 1/100 to 1/100,000

times.
Figure 3. 6205 2RS Type Tested Bearings.
The two bearings rotation shown in Table 1. Recorded

sounds are stored in the smartphone memory in wav file, PCM
format, 16 bits, mono, 44.100 Hz sampling rate, and 15
seconds duration each.
Figure 4. Normal Bearing Sound Power Spectrum in 0 – 160 Hz
TABLE I
AXIS ROTATIONS
Damage frequencies are calculated using Equation 6. until
No RPM Hz Equation 10. The damage frequencies and the harmonics are
1 1465 24.42 found between 89.92 Hz and 680.80 Hz. To find these damage
2 1590 26.50 frequencies possibility, power spectra of damaged bearing
3 1670 27.83 sound signal was zoomed in range beetween 80 Hz and 700
4 1740 29.00 Hz. Figure 5. shows the spectra with highest power value for
1465, 1650, 1670 and 1740 rpm axis rotation speed respectely.
VI. RESULT AND DISCUSSION
Natural frequencies of tested bearings is shown in Table

2. The calculation is taken from Equation 1. until 5. Number 1
– 4 indicate the axis rotation, 1465, 1590, 1670 and 1740 rpm
respectely. The presences of these natural frequencies
(beetween 0 until 150 Hz) will be investigated in the
frequency spectra.
TABLE 2
BEARING NATURAL FREQUENCIES
No fr (Hz) fa (Hz) fb (Hz) fc (Hz) fd (Hz) fdi (Hz)

1 24.22 9.99 67.08 89.92 128.06 129.10
2 26.24 10.82 72.68 97.42 138.74 140.12
3 27.59 11.38 76.41 102.43 145.88 147.17
4 28.94 11.94 80.15 107.44 153.02 153.34
Figure 5. Damaged Bearing Sound Power Spectra 80 Hz – 700 Hz on
Axis Rotation (a) 1465 rpm (b) 1650 rpm (c) 1670 rpm (d) 1740 rpm.
There are 3 significant frequencies group in the ormal
bearing power sound spectrum shown in Figure 6. Noise
The dominant frequencies, have highest power, marked
frequencies around 30 Hz indicate that rotation frequencies
with back dots, are 445.5 Hz, 474.4 Hz, 508 Hz, and 547.1 Hz
are modulated 1.6 times. Frequencies between 23Hz – 30 Hz
for axis rotation in 1465, 1650, 1670, and 1740 rpms. To
are rotation frequencies. The frequencies around 100 Hz are
determine the type of damages, the dominant frequencies will
same for every rpm rotation. Twice modulated of 50 Hz ac
be compared with ff (vibration frequencies caused by defect
power supply frequenccy is the source of this frequency.
on the bearing outer race) from Equation 8. Basic frequency
Bearing natural frequencies, fa, fb, fc, fd and fdi, have a very
and the harmonics are shown in Table 3.
small power compared with the power of noises, rotations,
and power supply frequencies and can not be seen in the

____________________________________________________________________________________________________________
TABLE 3 ACKNOWLEDGMENT
OUTER RACE DEFECTS VIBRATION AND THE 5 HARMONICS.
rpm ff 2*ff 3*ff 4*ff 5*ff The autor would like to thank to Prof. I.N.G Wardana,
1,465 89.92 179.83 269.75 359.67 449.58 University of Brawijaya and The Ministry of Research,
1,590 97.42 194.83 292.25 389.66 487.08 Technology and Higher Education, Republic of Indonesia
1,670 102.43 204.86 307.28 409.71 512.14 for supporting this research work.
1,740 107.44 214.88 322.32 429.76 537.20
REFERENCES
th
Table 4. shows the comparison beetwen 5 harmonics
form Table 3. with dominant frequenies investigated. [1] S. McInerny and Y. Dai, “Basic vibration signal processing for bearing
fault detection,” IEEE Trans. Educ., vol. 46, no. 1, pp. 149–156,
feb2003. [Online]. Available: http://dx.doi.org/10.1109/te.2002.808234
TABLE 4 [2] B. SAMANTA and K. AL-BALUSHI, “ARTIFICIAL NEURAL
DEFECTS FREQUENCIES COMPARISON. NETWORK BASED FAULT DIAGNOSTICS OF ROLLING ELEMENT
BEARINGS USING TIME-DOMAIN FEATURES,” Mechanical Systems
No RPM F (Hz) F (Hz) Difference and Signal Processing, vol. 17, no. 2, pp. 317–328, mar 2003. [Online].
Dominant 5th harmonic Hz % Available: http://dx.doi.org/10.1006/mssp.2001.1462
(Tabel 3) [3] D. Mba, “Acoustic emissions and monitoring bearing health,” Tribology
1 1465 445.50 449.58 04.08 09.07 Transactions, vol. 46, no. 3, pp. 447–451, jan 2003. [Online]. Available:
http://dx.doi.org/10.1080/10402000308982649
2 1590 474.40 487.08 12.68 26.03 [4] J. Rosero, L. Romeral, E. Rosero, and J. Urresty, “Fault detection in
3 1670 508.00 512.14 04.14 08.08 dynamic conditions by means of discrete wavelet decomposition for
4 1740 574.10 537.20 09.90 18.42 PMSM running under bearing damage,” in 2009 Twenty-Fourth Annual
Average 15.40 IEEE Applied Power Electronics Conference and Exposition. Instituteof
Electrical & Electronics Engineers (IEEE), feb 2009. [Online].
Available: http://dx.doi.org/10.1109/apec.2009.4802777
[5] M. Aswin, I. Wardana, Y. S. Irawan, and H. Suyono, “Bearing damage
VII. CONCLUSION detection based on sound signal,” AMM, vol. 548-549, pp. 698–702,
apr 2014. [Online]. Available:
http://dx.doi.org/10.4028/www.scientific.net/amm.548-549.698
Frequency spectra of the sound signal can be used as a tool [6] R. S. Wilson, “A derivation of the normal distribution,” 2016. [Online].
to obtain damage or defect type of a bearing. In this research, Available:
http://www.sonoma.edu/users/w/wilsonst/Papers/Normal/default.html
the damage bearing have one ore more defect in the outer race [7] Minebea. Vibration by Forced Rotation.
in contact with the balls with 15.4% recission. http://www.eminebea.com/en/engineering_info/bearing/ballbearings/cat
-3/004-15.shtml

____________________________________________________________________________________________________________
Design of Adaptive Sliding Mode Control for DC

Motor Speed Control
Ramadhani Kurniawan Subroto
Faculty of Engineering
University of Brawijaya
ramasubroto@ub.ac.id
Abstract— DC motor has been extensively used in many drawback, the sliding mode control parameters with PID
industrial applications for their accurate, simple, and continuous sliding surface will be adjusted by adaptive mechanism.
control characteristics. Since it plays important roles in many Adaptive control methodology with control parameters
applications, its control performance must be improved. Many updated online is a promising approach. Moreover the
control methods have been employed for DC motor speed
stability of the proposed controller is proved using the
control. The proposed controller has to be robust against various
disturbances such that the speed response can achieve the Lyapunov analysis and the performance of the resulting
desired one. The proposed adaptive sliding mode control is control system is demonstrated by theoretical and simulation
developed with adaptation tuning approach which deals with investigations.
unknown but bounded system uncertainties. Both regulatory and
tracking performance is guaranteed. System robustness, as well II. SYSTEM DESIGN
as stability, is proven by using Lyapunov theory. The simulation
results indicate that speed control of DC motor is improved by A. Dynamic Model of DC Motor
proposed strategy. The control of the motor’s speed with great precision is
required in most of its applications. A desired speed can be
Keywords— adaptive sliding mode, DC motor, Lyapunov achieved when a desired shaft position is tracked. The
theory, speed control reference signal determines the desired speed and the control
is designed to ensure the difference between the reference
I. INTRODUCTION input and system’s output ultimately approaches to its
DC motor are comprehensively used in various industrial minimum value, preferably zero. In this research, voltage is
applications such as electrical equipment, computer applied as an input to control permanent magnet the DC motor.
peripherals, robotic manipulator, actuators, steel rolling mills, The dynamical model of a DC motor consists of electrical
electrical vehicles, paper machine, etc. Its applications spread part and mechanical part. The loop equation for the electrical
from low horse power to multi-mega watt due to its wide circuit is
power, torque, speed ranges, high efficiency, fast response, di
u (t ) = e a = L a + Ria + eb (1)
simple, and continuous control characteristics [1]. dt
In the prior works, DC motor speed control problem has The mechanical torque based on Newton’s law is
been worked out by control using various control methods, dω
such as conventional PID control [2] and Sliding Mode J + Bω + Tl = Te = Ki a (2)
control [1,3-6]. The conventional PID control is unable to dt
overcome some related issues, such as modelling uncertainties Where u = e a is the armature winding input voltage;
/ parameter variations, load torque variations, and also high eb = K bω is the back-electromotive-force (EMF) voltage; L
non-linearity. is the armature winding resistance; i a is the armature winding
Sliding Mode Control (SMC) is a nonlinear control
current; R is the armature winding resistance; J is the system
approach that drives the systems state trajectory onto a
specified sliding surface and maintains the trajectory on this moment of inertia; B is the system damping coefficient;
surface for the subsequent time. However in conventional K and K b are torque constant and the back-EMF constant,
SMC design, a priori knowledge of the bounds on system respectively; Tl is the load torque; and ω is the rotor angular
uncertainties has to be acquired. As a result, the controller speed.
based on the above knowledge tends to be overconservative,
which may induce poor tracking performance and undesirable By letting x1 = i a and x 2 = ω , the electromechanical
oscillations in control signal. Adaptive sliding mode had been
dynamics of the dc motor can be described by the following
introduced by [7]. However, this control strategy was only
state-space description:
proposed to reduce high frequency switching by adjusting the
gain switching online. Thus, it needs more improvement to R K 1
x!1 = − x1 − b x 2 + u (3)
enhance the controller performance. To overcome this L L L

____________________________________________________________________________________________________________
K B 1 the tracking error e defined by (7) satisfies limt →∞ e(t ) = 0

x! 2 = x1 − x 2 − Tl (4)
J J J if the control law is given by
The parameters of the motor used in this paper are shown in 1
Table 1[8].
u (t ) =
K m λ2
[ ]
(a1λ2 − a 2 )ω! + λ2ω + a2 λˆ1e +
By converting state space equation (3) and (4), we can also (10)
represent it into transfer function (5) ⎛ a ⎞
Γˆ ⎜⎜ 2 ⎟⎟ sgn (σ )
Ω(s )
=
Km
(5)
⎝ K m λ2 ⎠
2
U ( s) a 2 s + a1 s + 1 where sgn(·) represents the signum function and the control
where gains λ̂1 and Γ̂ are estimated online according to adaptive
K JR + BL JL rules
Km = ; a1 = ; a2 = t
BR + KK b BR + KK b BR + KK b !
λˆ1 (t ) = −γσ ∫ e(τ )dτ (11)
TABLE I 0
DC MOTOR PARAMETERS ! 1
J Inertia 42.6 e-6 Kg-m2 Γˆ (t ) = σ (12)
L Inductance 170e-3 H
α
R Resistance 4.67 Ω Proof: To evaluate stability of closed loop system, an
B Damping coefficient 47.3 e-6 N-m-sec/rad estimation error for the desired control parameter Γd is
K Torque Constant 14.7 e-3 N-m/A
introduced below
Kb Back-EMF Constant 14.7 e-3 V-sec/rad ~
Γ(t ) = Γˆ (t ) − Γd (13)
In addition, a positive definite Lyapunov function candidate
is defined as
B. Sliding Mode Controller
1 1 ~
The basic concept in designing the Sliding Mode controller V = σ 2 + αΓ 2 (14)
2 2
is to define sliding surface and control signal. One should
Differentiating (14) respect to t gives
select the sliding surface such that the trajectory can be forced ~ ~!
and maintained on its equilibrium point. In this research, V! = σσ! + αΓΓ (15)
sliding surface of the controller is defined as: Taking the time derivative of the sliding variable defined in
τ
de(t ) (9) results in
σ = β1e + β 0 ∫ e(τ )dτ + (6)
dt ! t ⎛a 1 K ⎞
0 σ! = −ω! + λˆ1 ∫ e(τ )dτ + λˆ1e + λ2 ⎜⎜ 1 ω! + ω − m u ⎟⎟ (16)
where β1 , β 0 are sliding mode control parameters and 0 ⎝ a2 a2 a2 ⎠
selected by user. An error between actual rotor angular speed where the time variable is omitted for the clarity of
and its reference is defined as representation.
e = ωr − ω (7) Substituting the control law (10) into (16) gives
Since ω r is kept constant, thus its derivative ω! r = ω
!!r = 0 ! t
σ! = λˆ1 ∫ e(τ )dτ − Γˆ sgn (σ ) (17)
Therefore, by using Lyapunov theory for stability, it is 0
obtained the control signal as follows: ~! !
ˆ and substituting (11) – (12) and
Then, taking account Γ = Γ
⎛ a − β1 a 2 ⎞ ω β a a (16) into (15), a necessary calculation produces
u = ⎜⎜ 1 ⎟ω! +
⎟ + 0 2 e + K n 2 sgn (σ ) (8)
2
⎝ Km ⎠ Km Km Km ⎛t ⎞
V! = −γσ 2 ⎜⎜ ∫ e(τ )dτ ⎟⎟ − Γˆ d σ (18)
C. Adaptive Sliding Mode Controller ⎝0 ⎠
The SMC developed in the preceding section exhibits In order to make the closed loop system asymptotically
system stability and robustness. However, if the control stable, adaptation gain γ and desired switching gain Γd have
formulas (8) are used in practice, the upper bound of Tl is to be positive definite. One can deduce that the switching
unknown, and therefore, K n is often not easy to be found. variable σ → 0 as t → ∞ . According to the definition of σ
in (9), we can conclude that the tracking error satisfies
This section purpose is to adaptively tune the controller gain t
without knowledge of upper bound on the disturbance, which limt →∞ e(t ) = 0 and lim t →∞ ∫ e(τ )dτ = 0 and that ω → ω r as
is load torque in this case. 0
Theorem 1: For the system (5) with sliding surface given t → ∞ . Therefore, the proposed adaptive sliding mode
below controller guarantees a zero steady state tracking error.
t
de(t )
σ (t ) = e(t ) + λˆ1 (t ) ∫ e(τ )dτ + λ 2 (9)
0 dt

____________________________________________________________________________________________________________
III. RESULT AND ANALYSIS
The simulation is set up to illustrate the effect of proposed

control method for DC motor speed control, as explained in 350
the following sections. In this simulation scenario, DC motor
is simulated using MATLAB/SIMULINK under fully 300
controlled environments. The motor equations in (3) and (4)
are used as the main model, and it controlled by the Adaptive 250
Sliding Mode controller in (10) and simultaneously compared
with Sliding Mode controller in (8) under various cases. The 200
ω (rad/s)
system block diagram is illustrated in Fig. 1.
According to (18), in order to make the closed loop system 150
asymptotically stable, the parameters of adaptation gain have
to be selected to be positive definite, which means the range 100
values γ , λ , α > 0 . Therefore, by using trial and error method,
reference
the parameters of adaptive sliding mode are selected as 50
proposed method
follows γ = 5 × 10 −4 ; λ2 = 5 × 10 −2 ; α = 10 −6 , since those Sliding Mode Control
0
parameters are positive definite. The cases for system testing 0 0.5 1 1.5 2 2.5 3
can be summarized as follows: Time (seconds)
Fig. 3 Speed response for case 2
ωref Adaptive Sliding u ω
Case 2: Reference speed is subjected to change from 200
+_
Mode Controller
DC motor rad/s to 100 rad/s and then 300 rad/s with constant load torque
0.1Nm.
Speed response of DC motor for case 2 is shown in Fig. 3.
Fig. 1 Block diagram of system According to Fig. 3, one can conclude that both controllers
can follow the desired reference for tracking problem.
Case 1: Reference speed is kept constant and load torque
has been given a step change from 0.1 Nm to 0.6 Nm.
IV. CONCLUSIONS
According to simulation result of DC motor speed control,
250 performance of Adaptive Sliding Mode controller has shown
better than Sliding Mode controller in case of overcoming
load torque variations. Both controllers also can follow the
200
desired reference for tracking problem.
150
206
REFERENCES
204
[1] G. Murtaza and A. I. Bhatti, “Control of DC Motors using Sliding

ω (rad/s)
Mode”, Proceedings of 2012 9th International Bhurban Conference on

202
200
Applied Sciences and Technology (IBCAST), pp.37 – 42, Islamabad,

100 198
196
Pakistan, 2012
194
[2] G. Huang and S.Lee, “PC-based PID Speed Control in DC Motor”, in
192
International Conference on Audio, Language, and Image Processing,
50 0.98 1 1.02 1.04 1.06 1.08 1.1 1.12 1.14
reference
pp.400 – 407, Shanghai, China, July. 2008
[3] S.V. Ambesange, S.Y. Kamble, D.S. More, “Application of Sliding
proposed method
Mode Control for the Speed Control of DC Motor Drives”, in 2013
Sliding Mode Control
0 IEEE International Conference on Control Applications (CCA), pp.832
0 0.5 1 1.5 2 2.5 3 – 836, Hyderabad, India, August 28 – 30, 2013
Time (seconds) [4] Ch.U. Maheswararao, Y. S. K. Babu, and K. Amaresh, “Sliding Mode
Speed Control of a DC Motor”, in 2011 International Conference on
Fig. 2 Speed response for case 1 Communication Systems and Network Technologies, pp. 387 – 391,
Katra, Jammu, 3 – 5 June 2011
[5] N.Afrasiabi and M.H.Yazdi, “Sliding Mode Control for DC Motor
Speed Control”, Global Journal of Science, Engineering, and
The proposed controller can achieve 0.12 seconds to return
Technology, issue 11, pp.45 – 50, 2013
to its steady state and results steady state error as 0%, while [6] P.Ghalithath and S.S. Sankeswari, “Speed Control of DC Motor using
sliding mode controller cannot return to its steady state and Sliding Mode Control Approach”, IOSR Journal of Electrical and
results steady state error as 2.75%. Based on Fig. 2, one can Electronics Engineering, vol. 10, pp.17 – 22, July – Aug 2015.
[7] Y.J. Huang, T.C.Kuo, and S.H. Chang, “Adaptive Sliding Mode
conclude that the proposed method can follow the desired Control for Nonlinear System with Uncertain Parameters”, IEEE
reference though the presence of the disturbance, which is Transactions on Systems, Man, and Cybernetics Part B, vol. 38, no.2,
load torque variations. pp.534 – 539, April, 2008.

____________________________________________________________________________________________________________
[8] M.Y.Chow and Y.Tipsuwan, “Gain Adaptation of Networked DC Industrial Electronics, vol.50, pp.936 – 943, Oct. 2003
Motor Controllers Based on QoS Variations”, IEEE Trans. on

____________________________________________________________________________________________________________
A Study of Electric Field of Gaussian Silicon

P-N Junction
A.A.N.G. Sapteka#1, A.A.N.M. Narottama#2, I N. Sukarma#3, I G.P.M.E. Putra#4, M.L. Saragi*5
#
Electrical Engineering, Politeknik Negeri Bali
Denpasar, Indonesia
1
sapteka@pnb.ac.id
2
narottama@pnb.ac.id
3
sukarma@pnb.ac.id
4
mastawan@pnb.ac.id
*
Faculty of Information Technology & Computer, Universitas Mpu Tantular
Jakarta, Indonesia
5
mlsaragi@mputantular.id
Abstract— The quantity of electric field needs to be known in the doping concentrations of 1015 to 1017 per cm3 and the applied
design of P-N diode. In this paper we simulate, analyze and voltage is chosen from 0V to -3V. The applied voltage is
model the electric field of Gaussian Silicon P-N junction under connected to anode, meanwhile cathode is connected to the
reverse bias regime. For this purpose, we simulate P-N junction ground. We analyze the relationship between applied voltage,
with some configurations of donor and acceptor concentrations
dopant concentration of Boron (NA) as acceptor and
using Matlab and Comsol software. We analyze the value of
electric field of Gaussian doping profile at P-N junction and Phosphorus (ND) as donor and electric field for Gaussian
compare it with abrupt doping profile. Based on the simulation doping profile. In this research, we simulate P-N Junction as
result and analysis, we modify the equation of electric field for shown in Fig.1 and dopant concentration as shown in Fig. 2.
3 um
Gaussian Silicon PN junction from abrupt doping profile model.
Keywords— P-N Junction; Gauss; Electric Field; Diode; Reverse

2 um P-type
Bias.
I. INTRODUCTION
7 um
Besides abrupt and linear doping profile, research about

P-N junction is also performed on Gaussian doping profile. In N-type
1969, R.M. Allen studied about the summation of Phosphorus

doping to generate Gaussian doping profile by ion
implantation techniques [1]. In 1972, R.M. Warner Jr stated a
model of breakdown voltage of Gaussian Silicon diffused 5 um
junction [2]. In 2001, J. He et al. reported about effect due to Fig. 1 Geometric of P-N Junction
the lateral radius of main junction and width of the ring
junction on voltage and edge peak field profiles for planar
x1017 (cm-3)
junction with a single floating electrical field. The surface
concentration of the anode with a Gaussian doping profile [3].
In 2013, J. Huaizhou et al. proposed a Gauss simulation of
Phosphor coated light-emitting diode [4]. Our research group
also investigated about electric potential of Gaussian diode [5].
This motivated us to strengthen the knowledge about electric
field of P-N device. The purpose of this paper is to achieve
more accurate value of electric field of Gaussian Silicon P-N
junction.
II. METHODOLOGY
In this paper, we simulate, analyze and model the electric
field of Gaussian Silicon P-N junction under reverse condition.
We simulate the P-N junction with Boron (P-type) doping
concentration of 1016 to 1019 per cm3 and Phosphorus (N-type) Fig. 2 Concentration of P-N Junction

____________________________________________________________________________________________________________
After we obtain the electric field at the entire part of diode

we process the value of electric field at P-N junction as
shown in Fig. 3 and calculate its average value using Comsol.
We compare the average value of electric field of Gaussian
doping P-N junction with value of electric field of abrupt P-N
junction taken from calculation of Equation (1)-(3).
Fig. 5 Electric Field at ND of 1016 cm-3
Fig. 3 Electric Field at P-N Junction
III. RESULTS
We simulate the average electric field for some Fig. 6 Electric Field at ND of 1017 cm-3
configurations of dopant concentration, i.e.,
1. Donor concentration (ND) of 1015 cm-3 and acceptor For analyze purpose, we plot the average electric value of
concentration (NA) of 1016 to 1017 cm-3 as shown in Gaussian Silicon P-N junction for applied voltage (VA) from
Fig. 4. 0V to -3V as show in Fig. 7-10.
2. Donor concentration (ND) of 1016 cm-3 and acceptor
concentration (NA) of 1017 to 1018 cm-3 as shown in
Fig. 5.
3. Donor concentration (ND) of 1017 cm-3 and acceptor
concentration (NA) of 1018 to 1019 cm-3 as shown in
Fig. 6.
Fig. 7 Electric Field at VA = 0V
Fig. 4 Electric Field at ND of 1015 cm-3

