Performance Evaluation of MmWave in 5G Networks

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/339817188
Performance Evaluation of MmWave in 5G Networks
Thesis · September 2019

DOI: 10.13140/RG.2.2.33085.82403
CITATIONS READS
0 336
1 author:
Ahmed Nor
Polytechnic University of Bucharest
7 PUBLICATIONS 28 CITATIONS
SEE PROFILE
All content following this page was uploaded by Ahmed Nor on 10 March 2020.
The user has requested enhancement of the downloaded file.

Aswan University Faculty of Engineering
Performance Evaluation of MmWave in 5G

Networks
By
Ahmed Mohammed Nor El-Dien El-Zakaloby
B. Sc., Electrical Engineering, Aswan University, 2016
A Thesis Submitted to the Faculty of Engineering

Aswan University
(ASWU)
In Partial Fulfillment of the Requirements for the Degree of

Master of Science
In
Electrical Engineering
2019
Scanned by CamScanner
Declaration and Certificate of Originality
I certify that in the preparation of this thesis, I have observed the provisions of the scientific
code of ethics. Further; I certify that this work is free of plagiarism and all materials appearing
in this thesis have been properly quoted and attributed.
I certify that all copyrighted material incorporated into this thesis is in compliance with the
international copyright law and that I have received written permission from the copyright
owners for my use of their work, which is beyond the scope of the law.
I agree to indemnify and save harmless faculty of Engineering, Aswan University from any
and all claims that may be asserted or that may arise from any copyright violation.
I hereby certify that the research work in this thesis is my original work and it does not
include any copied parts without the appropriate citation.
Ahmed Mohammed Nor
2019
I
ABSTRACT
Internet of things (IoT) and tactile internet applications are widely spread recently. Also,
enormous speed communication networks become a persistent need nowadays due to the huge
traffic demand. Hence, researchers direct their interests towards 5G networks to tackle these
needs. Immigration to millimeter Wave (mmWave) band, i.e., 60 GHz band, seems to be a
promising candidate to enable 5G networks because it has a wide unlicensed available
spectrum. However, it faces big technical challenges that impede its work. MmWave signal is
highly attenuated in space due to propagation and penetration losses. Also, mmWave faces the
issue of high blocking probability due to human shadowing. Hence, directional transmission
with phased antenna array is defined by IEEE 802.11 ad standard as a beamforming training
(BT) process to overcome these issues and increase mmWave beam gain. However, BT is
highly complicated and needs long setup time to find out the mmWave beam that obtains the
maximum received power. Also, IEEE 802.11ad proposes using scheduled medium access
control (MAC) based on time division multiple access (TDMA) to share radio resources among
active mmWave users but it didn’t define a certain scheduling scheme to be used. Thus, a
performance evaluation of scheduling schemes is needed before implementing one of them in
mmWave network. Moreover, in concurrent transmission (CT) scenarios, selecting the
mmWave concurrent links that maximizing the total system data rate is another big problem.
Because, BT process is performed with no knowledge about mutual interference between
concurrent links hence losing the chance to select low interfering concurrent beams that provide
high data rate to the end user. In this thesis, a performance evaluation of different scheduling
schemes, e.g., round robin and proportional fairness scheduling, is studied in mmWave
networks under different mmWave channel conditions, i.e., with different blocking probability
scenarios, and various mmWave beamwidths regarding the total system data rate and fairness
between users. Also, the problem of selecting multiple mmWave concurrent links in downlink
scenario in wireless local area network is discussed. Where, a joint proportional fairness
scheduling (JPFS) optimization problem is formulated to maximize the total system data rate
with obtaining acceptable fairness among users. Then, an iterative search based JPFS is
proposed to find out a sub-optimal solution to this problem with a very low complexity
comparable to exhaustively search based scheme. This reduction in complexity is proven using
numerical analysis.
II
To all who have been a great support for me,
I dedicate this work.
III
ACKNOWLEDGEMENT
First of all, I thank Allah (subhanaho wa-taala) for endowing me with health, patience,
strength, knowledge, passion, and motiviation to complete this research and putting it in a
master-thesis form.
Also, I would like to express my sincere gratitude to my supervisors Dr. Osama Ahmed
Omer, and Dr. Hamada Ahmed Hamada, for their continuous support during my master study
and related researches. Also, I have to thank them for their patience, motivation, and immense
knowledge.
I would like to thank Dr. Osama Ahmed Omer for his expert advices and encouragement
throughout this research process. I would like to show my gratitude to him for guiding the work
during thesis development, giving several insightful discussions, motivating and enlightening,
all his efforts and suggestions are deeply appreciated.
Finally, I can’t deny the superior part from my family for standing beside me throughout
my research work and for completing this thesis. They have been my constant sources of
inspiration and motivation, for continuing research, to improve my knowledge and move my
career forward. I really appreciate it.
Eng. Ahmed Mohammed Nor
July 2019
IV
Table of Contents
Table of Contents
TABLE OF CONTENTS .................................................................................................................................... V
LIST OF FIGURES .......................................................................................................................................... VII
LIST OF TABLES ..............................................................................................................................................IX
LIST OF ABBREVIATIONS ............................................................................................................................. X
LIST OF SYMBOLS .......................................................................................................................................XIII
INTRODUCTION ......................................................................................................................... 2
1.1 BACKGROUND........................................................................................................................................ 2
1.1.1 MmWave Technology ...................................................................................................................... 4
1.1.1.1 Directional communication ...................................................................................................................... 5
1.1.1.2 Scheduling Schemes in MmWave............................................................................................................ 6
1.1.1.3 MmWave Concurrent transmission.......................................................................................................... 6
1.2 MOTIVATIONS ........................................................................................................................................ 7
1.3 THESIS OBJECTIVES ................................................................................................................................ 8
1.4 THESIS OUTLINE .................................................................................................................................... 9
LITERATURE REVIEW ........................................................................................................... 11
2.1 INTRODUCTION .................................................................................................................................... 11
2.2 MILLIMETER WAVE TECHNOLOGY ...................................................................................................... 11
2.2.1 MmWave physical layer ................................................................................................................. 12
2.2.2 MmWave Network Architecture..................................................................................................... 14
2.2.2.1 Beacon Interval ...................................................................................................................................... 14
2.2.2.2 Personal Basic Service Set ..................................................................................................................... 15
2.2.2.3 Network and Schedule Announcements................................................................................................. 16
2.2.3 MmWave Medium Access Control Layer ...................................................................................... 17
2.2.3.1 Contention based Medium Access ......................................................................................................... 17
2.2.3.2 Dynamic Channel Time Allocation........................................................................................................ 17
2.2.3.3 Pseudo-static TDMA Channel Time Allocation .................................................................................... 19
2.2.4 MmWave Beamforming Training................................................................................................... 20
2.2.4.1 Beamforming training based on mmWave channel estimation .............................................................. 21
2.2.4.2 Beamforming training without mmWave channel estimation ................................................................ 23
2.3 SCHEDULING SCHEMES IN MMWAVE .................................................................................................. 25
2.3.1 Channel Independent Scheduling Schemes .................................................................................... 26
2.3.2 Channel Sensitive Scheduling Schemes ......................................................................................... 27
2.4 MMWAVE CONCURRENT TRANSMISSION ............................................................................................ 27
2.4.1 IEEE 802.11ad Based WLAN with scheduled medium access control .......................................... 27
2.4.2 IEEE 802.11ad Based WLAN with random access scenarios ........................................................ 29
PERFORMANCE EVALUATION OF PROPORTIONAL FAIRNESS SCHEDULING IN
MMWAVE NETWORKS .................................................................................................................................. 32
3.1 INTRODUCTION .................................................................................................................................... 32
3.2 MMWAVE SYSTEM MODEL ................................................................................................................. 33
3.3 PROPORTIONAL FAIRNESS SCHEDULING IN MMWAVE......................................................................... 36
3.4 PERFORMANCE ANALYSIS.................................................................................................................... 38
3.4.1 Simulation scenario......................................................................................................................... 38
3.4.2 Results and discussion .................................................................................................................... 39
3.5 SUMMERY ............................................................................................................................................ 44
V
Table of Contents
JOINT PROPORTIONAL FAIRNESS SCHEDULING FOR MMWAVE CONCURRENT