____________________________________________________________________________________________________________
IV. DISCUSSION
As shown Fig. 4-6, it can be concluded that if the absolute

value of the applied voltage is greater, then the average
electric field at the junction will be even greater. Increasing
reverse voltage will widening the depletion region and
increasing the electric field. From Fig. 7-10, it also indicated
that higher concentration of dopant would cause higher
electric field. By comparing the simulation results with the
calculation of electric field on the abrupt junction as stated in
[6-7] using Equation (1) – (3), we find that Equation (2) need
to be remodelling by a coefficient to achieve accurate value of
electric field in Gaussian doping profile.
eVbi = 0.834eV (1)

Fig. 8 Electric Field at VA = -1V
(2)
(3)
where
e is magnitude of electron charge
Vbi is built-in voltage
ND is donor concentration
NA is acceptor concentration
is product of relative dielectric constant and free space
permittivity
Wp is depletion region at P-side of P-N junction
E is electric field.
Table I-IV show the differences between Gaussian and

Fig. 9 Electric Field at VA = -2V abrupt junction at VA = 0V to -3V.
TABLE I
ELECTRIC FIELD OF GAUSSIAN AND ABRUPT JUNCTION AT VA=0V
VA=0V
NA ND
(cm-3) (cm-3) Gauss Abrupt
Δ(V/m) Δ%
(V/m) (V/m)
1×1016 1×1015 -817138 -1518700 701530 46.19
16 15
5×10 1×10 -966540 -1577100 610560 38.71
16 15
10×10 1×10 -1011291 -1584900 573590 36.19
17 16
1×10 1×10 -2003376 -4802400 2799100 58.28
17 16
5×10 1×10 -2412617 -4987200 2574600 51.62
17 16
10×10 1×10 -2547755 -5011800 2464100 49.16
18 17
1×10 1×10 -4724115 -15187000 10463000 68.89
18 17
5×10 1×10 -5721840 -15771000 10049000 63.72
Fig. 10 Electric Field at VA = -3V
10×1018 1×1017 -6064637 -15849000 9784200 61.73

____________________________________________________________________________________________________________
Therefore, we proposed a modified equation for E in (3)

TABLE II
ELECTRIC FIELD OF GAUSSIAN AND ABRUPT JUNCTION AT VA=-1V as stated in (4), where c is a coefficient to adjust the value
electric field for Gaussian doping profile as stated in Table V
VA=-1V after quantitative analysis.
NA ND
Δ(V/m) Δ%
(V/m) (V/m)
16 15
1×10 1×10 -1588941 -2252100 663110 29.45
16 15
(4)
5×10 1×10 -1832882 -2338700 505820 21.63
TABLE V
10×1016 1×1015 -1901669 -2350300 448580 19.09 COEFFICIENT VALUE OF EQUATION (4)
1×1017 1×1016 -3723289 -7121600 3398300 47.72
VA (Volt) ND (cm-3) NA (cm-3) c
5×1017 1×1016 -4408125 -7395600 2987500 40.40 15 16 16
1x10 1x10 -10x10 0.5963
10×1017 1×1016 -4625002 -7432100 2807100 37.77
0 1x1016 1x1017-10x1017 0.4698
1×1018 1×1017 -8419683 -22521000 14101000 62.61
1x10 17 18
1x10 -10x10 18
0.3522
5×1018 1×1017 -10100000 -23387000 13287000 56.81 1x10 15 16
1x10 -10x10 16
0.7661
10×1018 1×1017 -10600000 -23503000 12903000 54.90 -1 1x1016 1x1017-10x1017 0.5804
17 18 18
1x10 1x10 -10x10 0.4189
TABLE III 15 16 16
1x10 1x10 -10x10 0.8393
ELECTRIC FIELD OF GAUSSIAN AND ABRUPT JUNCTION AT VA=-2V
-2 1x1016 1x1017-10x1017 0.6325
VA=-2V 17 18 18
NA ND 1x10 1x10 -10x10 0.4548
Δ(V/m) Δ% 15 16 16
(V/m) (V/m) 1x10 1x10 -10x10 0.8866
1×1016 1×1015 -2173742 -2799500 625750 22.35 -3 1x10 16 17
1x10 -10x10 17
0.6667
5×1016 1×1015 -2494292 -2907200 412910 14.20 1x1017 1x1018-10x1018 0.4806
16 15
10×10 1×10 -2580904 -2921600 340650 11.66
17 16
V. CONCLUSIONS
1×10 1×10 -5074234 -8852800 3778500 42.68
The simulation of electric field of Gaussian Silicon PN
5×1017 1×1016 -5965555 -9193400 3227800 35.11
junction shows different value compare with abrupt junction.
10×10 17
1×1016
-6240543 -9238800 2998200 32.45 Therefore we modify the equation of electric field to find
1×10 18
1×1017
-11400000 -27995000 16595000 59.28 more accurate value of electric field as stated in (4).
18 17
5×10 1×10 -13600000 -29072000 15472000 53.22 ACKNOWLEDGMENT
10×10 18
1×1017
-14300000 -29216000 14916000 51.05 The research of this paper was supported by Pusat
Komputer FTUI in developing Matlab computer programming.
TABLE IV REFERENCES
ELECTRIC FIELD OF GAUSSIAN AND ABRUPT JUNCTION AT VA=-3V [1] R.M. Allen, ―Summation of gaussians for ion-implantation profile
control‖, Electronics Letters, vol. 5, no. 6, pp. 111-112, 1969.
NA ND
VA=-3V [2] R.M. Warner Jr, ―Avalanche breakdown in silicon diffused junctions‖,
(cm-3) (cm-3) Gauss Abrupt Solid-State Electronics, vol.15, issue 12, pp.1303-1318, 1972.
Δ(V/m) Δ% [3] J. He, R. Huang, X. Zhang, Y.Y. Wang, and X.B. Chen, ―Analytical
(V/m) (V/m)
16 15 model of three-dimensional effect on voltage and edge peak field
1×10 1×10 -2674598 -3256200 581560 17.86 distributions and optimal space for planar junction with a single field
5×10 16
1×1015
-3065010 -3381400 316430 9.36 limited ring‖, Solid State Electronics, vol. 45, pp. 79-85, 2001.
[4] J. Huaizhou, S. Jin, K. Yuan, S. Cen, ―Modification to a single gauss
16 15
10×10 1×10 -3166899 -3398100 231240 6.80 simlation of a phosphor coated light-emitting diode‖, Optical
Engineering, vol. 52, issue 8, 2013.
1×1017 1×1016 -6245078 -10297000 4051800 39.35 [5] A.A.N.G. Sapteka, T. Abuzairi, and D, Hartanto, ―Effect of gauss
5×1017 1×1016 -7309201 -10693000 3383800 31.64 doping profile on electric potential of p-n diode‖, in International
Conference on Quality in Research (QiR), 2013, pp. 226-231.
10×1017 1×1016 -7631364 -10746000 3114500 28.98 [6] U. K. Mishra and J. Singh, Semiconductor Device Physics and Design,
18 17
Springer, Netherland, pp.146-215, 2008.
1×10 1×10 -14000000 -32562000 18562000 57.01 [7] S.M. Sze and K.K. Ng, ―Physics of Semiconductor Devices‖, John
5×10 18
1×1017
-16700000 -33814000 17114000 50.61 Wiley & Sons, Third edition, pp. 79-133, 2007.
10×1018 1×1017 -17600000 -33981000 16381000 48.21

____________________________________________________________________________________________________________
Predict Drug Addict Using Linear Discriminant

Analysis Based On Eyes and Faces Picture
Bagus Priambodo
University of Mercu Buana
Jl Meruya Selatan, Indonesia
1
bagus.priambodo@mercubuana.ac.id
Abstract— The drug problem in Indonesia has become a complex

issue, especially when it has entered the world of children. Data
Badan Narkotika Nasional in 2014 says there are 4.9 million
abusers of drugs (narcotics, alcohol, psychotropic or other
addictive substances) in Indonesia where most are in the
productive age of 15 to 49 years, and about 20-30% of this
Fig. 1 Evolution of drugs addicts faces [33]
amount was included category of children aged under 18 years.
Drugs inspection currently are not effective because only a few Information technology can be used to identify drug
students who examined his urine taken at random. Prediction addicts. To help parents and teachers predict their student that
positive junkie are needed, that only student that indicated drug
has problem with drugs. This paper will try to predict student
addict are selected to have urine test. Use of information
technology in the field of computer vision can be used for that indicated drug addicts using Linear discriminant analysis
selecting possible drug addict. Linear discriminant analysis are based on eyes and faces images.
used to predict drugs addict face based on eye and faces images. In what follows, we will discuss some related work in
The results of application based on the training data can predict section 3. In section 4, 5 we will give an overview to face
or forecast drugs addict but many test result show huge distance detection and face recognition. In section 6, we will explain
value it means the result is not reliable enough. Testing based on about the methodology of research, in section 7, we will
face has better result than based on eyes. The result from explain about the experiment and result. and finally in section
application can not be used to justified someone junkie or not, 8, we will conclude the paper.
but it can be a reference to give recommendation that someone
should have urine test or not. II. AIMS AND SCOPE
The research has the following aims:
Keywords— linear discriminat analysis; drug addict detection;  Predict drugs addict using linear discriminant
computer vision;
analysis based on eyes image.
I. INTRODUCTION  Predict drugs addict using linear discriminant
The drug problem in Indonesia has become a complex analysis based on faces .
issue, especially when it has entered the world of children.
Data BNN in 2014 says there are 4.9 million abusers [10] of Scope of this paper are as follow : training images are
drugs (narcotics, alcohol, psychotropic or other addictive limited only use thirty four image of evolution drugs addicts
substances) in Indonesia, mostly located in the productive face (junkie faces) taken from [25][26][27][28][29][33].
ages of 15 to 49 years, and about 20-30% of that amount still Thirty three images used as training data, and one image are
belongs to the category of children aged under 18 years. The used as data test. Also we add more faces picture as data test.
latest campaign of anti-drug [33] showed physical changes III. RELATED WORK
experienced by drug addict. Drug addicts increasingly looked
thin, due to malnutrition because the drugs suppress one's Face detection (face detection) is one of the crucial early
appetite and body begin to contract because of the lack of stage prior to the process of face recognition (face
proper nutrition. Abuse of drugs use constantly causing recognition). Face detection methods using Adaboost
terrible damage to the skin lesions caused by hallucinations of classifier with haar, in android device[15]. other detection
drugs abuse, with scratching their skin. Acne appears or method are local binary pattern with Histogram of Gradient
worsens. Obssesive skin-picking often couses meth users [1]. However, this method is less accurate for people who do
faces to be covered in small sores and scarring. More detail not pose fixed. To resolve this tree structure model was
can be seen in figure 1. developed to detect faces in people who do not pose fixed or
movable [22].
Face recognition, now has been applied to many
applications in the field. Some of the important aspects that

____________________________________________________________________________________________________________
affect the image of a human face including the lighting, facial features three-dimensional face [2], extract features with SVM
expressions and changes in attributes like moustache, beard method [9], with the approach of neural network [7], by
and glasses[20]. The usual method for face recognition among recognizing the combination of (clothes, hair, accessories face
them is to use eigen faces and fisher faces [6] [13]. ), and the method of local binary pattern (LBP) [16].
Recognizing gender through face recognition is successful Fisherfaces, maximizes the ratio of between-classes to
with various methods [11] also using linear disciminant within-classes scatter, instead of maximizing the overall
analysis, with backpropagation neural network [7], scatter. The idea is simple: same classes should cluster tightly
recognizing the hair and clothes [8], with support vector together, while different classes are as far away as possible
machine [9], and with local binary pattern (LBP) [16], from each other in the lower-dimensional representation. This
Recognize the success of gender through face recognition is was also recognized by Belhumeur, Hespanha and Kriegman
very impressive. It should continue with other recognize and so they applied a Discriminant Analysis to face
research, like research in this paper, recognize or predicts recognition in [3].
drugs addict.
VI. RESEARCH METHODOLOGY
IV. FACE DETECTION In this paper we do this steps as follows :
Face detection process is one of the most important parts of 1. Collecting faces drug addicts,
the face recognition system. Viola-Jones algorithm [19], [20] In this step, we gather thirty four evolution drugs addict faces
is the most widely used algorithms to detect faces. Face from [25][26][27][28][29][33].
detection process performed by classifying an image after a
classifier is formed of training data. Training data used by this
algorithm 5000 amounted face images and non-face image
9400 resulting in a system accuracy of 95% with a false
positive reading of 1: 14 084.
Image classification is done based on the value of a feature.
Using the feature is done because processing feature is faster
than the image processing by pixel. There are 3 types of
features based on the number of rectangles contained in it, as
Fig. 4 Drugs addict faces
can be seen in the figure below:
Fig. 2 Image classification based on square
To simplify the process of calculating the value of features,

Viola-Jones algorithm uses a media form an integral image.
Fig. 5 Person face before becoming drugs addicts
2. Setting up the training data (extraction fisherfaces)

At this step, data training of drug addicts face are extracted,
the extraction will be stored in the form of xml
Fig. 3 Integral image, left input image and right integral image. 3. Create application
Application are build based on training data
Integral image is an image that the value of each pixel is an 4. Preparing face data to test applications
accumulation of the pixel values above and left. For example, At this step the data will be prepared to test application. Face
pixels (a, b) have the accumulated value for all pixels (x, y) to trial is two faces, first is collection of drug addicts and
where x and y ≤ a ≤ b. second is face before them become drug addict. Also included
in test data, indonesian artist that arrested because of drugs.
V. FACE RECOGNITION
Facial recognition now able to recognize the face of a male
or female face automatically [6] [13]. This field of technology
is also able to recognize facial expressions [12] [14],
recognize facial attributes [5] also with video [17]. The
introduction of gender with facial recognition was also
successfully carried out [11], with the extraction of image
Fig. 6 Evolution drugs addict faces for testing, we call him xyz [28][29]

____________________________________________________________________________________________________________
Fig. 7 Evolution drugs addict faces for testing, angelina jolie health, and
angelina jolie act as drugs addict. [32] Fig. 12 Test result from application, test xyz faces health, application show
person is healthy
Fig. 8 Ahmad Albar, Indonesian musician arrested drugs (left) [23], Fariz
RM, Indonesian musician arrested drugs (right) [24]
Fig. 13 Test result from application, test xyz eyes indicate drugs addict,
application show person xyz are suspected drugs.
TABLE I
EXPERIMENT RESULT
No Based On Eyes
Name Result Distance
1 Mr Xyz health There is no indication 115.976
2 Mr Xyz drugs addict Suspected drugs 199.978
addict
3 Angelina jolie health There is no indication 3.97162
Fig. 9 Gatot Brajamusti (a), Indonesian artist arrested drugs [31], Gatot
4 Angelina jolie act as Suspected drugs 137.786
Brajamusti (b), Indonesian artist arrested drugs [30]
drug addict addict
5 Ahmad Albar musician Suspected drugs 104.587
caught drugs addict
6 Fariz RM, musician There is no indication 38.1394
cought drugs
7 Gatot Brajamusti (a), There is no indication 127.569
artist arrested drugs
8 Gatot Brajamusti (b), There is no indication 14.8016
artist arrested drugs
9 Bagus Priambodo, There is no indication 4.03713
Fig. 10 Bagus Priambodo, researcher researcher
TABLE II
5. Test applications EXPERIMENT RESULT
Application are test, first test will try recognize based on eye, No Based On Faces
and second test will try recognize based on face images Name Result Distance
1 Mr Xyz health There is no indication 134.088
VII. EXPERIMENT AND RESULT 2 Mr Xyz drugs addict Suspected drugs 0.0446251
addict
Application are test twice, first test will try recognize based
on eye images, and second test will try recognize based on 3 Angelina jolie health There is no indication 249.677
face images 4 Angelina jolie act as Suspected drugs 352.484
drug addict addict
5 Ahmad Albar musician Suspected drugs 126.669
caught drugs addict
6 Fariz RM, musician There is no indication 1.72563
cought drugs
7 Gatot Brajamusti (a) Suspected drugs 67.8692
addict
Fig. 11 Fisher faces image based on eyes and based on face, result from 8 Gatot Brajamusti (b) Suspected drugs 288.025
training data addict
9 Bagus Priambodo, There is no indication 80.6179
researcher

____________________________________________________________________________________________________________
[11] Ng, C. B., Tay, Y. H., & Goi, B. (n.d.). Recognizing Human Gender in
Computer Vision : A Survey.
Test results can be analyzed that application can detect drugs
[12] Pantic, M., & Rothkrantz, L. J. M. (2000). Automatic analysis of facial
addicted but less accurately, there are differences between the expressions: the state of the art. Pattern Analysis and Machine
results of test results by the eyes and face. The accuracy of the Intelligence, IEEE Transactions on, 22(12), 1424–1445.
test results can be seen based on the distance resulting from [13] Park, U., Klare, B., & Choi, H. (n.d.). Automatic Face Recognition  :
State of the Art Anil K . Jain.
the prediction process. If the distance < 100 can be concluded [14] Pfister, T., Li, X., Zhao, G., & Pietikainen, M. (2011). Differentiating
that the probability of the prediction is high. But also we spontaneous from posed facial expressions within a generic facial
cannot justified that if distance value > 100, results are not expression recognition framework. Proceedings of the IEEE
accurate. We need urine test to justified someone is drugs International Conference on Computer Vision, 868–875.
[15] Priambodo, Bagus. Deteksi Wajah Real Time Menggunakan OpenCV
addicted untuk Android. Sinaptika. 2014
[16] Shan, C. (2011). Learning Local Binary Patterns for Gender
VIII. CONCLUSIONS Classification on Real-World Face Images
Based on the experiment results, linear discriminant [17] Shreve, M., Godavarthy, S., Goldgof, D., & Sarkar, S. (2011). Macro-
and micro-expression spotting in long videos using spatio-temporal
analysis show it can predict but testing result show big strain. 2011 IEEE International Conference on Automatic Face and
distance value that means result not reliable enough. Testing Gesture Recognition and Workshops, FG 2011, 51–56.
based on face has better result than based on eyes. Of course [18] Syarizka, D. (2015). KECANDUAN NARKOBA: Kenali Gejala
the result can not be used to justified someone is addicted to Narkoba Pada Anak , 1–9.
[19] Viola, P., & Jones, M. (2004). Robust real-time face detection.
drugs, but it can be a reference to give recommendation that International Journal of Computer Vision, 57(2), 137–154.
someone should have urine test or not. Since many of the test [20] Viola, P., Jones, M. J., “Rapid Object Detection Using A Boosted
results show the value of distance are more than 100, test Cascade of Simple Features”, IEEE Conference on Computer Vision
result is not reliable enough. In the future we will add more and Pattern Recognition, Jauai, Hawaii, 2001.
[21] Yang, A. Y., Zhou, Z., Member, S., & Ganesh, A. (2012). Fast 1
training data face greater one hundred face. One person should -Minimization Algorithms For Robust Face Recognition.
has minimum five healthy face, and five data faces after arXiv:1007.3753v4 [cs.CV], 1–13.
becoming a drug addict. [22] Zhu, X., & Ramanan, D. (2012). Face detection, pose estimation, and
landmark estimation in the wild. Proc. Int. Conf. on Computer Vision
ACKNOWLEDGMENT and Pattern Recognition (CVPR).
[23] Artis Indonesia yang Pernah Terjerat Narkoba (2015) Available:
Support for this research has been provided by Kementerian http://hugsdontdrugs.blogspot.co.id/2015/06/artis-indonesia-yang-
Riset Teknologi Dan Pendidikan Tinggi Republik Indonesia, pernah-terjerat.html [Accessed: 19- Sept- 2016]
[24] Dua Kali Dipenjara, Ini Pelajaran yang Didapat Fariz RM (2015)
Research Dosen Pemula Grant. We are grateful for this
Available: http://showbiz.liputan6.com/read/2299175/dua-kali-
support. dipenjara-ini-pelajaran-yang-didapat-fariz-rm [Accessed: 19- Sept-
2016]
REFERENCES [25] Face it … the drugs don’t work (2011), Available:
[1] Ahmad, F., Najam, A., & Ahmed, Z. (2012). Image-based Face https://www.thesun.co.uk/archives/news/392819/face-it-the-drugs-
Detection and Recognition : “ State of the Art .” International Journal dont-work/ [Accessed: 19- Sept- 2016]
of Computer Science Issues (IJCSI), 3–6. [26] Faces of Meth (2006) Available: http://www.facesofmeth.us/main.htm/
[2] Ballihi, L., Amor, B. Ben, Daoudi, M., & Member, S. (2012). Boosting [Accessed: 19- Sept- 2016]
3-D-Geometric Features for Ef fi cient Face Recognition and Gender [27] Faces of Meth, 10 years later: Anti-drug campaign endures, as does
Classi fi cation, 7(6), 1766–1779. effect on addicts' lives.
[3] Belhumeur, P. N., Hespanha, J. P., & Kriegman, D. J. (1997). Available:http://www.oregonlive.com/portland/index.ssf/2014/12/faces
Eigenfaces vs.~{Fisherfaces}: Recognition using class specific linear _of_meth_10_years_later_a.html [Accessed: 19- Sept- 2016]
projection. Pami, 19(7), 711–720. [28] Horrifying And Gross. The Before And After Photos Of Drug
[4] Badan Narkotika Nasional (2015). Kenali ciri-ciri anak pengguna UsersAvailable: http://boredombash.com/before-and-after-drugs-
narkoba, 2014–2015. SIGKDD …, 1082–1090. mugshots/ [Accessed: 19- Sept- 2016]
[5] Cherniavsky, N., Laptev, I., Sivic, J., & Zisserman, A. (2012). Semi- [29] How methamphetamine destroys your face and physical appearance/
supervised learning of facial attributes in video. Lecture Notes in Available: http://abc7ny.com/society/photos-you-wont-want-to-try-
Computer Science (including Subseries Lecture Notes in Artificial meth-after-you-see-what-it-does-to-your-face/192488/#gallery-35
Intelligence and Lecture Notes in Bioinformatics), 6553 LNCS(PART [Accessed: 19- Sept- 2016]
1), [30] Polisi Indikasikan sejumlah Artis Kerap Pesta Narkoba di Rumah Gatot
[6] Jafri, R., & Arabnia, H. R. (2009). A Survey of Face Recognition Brajamusti (2016)Available: http://kriminalitas.com/polisi-indikasikan-
Techniques. Journal of Information Processing Systems, 5(2), 41–68. sejumlah-artis-kerap-pesta-narkoba-di-rumah-gatot-brajamusti/
[7] Jaswante, A., Khan, A. U., & Gour, B. (2014). Back Propagation [Accessed: 19- Sept- 2016]
Neural Network Based Gender Classification Technique Based on [31] Selain Narkoba, Aa Gatot Brajamusti Punya Satwa Langka Ilegal?
Facial Features, 14(11), 91–96. IEEE International Conference on Web (2016)Available: http://www.kapanlagi.com/showbiz/selebriti/selain-
Services, 297–304. narkoba-aa-gatot-brajamusti-punya-satwa-langka-ilegal-b14637.html
[8] Li, B., Lian, X. C., & Lu, B. L. (2012). Gender classification by [Accessed: 19- Sept- 2016]
combining clothing, hair and facial component classifiers. [32] Terungkap kisah angelina jolie sebagai pecandu narkoba Available:
Neurocomputing, 76(1), 18–27. http://jadiberita.com/34976/terungkap-kisah-angelina-jolie-sebagai-
[9] Moghaddam, B., & Yang, M.-H. (2000). Gender Classification with pecandu-narkoba.html [Accessed: 19- Sept- 2016]
Support Vector Machines. Proceedings of the 4th IEEE Int’ll Conf. on [33] The Horrors of Methamphetamines, Available:
Face and Gesture Recognition. http://www.rehabs.com/explore/meth-before-and-after-
[10] Narkoba, P. (2009). “ Upaya Penyelamatan Generasi Muda Melalui drugs/infographic.php#.V-CYZ6J97VM [Accessed: 19- Sept- 2016]
Penyuluhan Pengetahuan Bahaya dan Cara Penanggulangan
Penyalahgunaan Narkoba ,” (September).

____________________________________________________________________________________________________________
Design of a digital ECG signal generator on a low-

cost microcontroller
N. Sulistiyanto#1, O. Setyawati#2, R. A. Setyawan#3, M. Rif´an#4
#
Jl. MT Haryono 167 Malang 65145, Indonesia
1nnst@ub.ac.id
2osetyawati@ub.ac.id
3rarief@ub.ac.id
4rifan@ub.ac.id
Abstract— ECG signals represented by MIT BIH standard Arduino System) microcontroller. An analogue ECG signal is
database are widely used in digital signal processing, however, a generated by means of Pulse Width Modulation (PWM) and
reference ECG signal is required for analogue system. We present completed with a passive RC filter and variable resistor as
a prototype of an ECG signal generator based on a low cost attenuator. The aim of this study is to develop low-cost system
microcontroller. In this system the digital ECG signal is converted
to an analogue ECG signal by means of PWM. The output of the
which can be applied easily especially for biomedical
ECG signal generator is in accordance with the template signal electronics research field in development countries.
with the accuracy of approximately 99%.
II. MASTER ECG
Keywords— ECG signal, microcontroller, signal generator
ECG Record 100 (ML II) in the MIT-BIH Arrhythmia
Database [2] is used as the template signal. Record 100
contained of fluctuation of ECG signals’ baseline, however in
I. INTRODUCTION short interval some signals were relatively flat as shown in Fig.
1.
Electrocardiogram (ECG) signal consists of low frequency One beat of ECG signal consisted of 300 samples (Fig. 2),
component (P and T waves) and high frequency (QRS wave), was taken as the template. The template included P-wave,
hence the signal can be classified as non-stationer signal. Linier QRS-wave, and T-wave.
filtering generally decreases the ECG signal quality, for
instance noise reduction using low-pass filter can cause
significant distortion on QRS wave therefore it may result in
misinterpretation.
Front-end acquisitions system of ECG signal normally have
analogue amplifiers and analogue filters (RC) as bandwidth and
noise limiter. Frequency responses of the amplifiers and the
filters represent not the whole system performance. Responses
of the system to the real ECG signal will give much more
information regarding the system.
Despite the rapid development of digital system, the digital
signal processing for bio-signal, namely ECG, still requires
Fig. 1 ECG signal record 100 looks relatively flat
analogue system. To validate an amplifier or analogue filter
(and the whole system) ECG generator is required to produce a
standard signal.
Many methods to generate ECG signal are proposed. For III. SYSTEM DESIGN
instances, ECG generator based on standard ECG database is
developed [1-2] to validate commercial ECG record system ECG signal generator is designed using few external
controlled by MATLAB. ECG generator based on math model components as shown in Fig. 3. ECG template is implemented
and DAC 16 bit are published in [3-4], and based on parametric as constant arrays stored in ROM.
model is developed [5]. To produce sampling frequency of 360 Hz as specified in
In this research, ECG Generator based on standard database MIT-BIH database, each constant periodically is used for
[2] is implemented into ATMEGA328P (core of a low-cost

____________________________________________________________________________________________________________
determining PWM duty-cycle controlled by a timer. If the last happened due to the contaminated PWM frequency which was
constant is already used, automatically the program would not being properly filtered. However, the characteristics of
repeat the process of the first constant to produce the periodic ECG signal being produced was unchanged significantly.
ECG signal.
1
0.8
Normalized Amplitude
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
0 100 200 300 400 500 600 700 800 900
0 Sample
0 50 100 150 200 250 300
Sample
Fig. 4 Three consecutive template signals

Fig. 2 The ECG template signal
Low-pass filter worked to eliminate PWM frequency

components to obtain the ECG signal waveform. ECG
amplitude signal is decreased by attenuator to get the ECG
signal with an amplitude of 1 mV like real ECG signal.
Fig. 5 Output of the ECG signal generator
The template and experiment result of the QRS-wave are

Fig. 3 The microcontroller-based ECG Signal Generator depicted in Fig. 6 and Fig. 7, respectively. Both of the
waveforms showed no difference significantly, as well as the
slope of R-peak.
IV. RESULTS AND DISCUSSIONS The template showed normalized amplitudes of QR and RS
were around 0.97 and 0.89, respectively. Meanwhile time
duration of QR and RS were approximately 28 ms and 17 ms,
Three consecutive template signals are depicted in Fig. 4.
respectively.
Generally, the results are in accordance with the output of the
The generated QRS signal on oscilloscope were 0.96 and
real ECG signal generator (Fig. 5).
0.88 for the normalized amplitude of QR and RS, respectively,
The P- and T-waveform are defined very clear as well as the
while 30 ms and 15 ms are the duration time of QR and RS,
QRS-waveform. The only issue emerged here was the noise,
that was relatively high at the generator output generator. It was

____________________________________________________________________________________________________________
respectively. Therefore, the accuracy of the amplitude was

98.97%, and the duration time was 98.88%.
Hence, the waveforms produced by the generator can
represent the real ECG signal.
0.8
0.6
0.4 Fig. 7 The generated QRS signal
0.2
REFERENCES
0
0 10 20 30 40 50 60 70 80 90 100
[1] Á Sobrinho, P. Cunha, L. D. da Silva, A. Perkusich, T. Cordeiro and J.
Sample Rêgo, "A simulation approach to certify electrocardiography devices,"
2015 17th International Conference on E-health Networking,
Fig. 6 Template of the QRS signal Application & Services (HealthCom), Boston, MA, 2015, pp. 86-90. doi:
10.1109/HealthCom.2015.7454478.
[2] A. Goldberger, L. Amaral, L. Glass, J. Hausdorff, I. PCh, M. RG, M. JE,

M. GB, P. C-K and S. HE, “Physio Bank, PhysioToolkit, and PhysioNet:
Components of a New Research Resource for Complex Physiologic
V. CONCLUSIONS Signals. Circulation 101(23): e215-e220,” 13 June 2000. [Online].
The waveform generated by the ECG signal generator is Available: https://physionet.org/physiobank/database/mitdb/. [Accessed
21 August 2016].
generally in accordance with the template signal obtained from
the record 100 MIT-BIH database. The noise emerged in the [3] Z. Jun-an, “The Design of ECG Signal Generator using PIC24F,” in
generated ECG signal can be received, since the P-, QRS-, and International Conference on Advances in Engineering, Shanghai, China,
T-waveforms still can be recognized properly. 2011.
Noise can be reduced by increasing the PWM resolution and [4] J.-R. C. Chang, “Design of a Programmable Electrocardiogram
frequency using relatively high speed microcontrollers. Generator Using a Microcontroller and the CPLD Technology,” in 33rd
Annual Conference of the IEEE Industrial Electronics Society (IECON),
Taipei, November 2010.
[5] H. Al-Hamadi, A. Gawanmeh and M. Al-Qutayri, “An automatic ECG

generator for testing and evaluating ECG sensor algorithms,” in 10th
International Design & Test Symposium (IDT), IEEE Conference, 2015.
[6] C. Caner, M. Engin and E. Z. Engin, “The Programmable ECG

Simulator,” Journal of Medical Systems, vol. 32, pp. 355 - 359, August
2009.
.

____________________________________________________________________________________________________________
The Effect of Position of Anti Notch on the Surface

of Stator Teeth and Permanent Magnet to Cogging
Torque Reduction
Herlina Wahab#*1, Rudy Setiabudy#2, Ridwan Gunawan#3
#
Electrical Department, Universitas Indonesia
Depok, Indonesia 16424
*
Electrical Department, Universitas Sriwijaya
Inderalaya, South Sumatera
1
herlina41@ui.ac.id; herlinawahab@unsri.ac.id
2
rudy@eng.ui.ac.id
3
ridwan@eng.ui.ac.id
Abstract— In this paper, the author proposed the strategy to construction on the stator and the rotor side. Cogging torque
reduce the cogging torque using the anti-notch method which is reduction method that was developed lately, in [1] the
added on the surface of stator teeth and on the surface of the researcher develop the method by shaping the stator teeth
permanent magnet. It also investigates the effect of width using varying tooth tang radios and the reduction of the
variation of the anti-notch to cogging torque reduction. There
cogging torque is 31.9%. Notching method used by [2, 3]. in
are two shapes of anti-notch using in this paper, such as half
cylindrical namely 1st model for cylindrical anti-notch on the [2], the width and depth of notch was optimized using
surface of stator teeth, 3rd on the surface of the permanent response surface methodology to obtain a minimal cogging
magnet, and triangular namely 2nd for triangular anti-notch on torque values and the result is, cogging torque is reduced by
the surface of stator teeth and 4th on the surface of the 85 %. In [4], the cogging torque reduction is 76,6% with
permanent magnet. After simulated using FEMM 4.2, 2 D and forming a permanent magnet on the rotor with the
analysis the models, we found that the 1st model has the lowest arrangement forming the letter V. The author of [5]
cogging torque reduction with the 4 degrees of anti-notch width, Investigation the cogging torque reduction using the rotor
the same width of the stator slot, and by adding the anti-notch on
design techniques, reduction of cogging torque on each of
the surface of permanent magnet has no effect on the cogging
design are : by skewing the rotor tooth is 56%, tooth
torque reduction.
notching at the rotor is 43%, and pairing tooth of the rotor is
Keywords— Position, Anti-notch, Surface, Reduction, Cogging 21%. In [6], investigated the harmonic effect of radial
Torque asymmetry of magnet to cogging torque reduction. Using the
method of different laminate geometry [7], stated that cogging
I. INTRODUCTION torque peak almost canceled with this method.
In recent years, permanent magnet generators are widely Based on the previous research, the author developed the
used in industry, such as for wind turbine generator and others. method of cogging torque reduction using the anti-notch
This type of the generator was chosen because of it has many method. In this method, the author added a form of teeth on
advantages compared to induction type. Besides the the surface of stator teeth and magnet permanent. This
advantages, it also has disadvantages. One of the research is a continuation of research undertaken by the author
disadvantages is, there is electromagnetic torque in it. to obtain minimal cogging torque in permanent magnet
Electromagnetic torque consists of cogging torque and torque synchronous generator. In this paper, the authors propose the
ripple. Both of them caused noise, magnetic vibration and half cylindrical and triangular anti notch that is placed on the
make the lifetime of the generator shorter. The cogging torque surface of the permanent magnets on the rotor and different
also called as no load torque, it arises because of the width of it, to obtain a minimal cogging torque values and
interaction of electromagnetic fields generated by the compare it to position of anti-notch on the surface of stator
permanent magnets on the rotor with the stator slots. While teeth.
the torque ripple is the electromagnetic torque that occurs
when the generator has been given load. It is essential to II. COGGING TORQUE
reduce the cogging torque, the minimum cogging torque will Basically, the cogging torque can be determined in several
make the generator easy to rotate and produce the energy. ways, such as by determine the energy and co-energy in the
Research on the cogging torque has been done for few air gap by taking into account the degree of displacement of
decades by previous researchers. The common method to the rotor. The equation is as follows [8] :
reduce the cogging torque can be done by modifying the

____________________________________________________________________________________________________________
dW c Fig. 1 shows the initial design of magnet permanent

Tec  (1) generator (Reference model), before simulated by FEMM 4.2
d 2D, it is necessary to form a mesh for calculations and
where iterative process using finite element, as can be seen in figure
Tec is the electromagnetic cogging torque (Nm) below :
Wc is the co-energy (Joule)
θ is the mechanical angle of rotor position (degree).
The equation to determine the co-energy is as follows :
B2
Wc    d vr  ( J ) (2)
20
where
Bɵ is the flux density (T)
0 is the permeability
The electromagnetic torque effected by the volume of the
airgap (  g ), also the array of normal flux density (Bn) and the
tangensial array of flux density (Bt), the equation is as
follows :
L
 g μ 0 Sg
Tec  rBn B t ds
(3) Figure 2. Reference model with mesh
another way to determine the cogging torque is using the
fourier series, Fig. 2 represent the mesh of reference model, the smaller
 the mesh that is formed will affect the amount, timing, and
Tcog   Tmk sinnk  (4) accuracy of the iteration process. There are four proposed
k 1 models to simulate, for the position of anti-notch on the
where surface of stator teeth, the 1st model is half cylindrical, 2nd
Tmk is the fourier coefficient model is triangular. The position of anti-notch on the surface
k is an integer, and of the permanent magnet, a 3rd model is half cylindrical and
n = LCM (Ns, Np) 4th model is triangular. The models are simulated by FEMM
The hypotheses is, added the anti notch on the surface of 4.2, can be seen in the figure below :
stator and magnet permanent can reduce the cogging torque
by minimized the reluctance of the air gap.
III. DESIGN
In this paper, the author using the reference model of
radial flux permanent magnet generator model is 3 phase with
12/8 slot and pole, the material of magnet permanent using in
this model is NedFB 50 MgOe with 6 mm of the thickness, (a) 1st model (b) 2nd model
the same material, M19 steel for the stator and the rotor, 2 mm
of the air gap between the stator teeth and the magnet
permanent is 2 mm, the outer diameter of stator is 95 mm, the
inner diameter of the stator is 50 mm, 4 deg of slot width. For
the shaft, the author used 1018 steel. Reference model have
the 56210 nodes of mesh. This mesh is used as the basis for
the calculation using the finite element. (c) 3rd model (d) 4th model
Figure 3 Anti notch shape on the surface of stator teeth (a,b) and on the
surface of magnet permanent (c,d)
The models simulated by rotating the rotor every 1 degree

mechanical. The condition for the models devided into :
 Anti notch position on the surface of stator teeth. In this
position, the 1st and 2nd simulated with the different anti
notch width, eg. no anti notch (reference model), 2 degree,
Figure 1. Reference model and 4 degree mechanical.

____________________________________________________________________________________________________________
 Anti notch position on the surface of the permanent peak -0,045 Nm. Reduction of the cogging torque is 93.4% for
magnet of the rotor. For this position of anti notch, the the maximum and 97.8% for the minimum.
models simulated with the variation of anti notch width, Figure 4 (b) show the cogging torque value of the 2nd
eg. 4 degree, 3 degree, 2 degree, 1 degree and 0.5 degree model. In this model, the lowest cogging torque occurs on 2
mechanical. degree. It has 0.439 Nm for the maximum cogging torque and
-0.314 Nm for the minimum cogging torque. It has 75.8% and
IV. SIMULATION RESULT 81.1% reduction of cogging torque for the maximum and the
The simulation results that obtained in this paper is minimum peak.
cogging torque values, which are divided into two main
focuses, namely cogging torque values with half cylindrical B. Anti notch on the surface of permanent magnet
and triangular anti notch position in the stator tooth surfaces In this models, it has one anti notch at the center on the
and on the surface of the permanent magnet. surface of permanent magnet.
A. Anti notch on the surface of stator teeth 4,000
To simulate the models, the author added one anti notch at 3,000
the center of stator teeth on the surface of the stator teeth.
2,000
2,500
cogging torque (Nm)

2,000 1,000
1,500 0,000
Cogging torque (Nm)
1,000 -1 1 3 5 7 9 11 13 15
-1,000
0,500
0,000
-2,000 Rotor position (deg)
-1 1 3 5 7 9 11 13 15
-0,500 -3,000
-1,000
Rotor position (deg) -4,000
-1,500
Reference 4 deg 3 deg
-2,000
2 deg 1 deg cilinder 0,5 deg
-2,500
(a) Cogging torque of 3rd model
4,000
(a) Cogging torque of 1st model
3,000
2,500
2,000
2,000
cogging torque (Nm)
1,000
1,500
Cogging torque (Nm)
1,000 0,000
0,500 -1 1 3 5 7 9 11 13 15
-1,000
0,000 Rotor position (deg)
-1 1 3 5 7 9 11 13 15 -2,000
-0,500
-1,000 -3,000
Rotor position (deg)
-1,500
-4,000
-2,000
-2,500
2 deg 1 deg 0,5 deg
Reference 2 deg 4 deg (b) Cogging torque of 4th model
(b) Cogging torque of 2nd model Figure 5 Cogging torque value of the half cylindrical and triangular anti
Figure 4 Cogging torque value of the half cylindrical and triangular anti notch on the surface of permanent magnet
notch on the surface of stator teeth
In fig. 5 (a) and (b) shows the cogging torque wave almost
Fig. 4 (a) represent the value of cogging torque on 1st the same. Both of the models simulate with the variation
model. The reference has a maximum peak 2.178 Nm and the width of anti notch 0.5 degree, 1 degree, 2 degree, 3 degree
minimum peak -1.999 Nm. The lowest cogging torque occur and 4 degree to obtain the minimum cogging torque. In this
on 4 degree with maximum peak 0.143 Nm and the minimum position of anti notch, the minimum cogging torque occurs on
the 0.5 degree. Fig. 5 (a) and (b) shows the value almost the

____________________________________________________________________________________________________________
same with the reference model. On 3rd model maximum peak in Electrical Machines and Systems (ICEMS), 2015 18th International
Conference on, 2015, pp. 1475-1478.
is 2.156 Nm and the minimum peak -1.998 Nm, reduction on
[10] H. C. Yu, B. S. Yu, J. t. Yu, and C. K. Lin, "A Dual Notched Design of
it only 1.01% and 0.01% for the maximum and the minimum Radial-Flux Permanent Magnet Motors with Low Cogging Torque and
peak. The same things happened to 4th models, the maximum Rare Earth Material," IEEE Transactions on Magnetics, vol. 50, pp. 1-
peak has 2.170 Nm and -2.003 Nm for the minimum peak. 4, 2014.
[11] K. Dae-kyong, P. Young-Un, and C. Ju-Hee, "Cogging torque
Each of reduction cogging torque 0.38% and -0.18% for the
reduction of single-phase brushless DC motor with a tapered air-gap
maximum and the minimum peak. using optimizing notch size and position," in Energy Conversion
Congress and Exposition (ECCE), 2014 IEEE, 2014, pp. 2447-2453.
V. CONCLUSIONS
From the 4 models proposed to this research, the most
significant is the 1st model with highest cogging torque
reduction, 97.8%. This result is better than the cogging torque
reduction by [9] 12%, [10] 90.9%, [11] 91%. While adding
the anti notch on the surface of stator teeth, the volume of the
air gap decrease, so the cogging torque decrease too. In this
research we found that by added the anti notch on the surface
of the permanent magnet doesn't effect to the cogging torque
reduction. The lowest cogging torque wave almost the same
with the reference. For further research, the anti notch method
on the stator teeth can be combine with other method to
implement it on the permanent magnet.
ACKNOWLEDGMENT
The research was supported by the Directorate of Research
and Community Engagement - Universitas Indonesia, Final
Student Grant 2016 for Index Publication, contract no.
2115/UN2.R12/HKP.05.000/2016
REFERENCES
[1] A. N. Patel, "Influence of stator teeth shaping on cogging torque of
radial flux Permanent Magnet Brushless DC motor," in 2016 Biennial
International Conference on Power and Energy Systems: Towards
Sustainable Energy (PESTSE), 2016, pp. 1-4.
[2] K. Youngjun and L. Cheewoo, "Cogging torque reduction of brushless
DC motor for an electric wheelchair," in 2016 IEEE Transportation
Electrification Conference and Expo, Asia-Pacific (ITEC Asia-Pacific),
2016, pp. 954-957.
[3] J. H. Hu, L. Wang, J. B. Zou, and B. Zhao, "Cogging Torque
Reduction of Hybrid Excitation Flux Switching Motor," in 2015 Fifth
International Conference on Instrumentation and Measurement,
Computer, Communication and Control (IMCCC), 2015, pp. 1889-
1892.
[4] W. Ren, Q. Xu, Q. Li, and L. Zhou, "Reduction of Cogging Torque and
Torque Ripple in Interior PM Machines With Asymmetrical V-Type
Rotor Design," IEEE Transactions on Magnetics, vol. 52, pp. 1-5, 2016.
[5] L. Hao, M. Lin, D. Xu, N. Li, and W. Zhang, "Analysis of Cogging
Torque Reduction Techniques in Axial-Field Flux-Switching
Permanent-Magnet Machine," IEEE Transactions on Applied
Superconductivity, vol. 26, pp. 1-5, 2016.
[6] A. J. Pina, S. Paul, R. Islam, and L. Xu, "Effect of radial asymmetry of
magnets in surface-mounted permanent magnet motors on cogging
torque," in 2015 IEEE International Electric Machines & Drives
Conference (IEMDC), 2015, pp. 247-253.
[7] M. Caruso, A. O. D. Tommaso, R. Miceli, G. Schettino, and F. Viola,
"A cogging torque minimization procedure for IPMSMs based on
different laminate geometry," in 2016 Eleventh International
Conference on Ecological Vehicles and Renewable Energies (EVER),
2016, pp. 1-6.
[8] D. D. Hanselman, Brushless Permanent Magnet Motor Design, : The
Writer’s Collective, 2003.
[9] J. D. Seo, J. H. Yoo, and T. U. Jung, "Design on notch structure of
stator tooth to reduce of cogging torque of single-phase BLDC motor,"

____________________________________________________________________________________________________________
Drivetest Methods for