TRANSMISSION ............................................................................................................................................... 47
4.1 INTRODUCTION .................................................................................................................................... 47
4.2 THE PROPOSED SYSTEM MODEL .......................................................................................................... 48
4.3 JOINT PROPORTIONAL FAIRNESS SCHEDULING OPTIMIZATION PROBLEM ............................................ 49
4.4 PROPOSED IS BASED JPFS FOR MMWAVE CONCURRENT TRANSMISSION ........................................... 50
4.5 RESULTS AND DISCUSSION ................................................................................................................... 52
4.5.1 Performance Metrics ....................................................................................................................... 54
4.5.2 Simulation Results .......................................................................................................................... 55
4.6 SUMMARY............................................................................................................................................ 59
CONCLUSION AND FUTURE PERSPECTIVE .................................................................... 61
5.1 CONCLUSIONS ...................................................................................................................................... 61
5.2 FUTURE PERSPECTIVES ........................................................................................................................ 62
REFERENCES ................................................................................................................................................... 64
LIST OF PUBLICATIONS ............................................................................................................................... 68
VI
List of Figures
List of Figures
Figure 1-1: Cisco global mobile data traffic forecast, 2016-2021. ........................................... 2
Figure 1-2: Cisco forecast for number of connected IoT devices, 2015-2025. ........................ 3
Figure 1-3: Enabling 5G network through one of three axes. .................................................. 3
Figure 1-4: MmWave AP antenna beamforming. .................................................................... 5
Figure 1-5: MmWave concurrent transmission. ....................................................................... 7
Figure 2-1: IEEE 802.11ad packet structure. .......................................................................... 12
Figure 2-2: IEEE 802.11ad beacon interval structure............................................................. 14
Figure 2-3: Dynamic channel allocation. ................................................................................ 19
Figure 2-4: MmWave BT process a) SLS phase, b) MID sub phase and c) BC sub phase. ... 20
Figure 2-5: Non-optimal establishment of mmWave concurrent links based on IEEE 802.11ad
standard using scheduling MAC. ............................................................................................. 28
Figure 2-6: Optimal mmWave concurrent links. .................................................................... 28
Figure 2-7: Establishment of non-optimal mmWave concurrent links based on IEEE 802.11ad
using random access scenarios where Packet collision caused because of un-coordinated
mmWave BT. ........................................................................................................................... 29
Figure 2-8: Establishment of non-optimal mmWave concurrent links based on IEEE 802.11ad
using random access scenarios where Packet collisions caused because of non-optimal user
association. ............................................................................................................................... 29
Figure 3-1: MmWave network architecture............................................................................ 33
Figure 3-2: IEEE 802.11ad beacon interval structure............................................................. 34
Figure 3-3: Total system data rate of PFS and RR schemes with 30° beamwidth with/without
blocking probability against number of mmWave UEs. .......................................................... 40
Figure 3-7: Jain’s fairness index of PFS and RR schemes with 30° beamwidth with/without
Figure 4-1: Proposed WLAN Architecture. ............................................................................ 48
Figure 4-2: Study area............................................................................................................. 53
VII
List of Figures
Figure 4-3: The performance of the proposed IS-JPFS, ES-JPFS and IEEE 802.11ad based
WLAN (Conv.) in terms of total system data rate using different number of mmWave APs. 55
WLAN (Conv.) in terms of spatial reuse factor using different number of mmWave APs. ... 56
WLAN (Conv.) in terms of user’s fairness index using different number of mmWave APs. . 56
Figure 4-6: The total system complexity of the proposed IS-JPFS, ES-JPFS and the
conventional scheme when different number of mmWave APs are used. .............................. 57
Figure 4-7: Total system data rate of the proposed IS-JPFS, ES-JPFS and the conventional
scheme in different user densities scenarios. ........................................................................... 58
Figure 4-8: Spatial reuse factor of the proposed IS-JPFS, ES-JPFS and the conventional
Figure 4-9: Users’ fairness index of the proposed IS-JPFS, ES-JPFS and the conventional
VIII
List of Tables
List of Tables
Table 3-1: Study Parameters. .................................................................................................. 38
Table 4-1: Simulation Parameters. .......................................................................................... 53
IX
List of Abbreviations
A-BFT Association- Beamforming Training.
AGC Automatic Gain Control.
AoA Angle of Arrival.
AoD Angle of Departure.
AP Access Point.
APC Access Point Controller.
ATI Announcement Transmission Interval.
BC Beam Combining.
BHI Beacon Header Interval.
BI Beacon Interval.
BPSK Binary Phase Shift Key.
BT Beamforming Training.
BTI Beacon Transmission Interval.
CBAP Contention-based Access Period.
CEF Channel Estimation Field.
CMOS Complementary Metal Oxide Semiconductor.
CRC Cyclic Redundancy Check.
CS Compressive Sensing.
CSI Channel State Information.
CSMA/CA Carrier Sense Multiple Access with Collision Avoidance.
DTI Data Transmission Interval.
EDCA Enhanced Distributed Channel Access.
EDCF Enhanced Distributed Coordination Function.
ES-JPFS Exhaustive Search based Joint Proportional Fairness Scheduling.
FDBs Fingerprint Databases.
FI Fairness Index.
FIFO First Input First Output.
FST Fast Session Transfer.
GPS Global Positioning System.
X
HCF Hybrid Coordination Function.

IoT Internet of Things.
IP Internet Protocol.
IS-JPFS Iterative Search based Joint Proportional Fairness Scheduling.
JPFS Joint Proportional Fairness Scheduling.
LDPC Low density Parity Check.
LOS Line of Sight.
MAC Medium Access Control.
MCS Modulation and Coding Scheme.
MID Multiple Sector Identifier.
MIMO Multi Input Multi Output.
MmWave Millimeter Wave.
MR Maximum Rate.
MUSIC Multiple Signal Classification.
NAV Network Allocation Vector.
NP Non-Polynomial.
OFDM Orthogonal Frequency Division Multiplexing.
PBSS Personal Basic Service Set.
PCP PBSS Control Point.
PFS Proportional Fairness Scheduling.
QoS Quality of Service.
RAN Radio Access Network.
RF Radio Frequency.
RR Round Robin.
RRM Radio Resource Management.
RRS Radio Resource Scheduling.
RSS Received Signal Strength.
RX Receiver.
SC Single Carrier.
SLS Sector Level Sweep.
SNR Signal to Noise Ratio.
XI
SP Service Period.
SPR Service Period Request.
STF Short Training Field.
TDMA Time Division Multiple Access.
TRN Training Field.
TX Transmitter.
UE User Equipment.
VL Visible Light.
WLAN Wireless Local Area Network.
XII
List of Symbols
List of symbols
K Number of mmWave UEs.
M Number of mmWave APs.
N Number of time slots.
The instantaneous achievable data rate at user k from mmWave AP m at time