3G Network Quality Analysis
Jinantho P. H. S1, Sigit Kusmaryanto2, Dwi Fadilla Kurniawan3
1,2,3
Electrical Engineering Department, 1,2,3 Brawijaya University
Jl. MT Haryono 167 Malang
1
jp.habibullah@gmail.com
2
sigitkus@ub.ac.id
3
df_kurniawan@ub.ac.id
Abstract— Analysis of 3G mobile telecommunication network value of Ec / No. Strong signal received by the UE from the
quality is determined by the traffic indicators. It done through node-b each different from one another. This is because the
the method of OMC (Operation Monitoring Center). In the effect of damping due to losses in the propagation path
network control technology OMC done with computer obtained every user is different. RSCP scale between -120 - 0
programming can only be monitored remotely and analyze dBm, the greater value is the better result.
interference, but the handling and point problems cannot be
solved and accurately known. Drivetest is one method of 𝐸𝐸𝑏𝑏 /𝑁𝑁𝑜𝑜 is used as a parameter in digital Determining the
measuring the quality of the network is carried out using a rate of the data as well as parameter reference standard in
mobile phone and software TEMS Investigation 8.0.3. Drivetest
quality measurements for digital communications system
conducted to determine the quality of the base stations have been
built, to view and maintain the quality of the network in certain performance. Eb / No is the ratio between signal energy per
regions, and determine the cause of decline in the quality of a bit to noise.
network. Parameter-quality 3G network that can be known
When a signal, digital and analog, which has a binary
drivetest methods: RSCP and Ec/No. This study describes
drivetest methods on 3G networks at Brawijaya University. digital data is transmitted at a particular data rate, R, where R
can be calculated using the equation.
Keywords— drivetest, 3G, RSCP, Ec/No 𝐸𝐸𝑏𝑏 𝑆𝑆 𝑊𝑊
= .
𝑁𝑁𝑁𝑁 𝑁𝑁 𝑅𝑅
I. INTRODUCTION
Drivetest regularly conducted to determine the quality of an CSSR is the percentage rate of success to build a call in
area, with a large number of customers, a new frequency plan which the UE at the beginning of the signaling will acquire
is implemented, done specifically in places certain to know channels that will be used. CSSR used as measure the degree
the quality of service as well as some of the existing of availability of the network to which the
parameters. Drivetest can be done by static mode or carried telecommunications service provider provides its services in
out by a mobile or a path on the surrounding region. The the form of a voice call, video call, or SMS (Short Message
method used in 2 ways: Service). Percentage calculations CSSR obtained using
mathematical calculation as follows:
1. Idle Mode is drivetest done just to see the 3G signal
transmit power (RSCP) and quality Ec/No. 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝑥𝑥 100%
2. Dedicated Mode is drivetest done by performing a call so 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
that information can be obtained in addition to the
transmit (RSCP), quality (Ec/No), the success of the call Dropped Call Ratio (CDR) is the percentage ratio at the
(CSSR), the call failure rate (SCR), cell migration time of the call is built is the number of calls dropping out or
handover success ratio ( HOSR) can be obtained by fall when the conversation channel is used. Percentage
using drivetest[1] calculations CDR obtained using mathematical calculation as
follows:
Several parameters are used as a general reference to be
able to see the performance of the 3G network is like: RSCP, 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑒𝑒𝑒𝑒
Eb/No, Ec/No, Call Setup Success Ratio (CSSR), Call 𝐶𝐶𝐶𝐶𝐶𝐶 = 𝑥𝑥 100%
𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒ℎ𝑒𝑒𝑒𝑒
Dropped Ratio (CDR), Successfull Call Ratio (SCR), Call to
Interference (C/I) and Handover Success Ratio (HOSR). [2] Carrier to Interference ratio is the ratio value obtained
based on the value of Eb / No. The relationship between C / I
Received Signal Code Power (RSCP) is a strong signal and Eb / No based on mathematical :
receipt stating the amount of power on one code accepted by 𝐶𝐶 𝑅𝑅𝑏𝑏 𝐸𝐸𝑏𝑏
the UE which is one of the parameters that determine the = � �.� �
𝐼𝐼 𝑊𝑊 𝐼𝐼0

____________________________________________________________________________________________________________
Handover Success Ratio is the percentage of the UE experiment with the surrounding area is specified, then the
during a conversation in the mobile of the success rate of cell data processing to seek Carrier to Interference Ratio to
transfer process without termination. The criteria that led to mathematically, and then the poor network quality analysis,
the handover, among others: the level of acceptance (RSCP), after that drive performance improvements 3G network in
the reception quality (Ec / No), a distance of MS-BTS, power areas that experienced a decline in quality.
budget, Fast Upling handover (reduction in signal level
drastically) and calls traffic. [3] Percentage calculations on
Handover Success Ratio is obtained by using a mathematical
calculation as follows:
ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = 𝑥𝑥 100% Fig. 3. Block Diagram Testing
ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
A reference work to obtain data call attempt, call setup, call
II. RESEARCH METHODS blocked, call droped, droped calls, handover attempt,
handover success (Fig. 4.). Drivetest university area brawijaya
The data collection method using drivetest by performing by way of mobile with the state of the device is connected,
test calls at predetermined locations. The measurement then Mobile Station 1 (MS 1) to make calls to Mobile Station
technique is done by setting the phone in 3G mode in idle 2 (MS 2) with a state of motion, and then during a call MS 1
mode and mobilemode. Fig. 1. describes the phases of the secured channel to connect to the MS 2, if not there will be
network configuration user equipment in obtaining the signal blocked calls, if MS 2 successfully receive a call, then for 10
before they can access the network from a particular region[2]. seconds will maintain the call, after which if for 10 seconds
Arrangements in performing drivetest shown in Fig. 2. termination will occur dropped call.
Where the 3G signal emitted by BTS / Node B to be received
by the Mobile Station (MS) or a mobile phone, which will
transmit data on laptops integrated TEMS Investigation v
8.0.3. This arrangement scheme in phases for testing the
quality of the network at the university brawijaya.. This
measurement process is done with a mobile phase or move Fig. 4. Setting Measurement
with the manual. III. RESULT AND DISCUSSION
10 data taken from 100 times referring experimental data to
see how well these areas provide signal quality that can be
obtained phone is in motion. Of the 10 data showed the area
deteriorated due to obstructed experienced storey building in
the area of the Faculty of Economics and Business and the
Faculty of Law area. The more positive approach RSCP value,
the better the quality of the signal obtained UE shown in Fig.
5.
Fig. 1. 3G Network Configuration[3]
Fig. 5. The Sample Data of RSCP

Fig. 2. Testing Scheme
The more the value of Ec/No positive approach, the better
The steps in obtaining data RSCP and Ec/No to be obtained the 3G network that can be accessed by the UE is shown in
for the processing of the data analysis (Fig. 3.). The first step Fig. 6. 10 data taken from 100 times to see how well the
determines the BTS university brawijaya be traversed network can be accessed when user equip on the condition of
drivetest, device and user equipment is set up, locking 3G is in motion. Of the 10 data showed the area deteriorated
location drivetest, after it drivetest by means of a mobile to get due to obstructed storey building in the area experienced
RSCP data and Ec / No, data retrieval number 100X Widyaloka.

____________________________________________________________________________________________________________
only area experiencing Widyaloka 6 blocked call and one call

drop
ACKNOWLEDGMENT
I have taken efforts in this project. However, it would not
have been possible without the kind support and help of many
individuals and organizations. I would like to extend my
sincere thanks to all of them.
I would like to express member of Laboratorium of
Telecommunication on Electrical Engineering Department for
their kind co-operation and encouragement which help me in
Fig. 6. The Sample Data of Ec/No completion of this project.
My thanks and appreciations also go to my colleague in
Map and Event Measurement Drivetest shows the area of developing the project and people who have willingly helped
testing and the events that occurred at the time of me out with their abilities.
measurement drivetest using TEMS Investigation 8.0.3. Fig. 7.
REFERENCES
outlines the quality of a given network of providers Telkomsel
to user equipment. [1] Wardhana, Lingga, 2G/3G RF Planning and Optimization for
Consultant, Jakarta., 2011.
[2] Holma, Harri dan Toskala, Antti, WCDMA for UMTS, England : John
Wiley & Sons, Ltd., 2004.
[3] Anqiang, Jiao., W-Handover and Call Drop Problem Optimization,
Huawei Technologies Co., 2006.
[4] S. M. Metev and V. P. Veiko, Laser Assisted Microtechnology, 2nd ed.,
R. M. Osgood, Jr., Ed. Berlin, Germany: Springer-Verlag, 1998.
[5] Lee, William C.Y., Wireless and Cellular Telecommunication
Third Edition, Singapore: McGraw Hill Book, 2006.
[6] Huawei Technologies, eLTE2.2 DBS3900 LTE FDD Configuration
Principles, Shenzhen: Huawei Industrial Base, 2015.
[7] Dongwoo Kim, Dong Geun Jeong, Capacity Unbalance Between
Uplink and Downlink in Spectrally Overlaid Narrow-Band and
Wide Band CDMA Mobile System, IEEE Transact ion on Vehicular
Technolog y. Vol 49 no 4, 2000.
[8] Dongwoo Kim, II Gyu Kim, Dong Geun Jeong, Capacity
Analysis of Spectrally Overlaid Narrowband and Wideband CDMA
System for Future Mobile Communication Services, IEICE Trans.
Fig. 7. Map and Event Measurement Drivetest Commun Vol. E82-B No 8, 1999.
[9] Hyuck, M.Kwon, Capacity and Cut-off Rate of Coded FH/MFSK
IV. CONCLUSIONS Communication with Imperfect Side Information Generator, IEEE
Journal on Selected Areas in Communication,Vol 8 no 5, 1990.
Drivetest method is able to describe the quality of 3G [10] Ojanpera, R. Prasad, Overview of Air Interface Multiple Access
network. Based on the measurement data with the application for IMT-2000/UMTS, IEEE Commun. Mag., 1998.
keys on performance indicators using drivetest, areas [11] A.Bruce Carlson, Paul B. Crilly, Janet C. Rutledge,
Communication Systems, McGraw Hill, New York, 2002.
experiencing decline in quality that is at Widyaloka area. [12] Holma, Harry, LTE for UMTS OFDMA and SC-FDMA Based Radio
Widyaloka area showed a loss of quality compared to other 3 Access, United Kingdom: John Wiley & son Ltd., 2009
testing area, 100 times the measurement value of the CSSR
94%, the value of CDR 1:03%, and the value of SCR 89 073%

____________________________________________________________________________________________________________
Performance Analysis of Online Video Classification System
Based on Tag Using MapReduce for Internet Content Profiling
F. Astha Ekadiyanto1, M. Adhika Putra1,
1
Computer Engineering, Department of Electrical Engineering, Universitas Indonesia,
Kampus UI Depok 16424 INDONESIA
Abstract—This paper presents the research on Online II. BASIC THEORY

Video Classification System with Symbolic Distance and A. Big Data
Focal Point Method which is implemented using Map- Big Data has many definitions and keep on evolving.
Reduce from Hadoop framework for the purpose of Generally, it is described as a collection of data (structured
Internet Content Profiling. It identifies tags of the online and unstructured) in a very large size and contains potentially
video and compare them using co-occurrence matrix to significant information within [2]. Information from Big Data
find the symbolic distance value for the video. The Focal are obtained through a process of Big Data analysis that can
Point in this system is useful to improve accuracy and help decision making, process automation, and future science
focus of video classification. Experiments are performed view [3]. Doug Laney, industry analist, divide the concept of
to measure the processing speed of the classification Big Data into 4-Vs [4], they are:
system and search factors that can affect performance.  Volume, data size. When searching a unique information
Three scenarios are implemented based on Input Split from the object that it phenomenon will be predict,
size, amount of node, and YARN configuration over three statistic analysis is required from the very large data size.
different file sizes (500 MB, 1 GB, 1,5 GB) having 58718,  Velocity, the speed of data transmission to be processed.
119697, and 160395 number of tags consecutively. The Data transmission can happen from IoT device to storage
results of experiments with 500 MB, 1 GB, and 1,5 GB file and then the information can be obtained. With the
sizes indicate that by increasing the amount of node from growth of IoT application in the future, the increase of
two to three, processing speed is increased up to 16,3 transmission data is needed to giving a fast response
seconds. Optimizing YARN configuration can further quickly [5].
improve the speed up to 354,3 seconds. Furthermore,  Variety, the type of the data that will be analyzed. There
there was a tradeoff found during the experiment over the are various kinds of data to be stored in Big Data storage
size of InputSplit, where smaller size is preferable until system, where most of the data consists of unstructured
the mapper no longer able to handle the amount of the data (SMS, audio, video).
split which will cause trashing and increases the process-  Value, information that contained in the data. Not only
ing speed again. large amount of data storage capability, but also high data
Keyword— Big Data, 3V, Hadoop, Internet Content Profiling,
computation ability is required. A prediction model is
MapReduce, Classification. needed to accurately decide which technology fits best to
achieve optimal performance for the Big Data analysis to
I. PREFACE perform effectively [5].
A. Background
B. Hadoop
Concerns about information contained in the internet has Hadoop is a framework to distributed processing a large
brought the attention of governments. In Indonesia, the dataset and running from one server to a thousand computer
ministry of Communication and Information (Kominfo) has machine and also can process until petabytes data using a
published several regulations to handle negative contents. simple programming model. There are two main process,
Internet may have given people much benefit and easiness in MapReduce and Hadoop Distribution File System (HDFS) [6]
life [1], but it is also use to spread the dangerous information From architectural side, based on [2] and [7] Hadoop
in a very large amount that could impose social and moral consists of 2 main components:
issues and cause bad impacts to society. Therefore, a 1. MapReduce. An algorithm to do computation and data
sophisticated information filtering system based on its content sorting in its store.
is needed in order to reduce the bad impact of using internet 2. HDFS (Hadoop Distributed File System). A system with
without disturbing the easeness of internet usage. the function to store data over distributed file system.
In this research, a program that can classify online video
content is implemented on distributed processing using
C. Hadoop Distributed File System
Hadoop platform. In the experiments, different amount of
Hadoop Distributed File System (HDFS) is a system file
node, and different size of InputSplit are being investigated to
that store the data by splitting it into certain size and then put
analyze their impact to processing performance over different
it into storage cluster [8]. HDFS has three main components,
data sizes or tag sizes. We also attempt to fine tune the YARN
NameNode, DataNode, and Secondary NameNode [8] [9]
configuration to improve Hadoop performance.
[10].

____________________________________________________________________________________________________________
D. MapReduce Tabel 1 The Example of Co-Occurrence Matrix [14]
MapReduce is a programming model to process the data

on Hadoop. User assigns a map function for processing a
key/value pair and generate intermediate key/value. The
reduce function will combine all of intermediate value with its
intermediate key [8] [11].
E. YARN
YARN [12] is the 2nd generation MapReduce created to
solve the scale limitation on MapReduce in previous gene-
rations. YARN has the function to allocate the resource dyna-
mically to mapper and reducer. YARN separates JobTracker
into 2 main functions: resource management and job sched-
uling. The objective of YARN is for Hadoop to have a global
ResourceManager and Application Master per-application
[13]. With YARN, Hadoop can run multiple applications with In this reseach, the focal point as recommended in [14], is
the same resource management and therefore more robust and based on the frequency of tag’s appearance. Tags that become
reliable which also increases the scalability of the application. the focal point are the most frequent tag on the video data that
we have retrieved before. With focal point, the classification
III. IMPLEMENTATION DESIGN process is expected to be more focused and accurate. Pseudo-
A. HTML Data Retrieval process with Youtube API code of the process is shown in Figure 2.
Map
In this research, the data that will be processed are html Read youtube tag and url from html file
file container to a Youtube™ video. In order to retrieve the For i=0 to no_of_tag do
html data, Youtube API and Jsoup Java client side Library is For j=i+1 to no_of_tag do
used. The pseudocode of this program is shown in Figure 1. Generate pair <key=url, value=”tag[i], tag[j]”>
End for
End for
Set keyword for video search Reduce
Set max 50 video per result Split values into “splitTag”, “,” as delimiter
for i=0; i<=200; i++ If: splitTag[0] = focal_point||splitTag[1]=focal_point
Get search result from Youtube API symbolicValue=75
Read response from Youtube API if: value=co-occurrence matrix
For all video results do symbolicValue=+ matrix Value
Get VideoId if: value>75
Jsoup.Connect(“https://www.youtube/watch?v=” result=”Dangerous content”
+ VideoID) else
Download the response from URL result=”Safe Web”
End For Generate pair <key=rule, value=result>
Get NextpageToken
Figure 2. Pseudocode Video Classification
Go to next page search result
End For with Focal Point and Symbolic Distance
Figure 1. Pseudocode to retrieve data HTML Youtube C. Scenario of the Experiment
B. The Video Classification Algorithm in Internet Content The experiment uses three different html file sizes: 512
Profiling. MB, 1 GB, and 1,5 GB having amount tag 58.718, 119.697,
The classification process of Internet Content Profiling and 160.395 consecutively. The objective is to evaluate the
separates the video into dangerous content category (violent, best configuration for processing the various data sizes. In the
sexual, terrorism, etc) and non dangerous content category. It first scenario, the experiment will compare Hadoop perform-
is based on symbolic distance method which considers video ance using two and three nodes. The InputSplit size in
tags as symbols and will be evaluated to find out how similar InputFormat in this scenario is 64MB. The impact of the
a video symbolically with a provided category. number of processing nodes over different file sizes will be
The symbolic distances are then used to construt the co- evaluated.
occurrence matrix. Co-occurrence matrix (Table 1) is a In the second scenario, the purpose of the experiment is to
symmetrical matrix that shows the degree of occurrence of compare the Hadoop performance before and after a container
tags between a video and reference [14]. Every matrix is been configured. InputSplit size in InputFormat used in this
generated based on the classification category. Therefore, scenario is 64MB. The effect of memory allocation for
every category will have different matrix and every single container will be evaluated.
video will be fitted to all the matrices. In the third scenario, the experiment is performed over
In each co-occurrence matrix, the numbers of matching three different InputSplit sizes: 32 MB, 64 MB, and 128 MB.
tags pair are summed up. The higher the sum is, the more The impact of InputSplit size to the processing speed of
likely the video fits to the category. MapReduce data processing on Hadoop will be evaluated.

____________________________________________________________________________________________________________
IV. RESULTS
The experiment is performed in the Joint Innovation
Centre Laboratory – Department of Electrical Engineering,
Universitas Indonesia. Three machines are used for the
experiment, each has the specification of quad cores processor
with 2 GB RAM, 100 GB storage running CentOS 6.6, all
connected to a gigabit ethernet switch.
A. The Result of Video Classification System with Symbolic Processing Time (seconds)
Distance
Figure 3 shows the comparison between the total data that Figure 4. Comparison Chart of Data Processing Speed
Between Two and Three Nodes.
are processed and the result of the classification. Table 2
shows the detailed results of the classification process. For
500 MB html file with 3454 youtube pages, 877 video URL C. Second Scenario
or 25,3% are dangerous. In 1 GB html file with 7041 youtube The second scenario is used to find out the impact of
pages, 1791 video URL or 25,4% are dangerous. And for 1,5 memory allocation configuration for YARN. In this scenario,
GB html file with 9435 youtube pages 2414 video URL or the time of data processing using MapReduce on Hadoop is
25,5% are considered dangerous. measured before and after the YARN is configured. Three
different file size that will be processed are 500 MB, 1GB, and
1,5 GB with 58.718, 119.697, and 160.395 tags consecutively.
Memory allocation for ResourceManager is set of 75% from
physical memory on each node, the rest is for OS and other
application. The experiment is done 10 times.
Figure 5 shows that the configured YARN causes the
processing to be faster. In default condition, MapReduce run
inneficiently because the resource does not fit the Hadoop
requirements. Whenever the physical memory on server can’t
satisfy the request, it will generate some error when running
Dangerous content MapReduce, especially for file size more than 1 GB.
Figure 3 Comparison between Total Data with Dangerous Content Video
in the Classification Process
Tabel 2 Classification Result

File Total File Total Dangerous
Size HTML Tag Content
500 MB 3454 58718 877
1 GB 7041 119697 1791
1.5 GB 9435 160395 2414
Processing Time (seconds)
B. First Scenario Before configuration After configuration
First scenario is to find out the impact of the amount of Figure 5. Comparison Chart of Data Processing Speed Before and
After YARN Being Configured.
node to MapReduce performance on Hadoop with various file
size. Experiments are performed over two and three nodes and
When the memory allocation for ResourceManager and
iterated 10 times.
container matches the available physical resource, the
Based on chart shown in Figure 4, the usage of 3 nodes for
resource scheduling for MapReduce will be optimal,
processing file with size 1 GB dan 1,5 GB performs faster than
generates no error and even performs faster than before.
2 nodes. For 500MB file, the usage of 3 nodes does not have
YARN is sensitive to resource usage in Hadoop so when
significant effect to the performance. This may be related to
processing data with large size, YARN must be configured
the fact that the resource required for file processing is already
until there are no error in running MapReduce.
enough. When processing 1 GB file, the resource requirement
became significantly higher and additional node can improve D. Third Scenario
performance. For the larger 1,5 GB file, the reduction of Third scenario is used to evaluate the impact of InputSplit
processing time even become much more significant. size to MapReduce performance. In this scenario, the
MapReduce processing time will be measured over 3

____________________________________________________________________________________________________________
InputSplit sizes: 32 MB, 64 MB, and 128 MB for the 3 file

sizes 500 MB, 1 GB, and 1,5 GB each with the tag amount of
58.718, 119.697, and 160.395 consecutively. The experiment REFERENCES
is done 10 times.
[1] E. Turban, R. K. Rainer and R. E. Potter, Introduction to
Information Technology, New York: Wiley, 2005.
[2] D. V. T. Shweta Pandey, “Prominence of MapReduce in BIG
DATA Processing.,” in Fourth International Conference on
Communication Systems and Network Technologies, Bhopal,
India, 2014.
[3] J. G. a. D. Reinsel, “Extracting Value from Chaos,” June
2011. [Online]. Available:
https://www.emc.com/collateral/analyst-reports/idc-
extracting-value-from-chaos-ar.pdf. [Accessed 12 01 2016].
Processing Time (seconds) [4] F. Ohlhorst, Big Data Analytics: Turning Big Data into Big
Money, Wiley, 2013.
[5] M. W. a. R. H. G. A. Katal, “Big data: Issues, challenges, tools
Figure 6. Comparison Chart of Data Processing Speed and Good practices,” in Sixth International Conference on
using three InputSplit size. Contemporary Computing (IC3), Noida, 2013.
[6] “An introduction to Apache Hadoop for big data,” 29 August
Figure 6 shows that the fastest processing speed is 2014. [Online]. Available:
achieved when using 32 MB InputSplit size and the slowest is https://opensource.com/life/14/8/intro-apache-hadoop-big-
128 MB InputSplit. The size of InputSplit affects the amount data. [Accessed 12 01 2016].
of mapper to process the data. If there are too many mappers, [7] K. Grolinger, M. Hayes, W. A. H. A. L'Heureux, D. S. Allison
processing time become longer because there are too many and M. A. M. Capretz, “Challenges for MapReduce in Big
queues to process the data. When too small, every mapper will Data,” in IEEE World Congress on Services, Anchorage, AK,
2014.
need bigger resources and causes system error due to not
enough resource. Therefore, based on the experiment, the [8] C. Lam, Hadoop In Action, Stamford: Manning Publications
Co., 2011.
InputSlipt of 32MB is considered optimal.
[9] H. S. C. Shv, “The Hadoop Distributed File System,” in
V. CONCLUSION Proceedings of the 2010 IEEE 26th Symposium on Mass
Storage Systems and Technologies (MSST) , California, USA,
From the experiment results of Youtube video 2010.
classification system using symbolic distance method and
[10] T. White, Hadoop : The Definitive Guide 3rd Edition, USA:
Focal point in Hadoop framework, it can be concluded that:
O’reilly, 2012.
1. The accuracy of video classification system using
symbolic distance depends strongly on the completeness [11] S. G. Jeffrey Dean, “ MapReduce: simplified data processing
on large clusters,” in Proceedings of the 6th conference on
of occurrence tag for each content categories. Symposium on Opearting Systems Design & Implementation,
2. The number of nodes used in processing gives bigger San Francisco, CA, December 06-08, 2004.
impact for bigger payload size. In this experiment, a 500
[12] V. K. Vavilapalli, A. C. Murthy, C. Douglas, S. Agarwal, M.
MB file will have very small impact for an additional Konar, R. Evans, T. Graves, J. Lowe, H. Shah, S. Seth, B.
node usage than the other bigger payloads of 1GB and Saha, C. Curino, O. O'Malley, S. Radia, B. Reed and E.
1.5GB which reduces processing time by 5 seconds and Baldeschwieler, “Apache Hadoop YARN: yet another
38.8 seconds consecutively resource negotiator,” in Proceedings of the 4th annual
3. Matching memory allocation configuration for YARN Symposium on Cloud Computing, Santa Clara, California,
container can improve processing data performance October 01-03, 2013.
greatly. It can reduce the processing time by 47 seconds [13] J. D. Arinto Murdopo, “Next Generation Hadoop: High
for 500 MB file, 277,1 seconds for 1 GB file, and 354,3 Availability for YARN,” 2013.
seconds for 1,5 GB file. [14] M. E. A S Sharma, “Classification of Multi-Media Content
4. InputSplit size can also affect the data processing (Video’s on Youtube) Using Tags and Focal Points,”
performance on MapReduce. The InputSplit size should Unpublished manuscript, 2008.
be optimal in order to balance the number of mapper and
split tasks while still maintaining enough resource
allocation.