( )
slot n.
The bandwidth efficiency.
The SNR efficiency.
BW MmWave bandwidth in Hz.
( ) The SNR for user k in case of connect with AP m by beam ID .
The received power of user k which is far from mmWave AP m by distance

and connected to it using transmitted beam ID .
The noise power.
The transmitted power for mmWave AP in dBm.
( ) The mmWave average path loss in dB.
( ) The path loss at a reference distance .
The mmWave path loss exponent.
The mmWave shadowing term.
The standard deviation of .
( , ) The 3D beamforming gain for mmWave TX beam in dB.
The azimuth angle.
The elevation angle.
( ) The gains of TX beam in horizontal direction.
( ) The gains of TX beam in vertical direction.
The center angle of mmWave AP TX beam in horizontal direction.
XIII
List of Symbols
The center angles of mmWave AP TX beam in vertical direction.
The beamwidths of mmWave beam in the horizontal directions.
The beamwidths of mmWave beam in the vertical directions.
users' average data rate.
The vectors of all users' achievable data rates.
( ) The average achievable data rate that can be given to UE k till time slot n.
B The number of mmWave AP TX beams.
Spatial reuse factor.
. The complexity of the conventional IEEE 802.11ad based WLAN.
The complexity of ES-JPFS scheme.
The complexity of IS-JPFS scheme.
The average number of iterations performed by the proposed IS-JPFS scheme

to obtain the convergence point.
XIV
Chapter 1
Introduction
 Background
 Motivation
 Thesis Objectives.
 Thesis Outline.
1
Background
Introduction
1.1 Background
Mobile communication and internet of things (IoT) applications become highly common
recently and they are expected to be widely spread in the near future [1 - 4]. According to Cisco
global mobile data traffic forecast [5], the global mobile data traffic for a month will be nearly
49 exabytes by the year of 2021 as shown in Figure 1-1, and the annual traffic will exceed half
a zettabyte. Moreover, mobile traffic will represent 20 % of total internet protocol (IP) traffic,
and 78 % of the world’s mobile data traffic will be video which need a data rate in range of 20
Mbps to watch a 4K video. Also, 75 billion IoT devices are expected to be connected to the
Internet by 2025 as shown in Figure 1-2 where each device includes sensors that collect data,
interact with the environment, and communicate over a network. To cope with these expected
demands, researchers begin to direct their attention to 5G networks as a solution to provide
massive data rate up to multiple Gbps and ultra-low latency in the range of 1 ms with high
availability, spectral efficiency, energy efficiency, user experience, reliability and security [6 -
9]. This 5G networks can be enabled through one of three axes as shown in Figure 1-3. The
first one is network densifications using heterogenous networks and interworking among
licensed and un-licensed bands [10 - 12]. The second axis is increasing the spectrum efficiency
of the existing 4G network using massive MIMO and cognitive radios [13 - 15].
50
45 Mobile Data Traffic
40
Exabytes per Month
35
30
25
20
15
10
5
0
2016 2017 2018 2019 2020 2021
Year
Figure 1-1: Cisco global mobile data traffic forecast, 2016-2021 [5].
2
Background
90
80 Number of connected IoT devices 75.44
Connected IoT Devices in

70
60 51.11
billions
50
40 35.82
30 26.66
20.35
20 15.41
10
0
2015 2017 2019 2021 2023 2025
Year
Figure 1-2: Cisco forecast for number of connected IoT devices, 2015-2025 [5].
Macro BS
Small Cell
Spectrum Efficiency
• Massive MIMO
• Cognitive Radio Required Capacity UE
UE
UE
Current
Capacity
Spectrum Extension
MmWave Band Light Fidelity Band

Figure 1-3: Enabling 5G network through one of three axes.
Finally, immigration towards higher frequency bands such as the millimeter wave (mmWave)
band [16 - 19] and the visible light (VL) band [20 - 22]. As they have unlicensed wideband
spectrum. But, the release of the concept of tactile internet as one of the new innovations
beyond IoT imposes more challenges on data connectivity concerning both date rate and
latency time [23]. The tactile Internet will enable haptic interaction with visual feedback, with
technical systems supporting not just audiovisual interaction, but also that involving robotic
systems to be controlled with an imperceptible time-lag. A 1-millisecond end-to-end latency is
a predominant assumption in tactile internet applications, which includes, the backhauling
time, the cloud time, the radio access time and the terminals execution time. This puts a big
burden on the radio access network (RAN) to attain tens of Gbps speed with fractions of
milliseconds latency, which clearly opens the door for the era of beyond 5G [24, 25].
3
Background
Moreover, more than 80 % of user data traffic is generated indoors [5], and the main use cases
of the future tactile internet are mainly indoors. Hence, enabling 5G networks using mmWave
band with conventional schemes, i.e., applying channel independent scheduling schemes and
implementing conventional beamforming training algorithms, will be inefficient. Thus,
through this thesis, we present the performance evaluation of mmWave network using different
scheduling schemes, both channel independent and channel sensitive schemes. Also, we will
tackle the concept of mmWave concurrent transmission.
The use of the mmWave band, 30 ~ 300 GHz as enablers of future 5G networks, has much
interest in both academia and industry nowadays [16 - 22]. Because it has several advantages
compared with existing wireless communication technologies, which are the wide available
unlicensed spectrum, high data rate links, low latency and small element sizes, etc. However,
this technology comes with big issues. In the next sub section, a brief description about it will
be introduced with focusing on the main issues faced it.
1.1.1 MmWave Technology
MmWave, which works in 60 GHz band, occupies a good position in the wireless
communication society nowadays due to its wide bandwidth and low latency [16, 17]. IEEE
802.11ad [26] are the standard for 60 GHz wireless local area networks (WLANs) that is
defined to be used during working in mmWave technology. According to this standard, a
maximum end-user data rate of 6.7 Gbps can be achieved by access points (APs) using
orthogonal frequency division multiplexing (OFDM) modulations. However, mmWave band
suffers from harsh penetration and propagation losses due to its high operating frequency [27,
28]. Where according to Friis law, as the operating frequency of the wireless communication
technology increases, the attenuation of the signal highly increases. Moreover, mmWave
communication link has a high probability to be blocked due to human shadowing. Thus,
directional transmission using antenna beamforming becomes a necessity in this case [26], [28
- 30] as shown in Figure 1-4. This beamforming is widely utilized based on large-scale
mmWave antenna arrays to expand the coverage area of mmWave AP [30], thanks to the rapid
progress in complementary metal-oxide-semiconductor (CMOS) radio frequency (RF)
integrated circuits. Moreover, the interference in mmWave based networks can be cut down
based on these highly directional beams, hence it becomes a noise-limited network.
4
Background
MmWave AP
Antenna beam
User MmWave coverage area
Figure 1-4: MmWave AP antenna beamforming.
1.1.1.1 Directional communication
Beamforming process needs an estimation of mmWave channel to efficiently determine the

beam angles of departure (AoD) and angles of arrival (AoA) that can provide user with
maximum gain hence maximum received power [31 - 33].
However, there are difficulties in estimating the massive mmWave multi-input multi-output
(MIMO) channel. Hence, beamforming training (BT) using switched antenna arrays with a
predefined structured codebook is typically used for accomplishing mmWave beamforming
[34 - 41]. IEEE 802.11ad standard defines an exhaustive search BT [26], in which the
transmitter (TX) and receiver (RX) search over all available beam patterns to find out the best
one for constructing the mmWave link. Obviously, sharp (pencil) beams guarantee higher gain
(i.e., wider coverage) but at the expense of extremely high BT complexity. For instance, if 3D
beamforming using 5° (2.8°) beamwidth is utilized, mmWave AP should examine nearly 1200
(3700) different TX beam patterns to cover all its surrounding area. This huge number of beams
switching increases setup time and consumes high power. Also, the overall system throughput
is decreased particularly if short beacon frames are used in mobility scenarios. Hence, initial
access, i.e., discovery and association, consumes very long time and high power due to the
exhaustively search BT. This issue opens the door to the researchers to design BT schemes
other than the conventional exhaustive search scheme.
Motivated by the fact that mmWave communication is almost a line-of-sight (LOS)

communication system, UE localization can play an important role in reducing the complexity
of mmWave BT [42 - 52]. This can be obtained by limiting the searched beams to that only
can cover the area within which the UE is most properly located. Obviously, the complexity of
5
Background
location-based BT is mainly depending on the accuracy of the used localization technology. In

other words, if the localization error is much larger than the used mmWave beamwidth, still
higher number of beams switching are needed. Consequently, other radio resource management
(RRM) can be effectively achieved, such as mmWave concurrent transmissions, beams mutual
interference mitigation, fast session transfer (FST), etc.
1.1.1.2 Scheduling Schemes in MmWave
As it is defined by IEEE 802.11ad standard [26], mmWave AP has the responsibility of RRM
functions such as beamforming, user allocation, handover schemes, mobility management and
radio resource scheduling (RRS). It utilizes hybrid medium access control (MAC) protocols to
handle these functions. For MAC layer, IEEE 802.11ad supports contention-based access, i.e.,
random access using carrier sense multiple access, scheduled channel time allocation, and
dynamic channel time allocation. The last two methods depended on using time division
multiple access (TDMA) to share radio resources among active mmWave users. Scheduled
MAC is more provable than contention-based scheme because it is more power efficient than
random access and eliminate collisions between active users thus, we will use it through this
thesis. The final aim from designing a TDMA scheduler is to efficiently distribute time slots
between users that associated with a certain mmWave AP. Hence, utilizing an efficient
scheduling scheme along with a robust modulation technique in RRM improves the entire
system performance. There are several scheduling schemes have been implemented in TDMA
systems which will be discussed in detail in the next chapter. Although, mmWave standard
discussed MAC protocol, it didn’t standardize radio resources allocation scheme to be
implemented in mmWave networks. Thus, there is a high need to answer the question of how
a certain scheduler performance will be if it is used in a mmWave network under different
conditions and scenarios.
1.1.1.3 MmWave Concurrent transmission
Because the coverage area of mmWave AP is so limited due to its natural characteristic, a
multiple number of mmWave APs should be deployed to fully cover a typical indoor
environment, e.g., conference hall, stadium, and hotel's lobby, utilizing spatially concurrent
mmWave links inherent in directional transmissions. In concurrent transmission scenarios [53
- 56], all AP-user equipment (UE) configurations using all possible TX/RX beam patterns
6
Motivations
should be examined to find out the optimal mmWave concurrent links having lowest mutual
interferences. Hence, multi-Gbps data rates can be provided for all mmWave UEs inside the
indoor area simultaneously. But users’ association using the exhaustive search BT and based
on the maximum received power [26] prevent mmWave APs from establishing optimal
concurrent links with the UEs as shown in Figure 1-5. Moreover, this turns to be
computationally infeasible especially using too sharp beams.
MmWave AP 1 MmWave concurrent links MmWave AP 2
UE
UE MmWave AP 3
UE
UE UE
Figure 1-5: MmWave concurrent transmission.
MmWave APs may start transmitting directional beacon frames for performing the exhaustive
search BT simultaneously, or an AP may perform the exhaustive search BT while its nearby
APs are involved in directional data transmissions, causing a lot of packet collisions and
interferences especially with the existing data links. Moreover, as users are associated based
on maximum received power and they are randomly accessing the medium, high mutual
interferences may occur between the randomly established links. This highly deteriorates
spatial reuse characteristics of mmWave WLANs getting it unable of providing massive data
transmissions for all users inside the indoor area if working solely. The aforementioned issues
will be discussed in more details in the literature review chapter.
1.2 Motivations
Because of the large swath spectrum and low latency that provided by mmWave band, the
usage of this technology becomes a persistent demand to enable 5G networks. However, it
faces many technical issues that hinder its working and reduce its performance in terms of total
system data rate, delay, packets losses, etc. Also, the highly effect of blocking probability in
mmWave communication opens the door for researchers to study its effect over the mmWave
channel in different scenarios. The ratified mmWave standard didn’t define a specific
7
Thesis objectives
scheduling scheme to be used in mmWave based 5G network. Thus, this inspired us to evaluate
the performance of different scheduling schemes in mmWave.
Moreover, in indoor areas such as stadiums, conference halls, and hotels’ lobbies, there is a big
need of multiple massive data rate links to handle the users’ requirements of IoT, tactile internet
and cloud computing applications. But these multiple links produce high interference between
each other. Hence our research is directed to solve the problem of mmWave concurrent
transmission in indoor area. Because this issue highly reduces the performance of the mmWave
network specially in crowded environments.
1.3 Thesis objectives
This thesis tackle mmWave technology when it is used as a data layer for downlink connection.
It is categorized into two main topics: the first one interests in discussing the concept of radio
resource scheduling (RRS) where the aim of RRS is defining the best scheduling criteria to be
used in mmWave to efficiently schedule mmWave UEs along available resources. While, the
second topic is discussing the problem of selecting the mmWave concurrent links and how it
can be solved.
The contributions of this thesis are:
 Firstly, the performance evaluation of radio resource scheduling schemes in mmWave

is presented in terms of total system data rate and fairness between users. This study
proves that the channel sensitive scheduling schemes, e.g., proportional fairness
scheduling, is more efficient than channel independent schemes, e.g., round robin, when
they are used in mmWave networks. Because mmWave channel is highly dynamic
channel as it will be discussed in the next chapter.
 Secondly, a joint proportional fairness scheduling optimization problem is formulated.
Because, the solution of this problem using the exhaustive search scheme is highly
complex. Thus, an iterative search based joint proportional fairness scheduling scheme
is introduced to highly reduce the complexity of selecting the optimum mmWave
concurrent links that can be scheduled at the same time. This scheme can find a sub-
optimal solution to this problem with low complexity which will be proven using
numerical analysis.
8
Thesis Outline
1.4 Thesis Outline
In Chapter 2 a brief introduction about the previous work in 5G networks is presented. Firstly,
we show the work that has been achieved to enable 5G networks using mmWave band
subsequently discussing the main concepts that have been tackled in IEEE 802.11ad standard
as physical layer, network architecture and medium access control layer. Also, the challenges
faced mmWave such as the beamforming training is introduced. Then, scheduling schemes,
both channel independent and channel sensitive, is presented in detail. Also, we present the
importance of providing network with multiple mmWave concurrent links which namely,
mmWave concurrent transmission, and present the schemes used to enable it.
In Chapter 3, the performance evaluation of radio resource scheduling schemes in mmWave is

presented in terms of total system data rate and fairness between users under different channel
scenario, i.e., with considering blocking probability and without considering it. This study
concludes to that the channel sensitive scheduling schemes, e.g., proportional fairness
scheduling, is more efficient than channel independent schemes, e.g., round robin, when they
used in mmWave networks under the same conditions.
Chapter 4 presents the problem of selecting the mmWave links that used for mmWave
concurrent transmission and puts this problem as a joint proportional fairness scheduling
optimization problem. Then, it provides a direct solution to this problem based on exhaustive
search. But this solution needs a long time and high complexity. Hence, an iterative search
scheme is proposed to find a sub-optimum solution to this problem with reducing complexity
and obtain nearly the same performance of exhaustive search-based solution.
Finally, in Chapter 5, the main contributions of this thesis are summarized as follows:
conducting the performance of scheduling schemes in mmWave network under different
channel scenarios and using various mmWave beamwidths, improving the performance of
mmWave concurrent transmission based on proposing an iterative search based joint
proportional fairness scheduling scheme. Furthermore, the future work is described.
9
Chapter 2
Literature Review
 Millimeter Wave Technology.