____________________________________________________________________________________________________________
Traffic Monitoring System for Wireless Network

Using Embedded System
Heru Nurwarsito1, Gembong Edhi Setyawan2, Kautsarani Permata Alam3
123
Faculty of Computer Science, University of Brawijaya, Malang, Indonesia
1
heru@staff.ub.ac.id, 2gembong@ub.ac.id, 3kautsaranipermata@gmail.com
Abstract— By increasing of technology development at the age of The traffic monitoring system design on wireless network
information and globalization, the wireless technology users are is using Raspbian Wheezy operating system, the Linux
also increasing. Rising numbers of wireless technology user in a operating system specifically developed to work on
certain area causing a density of the data traffic, so that required Raspberry-Pi. Where this Raspberry-Pi device also utilized a
a media in order to do network monitoring. In this research, the
traffic monitoring system is designed using raspberry-pi for
device in a Wi-Fi dongle form as transmission media in order
wireless network capacity compiling. The system design utilized to access an access point through wireless network. This
SNMP protocol (Simple Network Management Protocol) as system developed by Phyton programming language using
linking and exchanging media in order to do wireless network pySNMP library to SNMP protocol application. The traffic
monitoring. The system implementation utilized Phyton monitoring system device built capable to read information
programming language with pySNMP library. Based on data on wireless network, that is traffic amount passed by
connection and traffic calculation testing, system was capable to utilizing SNMP protocol.
capture, connect and also read wireless network traffic capacity. On the previous research, developed a research about
The test results of calculating traffic capacity using raspberry-pi monitoring traffic on internet network using ENTM system
have relatively smaller value than the Mikrotik router and have
utilized open source code embedded Linux as operating
relatively greater value relatively than the dude. The monitoring
accuracy comparison results between raspberry-pi and the system [1], while on the similar research contained about
mikrotik router shown the value: Tx accuracy was 0.987 and Rx building an online network management system based on PHP
accuracy was 0.991, while comparison with the dude shown the utilized android operating system [2].
value: Tx accuracy was 1.0007 and Rx accuracy was 1.0022.
II. THEORY AND RELATED RESEARCHES
Keywords— Traffic monitoring; SNMP protocol; Wireless Raspberry-Pi is microcomputer device in credit card sized
network; Raspberry-Pi; embedded system. which easily designed and implemented. This research, is
using Raspberry-Pi type B which has 512MB RAM memory,
I. INTRODUCTION 2 ports USB and Ethernet. System design is using Raspbian
At the age of information and globalization, the world of Wheezy operating system, the Linux operating system
communication increased very dramatically, marked by arise specifically developed to work on Raspberry-Pi.
of many new communication tools with various features. One Raspberry-Pi also utilized a device in a Wi-Fi dongle form
aspect of technological developments is the increasingly as transmission media in order to access an access point
widespread of internet with wireless technology using. Rising through wireless network. While on the implementation, this
numbers of wireless technology user in a certain area can system developed by Phyton programming language using
cause a density of the data traffic, so that required a media in pySNMP library to SNMP protocol application. Raspberry-Pi
order to do monitoring on the network. Then built a type B shown in Fig. 1.
monitoring system on wireless network use for compiling an
information in traffic, include to monitor the traffic condition
on network. The monitoring system on internet network
existence is important for comprehend network behaviour so
the administrator can react properly and helping design also
give network future with more efficient [1].
In this research, monitoring traffic system built using
Raspberry-Pi by utilizing SNMP protocol (Simple Network
Management Protocol) as linking media between transmitter
and receiver in communicating and also exchanging
information. While the SNMP protocol designed for giving an
ability to user/administrator to monitor and arrange the
computer network systematically from distance or commonly
called as remotely [2]. Fig. 1. Raspberry-Pi type B

____________________________________________________________________________________________________________
SNMP is an application protocol on TCP/ IP network A. Work Method Program Design

handled a network management. This protocol designed so To be able to do traffic amount-counting process on
that user can easily to monitor computer network condition. wireless network need a program addressed to read the traffic
Network condition monitoring can be done by collecting object value. Which the counting work system based on
information values from network condition with remotely overall data packets transmission process in a network. The
method or using one spot observation. program flow chart shown in Fig. 3.
This research utilizing SNMP protocol as linking media
between transmitter and receiver in communication and also Start
information exchanging so can do a monitoring towards

wireless network. SNMP divided into three main elements [3
Network monitoring purpose is compiling information Connect to wireless Timeout sign on
access point
about status and behaviour on a network. The network
SNMP while
unconnected
monitoring utility are [5]:

Fail
Fail
• Detecting the down and up of a link and inform it. Connect to wireless
acces point
• Monitoring good and bad line in a network. Count amount Succeed

traffic
• Knowing the number of data packets passed (transmitted Succeed
Display
Result
and received). Check availability
counting
amount traffic
WLAN is local area network without cable which the OID
transmission media utilized radio frequency (RF) and infrared End
(IR) to give a network connection to the whole users in Fig. 3 Work Method Program
surrounding area.
System design in this research required router wireless. B. System Implementation
Router wireless is a device functioned to forward packets Implementation based on desired design that predetermined
from a network to another (either LAN to LAN or LAN to before. Implementation process done gradually so the
WAN). researcher can repair it if errors happened on system. This
research implementation stages shown in Fig. 4.
III. SYSTEM DESIGN AND IMPLEMENTATION Start
On this Monitoring system design needed several

supporting factors in some devices form those are hardware Setting up the Raspberry Pi devices and
SD cards, as well as wifi dongle for
and software. All of the software and hardware needed have wireless connection
already predetermined before. On the system design stage,
specified the MIB contained on SNMP agent will be analysed.
Then using Raspbian Wheezy operating system and Installing and configuring software on Raspberry
Pi
supporting software installation like Phyton for programming
language. Monitoring system design include SNMP value Installation and configuration of network devices
as a supported device
retrieval program implementation also display passed traffic No
value or amount results. Implementation and configuration program on the

Raspberry Pi
Network Topology Design
Network topology design is an illustration how the
embedded system devices working in wireless network system. testing the system
The research implemented on Wireless Local Area Network

Yes
(WLAN). Traffic monitoring system topology on wireless
network using Raspbery-Pi shown in Fig. 2. End
Raspberry Pi Fig. 4 System Implementation Flow Chart

Monitoring process
IV. TESTING AND ANALYSIS

User connected to On system testing there are scenario by sending data
access point
packets between user and server so then knew the traffic value
wireless
access point
passed on access point through the wireless network. Testing
Internet process done by two stages that are connectivity testing and
LAN Port 2 LAN Port 1
traffic amount calculation testing on wireless network.
Testing process done by connecting to access point through
Mikrotik
Router wireless network and calculating the traffic amount passed so
then analysing.
Fig. 2 Traffic Monitoring System Topology

____________________________________________________________________________________________________________
A. Connection Testing Difference value of Raspberry-Pi with Mikrotik Router:

Connection testing done for fixing that system devices built Tx = 0.077 MB (0.034) and Rx = 0.018 MB (0.014)
can be connected to internal network, then in a future can Difference value of Raspberry-Pi with The Dude:
Tx = -0.012 MB (-0.005) and Rx = -0.009 MB (-0.007)
monitor towards desired access point. Testing did by
simulation, which is scanning IP address of access point
The Testing 2 mechanism results shown in Table II.
device and pc as user connected through Mikrotik router. The
scanning results shown in fig. 5 and fig. 6. TABLE II. TESTING 2 RESULT
Fig. 5 Scanning IP address by 1 user
Difference value of Raspberry-Pi with Mikrotik Router:

Tx = 0.156 MB (0.014) and Rx = 0.029 MB (0.009)
Difference value of Raspberry-Pi with The Dude:
Tx = 0.006 MB (0.0005) and Rx = 0.002 MB (0.0007)
The Testing 3a mechanism results shown in table III.

TABLE III. TESTING 3A RESULTS
Fig. 6 Scanning IP address by 3 users
B. Traffic Calculating Testing

Traffic calculating testing done to know that system built
was able to give traffic amount calculating results. Traffic
amount calculating test consist of four mechanisms, those are
testing 1: when access point not connected to internet, testing
2: when access point connected to internet, testing 3a: when 1
user browsing, testing 3b: when 3 users are browsing, testing
4a: when 1 user downloading and testing 4b: when 3 users are
downloading.
The Testing 1 mechanism results shown in table I.
TABLE I. TESTING 1 RESULTS
Difference value of Raspberry-Pi with Mikrotik Router:

Tx = 0.225 MB (0.009) and Rx = 0.060 MB (0.014)
Tx = 0.146 MB (0.006) and Rx = 0.026 MB (0.006)
The Testing 3b mechanism results shown in Table IV.

____________________________________________________________________________________________________________
TABLE IV. TESTING 3B RESULTS Difference value of Raspberry-Pi with Mikrotik Router:
Tx = 0.470 MB (0.003) and Rx = 0.010 MB (0.0005)
Tx = -0.280 MB (-0.002) and Rx = -0.100 MB (-0.005)
From the four testing mechanisms done using Raspberry-

Pi, mikrotik router and the dude. It shown that monitoring has
no big difference results. Raspberry-Pi, mikrotik router and the
dude monitoring results accuracy comparison, i.e.:
Accuracy of Raspberry-Pi with Mikrotik Router
Tx Accuracy
= 1 - (0.034+0.014+0.009+0.003+0.014+0.003)/6
= 1 – 0.013 = 0.9870
Rx Accuracy
= 1 – (0.014+0.009+0.0014+0.0005+0.013+0.0005)/6
Difference value of Raspberry-Pi with Mikrotik Router: = 1 – 0.009 = 0.9910
Tx = 0.080 MB (0.003) and Rx = 0.010 MB (0.0005) Accuracy of Raspberry-Pi with The Dude
Difference value of Raspberry-Pi with The Dude: Tx Accuracy
Tx = -0.110 MB (-0.003) and Rx = -0.070 MB (-0.006) = 1–(0.005+0.0005+0.006+(-0.003)+(-0.0008)+(-0.002))/6
= 1 – (- 0.0007) = 1.0007
The Testing 4a mechanism results shown in table V. Rx Accuracy
TABLE V. TESTING 4A RESULTS = 1 – (0.007+0.0007+0.006+(-0.006)+(-0.002)+(-0.005))/6
= 1 – (- 0.0022) = 1.0022
V. CONCLUSIONS
Conclusions of this research are:
• Capable to design and build traffic monitoring system
on wireless network using Raspberry-Pi that can work
well. Those things proved by system testing to access
point results showed system capable to read traffic
amount on wireless network.
• The traffic amount calculating test results using
Raspberry-Pi have relatively smaller value than the
Mikrotik router and have relatively greater value than
the dude.
• The results monitoring accuracy comparison between
Difference value of Raspberry-Pi with Mikrotik Router: Raspberry-Pi and Mikrotik router shown the value: Tx
Tx = 1.466 MB (0.014) and Rx = 0.137 MB (0.013) accuracy was 0.987 and Rx accuracy was 0.991, while
comparison with the dude shown the value: Tx accuracy
Tx = -0.074 MB (-0.0008) and Rx = -0.023 MB (-0.002)
was 1.0007 and Rx accuracy was 1.0022.
The Testing 4b mechanism results shown in Table VI. REFERENCES
TABLE VI. TESTING 4B RESULTS [1] Rahman, M.D Mostafijur.: Network Traffic Monitoring System Based
On Embedded Linux And Single Board Computer. Malaysia. School
Of Computer and Communication Engineering, Universiti Malaysia
Perlis. 2009.
[2] Kakanakov, Nikolay., Elena Kostadinova.: Using SNMP for Remote
Measurement and Automation. Bulgaria. Department of Computer
Systems and Technologies, Technical University of Sofia. 2007.
[3] Shiddiqi, Ary M., Andhika Panji N.: Sistem Monitoring Jaringan
Dengan Protokol SNMP Menggunakan Piranti Bergerak. Surabaya.
Institut Teknologi Sepuluh Nopember (ITS). 2011.
[4] Pradikta, Reza., Achmad Affandi, Eko Setijadi.: Rancang Bangun
Aplikasi Monitoring Jaringan dengan Menggunakan Simple Network
Management Protocol. Surabaya. Institut Teknologi Sepuluh
Nopember (ITS). 2013.
[5] Stallings, William.: Data and Computer Communications. Prentice
Hall PTR. Page 45. 1999.

____________________________________________________________________________________________________________
Design of An Automation System

for Hydroponic Plantation
Based on the Ebb & Flow Method
Wahidin Wahab1, Andhika Kumara Djaffri 2, Winarsita Gilang Sotyaji3
Department of Electrical Engineering, Universitas Indonesia
Kampus UI Depok, 16424, Indonesia
1
wahidin.wahab@ui.ac.id
2
dhika_djaffri@yahoo.com
3
winar.gilang@hotmail.com
Abstract - Hydroponic is a promising agriculture technology and optimizing hydroponic plants is designed. The system is based
has a wide possibility for improvement. Having with such a on microcontrollers, sensors, and actuators that adjusts pH
concept, some of the farmers in Indonesia or other foreign value, nutrition concentration, and lighting with user friendly
countries, are starting to use this hydroponic technology in their operation and affordable price.
farming routines. Nowadays, hydroponics are not only applied
by farmers but also by some plant hobbyists. II. SYSTEM DESIGN
Though, in real-life application, some users tend to have some
difficulties when it comes to getting the desired result for their A. The Ebb and Flow Method
plants, because of the lack of supervision for nutrition, acidity, or The Ebb & Flow or The Flood & Drain Method is a
lighting, which are essential for hydroponic plantation. In this hydroponic growing system where plants get the water,
paper, we designed a system that can help users to maintain their nutrition, and oxygen through a reservoir that is pumped in to
plants’ nutrition, solution acidity, or the lighting, using the help the media in the grow tray. In a certain time, the water
of microcontroller, sensors, and actuators, and also a user together with the nutrition will be drained back to the
interface that can help the user to input the pH data that they
reservoir through an overflow channel. The flood and drain
desired. The result of this research is to prove that the system
can successfully carry-out the preloaded functions and reach the
period can be adjusted using a timer eliminating the risk of
goal, such as pumping nutrition, lighting adjustment, and plant dehydration or overwatered. Fig. 1 shows the detailed
keeping the value of acidity (pH) well in a long run. diagram of the Ebb & Flow method. [5][6]
Keywords - hydroponic, nutrition, acidity, lighting, automation,

microcontroller, actuator, sensor, user interface
I. INTRODUCTION
The current trend on hydroponics technology in farming is
heading towards a promising growth with applications mainly
Fig. 1. The Ebb & Flow Method
in the vegetable industry [1]. Several Hydroponic methods can
be applied according to resource availabilities and needs,
B. Overall Design
which are not only done by farmers, but also hobbyist in
The system involves the control of pH, nutrition, and
regular households [2][3]. Although the technology is fairly
lighting to support the plant growth. The system also controls
easy to be implemented, household hydroponic farmers
water flow using a water pump that runs with a timer, air
occasionally find it difficult to get good results due to the
concentration in the water reservoir, and stirring motor to mix
negligence of maintaining the constant nutrition value, ground
the pH regulating solution and nutrition in the reservoir. The
acidity, as well as lighting, which are important aspects in the
system is equipped with an LCD display and Keypad for users
farming [4]. For this reason, a technology to support the
to input a required set point for the pH level, while other
maintenance of hydroponic plants automatically, is developed
aspects are controlled automatically in a certain amount of
so that farmers can achieve the highest quality with minimum
time using the microcontroller internal timer. Keypad, LCD
effort. The system aims to broaden the scope of farmers so
Display, and pH sensors are connected directly with the
hydroponics can be done by anyone, anywhere and anytime.
microcontroller (ARDUINO UNO) while the micro gear
In this paper, an automation system that can simplify
pumps are powered by an external DC source of 12V and the
hydroponic farming in monitoring, maintaining, and
aerator is controlled by an external AC source (220V).

____________________________________________________________________________________________________________
PC81
7 TIP120
Fig 5. Micro Gear Pump Control Electronic
The system use optocouplers PC817 for isolating the logic

Fig. 2. Overall System Diagram circuit (Arduino) from the power circuit (Mini Micro Gear
Pump DC RS-360SH) so that reverse current from the motor
does not damage the microcontroller. TIPI120 transistors are
used to control the power current of the motor using low
powered signal from the microcontroller.[7]
3) LED Strip Control Configuration: The

configuration in fig. 6 controls the color spectrum emitted by
the LED, its intensity, as well as the duration of the lighting.
Fig 3. Actual System Display Fig 6. LED Strip Electronic Configuration
C. Electronic Design
1) User Interface Configuration: The configuraton in 4). Water Pump Control Configuration
Fig.4 shows how the LCD and Keypad is connected to the
microcontroller The configuration in fig. 7 below is made to control the on/off
period of the pump using a timer generated inside the
microcontroller. The power source of the pump is a 12V DC
connected through a relay driven by a seried transistor to be
controlled by the low powered signal from the
microcontroller.
Fig. 4 User Interface Electronic Configuration
The design uses I2C Serial Interface for the LCD Display
with analog pin A4 and A5 connected.
2) Micro gearp pump control configuration: The micro

gear pump used to control the pH level is connected with three
bottles (nutrition, pH - UP, pH - Down) Fig 7. Water Pump Control Electronic Configuration

____________________________________________________________________________________________________________
D. Control System Design III. EXPERIMENTAL RESULTS

1) pH Control: The pH Sensor senses the pH value A. pH Control Observation
through levels of voltage from -414.2 mV to 414.2mV. The The proposed system has successfully managed to read the pH
level of voltage that the sensor measures are then converted to value of three different kinds of solution. The graph in fig. 10
pH value according to table 1. shows that the pH reading from the sensor has a settling time
TABLE I at about 10 seconds and considered quick enough for control.
RELATIONSHIP BETWEEN PH AND VOLTAGE LEVEL DETECTED BY THE SENSOR
The data of pH that is acquired from the sensor are then

compared with the desired pH set point that the user inputs in
the keypad as such in fig. 8 The sensor reads the pH in 20ms
interval and updates the voltage as well as the pH level
display in the LCD in 800ms interval [8]. When the actual pH
is higher than the set point, the pump will release Phosphoric
Acid (H3PO4) 10% solution, and when it is lower than the set Time(s)
point it will release Sodium Hydroxide (NaOH) 10% solution. Fig. 10 graph of pH reading of NaOH(top), water(middle), and
H3PO4(bottom)
Controller
Fig 8. pH control closed-loop block diagram
2) Nutrition Control: The system controls the nutrition

without the use of any sensor. It controls the level of nutrition
given via interrupt timer for every 12 hours with 3 seconds
actuating time. The duty cycle used in every actuation is
100%.
Time(s)
Controller
Fig. 11 graph of pH regulation by the automation system
Fig 9. Nutrition control block diagram Fig. 11 shows the system managed to automatically adjust
the pH to reach a set point of 6.1. The pH achieved the desired
3) Lighting Control: Like nutrition control, the LED value with a settling time of around 100 seconds. The micro
system uses an open loop control system without any sensors. gear pump actuator pumps mainly H3PO4 solution in the
Using timer, the LED changes colors every 10 days from blue experiment to achieve the set point. Each pump running for 3
(vegetative period), to red (generative period), and to purple seconds add about 2ml-3ml of the solution and the mixer is
(flowering period). Then the system repeats the LED used to help distributed it uniformly in the reservoir.
sequence every 30 days.
B. Nutrition Control Observation
Controller The graph of the experiment in fig. 12 shows the relationship
between the PWM value vs time for the gear pump to regulate
the amount of nutrition. For simulation purposes, the gear
Fig 9. lighting control block diagram
pump is made running for 1 second in every 2 hours (7200
seconds). It is shown that the system manages to automate the

____________________________________________________________________________________________________________
nutrition regulation in the desired amount of time period as IV. CONCLUSIONS

expected. In this paper, Automation System for Hydroponic
Plantation Based on Ebb & Flow Method is designed and
build. The system shows promising automated control of
the aspects necessary in hydroponics farming using the
Ebb & Flow method (pH, nutrition, lighting, and water
flow control), providing the simplification hydroponics
farming process. There is still yet many room for
improvements in the system such as more accurate
controlling of the pH using PID or Neural Network and
closed loop control of other aspects beside the pH may be
developed in the future.
REFERENCES
[1] (2014) PRNewswire. [Online]. Available:
http://www.prnewswire.com/news-releases/growth-in-the-hydroponics-
food-industry-set-to-outpace-global-markets-by-80-241264701.html
[2] Karsono, Sudibyo, Sudarmodjo dan Yos Sutioso. 2002. Hidroponik
Time(s) Skala Rumah Tangga. Jakarta : Agro Media Pustaka
[3] Karsono, Sudibyo, et al. 2002. Hidroponik Skala Rumah Tangga.
Fig. 12 graph of Nutrition regulation by the automation system Jakarta : Agro Media Pustaka
[4] Sutioso, Ir. Yos. 2004. Hidroponik Ala Yos. Jakarta : Penebar Swadaya
[5] (2016) SA Horticulture [Online]. Available:
C. LED Strip Performance http://sahorticulture.co.za/six-hydroponic-systems/
[6] Resh, Howard M. 1987. From Hydroponic Food Production.
For simulation purposes, instead of every 10 days we simulate Woodbridge Press
the period of lighting to 10800 seconds (4 hours). The system [7] Kleitz, William. 1996. Digital Electronics A Practical Approach.
shows that the automation of the lighting has been done Fourth Edition. New Jersey : Prentice-Hall International, Inc.
[8] Bentley, John P. 1995. Principles of Measurement System. Third
successfully, marked by the graph in fig. 13. From the first Edition. Singapore : Longman Scientific & Technical
second until the 10799th second the red LED is ON using
maximum PWM then descends in the 10800th second. The
blue LED then turns on from the 10800th second until the
21599th second. Finally, the red LED turns on again together
with the blue LED to create purple spectrum in the 21600th
second until the 32400th second.
Time(s)
Fig. 13 graph of LED strip performance

____________________________________________________________________________________________________________
APT Detection Design