 Scheduling Schemes in MmWave.
 MmWave Concurrent Transmission.
10
Introduction
Literature Review
2.1 Introduction
This chapter presents a summary about mmWave technology for 5G networks. Firstly, a brief
introduction about mmWave technology is given. Then, the main concepts in mmWave
standard, i.e., IEEE 802.11ad, that will be used through this thesis, is presented in some details.
These concepts are physical layer, network architecture, medium access control layer and
beamforming training. Secondly, the previous works in scheduling schemes that can be used
in mmWave network are introduced where we classified schemes based on its relationship with
mmWave channel to channel independent schemes and channel sensitive schemes. Finally,
mmWave concurrent transmission is presented where it is used to provide multiple massive
speed data rate links to UEs which will become a dominant need in the future.
2.2 Millimeter Wave Technology
MmWave is the wireless communication technology that can work in millimeter wavelength
bands, e.g., 28 GHz, 60 GHz, and 73 GHz. These bands are characterized by wide and
unlicensed available spectrum in range of 2 GHz. Hence, mmWave can provide multi Gbps
data rate. Thus, mmWave attracts high attentions from industry, academia, and standardization
societies [16 - 19]. According to the mmWave standard, IEEE 802.11ad [26], a maximum end-
user data rate of 6.7 Gbps can be achieved by mmWave access point (AP) using orthogonal
frequency division multiplexing (OFDM) modulations. But, 60 GHz band differs from 2.4 and
5 GHz bands in terms of signal propagation where mmWave signal suffers from high
penetration and propagation losses, e.g., oxygen absorption [27, 28]. Also, mmWave
communication has a quasi-optical propagation behavior where the line of sight (LOS) path
considered as the dominated received signal path [28]. Hence, mmWave signal is more likely
to be blocked during propagation due to human shadowing.
Thus, it is a necessary to apply directional transmission with the aid of antenna beamforming
[29, 30]. Moreover, because of the increasing demand of high-speed data rate links, limitation
in mmWave transmission range and a high mmWave blocking probability must be handled
11
Millimeter Wave Technology
specially in indoor areas. Thus, a WLAN with multiple mmWave APs must be installed in
indoor areas to fully cover it by establishing multiple spatially mmWave concurrent links [53,
55]. In the next sub sections, we will discuss the concepts of mmWave including physical layer,
network architecture, medium access control layer and beamforming training in more details.
2.2.1 MmWave physical layer
The ratified standard, IEEE 802.11ad [26], presents three different physical (PHY) layers for
different application use cases. For low signal to noise ratio (SNR) operation, e.g.,
beamforming, the Control PHY is designed. While, the single carrier (SC) PHY enables power
efficient and low complexity transceiver implementation. The low-power SC PHY option
replaces the low-density parity check (LDPC) encoder by a Reed-Solomon encoder for further
processing power reduction. On the other side, the OFDM PHY provides high performance in
frequency selective channels achieving the maximum 802.11ad data rates up to 7 Gbps.
Despite having different PHYs, all of them share the same packet structure with common
preamble properties. Specifically, the same Golay sequences are used for the preamble training
fields. Also, a common rate 3/4 LDPC structure is used for channel encoding. Moreover,
802.11ad defines a single bandwidth of 2.16 GHz, which is 50 times wider than the channels
available in 802.11n and roughly 14 times wider than the channels defined in 802.11ac. The
single IEEE 802.11ad packet structure is shown in Figure 2-1. The packet consists of typical
IEEE 802.11 elements, e.g., a short training field (STF) and a channel estimation field (CEF)
that is also used for auto-detection of the PHY type. They are followed by the PHY header and
the PHY payload that is protected by a cyclic redundancy check (CRC). Finally, optional
automatic gain control (AGC) and training (TRN) fields, unique to IEEE 802.11ad, might be
appended. These are used for the beamforming mechanism that will be described.
PHY Payload
PHY MAC
STF CEF MAC Payload CRC AGC TRN
Header Header
Figure 2-1: IEEE 802.11ad packet structure.
12
To provide robust discovery and detection, the control PHY has a longer STF than the SC and
OFDM PHYs, comprising 48 Golay sequences, each 128 samples long. The SC and OFDM
PHY only use 17 Golay sequences for the STF. The channel estimation field that follows the
STF has 9 Golay sequences. The OFDM PHY uses a different combination of Golay sequences
in the CEF to distinguish between OFDM and single carrier modulation.
The Control PHY defines modulation and coding scheme (MCS0). It implements a 32 sample
Golay spreading sequence along with rate 1/2 LDPC encoding (spread from the common rate
3/4 LDPC code) to extend range and reliability for management frames, giving a throughput
of 27.5 Mbps. The Control PHY uses π/2-differential binary phase shift key (BPSK)
modulation to further enhance robustness to distortion like phase noise. The mandatory Control
PHY defines the minimum rate that all devices use to communicate before establishing a high
rate beamformed link. It is used for transmitting and receiving frames such as beacons,
information request and response, probe request and response, sector sweep, sector sweep
feedback, and other management and control frames.
The SC PHY (MCS 1-12) and low power SC PHY (MCS 25- 31) allow for low complexity
and energy efficient transceiver implementations with a throughput of up to 4.62 Gbps. The
lowest SC data rate is 385 Mbps (MCS 1). It is implemented using BPSK modulation and rate
1/2 code with a symbol repetition of two. All modulation types use π/2 rotation to reduce the
peak-to-average power ratio for BPSK and to enable gaussian minimum shift keying (GMSK)
equivalent modulation. To provide interoperability between different device types, MCS 1-4
are mandatory for all devices. These four MCS’s are all based on π/2-BPSK modulation. MCS
2, 3, and 4 use code rate 1/2, 5/8, and 3/4, respectively.
The OFDM PHY (MCS 13-24) is an optional mode for maximum throughput at the cost of a
more complex and energy intensive transceiver structure. The OFDM PHY type utilizes 64-
QAM and a rate 13/16 code to achieve the highest 802.11ad data rates of up to 6.75 Gbps. In
order to keep transceiver complexity and energy consumption low, mobile and low-cost
devices are likely to implement only single carrier PHYs. In contrast, stationary devices with
fixed power supply and high throughput requirements (access points, wireless displays)
implement the full spectrum of MCSs including complex OFDM transceivers.
13
2.2.2 MmWave Network Architecture
This sub section describes the changes made in mmWave network architecture that defined by
IEEE 802.11ad standard. First, the changes to the beacon interval (BI) is introduced. Then, a
new network type, which is personal basic service set (PBSS), is presented. After that, the
description of the network and schedule announcement mechanisms is introduced.
2.2.2.1 Beacon Interval
IEEE 802.11 standard in low frequency bands, e.g., 2.4 and 5 GHz, organizes the medium
access through periodically reoccurring beacon intervals that are initiated by a single beacon
frame transmitted omnidirectionally by the access point (AP) or coordinating station. The
beacon announces the existence of a Wi-Fi network and carries further management data. The
rest of the BI is used for data transmissions between stations. The length of a BI is limited to
1000 ms, but typically chosen in the range of 100 ms. While longer BI durations increase the
connection delay for nodes waiting for the beacon, a longer interval reduces management frame
transmission and increases throughput.
The IEEE 802.11ad standard extends this concept in several ways to cope with the challenges
of mm-Wave propagation. First, a beacon interval as shown in Figure 2-2 is initiated with the
beacon header interval (BHI) that replaces the single beacon frame of legacy Wi-Fi networks.
The BHI facilitates the exchange of management information and network announcements
using a sweep of multiple directionally transmitted frames. The BHI sweeping mechanism
overcomes increased attenuation and unknown direction of unassociated devices. Additional
functionality of the BHI is described later. The BHI is followed by a data transmission interval
(DTI), which can implement different types of medium access. The schedule and medium
access parameters, which are necessary for stations to participate in a BI, are announced by the
central network coordinator, the PBSS Control Point (PCP) or AP, during the BHI. This ensures
that stations receive this information even though no broadcasting mechanism is available.
Beacon Interval (BI)

Beacon Header Interval (BHI) Data Transmission Interval (DTI)
BTI A-BFT ATI CBAP or SP CBAP or SP … CBAP or SP Time
Figure 2-2: IEEE 802.11ad beacon interval structure.
14
A typical beacon interval, consisting of BHI and DTI, is shown in Figure 2-2. The BHI consists
of up to three sub-intervals. First, the beacon transmission interval (BTI) comprises multiple
beacon frames, each transmitted by the PCP/AP on a different sector to cover all possible
directions. This interval is used for network announcement and beamforming training of the
PCP/AP’s antenna sectors. Second, the association beamforming training (A-BFT), is used by
stations to train their antenna sector for communication with the PCP/AP. Third, during the
announcement transmission interval (ATI), the PCP/AP exchanges management information
with associated and beam-trained stations. While communication during BTI and A-BFT uses
MCS 0 to increase range for untrained beams, communication during the ATI takes place with
beam-trained stations and thus is more efficient.
The DTI comprises of one or more contention-based access periods (CBAPs) and scheduled
service periods (SPs) where stations exchange data frames. While in CBAP multiple stations
can contend for the channel according to the IEEE 802.11 enhanced distributed coordination
function (EDCF), an SP is assigned for communication between a dedicated pair of nodes as a
contention free period.
2.2.2.2 Personal Basic Service Set
IEEE 802.11ad introduces the PBSS, where nodes communicate in an ad-hoc like manner.
However, one of the participating nodes takes the role of the PBSS Control Point. This PCP
acts similar to an AP, announcing the network and organizing medium access. This centralized
approach allows the directional network and schedule announcement process described in the
next section to be used for an ad-hoc like network. The PBSS network has been introduced to
satisfy new applications targeted by IEEE 802.11ad such as, for example, wireless storage and
peripherals or wireless display usage. For these applications, usually no preinstalled
infrastructure exist, and communication takes place between a set of personal devices.
An ad-hoc like network with a centralized controller poses two main challenges. First, for
energy-constrained devices, increased power consumption at the PCP penalizes a single device
while a fair sharing of the energy costs is desirable. Second, outage of the PCP paralyzes the
entire PBSS. To respond to these challenges, a PCP Handover procedure is defined. This
procedure can be used for explicit (initiated by the current PCP) or implicit (after PCP becomes
unavailable) handovers. Further, when selecting between a set of possible PCPs the unique
15
capabilities of PCP candidate stations are considered to choose the PCP providing the most
complete number of services to the network.
2.2.2.3 Network and Schedule Announcements
Network announcements in legacy IEEE 802.11 are traditionally propagated periodically,

using beacon frames by the AP. Due to the limited antenna gain of quasi-omnidirectional
mmWave transmissions the coverage range is severely restricted. Consequently, the beacon is
sent as a series of directionally transmitted beacon frames. To have the largest possible range,
the beacon frames are transmitted at the most robust MCS (MCS 0).
IEEE 802.11ad also specifies additional signaling for network scheduling and beam training
appended to every beacon frame. Collectively, this results in a significantly increased overhead
in comparison to legacy WiFi. Thus, it becomes critical to control the amount of information
that is transmitted in each BTI. In addition, transmissions during the A-BFT, which also use
(MCS 0), create overhead reoccurring with every BI where the A-BFT is present. The overhead
problem gets especially relevant when short BI durations are applied for delay critical
application as, for example, video streaming.
The IEEE 802.11ad amendment defines number of counter strategies. First, it is possible to
split a beacon sweep over several BIs. This, however, increases the time a node needs to setup
its link to the PCP/AP, as not every direction is served at every BI. The result is an increased
association delay. Second, it is possible to periodically schedule BIs without A-BFT, which
also results in additional association delays. Third, IEEE 802.11ad introduces the ATI. During
the ATI, beam-trained and associated nodes can be served with management data using
individually addressed directional transmitted frames encoded with a more efficient MCS.
Thus, it is possible to move information from the spectrally inefficient beacon frames to the
frames transmitted during the ATI, limiting beacons to the minimal information necessary.
Also, for beacon intervals with split beacon sweeps, stations that do not receive a beacon, miss
network and timing information. Without this information, stations cannot participate in a BI.
Implementing an ATI solves this problem, as scheduling and management information is
transmitted individually to associated stations.
16
2.2.3 MmWave Medium Access Control Layer
In contrast to legacy Wi-Fi, IEEE 802.11ad uses a hybrid MAC approach to address its various
use cases. The standard supports contention-based access, scheduled channel time allocation
and dynamic channel time allocation. The latter two schemes correspond to TDMA and polling
mechanisms. The polling-based access shares similarities with the IEEE 802.11ad PCF mode
but is adapted to directional transmissions and provides a higher flexibility when it comes to
distribution of resources among the nodes. The scheduled allocation mechanism extends the
traffic stream concept known from the IEEE 802.11 hybrid coordination function (HCF) to
request time shares of the DTI for TDMA like medium access. In the following, the three
methods are described in detail.
2.2.3.1 Contention based Medium Access
Medium access in CBAPs follows IEEE 802.11 enhanced distributed channel access (EDCA),
including traffic categories to support quality of service, frame aggregation and block
acknowledgments. However, when using contention-based access with directional antennas,
the problem of deafness arises. A deaf node does not receive directionally transmitted
information due to misaligned antenna patterns. A detailed description of the effect can be
found in [6].
While the beam training process in IEEE 802.11ad prevents deafness for intended
transmissions, it poses a problem for carrier sensing during contention-based access and can
lead to increased collisions. A further problem for contention-based access is that a receiver
typically does not know from where a signal comes. Thus, usage of quasi-omni-directional
beam patterns is necessary, which reduces link budget and throughput. The contention based
medium access in IEEE 802.11ad is adapted for directional medium usage and multi-MAC
usage. This includes support for multiple network allocation vector (NAV) timers (one per peer
station), which allows a transmission to be initiated to a peer device where the NAV for that
device is zero, even though the NAV for another peer device might be nonzero. Details about
802.11 EDCA and its use in 802.11ad can be found in [3, 8].
2.2.3.2 Dynamic Channel Time Allocation
IEEE 802.11ad defines a dynamic channel time allocation mechanism that implements polling-
based channel access. Dynamic channel time allocation is an extension of the IEEE 802.11
17
PCF mode. It provides higher flexibility in resource allocation (polled stations request channel
time instead of just transmitting one frame) and adaptation to directional communication.
Polling based channel access brings several advantages for mm-Wave communication. First,
due to the centralized approach with a PCP/AP, stations are aware of the direction of incoming
signals. Thus, the deafness problem that affects contention-based access is prevented and quasi-
omni directional receive patterns can be avoided. Second, centralized scheduling at a PCP/AP
also helps to efficiently react to burst downstream traffic, as dynamic scheduling can be adapted
in the course of a BI.
In contrast, pseudo-static scheduling described in the following section, can only announce
modified allocation parameters with the beginning of every BI. When applying the dynamic
allocation mechanism, the medium access during DTI is organized as follows. The PCP/AP
acquires the medium and sends a series of polling frames to associate UEs. This is answered
with a block of service period requests (SPR) used by the polled stations to request channel
time. The PCP/AP allocates the available channel time according to these requests, announcing
each allocation with a separate grant period, consisting of individual Grant frames for the
stations involved in the allocation.
IEEE 802.11ad foresees integration of the dynamic allocation mechanism into both CBAPs
and SPs. When integrated in a CBAP, associated stations try to acquire the medium and may
interfere with the dynamic allocations. To prevent this, the PCP/AP makes use of prioritized
medium access using the short PIFS inter frame spacing and the channel is protected by
extension of the frame duration fields. This extension causes nodes that overhear a frame to
assume the channel to be occupied until the time specified in the duration field. This mechanism
is used, such that Polling and SPR frames protect the polling phase, while every dynamic
allocation is protected by its preceding grant frames.
To simplify the scheduling mechanism and reduce implementation complexity, dynamic