Using DNS Traffic Analysis
Muhammad Salahuddien Manggalanny#1, Kalamullah Ramli#2
Department of Electrical Engineering, University of Indonesia
Kampus Baru UI Depok 16424 Indonesia
1
aku@pataka.net
2
k.ramli@ee.ui.ac.id
Abstract— Advanced Persistent Threat (APT) combined different detection process, IDS compared security parameter databases
types of targeted vulnerability exploits to the specific object using with actual condition to determined anomaly or rules violation.
malware. This kind of attack would be extremely dynamic, so that For misuse detection, IDS uses pre-defined signature or rule
conventional anomaly signature-based detection system will failed set of known anomalous characteristic. Effective in a real time
to identify. To reduce false positive/negative detection result, this
system, but signature-based IDS has fundamental drawbacks.
combination method design proposed more effective DNS Traffic
based analysis as an advancement model for APT detection. Rule sets must be updated to follow dynamic changes of threat.
To narrow time gap between the emergence of new threat and
Keywords— APT, Anomaly Detection, DNS traffic analysis rule update speed, modern IDS connected to the Global Threat
Management System. An online signature reference exchange
I. INTRODUCTION to increase recognition time, prevent new threat from spreading.
Based on Annual Internet Security Report in 2014 and 2015 B. APT Detection
released by ID-SIRTII/CC, shown a significant increase of
Unlike generic attack techniques, targeting widespread non-
APT threat through malware combined attack and recently
specific object that are more easily to prevent (predictable, high
achieved second largest position in term of volume.
profile action) through perimeter security (firewall, IDS, IPS)
SECURELIST website, indicating continuous evolution of
countermeasures. APT will deceived any kind of conventional
APT modus. It directly affect ability of detection technologies
intrusion detection through insider threat (people) weaknesses
to adapt rapid changes. APT are more targeted, sophisticated,
and social engineering. APT aimed for undetected long term
following the dynamic of object behaviour. Signature-based
exploitation and maintain covert remote access backdoor.
IDS received high level of false result, failed to identify APT.
Conventional detection method (rule-based, pre-defined
Malware used by the attackers was modified every time it
pattern signature database) are not effective to identify APT. It
launch and it aimed to the object with a very personal profile.
required long-term analysis the entire system and network logs.
By speeding up formulation of signature-based update will not
Including behaviour profiling investigation to identify anomaly
match APT growth and malware evolution acceleration rate.
activity correlates with APT characteristic. Combination of log
More specific detection method needed to address this issue.
analysis tools: SIEM, IDS, can be used with forensic analysis
II. ADVANCED PERSISTENT THREAT techniques and sandboxing to investigate suspected APT traffic.
Another workaround is: APT threat are closely associated to
Is a method of cyber-attacks with stealth level (the ability to malware distribution pattern and botnet activity. An effort of
hide and avoid detection) to maintain access and control in a identification, mapping out DNS query traffic pattern - which
long term by combining variety of techniques to achieve are directly representing user behaviour characteristics - can be
highest success rate of very specific target compromising.
used as promising alternative technique to investigate user
APT interest can be expanded later, after successful system
activity, specifically correlated with malware and botnets.
compromised and financial goal was reached, allows repeated
future exploitation. APT did not cause interference, disruption,C. APT Classification
damaging target system. Detectable or identifiable symptoms Classification is a mathematical model to assess and classify
of compromising are very rare to be found. data (cluster) into a number of data classes. Generally defined
A. Intrusion Detection as the process of identifying target function “f” that maps each
set of attributes (features) “x” into one of classes that have been
Generic IDS uses structured packet data (header, payloads) determined as “y”. The results of this process can be used to
analysis to determined malicious activity (anomaly and misuse) predict and define a new expected class, i.e. malicious DNS
based on pre-defined category: normal, malicious and unknown.
traffic patterns related to malware distribution and botnet
In order to identify anomaly, IDS works in two phases which
activity that are strongly indicating a presence of an APT attack.
are learning and detection. In the learning process, IDS draw a
Some of popular Data Mining classification algorithm are:
profile and behaviour, based on event log historical statistic
Artificial Neural Network (ANN), K-Nearest Neighbour (K-
model to determined normal or unwanted activity. In the
NN) and Support Vector Machine (SVM). Most of formula was

____________________________________________________________________________________________________________
widely used to detect general anomaly traffic. For DNS traffic In (2012) research reported botnet DNS traffic identification
analysis, classification formula that are effectively proven is based on the group's activities [7]. In (2014) similar approach
Shared Nearest Neighbour (SNN) clustering model [14]. was conducted with Kindred Domains technique [15]. In (2014)
mapped DNS traffic behaviour based on properties [11]. Recent
D. SNN Formula study in (2015) observed the group's activities and errors
Domain name continue to grow, not possible to study DNS history in DNS traffic logs to detect botnet-flux domain [21].
query traffic characteristics manually. Automatic clustering is APT detection methods today, more focused on DNS traffic
required to classify domain that has DNS query traffic similar behaviour research. Study in (2007) found a correlation of DNS
to one group (cluster) of characteristics. The DNS query traffic traffic behaviour with the motivation behind them [3]. The
that does not have the same characteristics will be separated to observations made in (2011) detect the presence of malicious
another groups (cluster). This model is to simplify analysis, i.e. domains based on DNS traffic behaviour analysis [5]. More
extraction of specific user behaviour characteristics, identifying specifically research in (2013) conducted an empirical analysis
an anomalous behaviour characteristics such as APT. of APT malware specifically targeting Internet Banking [10].
The use of SNN clustering formula to find DNS query traffic In (2013) research exposes DNS service involvement in
pattern anomaly was introduced in (2013)[14]. This approach, malware activity Command and Control (CNC) [9]. While
obtained more accurate grouping, provide information insight research in (2014) shows the DNS-based security approach are
to identify malicious traffic. Combining with dynamic analysis, very effective to employ in a large-scale network [17].
sandboxing [6], backscatter [19]. This model can also be used Alternative detection techniques through DNS observation
for another DNS security research, i.e. DGA botnet activity approach was developed recently. Studying the taxonomy of
profiling. SNN has the flexibility to handle high volume of data, the motivations behind the behaviour of DNS related threats
high speed large networks, significantly reduce computation. was doing in (2007) [3]; classification techniques (clustering)
DNS query traffic based on context awareness also conducted
E. Data Processing
later in (2008) [4]; and tracking the characteristic pattern of
Another strong point for SNN classification – compared to behaviour-based access control traffic carried in (2013) [13];
Gaussian Dataset and Jarvis – Patrick Clustering, is it ability to with DNS query traffic pattern recognition in (2013) [14].
handle high dimensional data. Which is match to DNS query We can estimates characteristics of DNS query access using
traffic characteristic – that has many (domain name) object, active probing techniques (2015) [20]; the latest APT malware
various query time interval, complex parameter, variable infection techniques detection using DNS in (2015) [18]; has
dimension, character attributes (related to suspected activity). contribute am increase of detection capability to identify where
With SNN, degree of similarity between two objects can be the APT activity is. It is an effective tools for APT analysis.
measured with the proximity of other objects neighbouring and However, it turns out that, approach based on the detection
intersect with it. The density of an object is defined by the of malicious DNS traffic analysis also has several problems and
number of similarities (characteristics) with the closest downsides to consider, as revealed in a research in (2015) [22].
neighbour. SNN will eliminate noise (objects with low density) To get a better visibility in DNS traffic analysis, research in
and form new cluster by associating an object that does not (2014) offers a graphical visualization models called GMAD
have a noise with the main object (core, objects with high [16]. Visualization modelling with different data bases
density). combine with log mining techniques also offered with different
Previous study shows level of DNS query traffic noise on approach (2015) [23]. Traffic Visualization is not something
large scale live network is very high. Resulting temporal auto new. In fact in (2006) it has been used to facilitate DNS traffic
correlation phenomenon. i.e. volume of traffic repeated in short analysis [2].
time, tend to be correlated, resulting similar measurement bias.
To reduce these symptoms, it is necessary to do normalization IV. CONTRIBUTION
of data among other z-score transformation formula [14]. Previous study showing successful result in detecting APT
Malware activity through properties characteristic method [18]
III. PREVIOUS STUDY
and detecting APT Malware for authoritative NS through DNS
To track and identify malware can be done in several ways: Backscatter analysis [19], botnet detection based on the history
by analysing traffic packet in-depth as has been reported in of group activities and failures in DNS traffic [21].
research (2011) [6]; using SANDNET (malware environment More deeper, using Behaviour Clustering method to analyse
analysis platform). With this technique, researchers conduct Malicious Domain Behaviour [5] and DGA profiling, hierarchy
dynamic malware analysis in a longer time span, especially to clustering, for DNS recursive and authoritative services [15].
observe the behaviour of DNS and HTTP protocols. Then, sorting suspected malicious activity by Labelling DNS
Malicious activity associated with DNS traffic activity, also traffic, and do evaluation based on black/white list [22].
can be detected using DNS backscatter technique as has been To justify more accurate Threat Detection we study previous
reported in research (2015) [19]. Previously, research in (2005) successful approach in Bro IDS, protocol analyser and mapping
[1] was applied data mining techniques to identify DNS traffic properties [11] and SNN clustering, trend prediction [14] with
pattern indicating the existence of worm infected hosts that Active probing, for real world traffic implementation [20].
does unsolicited mass-mailing (SPAM).

____________________________________________________________________________________________________________
Most of study conducted in a limited, simulated environment In the second part of the experiment, we conduct:
i.e. small campus network. Proposed design continue previous • Processing RAW data: normalisation, implementing
works in a live data at national scale network using combination SNN formula combined with reputation engine
of experiment: APT Malware Activity Detection, Clustering • Examining detection result through comparison analysis
Analysis and Threat Prediction in order to have better visibility with an existing detection system as a validation method
for sophistication of APT Malware specifically in Indonesia • There are two potential existing benchmark source as
Network and to reduce False Positive/Negative misdetection. comparison which is Big Data from DAMBALLA or
Preliminary test on existing ID-SIRTII Mata Garuda IDS by SPLUNK and event correlation analysis from OSSIM.
implementing DNS attributes characteristic rules adjustment, If the final result is better, then the analysis was concluded.
shows promising result which is an increase of successful DNS
malicious traffic detection from March to June 2016 period. A. Data Validation
TABLE I Clustering analysis process using SNN will classify object
DNS BAD TRAFFIC DETECTION with pre-defined anomalous characteristic. The result expected
is a better identification design and to map the existence of APT
January February March
malware and its CNC. To examine accuracy, effectivity and
Port % Port % Port %
performance of this experiment in the real world, we validate
53 7% - - 5060 44%
and compare the result with an existing benchmark source
4444 4% - - 1701 14%
which is Big Data analysis from DAMBALLA or SPLUNK
1434 3% - - 53 7%
and event correlation analysis OSSIM hosted by ID-SIRTII/CC.
April May June
Port % Port % Port % B. Reputation Engine
53 77% 53 53% 161 48%
After RAW data being normalized and clustered with SNN
161 17% 161 39% 53 40%
method, we sanitize the result with existing Reputation Engine
5060 2% 2055 2% 2055 3%
hosted by ID-SIRTII/CC that employ pre-design filter based on
V. DETECTION DESIGN anomalous attribute characteristic. If the result data match with
some of those attribute, it means strong indication of an APT
Our research framework to detect APT malware divided into malware and its CNC existence. There are three attribute label:
two part of experiment: data collection, analysis (evaluation).
TABLE II
WHITE LIST ATTRIBUTE
No Rule Set
1 .id domain names
2 Indonesia IP allocation
3 Used by exclusion list
TABLE III
BLACK LIST ATTRIBUTE
No Rule Set
1 Used by anti-virus software
2 Used by anti-SPAM RBL
3 Used by DNSBL, Phishing list
4 RFC/standard best practice violation
5 Unknown TLD, NXDOMAIN, DGA
6 SERVFAIL error, Refused error, others error
7 REGEX, random words, name string filter
8 Reported event quarantine data
Fig. 1 Detection System Configuration Design
TABLE IV
ANOMALY ATTRIBUTE
The first is data collection process:
• DNS query traffic RAW data obtained from IDS (Mata Feature Set No Feature Name
Garuda Project) installed at DNS NAWALA, an open Domain Name 1 Contain famous name
resolver and filtering system to represent DNS resolver 2 Contain particular name
• Other DNS query traffic RAW obtained from IDS (Mata 3 Contain phishing name
Garuda Project) installed at PANDI– the CCTLD of .id DNS Answer 4 Silent IP
domain, .id-root, L-root to represent DNS recursive data 5 Number of distinct resolved IP
The second data collection will be obtained from the DNS 6 Number of distinct countries
7 Number of domains share the same IP with
backscatter produced by root-DNS (.id-root and L-root).
8 IP in same Class B of known CNC server
The third data collection will be obtained from DNS error
Time Value 9 Daily similarity
log that has been hosted at ID-SIRTII/CC.
10 Same query numbers in same time window

____________________________________________________________________________________________________________
11 Very low frequency query [2] P. Ren, J. Kristoff, and B. Gooch, "Visualizing DNS traffic," presented
TTL Value 12 Average TTL value at the Proceedings of the 3rd international workshop on Visualization
for computer security, Alexandria, Virginia, USA, 2006.
Active Probe 13 Web server or others [3] N. Chatzis, "Motivation for Behaviour-Based DNS Security: A
14 WHOIS information Taxonomy of DNS-Related Internet Threats," in Emerging Security
Information, Systems, and Technologies, 2007. SecureWare 2007. The
C. Comparison: Anomaly Matrix Evaluation International Conference on, 2007, pp. 36-41.
First is to map the calculation output into existing anomaly [4] D. Plonka and P. Barford, "Context-aware clustering of DNS query
traffic," presented at the Proceedings of the 8th ACM SIGCOMM
detection matrix, using predicted labels and true labels: conference on Internet measurement, Vouliagmeni, Greece, 2008.
[5] S. Hao, N. Feamster, and R. Pandrangi, "Monitoring the initial DNS
Precision = TP / (TP+FP) behavior of malicious domains," presented at the Proceedings of the
Accuracy = (TP + TN) / (TP + TN + FP + FN) 2011 ACM SIGCOMM conference on Internet measurement conference,
Berlin, Germany, 2011.
[6] C. Rossow, C. J. Dietrich, H. Bos, L. Cavallaro, M. v. Steen, F. C.
Then evaluate with Matthews Correlation Coefficient: Freiling, et al., "Sandnet: network traffic analysis of malicious
software," presented at the Proceedings of the First Workshop on
TP x TN − FP x FN Building Analysis Datasets and Gathering Experience Returns for
𝑀𝐶𝐶 = Security, Salzburg, Austria, 2011.
(TP + FP)(TP + FN)(TN + FP)(TN + FN) [7] H. Choi and H. Lee, "Identifying botnets by capturing group activities
in DNS traffic," Computer Networks, vol. 56, pp. 20-33, 1/12/ 2012.
[8] S. Marchal, J. Francois, C. Wagner, R. State, A. Dulaunoy, T. Engel, et
TAPI = True Positives, FP = False Positives, TN = True al., "DNSSM: A large scale passive DNS security monitoring
Negatives and FN = False Negatives; output anomaly or normal. framework," in Network Operations and Management Symposium
(NOMS), 2012 IEEE, 2012, pp. 988-993.
TABLE V
[9] X. Kui, P. Butler, S. Saha, and Y. Danfeng, "DNS for Massive-Scale
ANOMALY DETECTION RESULT MATRIX
Command and Control," Dependable and Secure Computing, IEEE
TRUE FALSE Transactions on, vol. 10, pp. 143-153, 2013.
[10] Andr, #233, R. A. Gr, #233, gio, D. Sim, et al., "An empirical analysis
POSITIVE ( + ) Rule match; detected Rule match; not of malicious internet banking software behavior," presented at the
detected Proceedings of the 28th Annual ACM Symposium on Applied
NEGATIVE ( – ) Not match, not Not match, Computing, Coimbra, Portugal, 2013.
detected detected [11] T. Callahan, M. Allman, and M. Rabinovich, "On modern DNS behavior
and properties," SIGCOMM Comput. Commun. Rev., vol. 43, pp. 7-15,
2013.
If resulted experiment producing more True Positive and/or [12] Y. Kazato, K. Fukuda, and T. Sugawara, "Towards classification of DNS
reduce False Positive/Negative findings; then it was concluded. erroneous queries," presented at the Proceedings of the 9th Asian
Internet Engineering Conference, Chiang Mai, Thailand, 2013.
VI. CONCLUSIONS [13] D. Herrmann, C. Banse, and H. Federrath, "Behavior-based tracking:
Exploiting characteristic patterns in DNS traffic," Computers & Security,
Drawback of signature-based detection approach is inability vol. 39, Part A, pp. 17-33, 11// 2013.
to identify new threats that have not been found, not yet defined. [14] W. Ruan, Y. Liu, and R. Zhao, "Pattern Discovery in DNS Query
Growth of malware continue to evolve dynamically, and more Traffic," Procedia Computer Science, vol. 17, pp. 80-87, // 2013.
complex using encryption, employing VPN / peer to peer tunnel [15] M. Thomas and A. Mohaisen, "Kindred domains: detecting and
clustering botnet domains using DNS traffic," presented at the
access to the CNC. This profile are not yet implemented into Proceedings of the 23rd International Conference on World Wide Web,
current dataset reference (DARPA KDDCUP ’99) as primary Seoul, Korea, 2014.
source of most research evaluation for the past of 15 years. [16] J. Lee and H. Lee, "GMAD: Graph-based Malware Activity Detection
Most of anomaly detection approach can be directly applied by DNS traffic analysis," Computer Communications, vol. 49, pp. 33-
47, 8/1/ 2014.
to the DNS traffic study. Machine learning IDS, classification [17] M. Janbeglou, H. Naderi, and N. Brownlee, "Effectiveness of DNS-
of anomalous behavior characteristics, labeling, DPI, dynamic Based Security Approaches in Large-Scale Networks," in Advanced
analysis (sandbox, forensics), reputation engine (black/white Information Networking and Applications Workshops (WAINA), 2014
list). Recent study demonstrated effectiveness of combination 28th International Conference on, 2014, pp. 524-529.
[18] G. Zhao, K. Xu, L. Xu, and B. Wu, "Detecting APT Malware Infections
techniques to improve detection accuracy of APT malware. Based on Malicious DNS and Traffic Analysis," Access, IEEE, vol. 3,
This design proposes new combination approach that are pp. 1132-1142, 2015.
more effective to detect DNS Malware and CNC activity. [19] K. Fukuda and J. Heidemann, "Detecting Malicious Activity with DNS
Backscatter," presented at the Proceedings of the 2015 ACM Conference
ACKNOWLEDGMENT on Internet Measurement Conference, Tokyo, Japan, 2015.
[20] X. Ma, J. Zhang, Z. Li, J. Li, J. Tao, X. Guan, et al., "Accurate DNS
This works are partially supported by ID-SIRTII/CC in a query characteristics estimation via active probing," Journal of Network
favour to enhance core IP infrastructure resilience in Indonesia. and Computer Applications, vol. 47, pp. 72-84, 1// 2015.
[21] R. Sharifnya and M. Abadi, "DFBotKiller: Domain-flux botnet detection
REFERENCES based on the history of group activities and failures in DNS traffic,"
Digital Investigation, vol. 12, pp. 15-26, 3// 2015.
[1] K. Ishibashi, T. Toyono, K. Toyama, M. Ishino, H. Ohshima, and I. [22] M. Stevanovic, J. M. Pedersen, A. D'Alconzo, S. Ruehrup, and A.
Mizukoshi, "Detecting mass-mailing worm infected hosts by mining Berger, "On the ground truth problem of malicious DNS traffic
DNS traffic data," presented at the Proceedings of the 2005 ACM analysis," Computers & Security, vol. 55, pp. 142-158, 11// 2015.
SIGCOMM workshop on Mining network data, Philadelphia, [23] Q. Lai, C. Zhou, H. Ma, Z. Wu, and S. Chen, "Visualizing and
Pennsylvania, USA, 2005. characterizing DNS lookup behaviors via log-mining," Neurocomputing,
vol. 169, pp. 100-109, 12/2/ 2015.

____________________________________________________________________________________________________________
Design of Sensor Reading Visualization on

Line Follower Robot
Akhmad Zainuri 1, Ahmad Sirojuddin 2, Ponco Siwindarto 3
1,2,3
Department of Electrical Engineering, Faculty of Engineeing, Brawijaya University,
1 akhmad.zainuri@ub.ac.id
Jln Veteran 65145, Malang, Indonesia
Abstract— Line follower robot is a robot that runs follow the line so can be display on personal computer. The system can
on the track. It has a line sensor that can sense the position of the displays every unit of sensor reading on PC screen. The
robot relative to the line. Frequent errors by the line sensor readout system consist of several part: line sensors, microcontroller to
caused by various factors both from within and from outside. The process data and converts it into data packets, data
troubleshooting process in this line sensor reading is difficult
transmission device, and computer program as a viewer the
because of the results that is rapidly changing. Therefore, in this
research a visualization system that can record sensor readings and result of the readings.
visualize the dynamics of the line sensor readings on the PC has
been designed. The system consists of several parts, including a line II. SYSTEM DESIGN
sensor as the input of the system, the microcontroller to process A. Diagram Block System
data readings and converts it into data packets, bluetooth module
HC-05 as the sender of data, visual and computer program as a The system diagram block for all system part is shown at
viewer of the readings. figure 1.
Keywords— Line sensor, wireless communication, visualization,

ADC.
I. INTRODUCTION
Line follower robot is a robo car that can follow a path. The
path can be visible like a black line on the white surface (or
vice-verse). It is an integrated design from the knowledge of
Mechanical, Electrical and Computer engineering [1]. The
specifications of the line are in terms of its width and its color
with respect to the color of the background. It presents a Fig. 1 Diagram Block System
comprehensive idea using which the threshold value for the
detection of the line from the background [2]. Therefore, this Line follower robot position relative to the line is read by
sensor role is very important for the robot. Unfortunately, the line sensor. In this research, the sensor line consists of 12
error reading often occur when this sensor works. Something units TCRT5000 light sensor circuit. Each TCRT5000 unit
that causes this problem is ADC value read by sensor is light sensor circuitry generates a certain voltage value which
different with actual value. represents the amount of reflected light received by the
Robot designer difficult to detect or troubleshoot these receiver section in TCRT5000 light sensor. Twelve analog
problem. This all are caused by that line sensor reading data coming from the line sensor is read by the
process is not visible by human eye. Other problem is microcontroller ADC pin. Before reading the ADC value, the
difficult to implement measuring instrument when robot is at microcontroller first set on a multiplexer selector pin IC on the
track, especially when robot runs following a line. line sensor to select the sensor that will be read by the
Teleoperated applications controlling a movable robot microcontroller ADC pin. Then the analog data from each
prototype in an easy, direct way, as this prototype must move circuit unit TCRT5000 read by the microcontroller, the data is
in places with difficult access and/or human-health processed and formed into packets of data in the form of
threatening, through wireless communication such as via string. This string is then transmitted by the microcontroller to
bluetooth communication [3]. The technology development in the Bluetooth module HC-05 are connected via USART serial
mixed analog-digital signal system and wireless communication via pin TX. The data packet is then
communication provide advantages in instrumentation design transmitted by the bluetooth module and received by the
[4]. Bluetooth module HC-05 that is connected to the computer.
Therefore, line sensor reading visualization system is Both Bluetooth module has been paired before. After received
designed to read the dynamical changing of line sensor value by the computer, the data packets broken down into units of

____________________________________________________________________________________________________________
data each light sensor TCRT5000. The data is then visualized D. Software Designing for The Microcontroller
on a computer screen with several modes of visualization Microcontroller has two modes during operation:
using software that was created earlier. Not only be displayed calibration mode and data transmission mode. Flowchart for
on PC's screen, but these data can also be saved into a data both mode is shown in Figure 4.
logger file during the visualization process. At other times, the
data logger files can be visualized back on the computer
monitor to ease further analysis.
B. Line Sensor Circuit Designing
Light sensor circuit schematic and multiplexer are shown in
Figure 2.
Fig. 2 Light Sensor Circuit Schematic
The sensor consists of 12 units TCRT5000. The TCRT5000

transmitter is connected with 330 Ω resistor to produce an IR
light. The TCRT5000 receiver is connected with 10 kΩ to
produce a voltage output difference when it gets a bit and a lot
of IR light. The twelve output voltage is connected with two
multiplexer IC 4051 (every one multiplexer is connected with Fig. 4 Microcontroller Flow Chart for calibration mode (a) and data
6 output voltage) and common channel multiplexer is transmission mode (b)
connected with a variable resistor in order to drop the voltage
The purpose of the calibration mode is to get the black and
linearly from 0-5 V to 0 to 3.3 V.
white threshold value that used as a reference in determining
C. Bi-directional Logic Level Shifter Circuit Designing the logic condition sensor. These threshold values are then
transmitted to a computer via serial communication. Data
Bi-Directional Logic Level Shifter circuit is shown in
transmission mode is a mode when the microcontroller
Figure 3. This circuit is needed so that the output voltage of
sending the line sensor readings to a computer so it can be
3.3 V microcontroller can control multiplexer I/O pin of
visualized on a computer's monitor. The microcontroller
CMOS IC that has VIH 5 V. FET 2N7000 is used because it
continuously read sensor line, converting the data into a data
has a threshold voltage below 3.3 volts and is suitable for fast
packet, and sends it to the Bluetooth module through the
switching applications [5] .
USART communication.
E. Software Designing for Visualization
This research uses Java SE 8 as GUI computer program for
visualization [6]. The computer program has two modes of
operation: a data stream visualization mode and visualization
of data logger mode. Flow chart of the two modes is shown in
Figure 5. The basic difference of the two modes is the source
of the data to be visualized on a computer monitor. In the data
streams visualization mode, data shown on monitor screen is
Fig. 3 Bi-Directional Logic Level Shifter Circuit the data sent by the microcontroller via wireless
communication.