allocations are scheduled back to back, with every allocation immediately following its Grant
period. To reliably reach the nodes that are involved in an allocation, individual directional
frames are transmitted during the grant period. In case of an allocation between PCP/AP and a
station, only one grant frame is sent to the non-PCP/AP station. When not, all available channel
time is allocated dynamically, the PCP/AP can repeat the entire polling process. In case of
18
integration into a CBAP, remaining channel time can also be used for carrier sense multiple
access with collision avoidance (CSMA/CA) access.
An example for three polled stations is shown in Figure 2-3. The PCP/AP commences a polling
phase at the beginning of the DTI, transmitting a polling frame for every associated station,
which is answered with a series of three SPRs by the stations. The second station requests
communication with another nonPCP/AP station while stations one and three intend to
communicate with the AP (not shown). The resulting allocations are scheduled back to back;
each proceeded by a grant period. In case of communication with the AP, the grant period
consists of one frame, otherwise of two. The time until which preceding frames protect the
channel is indicated by separating lines.
Polling Phase
for 3 stations Grant Period
Allocation Allocation Allocation

BHI PP SPRs
STA-AP STA-STA STA-AP
Channel protection points indicated by frame duration fields
Figure 2-3: Dynamic channel allocation.
2.2.3.3 Pseudo-static TDMA Channel Time Allocation
During pseudo-static channel time allocation, SPs that reoccur every beacon interval are
dedicated exclusively to a pair of communicating nodes. Accessing the channel using this
TDMA mechanism provides reliability and is the best way to comply with quality of service
demands. Further, the schedule of SPs is propagated by the PCP/AP to all associated stations.
Thus, every node that is not communicating during a SP can go into sleep mode, which allows
for efficient power saving.
For pseudo-static medium allocation, the concept of traffic streams for IEEE 802.11 HCF, as
described in [8], is extended. A traffic stream is defined as a flow of MAC service data units
that is to be delivered subject to certain quality-of-service parameters, characterized by a traffic
specification. The IEEE 802.11ad amendment defines stations to use traffic specifications to
request scheduling of pseudo-static channel allocations at the PCP/AP. A requesting station
19
defines the properties of its traffic demand in terms of allocation duration and isochronous or
asynchronous traffic characteristic. Calculating the allocation duration, requires a complete
beam trained link with known rate between source and destination. An isochronous traffic
stream results in pseudo static SP allocations that satisfy a constant rate of reoccurring payload
(typical, e.g., for wireless display applications) with certain latency demands. Asynchronous
traffic streams, in contrast, satisfy non-reoccurring payload demand. A typical example
application are rapid file downloads. The actual schedule that includes the requested allocations
is broadcasted by the PCP/AP in an extended schedule element in the next BTI or ATI.
2.2.4 MmWave Beamforming Training
MmWave beamforming training is a main concept in mmWave transmission hence the ratified
mmWave standard, IEEE 802.11ad standard [26], takes it into consideration. IEEE 802.11ad
standard defines exhaustive search BT, where mmWave AP (UE) searches on all available
mmWave TX (RX) beams to find out the best mmWave TX (RX) beam. These TX and RX
beams utilized as a TX/RX beam pair to establish a mmWave connection between AP and UE.
This MmWave BT process split into three phases the sector level sweep (SLS) phase, the
multiple sector identifier (MID) sub phase, and the beam combining (BC) sub phase.
TX: Beam scan RX: Quasi-omni TX: Quasi-omni RX: Beam scan
UE (RX) AP (TX)
AP UE (RX)
(TX)
(a) (b)
TX: Beam scan (for selected beams) RX: Beam scan (for selected beams)
AP (TX)
UE (RX)
(C)
Figure 2-4: MmWave BT process a) SLS phase, b) MID sub phase and c) BC sub phase.
20
As shown in Figure 2-4. (a), the mmWave AP steers its TX beam to all directions while UE
fixed its antenna in quasi-omni mode hence the appropriate TX beams can be selected. Then,
in the MID sub phase as shown in Figure 2-4. (b), the mmWave UE steers its RX beam to all
directions while AP fixed its antenna in quasi-omni mode hence the appropriate RX beams can
be selected. These suitable TX and RX beams is used in BC sub phase as shown in Figure 2-4.
(c), to find out the optimum TX/RX beam for establishing high speed mmWave connection
between AP and UE.
Although exhaustive search BT guarantee finding out the best TX/RX beam pair that achieves
the maximum received power between mmWave AP and UE, it needs high setup time and
power consumption specially when sharp beams, i.e., beams with small beamwidth, are used.
For instance, if 3D beamforming using 5° (2.8°) beamwidth is utilized, mmWave AP should
examine nearly 1200 (3700) different TX beam patterns to cover all its surrounding area. This
huge number of beams switching increases setup time and consumes high power. Moreover,
the overall system throughput is decreased particularly if short beacon frames are used in
mobility scenarios. Hence, initial access, i.e., discovery and association, consumes very long
time and high power due to the exhaustively search BT. These issues open the door to the
researchers to design BT schemes other than the conventional one. Generally, these BT
schemes can be classified into two main categories as following:
2.2.4.1 Beamforming training based on mmWave channel estimation
MmWave BT based on channel state information (CSI) estimation aims to estimate the main
features of the mmWave channel, e.g., angle of departure (AOD), angle of arrival (AOA), path
gains, etc., using BT then steers the TX/RX beams accordingly. Based on the mmWave CSI
estimation, different BT schemes were proposed in literature, e.g., hybrid analog/digital BT
based channel estimation [31 - 33] and 2-D beam-space multiple signal classification (MUSIC)
BT [57]. Also, compressive sensing begins to occupy a main position as a scheme to estimate
the mmWave channel using its sparsity nature as presented in [58, 59], hence the mmWave BT
complexity can be relaxed while improving its performance. Moreover, location-based BT
using compressive sensing (CS) based channel estimation is proposed in [49, 50]
Hybrid analog/digital BT using CS based channel estimation is proposed in [31]. In this

algorithm, antenna beamforming design is depended only on the quantized beam steering
21
directions to directly approximate the channel’s dominant singular vectors instead of designing
the precoding vectors as linear combinations of the steering vectors of the known angles of
arrival/departure. Hence, it implicitly considers the hardware limitations. However, this
technique suffers from the problems of low coverage, high outage and low spectral efficiency
because it is based on adaptive beamwidth BT.
CS using continuous basis pursuit (CBP) has been proposed in [58] to overcome the
performance degradation of the BT scheme given in [31] while obtaining better performance,
but it still has a low coverage probability. In [57], a 2-D beam-space MUSIC BT is used to
enhance the BT performance compared to the schemes based on CS based channel estimation,
but this scheme still suffers from high complexity.
Location-based mmWave channel estimation using CS has been proposed in [49] and [50] to
efficiently estimate the mmWave channel while reducing the required BT complexity. In [49],
the UE localization error is used to define the searching angle of the sensing matrix, i.e., the
ranges of the AoDs and AoAs of the estimated channel. Based on this, a single level BT is
conducted with all measurement beams are located within the defined ranges of AoDs and
AoAs while using a fixed beamwidth value.
In [50], instead of only using a single level BT, a multi-level BT is conducted using multi-
resolution sensing matrices, to estimate the mmWave channel using CS. In which, the
beamwidth of each beam searching level is also adjusted based on the uncertainty of UE
position area. Although the scheme given in [50] is much better than that given in [49] from
complexity and performance point of views, it suffers from a near-far problem. In which, as
the UE becomes far from the mmWave AP, the uncertainty of UE position area may not cover
all mmWave channel paths resulting in losses in the estimated channel and then poor
performance. The situation becomes more worse if this estimated channel path becomes
blocked as no alternative paths are estimated.
Hence, authors in [60] propose an adaptive location-based mmWave beamforming using CS