____________________________________________________________________________________________________________
B. Testing of Line Sensor Accuracy and Precision.

The purpose of this test is to determine the accuracy of
sensor readings visualization system that has been designed.
Sensors are placed such that only two sensors in the center
right on the line. While the other sensor is above a white
surface. The test is performed 20 times. Based on data
obtained from testing, the accuracy of readings to the black
line at 95%. To determine the level of line sensor precision
that has been designed. Magnitude which is used to measure
the level of precision in this test is the coefficient of variation
[7]. Tests are done on the white surface of track. In each
experiment, an ADC data value taken 5 times, at different
places on the track. The test result of data precision level of
line sensor reading are shown in Table 1.
TABLE 1 DATA RESULT FROM PRECISION TESTING WHEN SENSOR IS PLACED
ABOVE WHITE SURFACE
ADC Value Standard

C.V. (%)
1 2 3 4 5 Deviation
0 41 40 40 40 41 0,54772 1,35575
1 41 40 41 40 40 0,54772 1,35575
2 21 21 22 21 21 0,44721 2,10950
3 20 20 20 19 19 0,54772 2,79450
4 40 39 40 39 39 0,54772 1,39016
Sensor
5 41 41 41 41 41 0,00000 0,00000
6 41 41 41 40 41 0,44721 1,09611
7 21 21 22 21 22 0,54772 2,55945
Fig. 5 Computer Program Flow Chart for Stream Data Visualization Mode 8 40 39 41 39 39 0,89443 2,25865
Computers are constantly waiting for data sent from the 9 22 21 22 21 21 0,54772 2,55945
microcontroller. If a data packet is received, then the data is 10 39 39 39 39 39 0,00000 0,00000
broken down into data for each sensor unit. Furthermore, the 11 22 21 21 21 22 0,54772 2,55945
Mean 0,47 1,67
data is visualized on a computer monitor. These data are also
stored in the data logger files if this feature is enabled. While Based on Table 1 it can be concluded that each sensor unit
the data logger visualization mode, data shown on computer has a level of precision that is different from one another. The
screen shown is data from sensor readings stored on the average value of coefficient of variation for the white surface
computer. is 1.67%. This value is good enough to be applied to the line
sensor line follower robot.
III. RESULTS AND ANALYSIS
A. TCRT5000 Light Sensor Circuit Testing C. Visualization of Line Follower Robot Sensor Readings
This testing is for knowing sensor output voltage when Figure 7 displays the data streams transmitted by the
placed on black and white surface. TCRT5000 light sensor microcontroller to computer during data transmission. This
testing circuit is shown in Figure 6. Based on data obtained image is similar to the virtual terminal software such as
from test, the average voltage for a black surface is 2.67 V and Parallax or TeraTerm. However, this data stream visualization
160.9 mV for white. program designed specifically for the visualization of sensor
values robot line follower. The x-axis is an index of data
packets, while the y-axis is the value of the ADC. In this
graph, there are 12 curves showing the line sensor readings
from the beginning to the end of the transmission. There is
also a cursor which helps users to do the reading chart.
Fig. 6 TCRT5000 Light Sensor Testing Circuit

____________________________________________________________________________________________________________
Fig. 9. Animation display line sensor readings
Fig. 7 Visualization graph of line sensor value

IV. CONCLUSION
According to the design, measurement and analysis it can
Each unit light sensor has three pieces of data value, white be concluded that the device can run very well in accordance
threshold, black threshold, and the ADC value. White with the design. Based on the testing results and analysis of
threshold value represented by the red vertical bars. Black this study, it can be concluded that,
threshold value is represented by the blue vertical bars. While 1. The line sensor has been designed has the following
the value of ADC that is being read is represented by the characteristics: has an accuracy rate of 95% and the
green vertical bars as shown in Figure 8. level of precision that is quite good with a value of
coefficient of variation of 1.67%.
2. Based on the images of the results of testing the
overall system, it can be concluded that the results of
this research have been run well.
REFERENCES
[1] Islam M. S. and Rahman M. A.2013. Design and Fabrication of Line
Follower Robot. Asian Journal of Applied Science and Engineering,
Volume 2, No 2 (2013). ISSN 2305-915X
[2] Supe ,T. and Joy, A. Design of a line following sensor for various line
specifications . International Journal of Computer Engineering and
Technology (IJCET), Volume 4, Issue 6, November - December (2013),
[3] Silva, J.F.M.C. Santos, D.M.S. Marques, V.C. Oliveira, K.D.
Rodrigues, T.O. Texeira, R.G.F. Menezes, J.W.M. Silva, F.D. (2012).
A Study of Bluetooth Application for Remote Controlling of Mobile
Embedded Systems. Computing System Engineering (SBESC), 2012
Fig. 8. Bar chart of visualization sensor line value Brazilian Symposium on, pp. 116 - 116.
[4] A. Sudarmaji, A. Kitagawa, and J. Akita, “Design of wireless
measurement of soil gases and soil environment based on
Figure 9 is an animated display line sensor readings. There Programmable System-on-Chip (PSOC),” in Proceedings of the
are 12 pieces of blue dots that represent the position of each International Symposium on Agricultural and Biosystem Engineering
sensor light. Each blue dot is in the box that can be white, (ISABE ’13), pp. E5-1–E5-13, 2013.
[5] Schutte, Herman. 1997. Bi-Directional Level Shifter for I2C-Bus and
black, or red. If a sensor is declared to be above the white Other System – Application Note. Eindhoven : Philips Semiconductor.
surface by a visualization system, the box surrounding the [6] Schildt, Herbert. 2007. The Complete Reference Java Seventh Edition.
blue dot will appear white. If a sensor is declared to be above New York: McGraw-Hill Companies Inc.
the surface of the black box surrounding the blue dot will be [7] Chapra, Steven C dan Canale, Raymod P. 2010. Numerical Methods
for Engineers Sixth Edition. New York : Mc-GrawHill Companies Inc.
black. If the declared value of a sensor is between the
threshold values white and black threshold value, then the
sensor concerned is declared to be in forbidden area and the
box will be colored red. Based on the images of the results of
testing the overall system, it can be concluded that the results
of this research have been run well.

____________________________________________________________________________________________________________
3D Human Brain Tumor Detection from MRI Image Using K-Means Clustering, Non-
Local Means, Thresholding and Marching Cube
Dodi Sudiana, Farid P Putera, Basari
Abstract: Medical image enhancement especially for human brain imageries is rapidly
developed, including 3D modeling. This research is aimed to reduce the error of diagnosis
process and facilitate brain tumor detection using 3D approach. In the process, 3D brain from
MRI imageries is analyzed to detect brain tumors. MRI image is converted from MINC format.
Then, the image is classified to detect objects using K-Means Clustering to divide each part of
brain. Filtering is performed using Non-Local Means to remove noise from previous processes.
The result of imageries are segmented to enhance and support reconstruction process using
Thresholding. Finally, 3D image reconstruction is performed using Marching Cube method.
The accuracy of brain tumor detection is evaluated of resolution reduction, non-tumor image
testing, comparison testing, modifying image format, and adding noise. The accuracy rate of
brain tumor detection is 100% for PNG format and 512×512 resolution, 97.7% for 256×256
resolution, 96.9% for non-tumor image and 97.96% for comparison between ideal image and
reference data. PNG format has better accuracy with JPEG by 4% improvement. The accuracy
of adding noise is 87.6%, 83.6%, and 74.5% for 0.01, 0.05 and 0.09 density level, respectively.
Keywords: Magnetic resonance imaging; brain tumor; k-means clustering; non-local means,
marching cube
(Under review at Jurnal Teknologi)

____________________________________________________________________________________________________________
A Review on the Achievable Sum Rate in Massive MIMO for 5G Wireless Cellular
Networks
Adeeb Salh1, Lukman Audah2, Nor Shahida Mohd Shah3, Shipun Anuar Hamzah4, Jiwa
Abdullah5
1,2,3
Universiti Tun Hussein Onn Malaysia, Malaysia;
4,5
Universiti Tun Hussein Onn Malaysia,Faculty of Eletrical and Electronic Engineering
(FKEE), Malaysia
Abstract: Massive multi- input multi- output (MIMO) has become an important technique for
the fifth generation (5G) cellular networks, where energy and spectral efficiencies become an
important design criterion for sustainable evolution. The achievable optimal sum rates of
MIMO using the minimum mean square error (MMSE) receivers can be done by portraying
the asymptotic multiplexing diversity at the receivers. Meanwhile, the energy and spectral
efficiency systems are able to work at high signal-to-noise ratio (SNR) systems and used to
suppress the robust interference signals connecting the closest tiers of neighbouring cells. We
conclude that the MMSE is better than ZF in perfect CSI for approximately 20% of the
achievable sum rate.
Keywords: 5G networks; massive MIMO; energy efficiency; spectral efficiency

____________________________________________________________________________________________________________
A Review on the Japan National Standards of Radio Frequency and Microwave

Attenuation
Anton Widarta
(National Metrology Institute of Japan, NMIJ/AIST, Japan)
Abstract: Attenuation is one of the fundamental measurement quantities that characterize

electromagnetic waves propagating in transmission lines of radio frequency (RF) and
microwave (MW) networks. At the National Metrology Institute of Japan (NMIJ), accurate
attenuation calibration systems operated in the frequency range of 10 MHz to 50 GHz have
been developed and used as a national standard of attenuation. This paper summarizes the
characteristic of the current calibration systems and discusses some of the measurement
techniques used to obtain the maximum results. The calibration capabilities range from 0 dB
to 110 dB with the expanded uncertainties from 0.002 dB to 0.068 dB.
Keywords: Attenuation; measurement; standard; microwave; uncertainty

____________________________________________________________________________________________________________
Analysis of Bidirectional VSI as Regenerative Braking Converter for BLDC Motor
Rini Nur Hasanah, Victor Andrean, Soeprapto Soeprapto and Hadi Suyono
Brawijaya University, Indonesia
Abstract: A plug-in electric vehicle is any motor vehicle which can be recharged from any
external source of electricity. The electricity is stored in the rechargeable battery packs to drive
or to contribute to drive the wheels. The limited capacity of battery is the principal cause of the
limited distance covered by the vehicle. Regenerative braking method is one way to recover
energy wasted during braking and to utilize it to reduce the energy consumption from battery.
In this way, it brings benefit to increase the battery efficiency while providing vehicle
deceleration. Other possible advantages include reducing maintenance cost because of the
wearing out of friction pads and/or brake shoes and providing smoother braking torque. This
paper analyzes two methods of regenerative braking using bidirectional voltage-source inverter
being object of many researches nowadays. Characteristics of each method and its
implementation are presented. Just by interchanging the switching order of inverter, motor
torque can be controlled and energy can be recovered to charge the battery.
Keywords: Bidirectional Voltage-Source Inverter (VSI); Brushless DC Motor; Plug-in

Electric Vehicle (PEV); Regenerative Braking System (RBS); Six-step Method

____________________________________________________________________________________________________________
Data-Logging in an Acidity Measurement System Using pH-Sensor with Circular

Nickel-Wire Electrodes
M. Julius St. (Brawijaya University; Faculty of Engineering, Indonesia); Sudjito Suparman,

Eko Siswanto and Denny Widhiyanuriyawan (Brawijaya University, Indonesia)
Abstract: Acidity or alkalinity of a solution can be known from its pH value. Conventionally,
litmus paper has been widely used to indicate the pH level, but it would require a lot of time
besides of the possible inaccuracy. A better device which is able to give result more quickly
and accurately is required. This paper presents a design of special pH-meter to measure the
acidity of a solution in real-time by taking the advantage of a data-logging system. The
designed system enables the addition of a memory card to store the measurement results. The
pH-level measurement has been performed based on the use of a pHsensor being made of
nickel. A microcontroller has been used to manage and control the operation and the
performance of system. The measurement reading results have been presented using a 16x2
LCD-display, whereas the related data have been stored using a 2GB microSD memory card.
The research results show that the pH-meter data-logging system worked well in measuring the
acidity of a solution continuously and that the reading data could be stored in a memory card.
The designed sensor system could detect the 1-7 pH-level with an average error of 0.040655
%. The memory card of 2GB could be used for a total reading data during around 744 days.
Keywords: data logger; nickel pH-sensor; SD card

____________________________________________________________________________________________________________
DESIGN OF E-SLOT MICROSTRIP ANTENNA FOR WLAN 2.4 GHz
Ruth Anindita (University of Brawijaya, Indonesia);

Onny Setyawati (Brawijaya University, Indonesia);
Rahmadwati Rahmadwati (University of Brawijaya, Indonesia)
Abstract: In wireless communication, the role of antenna is very necessary as a medium to

transmit and receive electromagnetic waves which contain information signals. Development
of wireless communication requires low profile, light weight, and easy-to-fabricated antenna,
such as microstrip antenna. E-slot microstrip antenna has an optimal performance with high
bandwidth, high gain, and low return-loss. In addition, E-slot is also simple in analysis of
geometry. In this paper, Method of Transmission Line and model simulation by CST Studio
Suite Student Version Software are used. The results of the model simulation show the value
of S11 of -16.514 dB, VSWR of 1.351, and gain of 3.433 dB.
Keywords: microstrip antenna; E-slot; model simulation

____________________________________________________________________________________________________________
Design of Floating Waterwheel for Harnessing River-Flow Surface-Stream Energy
Suparman Suparman and Hadi Suyono (Brawijaya University, Indonesia); Rini Nur Hasanah
(Brawijaya University; Faculty of Engineering, Indonesia)
Abstract: This paper proposes a method to harness low-capacity hydro-energy normally

abundant and abandoned in many shallow rivers and irrigation systems. It is realized by
designing a portable floating waterwheel equipped with a generator system capable of
producing electricity sufficient to supply a single small household. The design has been based
on the optimization of the desired power and water velocity, under the constraints of the cost
and materials availability in the market. Simulation on two different designs of waterwheel is
presented to observe and examine the influence of the blades number and the available water-
flow speed on the generated power. It has been concluded that by taking the benefit of
minimum required water velocity of 0.5 m/s, a low-capacity generator of ±500VA could
provide sufficient power for a simple household using a waterwheel. The design and
development included two portable floating waterwheel models i.e. 8-blade and 12-blade. An
8-blade waterwheel could produce power of 18.72W at rotating speed of 8.6 rpm, whereas a
12-blade waterwheel would produce the power of 28.72W at the same rotation speed.
Keywords: energy harnessing; floating-waterwheel; river-flow; surface-stream

____________________________________________________________________________________________________________
Design & Implementation of Wildlife (Tiger) Surveillance System to Prevent Poaching
Muhammad Muzakkir Mohd Nadzri1, Afandi Bin Ahmad2, Abbes Amira3
1,2
Universiti Tun Hussein Onn Malaysia (UTHM), Malaysia
3
Qatar University, Qatar
Abstract: This paper presents the design and implementation of wildlife (tiger) surveillance
system to prevent poaching. In this study, inputs based on the motion of surrounding related
with the destruction of the habitat and poaching activities were detected using a motion sensor
and camera. Images of these activities were captured and analysed using computer vision
system toolbox available in MATLAB & Simulink. A Raspberry Pi acts as a host with a
graphical user interface (GUI) that configures the systems and its peripheral. Images were
transmitted to the nearest wildlife agencies station using Wi-Fi 802.11n (2.4GHz) transmission.
Results obtained show that the proposed system capable to capture and process various size of
images. It is worth mentioning that transmitted images have been through various processes
and finally the decompressed images were obtained via host computer at the authority's station.
With the proposed solution, issue such as human-based system (which is time consuming) has
been solved.
Keywords: Wildlife surveillance;real-time monitoring;computer vision

____________________________________________________________________________________________________________
Directed Diffusion Based Routing Protocol Wireless Sensor Networks for A Forest
Preventive Fire System At Tesso Nilo National Park
Indra Yasri
(Universitas Riau, Indonesia)
Abstract: An alternative approach to achieve energy efficiency in Wireless Sensor Network

(WSN) using Directed Diffusion (DD) based routing protocol is presented in this paper. There
are 4 stages of scenarios involved in this approach. It's started by appointing of sampling area
through GPS coordinate. The sampling area is determined by optimization processes from
500m x 500m up to 1000m x 1000m with 100m increment in between. The next stage is sensor
node placement. Sensor node is distributed in sampling area with three different quantities i.e.
20 nodes, 30 nodes and 40 nodes. One of those quantities is choose as an optimized sensor
node placement. The third stage is to implement all scenarios in stages 1 and stages 2 on DD
processes. In the last stage, the evaluation process to achieve most energy efficient in the
combination of optimized sampling area and optimized sensor node placement on Direct
Diffusion (DD) routing protocol. The result shows combination between sampling area 500m
x 500m and 20 nodes able to achieve energy efficient to support a forest preventive fire system
at Tesso Nilo National Park.
Keywords: Routing Protocol; Wireless Sensor Network; Direct Diffusion; Forest Preventive
Fire System; Tesso Nilo National Park

____________________________________________________________________________________________________________
Growth of Vertically Aligned ZnO Nanorods Array and Its Application to Perovskite-type Solar
Cells
Murakami Kenji, Albertus Bramantyo and Masayuki Okuya (Shizuoka University, Japan); Nji
Poespawati and Arief Udhiarto (Universitas Indonesia, Indonesia)
Abstract: Well vertically aligned zinc oxide (ZnO) nanorods have been grown for the application of
perovskite-type solar cells (PSCs). For the growth of ZnO nanorods on glass substrates, a formation of
ZnO seed layer is necessary as a buffer layer. Spin coating and water bath methods are used for
preparation of the ZnO seed layer and the ZnO nanorods, respectively. Very critical factor to grow the
vertically aligned ZnO nanorods with high surface area is the ZnO seed layer. The seed layer with
smooth surface and high seed density are necessary to grow them. The rotational speed (v) and the
number of spin cycles (n) could control the smoothness and seed density of the seed layer. Surface
area of the nanorods becomes higher for the thinner nanorods array which could be grown the seed
layer with high density. The optimized ZnO seed layer is obtained with v=3000 rpm and n=3. The well
vertically aligned nanorods are grown along to [101] direction on the glass substrate. Finally, the
perovskite layer is formed on the vertically aligned ZnO nanorods array for the application of PSCs.
Keywords: ZnO; Seed layer; Nanorods; Solar cells
(This paper will appear in Makara Journal of Technology 2017)

____________________________________________________________________________________________________________
Ignition Timing Prediction of a Four Strokes Engine Using ANFIS Predictor
Muhammad Aziz Muslim, Dwi Fadilla Kurniawan and Goegoes Dwi Nusantoro
Abstract: Engine Control Unit (ECU) is main part of modern internal combustion engine.
Using information provided by many sensor attached surrounding the engine, ECU produce
appropriate control signal to actuators. One of the most important signal is ignition timing
signal. This signal will start combustion process in an internal combustion engine, ensure the
engine keep working. This paper proposes ignition timing prediction of a four strokes engine,
i.e. a motor cycle, using Adaptive Neuro Fuzzy Inference System (ANFIS). Fuzzy System (FS)
is widely known has best capability in approximate reasoning but suffers from learning
capability. On the other hand, Artificial Neural Networks (ANNs) despite of its black box
interpretation has capability of learning. ANFIS is a predictor constructed from hybridization
of FS and ANNs. It is hope that ANFIS can predict the ignition time through its reasoning and
learning capability. Recorded sensory data provided from a four stroke engine is a cyclic time
series data. To predict proper ignition time, the ANFIS must be able to understand engine cycle
inferred from sensory signals. Experimental results show that ignition prediction using ANFIS
has good accuracy with considerably small error rate in term of signal amplitude. Nevertheless,
in term of ignition timing, ANFIS provide almost perfect information on ignition timing.
Keywords: Engine Control Unit (ECU); Ignition Timing Prediction; Adaptive Neuro Fuzzy
Inference System (ANFIS)

____________________________________________________________________________________________________________
Injection Impact of Hybrid Embedded Generation in Distribution System
Hadi Suyono, Rini Nur Hasanah, Panca Mudjirahardjo, Muhammad Fauzan Edy Purnomo
(Faculty of Engineering, Brawijaya University, Indonesia)
Abstract: Development and needs of the electrical energy continues to increase at this time
requires continuity of service of fulfilled technical and economical. Therefore, it is necessary
to discover other alternative energy sources that not only to generate the power but also has
minimum impact to the environment. The development of new and renewable energy including
solar energy, hydro power, wind energy, biomass, and other sources are the best solution. The
power generation system developed now days includes a small-scale in terms of power output
but is expected to be massive in terms of numbers. Photovoltaic (PV) as a source of solar energy
in general has not been widely used in Indonesia, contrasting in modern countries, PV has been
installed in every house for internal needs and the surplus is exported to the grid. In addition,
the micro/mini hydro power has been extensively applied in the distribution system. Therefore,
the steady state and dynamic analyses are essential to determine the optimal size of hybrid
micro hydro and PV powers injection such that the system power loss and voltage profile
satisfied with standards. To model the proposed hybrid PV and micro hydro power, the real
distribution system is implemented. The steady state analysis before and after injection of
micro power plants and PV were analyzed in this study. Based on the analysis shows that the
injection of hybrid micro hydro and PV give a better impact in term of power losses,
improvement of voltage profile compared with system before the injection.
Keywords: hybrid distributed generation; micro hydro power; photovoltaic; steady state
analysis

____________________________________________________________________________________________________________
Mitigating BER of FSO Communications by Minimizing the Temporal Fluctuation in

Signal Spectral
Ucuk Darusalam
(Universitas Nasional, Indonesia)
Abstract: Optical propagation of Free-Space Optical (FSO) Communications on atmospheric

turbulence exhibits deteriorated of signal spectral. In order to recover it, we propose an Optical
Spatial Filter (OSF) as detection method. It aim to increase intensity of signal spectral or signal
power by suppressing noise. We compare results of measurement that are original signal, signal
at non-turbulent and signal at turbulent media. We also compare FSO performances in three
schemes that are fiber-, pinhole-fiber-, and OSF-detection method. OSF is intalled in rear focus
spot of receiver lens before photodetector. The experiment results show significant
improvement in suppressing spatial noise by OSF. 〈SNR〉 increase that are 37.5 dB, 38.5 dB,
38.7 dB, and 39.2 dB for D_P1, D_P2, D_P3, and D_P4, respectively. 〈BER〉 are produced
more lower that are 10-9 to 10-12 by OSF. Those performances improvement is achieved
higher compared to fiber- and pinhole-fiber-detection method. The outcome of this work is
recovering signal spectral that is deteriorated caused by turbulence effects. Hence OSF can be
implemented in WDM of FSO in order to enhance 〈SNR〉 and 〈BER〉 performances.
Keywords: optical spatial filters; beam wander; scintillation; turbulence effects; free-space
optical communication

____________________________________________________________________________________________________________
Modelling of a Four Branch Power Splitter Using III-nitrides Semiconductor for Long
Wavelength Optical Telecommunication
Retno Purnamaningsih and Nji Poespawati (Universitas Indonesia, Indonesia) ;

Elhadj Dogheche ((IEMN CNRS UMR 8520), Cite’ scientifique, Villeneuve d’ Ascq, France);
Dimitris Pavlidi (Boston University, USA)
Abstract: In this paper, we propose a model of four branch optical power splitter based on III-
nitrides. The design consists of large cross section rib waveguides and a conventional rectangular
multimode interference (MMI) structure. The MMI structure is chosen due to their attractive
performances, such as compactness, low excess loss and ease to fabricate. The intension of this
modelling is to be used in long wavelength telecommunication wavelength, 1.55 µm. Optimization
of the geometrical structure parameters for the design is conducted theoretically using 3D FD-
BPM method. It is found that the power splitter shows excess loss of 0.66 dB and imbalanced of
0.001 dB. It means that this proposed model has a potential application in the future
Keywords: GaN; MMI structure; optical power splitter; optical communication; four branches

____________________________________________________________________________________________________________
Numerical Examination on Transmission Properties of FBG with GPU Support
Angger Abdul Razak and Mitsuhiro Yokota
(University of Miyazaki, Japan)
Abstract: It is important to examine Fiber Bragg grating (FBG) for the design of the practical
use. Finite difference time domain (FDTD) method is used to examine the transmission
properties. FBG is constructed as three-dimensional periodic structures. Graphic processing
unit (GPU) calculation is proposed to shorten simulation time. Results of FBG transmission
properties from Central Processing Unit (CPU) and GPU calculation was compared.
Comparison of simulation time between CPU and GPU calculation also observed. Up to 100x
faster simulation time by using GPU computation was obtained with indistinguishable results
in transmission properties compared to CPU calculations. Some simulation with parameter
changes to the grating structures also conducted. Higher grating refractive index value affect
in smaller value of transmission amplitude in specific frequency, and the center of transmission
frequency drop is slightly shifted to lower frequency value. Longer grating period will reflect
lower frequency. Longer total grating length will affect to the lower transmission value.
Chirped FBG (CFBG) show a wider broadband reflect compared to the uniform FBG type,
with a weaker reflectance.
Keywords: FBG; FDTD; GPU Computing

____________________________________________________________________________________________________________
Performance Evaluation of Hypertext Transfer Protocol Over Terrestrial and Satellite

Networks
Liyana Zakaria1, Lukman Audah2, Ansar Jamil3, JiwaAbdullah4
1
TM Berhad, Malaysia
2,3,4
Universiti Tun Hussein Onn Malaysia, Malaysia
Abstract: Hypertext Transfer Protocol (HTTP) is the backbone of the World Wide Web to
deliver messages between Web browsers and Web servers. Currently there are three versions
of HTTP, namely HTTP/1.0 and HTTP/1.1; also HTTP/2, which was officially released in
2015. Since HTTP/1.1 supports the use of pipeline requests and because of the persistent
connections that are enabled by default, HTTP/1.1 seems to be more efficient than HTTP/1.0.
Therefore, we study end-to-end QoS performance parameters of HTTP/1.1 version over
terrestrial links and satellite networks using Network Simulator-2 (NS2). Simulations were
developed that consist of four ground nodes including a HTTP server, a HTTP client, and a
pair of server and client using background traffic that is aimed to congest the Internet network.
Additionally, we present the performance evaluation and analysis of two different types of
transmission medium: terrestrial links and satellite networks. Three types of satellite are used,
namely LEO, MEO and GEO satellites. The results indicate that the terrestrial connection
produces better QoS performance compared to satellite networks, in terms of lower delay, jitter
and packet loss ratio; but higher throughput against the increment of packet rate from the server
to the client. Future works will involve better routing and queuing management to achieve
better QoS performance.
Keywords: HTTP/1.1; quality of service (QoS); terrestrial networks; satellite networks

____________________________________________________________________________________________________________
Power Optimization for Tunable Microwave Generation Based on Mixing Two DFB
Laser Using Polarizing Effect
Nursidik Yulianto1, Purnomo Sidi Priambodo2, Bambang Widiyatmoko3
1,3
Indonesia Institute of Sciences (LIPI), Indonesia
2
Universitas Indonesia, Indonesia
Abstract: The generation of tunable microwave by mixing two optical wave DFB laser and
heterodyne detection technique has been performed. Microwaves that have broadband
characteristics, low loss, stable and high power to be a challenge for the development source
of communication in the future. One of the effects on the optical laser source for microwave
generation is optical polarization. By arranging the optical polarization of laser sources are
expected to increase the power of micro-wave generator. From the experimental results
proved that setting the direction of polarization of two laser as optical sources can increase
the generation of microwaves up to 2000 percent in the same injection current.
Keywords: Photomixing; Polarization; Microwave

____________________________________________________________________________________________________________
Requirement Analysis on Received RF Voltage Drive for Electro-Optic Modulator in

Distributed Antenna System
Yus Natali (Akademi Teknik Telekomunikasi Sandhy Putra Jakarta, Indonesia); Purnomo
Priambodo (Universitas Indonesia, Indonesia)
Abstract: Integrated antenna for Electro-Optic Modulator has become an applicable solution
in Distributed Antenna System (DAS). DAS, also known as In Building Coverage/System
(IBC/S), mostly still uses coaxial cable as a transmission medium in the building. This paper
proposed to use the electro-optic modulator as Photonic Antenna in DAS. The research has
been conducted theoretically based on empirical reference data to design the DAS
configuration. Such data was simulated using the Pathloss software to calculate the back
calculation power in DAS. Furthermore, DAS configuration system in this research was
designed by using Mach-Zehnder modulator which was simulated using OptiSystem. Based on
the result of the modelling system, simulation measuring result showed that the use of the
electro-optic modulator has further made output power more efficient up to a photodetector.
Pathloss simulation result showed that the power obtained by RF donor antenna was -39.13
dBm; while OptiSystem simulation result showed that photodetector output power in the form
of RF signal was -31 dBm. Power efficiency was obtained around 8 dB if compared with donor
antenna output power
Keywords: Mach-Zehnder modulator; Distributed Antenna System; RF Antenna Electro-

Optic Modulator

____________________________________________________________________________________________________________
Single and Coupled-Dopants as Quantum Dots for Room Temperature Single-Electron

Tunneling
Daniel Moraru (Shizuoka University & Research Institute of Electronics, Japan); Arup
Samanta and Michiharu Tabe (Shizuoka University, Japan)
Abstract: Transistors containing in the channel a few dopant-atoms work as single-electron

transistors with the dopants working as quantum dots (QDs). This system offers the ability to
control carrier transport to the level of single atoms and single electrons. However, typical
dopants in Si (such as phosphorus, P) have small barrier height and cannot sustain tunneling
operation at practical temperatures. Here, we first overview the limitations for the cases of
single-dopant transistors. Then, we discuss an alternative of using a few strongly-coupled
donors to form quantum dots (QDs) with larger barriers, allowing tunneling operation at room
temperature. Guidelines for room-temperature single-electron tunneling operation will be
outlined.
Keywords: single dopant transistors; a-few-donor QD; single electron tunneling; silicon nano-
devices

____________________________________________________________________________________________________________
Singly-fed Circularly Polarized Triangular Microstrip Antenna with Truncated-tip for

CP-SAR Application
Muhammad Fauzan Edy Purnomo, Hadi Suyono, Dwi Kurniawan, Ali Mustofa and Sholeh
Pramono, Rahmadwati Rahmadwati
(University of Brawijaya, Indonesia)
Abstract: The development of radar technology, SAR (Synthetic Aperture Radar) and
geostationary satellite relatively fast and demanding needs of communication facilities and
infrastructure that has a variety of platforms and imaging of high quality, which can generate
data processed with high resolution and a better image for all types of terrain explored. In this
paper, it obtain a basic construction for CP-SAR (Circularly Polarized-Synthetic Aperture
Radar) geostationary satellite with compact, small, and simple configuration i.e. the truncated-
tip hexagonal (c3) and scope (c2) microstrip antennas with low-power in S-band (2.5 GHz -
2.7 GHz). The performance characteristics of these antennas are circular polarization, both
circular to the left or to the right, making them easier to capture images of objects on earth for
processing information content that can be utilized. These antennas are made from the type of
microstrip antenna in accordance with the technical specifications desired. The performance
results almost the same, except of S-parameter and axial ratio in radiation pattern that the
truncated-tip c2 antenna smoother than c3 antenna.
Keywords: Geostationary satellite; CP-SAR; Truncated tip

____________________________________________________________________________________________________________
Star Patch Antenna for ISM Band Applications
Rudy Yuwono and Faisal Ramadhan (University of Brawijaya, Indonesia); Fauziahanim Che
Seman (Universiti Tun Hussein Onn Malaysia, Malaysia)
Abstract: This paper proposes star patch antenna for ISM Band Applications. The dimension
of the antenna is optimized using commercially available software, Computer Simulation
Technology (CST) Microwave Studio. The antenna is fabricated on two substrate materials,
FR4 and Polyimide Film. The antenna provides reflection loss of 27.67 dB at 2.45 GHz with
610 MHz in -10 dB bandwidth. The performance of the antenna in terms of reflection loss S11,
far field plot and antenna gain was conducted in semi-anechoic chamber. It can be observed
that the measured performance of the fabricated antenna is in very close agreement with the
computer model.
Keywords: Star patch; ISM Band; Polyimide Film

____________________________________________________________________________________________________________
Study of Design of A Double-Sided LIM (DSLIM) with Ladder Secondary for Slow Speed
Purpose
Mochammad Rusli
Abstract: The linear induction motors are one of the famous kind of linear electrical motors, which
are produced commercially by many countries and are realized in many applications. A Linear
Induction motors(LIMs) are, for instance, mostly used in machine tools, linear tables, textile tools,
saws, separators, operation of sliding doors, transportation systems and many things. For slow speed
purposes, many researcher employed the linear motor with permanent magnet. However if it is used
in machine tool, permanent magnet might attract the metal durst, so the maintenannce of motor
could not be ignored. The one of alternative is the linear induction motor. For slow speed application,
the linear induction motor can provide a great big thrust if the secondary part use the ladder-bar, so
the air gap length of motor can make as small as possible. The consequence is the useful thrust will
increasing if the air gap length so smaller. This paper describes the standrad design open-slot Double
Sided Linear Induction Motor (DSLIM).
Keywords: Double-Sided Linear Induction Motors; Slow speed; Cogging Force

____________________________________________________________________________________________________________
The Evaluation of Tropical Tuber Properties Using Far Field Method
Onny Setyawati and Nanang Sulistiyanto
Abstract: Detection of the frequency response of tropical tubers has been evaluated using the
far-field method. The tropical tubers used as targets in this research are Porang
(Amorphophallus muelleri Blume) tuber, Iles-iles (Amorphophallus variabilis) tuber, and
sweet potatoes-Cileumbu. A horn reflector, appropriate to the minimum range, is designed as
the guide of the electromagnetic wave. The experiments were performed by variation of the
distance between the horn mouth and the target material (tropical tubers), the optimum distance
of 6 up to 10 cm resulted in relatively high reflected power received by the antenna. Different
thickness of the sliced tubers influenced the reflected power. The results of the measurement
show that Porang, Iles-iles and sweet potatoes-Cileumbu can be detected at range frequency of
1.745 - 1.88 GHz, 2.185 - 2.21 GHz, and 2.055 - 2.17 GHz, respectively. These results will
give benefit for modelling the sensor system, for detecting the tubers underground, wherein
working frequency of the device must be selected carefully and the tubers properties should be
evaluated.
Keywords: Far field method; Porang tuber; Range of frequency; Tropical tubers

____________________________________________________________________________________________________________
Conpot Analysis: Low-Interaction Honeypot as Attack Detection System on Industrial

Network Control System / SCADA
Anak Agung Putri Ratna, Mahardianto Yudha Bestari, Ihsan Ibrahim, Prima Dewi
Purnamasari and Fransiskus Astha Ekadiyanto, Muhammad Salman
Faculty of Engineering, University of Indonesia, Indonesia
Abstract: The rise of the attack on SCADA systems on modern industries currently attracts
attention for developers to think about the solution of the problem. The purpose of this paper
is to discover one solution to improve network security systems ICS / SCADA, by designing a
schematic design of Conpot. Honeypot is a system that deliberately targeted by attacks such as
Stuxnet to distract the attacker from the real system. Conpot will be simulated by virtualizing
the original SCADA device and run existing protocol such as Modbus TCP. Based on the
analysis of the system and the results of the trials, Conpot may be one solution to improve the
security of the original system, because Conpot has the reliability in detecting attacks into the
Modbus network. Meanwhile, responsive test obtained response time of 0.2521 seconds for
one attacker and 0.2582 seconds for two attackers. The results also show that the installation
of Conpot does not affect the system and network performances.
Keywords: Stuxnet; Network Security; SCADA; Low-Interaction Honeypot; Conpot

____________________________________________________________________________________________________________
GRATITUDE FOR REVIEWERS
We would like to thank all reviewers for having given their invaluable support:
Dr. Catur Apriono (University of Indonesia, Indonesia)

Dr. Muhammad Aziz Muslim (Brawijaya University, Indonesia)
Dr. Anayat Ullah Baloch (Balochistan University of IT, Engineering and Management
Sciences – BUITEMS, Pakistan)
Mr. Sudeep Bhattarai (Virginia Tech, Virginia)
Mr. Ahmet Elbir (Duzce University, Turkey)
Mr. Rahul Gore (ABB Corporate Research Center Bangalore, India)
Dr. Rini Hasanah (Brawijaya University, Indonesia)
Mr. Angger Abdul Razak (University of Miyazaki, Japan)
Dr. Nabihah Ahmad (Universiti Tun Hussein Onn Malaysia, Malaysia)
Prof. Mudrik Alaydrus (Universitas Mercu Buana, Indonesia)
Mr. Primatar Kuswiradyo (Brawijaya University, Indonesia)
Dr. Siti Zarina Mohd. Muji (Universiti Tun Hussein Onn Malaysia, Malaysia)
Dr. Panca Mudjirahardjo (Brawijaya University, Indonesia)
Mr. I Gde Dharma Nugraha (Chonnam National University)
Dr. Rosli Omar (Universiti Tun Hussein Onn Malaysia, Malaysia)
Dr. Purnomo Sidi Priambodo (University of Indonesia, Indonesia)
Mr. Muhammad Fauzan E.P. (Brawijaya University, Indonesia)
Mr. Abdul Rajab (Kyushu Institute of Technology, Japan)
Mr. Mochammad Rif'an (Brawijaya University, Indonesia)
Ms. Mia Rizkinia (University of Indonesia, Indonesia)
Mr. Mochammad Rusli (Brawijaya University, Indonesia)
Mr. Annasamudram S R Sekhar (G V P College of Engineering, India)
Dr. Onny Setyawati (Brawijaya University, Indonesia)
Mr. Nanang Sulistiyanto (Brawijaya University, Indonesia)
Dr. Alireza Tavousi (University of Tabriz, Iran)
Prof. Peter Roessler (University of Applied Sciences Technikum Wien, Austria)
Prof. Hiroki Tamura (University of Miyazaki, Japan)
Dr. Hadi Suyono (Brawijaya University, Indonesia)
Dr. Arief Udhiarto (University of Indonesia, Indonesia)
Dr. Gunawan Wibisono (University of Indonesia, Indonesia)
Mr. Rudy Yuwono (Brawijaya University, Indonesia)
Prof. Roshdy Abdel Rassoul (Arab Academy for Science and Technology, Egypt)
Mr. Mohamad Hairol Jabbar (Universiti Tun Hussein Onn Malaysia, Malaysia)

____________________________________________________________________________________________________________
GRATITUDE FOR PARTICIPANTS
We would like to thank all participants for having given their invaluable contribution:
 UIN Maliki Malang, Indonesia

 UIN Sunan Ampel Surabaya, Indonesia
 Universitas Surabaya, Indonesia
 Tax Education & Training Center, Ministry of Finance, Indonesia
 Institut Teknologi Sepuluh Nopember Surabaya, Indonesia
 Politeknik Perkapalan Negeri Surabaya, Indonesia
 Brawijaya University, Indonesia
 University of Surabaya, Indonesia
 Indonesia Institute of Sciences (LIPI), Indonesia
 National Metrology Institute of Malaysia, Malaysia
 National Metrology Institute of Japan, NMIJ/AIST, Japan
 Universitas Indonesia, Indonesia
 California Polytechnic State University, USA
 Politeknik Negeri Malang, Indonesia
 Microelectronic and Nanotechnology-Shamsuddin Research Centre Malaysia,
Malaysia
 Qatar University, Qatar
 Shizuoka University, Japan
 TM Berhad Malaysia, Malaysia
 Center for Science and Technology Accelerator Indonesia, Indonesia
 Universitas Mercu Buana, Indonesia
 University of Miyazaki, Japan
 BPPT, Indonesia
 Politeknik Negeri Batam, Indonesia
 Universitas Widyagama Malang, Indonesia
 University of Leeds, England
 University of Al Azhar Indonesia, Indonesia
 Universitas Riau, Indonesia
 Institut Teknologi Bandung, Indonesia
 Universitas Trisakti, Indonesia
 Universitas Nasional, Indonesia
 Politeknik Caltex Riau, Indonesia
 Universitas Sebelas Maret, Indonesia
 University of Mataram, Indonesia
 University of Lampung, Indonesia
 University of Alberta, Canada
 Akademi Teknik Telekomunikasi Shandy Putra Jakarta, Indonesia
 Universitas Mpu Tantular, Indonesia
 NSHM India, India
 Jadavpur University, India
 Bankura Unnayani Institute of Engineering, India
 Politeknik Elektronika Negeri Surabaya, Indonesia
 Politeknik Negeri Bali, Indonesia

Proceeding 3rd ICNERE and 8th EECCIS 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Proceeding 3rd ICNERE and 8th EECCIS 2017

Uploaded by

Copyright:

Available Formats

Joint International Conference

3rd ICNERE and 8th EECCIS

October 31 – November 2, 2016

Organized by: Supported by:

The 3rd International Conference on Nano Electronics Research and Education

Copyright © Electrical Engineering Department, Enginering Faculty,

Joint International Conference:

The 3rd International Conference on Nano Electronics Research and Education

"Green Technology towards a Better Environment”

October 31 –November 2, 2016

Universitas Brawijaya PSeeMRG

©ICNERE-EECCIS 2016 iii

TABLE OF CONTENTS .................................................................... iv

ABOUT THE CONFERENCE

 Proceeding ICNERE - EECCIS 2016

This joint conference topics are include, however not limited to :

Assalamu’alaikum warahmatullahi wabarakatuh

Wassalamu’alaikum warahmatullahi wabarakatuh

Dr. Ir. Pitojo Tri Juwono, MT.

Prof. Dr. Harry Sudibyo S, DEA

©ICNERE-EECCIS 2016 vii

International Advisory Board

©ICNERE-EECCIS 2016 viii

o Adharul Muttaqin (UB-Indonesia)

Student centered design of CMOS programmable biosignal recorders

National Sun Yat Sen University, Taiwan

Prof. Robert Rieger, Ph.D

The Future of Nanotechnology for Medical Applications

Keywords: carbon-encapsulated magnetic nanoparticles; plasma surface functionalization; carbon nanotube

Prof. Masaaki Nagatsu

©ICNERE-EECCIS 2016 xii

©ICNERE-EECCIS 2016 xiii

Plasma processing technology for future solar device fabrication

Universiti Tun Hussein Onn Malaysia

Ass. Prof. Nafarizal Nayan, D. Eng.

He headed Microelectronic Laboratory and Microelectronic and Nanotechnology – Shamsuddin Research

©ICNERE-EECCIS 2016 xiv

ICNERE_EECCIS 2016 SCHEDULE

Parallel Session (Monday - October 31, 2016)

ROOM 1 Session Chair : Lukman Audah; Anton Widarta

09:15 - 09:30 1570298511 Performance Evaluation of Hypertext Transfer Protocol Audah

©ICNERE-EECCIS 2016 xvi

ROOM 2 Session Chair: Angger Abdul Razak; Yuning Widiarti

©ICNERE-EECCIS 2016 xvii

ROOM 3 Session Chair: Rudy Yuwono; Ucuk Darusalam.

©ICNERE-EECCIS 2016 xviii

ROOM 4 Session Chair: Yus Natali; Ruth Anindita

©ICNERE-EECCIS 2016 xix

Program Session (Monday - October 31, 2016)

10:30 - 10:45 Dance performance

11:45 - 13:15 LUNCH

13:15 - 13:45 Keynote Speaker 2 : Prof. Dr. Masaaki Nagatsu.

14:15 - 14:30 COFFEE BREAK

Parallel Session (Monday - October 31, 2016)

15:00 - 15:15 1570294013 Integration of Distributed Generation in Radial Nadhiroh

©ICNERE-EECCIS 2016 xxi

ROOM 2 Session Chair: Dikpride Despa; Ida Bagus Fery Citarsa

©ICNERE-EECCIS 2016 xxii

ROOM 3 Session Chair Hadi Suyono; Setyo Nugroho.

©ICNERE-EECCIS 2016 xxiii

ROOM 4 Session Chair: Catur Apriono; Nabihah Ahmad.

18:30 - 20:30 GALA DINNER

©ICNERE-EECCIS 2016 xxiv