based channel estimation to overcome on the limitations of the previous two scheme. In which,
both the statistics of the angular spread of the mmWave channel and the uncertainty of UE
position area are taken into consideration while constructing the multi-resolution sensing
matrices used for the adaptive multi-level BT process. Where, the uncertainty of UE position
22
area is used to define the number of used BT levels besides determining the initial ranges of
the AoDs and AoAs of the estimated channel at each BT level. At each BT level, the adaptive
BT process is conducted till the expected number of channel paths is achieved or the maximum
expected angular spread of the estimated channel is reached. Then, the search is conducted
among the different BT levels till a predefined received power at the UE is obtained, or the BT
level with the narrowest available beamwidth is attained. This scheme is used to reduce the
number of used BT vectors while estimating enough channel paths and obtaining a sufficient
received power at the UE, even if the UE is located far from the mmWave.
2.2.4.2 Beamforming training without mmWave channel estimation
Although, BT schemes based on channel estimation provide a good performance and

acceptable complexity, it is still in-efficient to estimate channel paths to design beamforming
antenna. Moreover, this process needs high overhead on the system in the form of information
exchange between mmWave AP and UE. Hence, estimating the whole mmWave CSI for
constructing the beamformer seems to be in-efficient solution. This is especially when using
pencil beams, which is the main concern of this study. For example, when using 5° beamwidth
to cover a searching angle of 180° using 3D beamforming, about 600 antenna elements are
required at both TX and RX. This necessitates the estimation of 600 × 600 MIMO CSI for
efficiently constructing the TX/RX beamformer, which is infeasible in real scenarios. Thus,
schemes that based on the predefined codebook, can provide efficient BT process specially
when sharp beams are used [60].
Iterative search BT scheme [35] relaxes the BT complexity over the exhaustive search BT by
using hierarchical multi-level beam widths in the BT process. This scheme begins by using the
widest available TX/RX beams, i.e., quasi-omni antenna patterns, for covering the space
around a WiGig AP/UE. Thus, narrower beams having higher beamforming gains are used to
scan the space covered by the widest TX/RX beams. This process is repeated till reaching the
finest available TX/RX beams for constructing the mmWave link. Hence, this scheme divided
into three levels, AP to UE linking, sector level searching, and beam level searching. Although
this BT strategy highly relaxes the BT complexity compared to the exhaustive search BT, it
suffers from outage. This is because, the wider beams used at the earlier stages have lower
beamforming gain causing AP/UE misdetection. Moreover, in the pencil beams scenario, it
still needs a high number of beams switching increasing its BT complexity.
23
Numerical search, such as the Rosenbrock algorithm proposed in [38, 39] is used to reduce the
mmWave BT complexity compared to the exhaustive/iterative search BT schemes while
obtaining a sufficient beamforming gain. This scheme based on defining the beam switching
process as a global optimization problem in a two-dimension plane that is formed by the
potential beam pattern index. But the analytical gradient information of this objective reward
function is unavailable, hence Rosenbrock numerical algorithm is adopted to implement beam
searching process by implicitly approaching and exploiting the gradient descent direction
through the numerical pattern-search mechanic. Inspired by the appealing conception of small-
region dividing and conquering, this pre-search algorithm can efficiently reduce the search
scope and hence improve the success probability. However, still 90 beam switchings are needed
to obtain the same performance as the exhaustive search BT using narrow beams of 5°
beamwidth, which is still too large number.
The Turbo-like BT scheme, using the Tabu search was proposed in [40] based on the
information reciprocity between the mmWave AP and UE. Hence, this scheme increases the
system overhead. Moreover, the complexity of the technique depends on the probability of
selecting good initial solution and stopping criterion. Thus, non-proper selection of these
parameters may result in highly complicated BT process. In [41] another scheme is discussed
where coding the mmWave beams outperforms the exhaustive search BT in reducing the
mmWave BT complexity using multiple beams with unique signatures simultaneously.
However, this scheme is non-applicable to non-uniformly weighted array.
Because of the essential demand of using directional antenna array and motivated by the wide
spread of user context information services, BT schemes based on these services are widely
used nowadays in determining the best TX/RX beam directions [42 - 52]. If mmWave AP (UE)
has a full knowledge about the actual location of mmWave UE (AP), it just needs to adjust its
TX (RX) beam towards UE (AP) directly without any further step for establishing the mmWave
downlink (uplink), respectively. Hence, user context information, e.g., localization, based BT
schemes highly reduce the computational complexity of mmWave BT. These schemes use the
localization services to know the position of UE then restricting the searching space in a group
of TX/RX beams instead of searching over all available TX/RX beams.
In [44], global position system (GPS) is used to determine UE location hence enabling a LOS
mmWave beamforming. However, it faces two main issues, high energy consumption that
24
Scheduling Schemes in MmWave
makes GPS based BT scheme unsuitable for the limited power resources of the UEs, and GPS
is only valid for outdoors environment. To overcome to these challenge, the low energy
consumptions terrestrial networks, i.e., LTE and Wi-Fi networks, can be used for UE
localization as discussed in [46] hence enabling fast BT.
However, these schemes suffer from high localization errors comparable to the coverage of
mmWave beam specially if sharp beamwidth is used, e.g., the localization accuracy using Wi-
Fi technology reaches on the average to 2.5m. This error necessitates estimating a large number
of TX/RX beams for conducting BC sub phase to find out the best pair. Also, this number of
estimated TX/RX beams is incredibly increased when using pencil beams. The authors in [47]
used Wi-Fi received signal strength (RSS) fingerprints to obtain a high accurate localization
service in indoor environment to highly relaxing the mmWave BT complexity. However, the
scheme needs a labor-intensive work to construct mmWave and Wi-Fi radio maps to be stored
in the form of fingerprint databases (FDBs). Moreover, the scheme is only valid for highly
static environment, as the FDBs should be frequently constructed if the environment changes.
2.3 Scheduling Schemes in MmWave
The scheduler function in the MAC layer of the mmWave AP is a promising research topic
because efficient scheduling guarantee saving precious mmWave band [61, 62]. This function
is to divide radio resources, i.e., time slots in TDMA scenario, between user equipment (UEs)
which associated to that AP and introduce a scheme for resource assignment. The total system
performance, the data rate assigned to each UE and the fairness between UEs is affected by the
scheme used for scheduling. Hence, an evaluation of the efficiency of various scheduling
schemes is needed before any practical deployment of one of them in mmWave networks. To
efficiently evaluate system performance, mmWave channel conditions with considering the
parameters that have high influence on it, e.g., beamforming with different beamwidths and
blocking probability, must be taken into consideration.
A comprehensive study about different wireless networks scheduling schemes is discussed in

[63]. Scheduling schemes can be categorized to two branches namely channel independent
scheduling and channel sensitive scheduling.
The schemes in first category don’t take the channel effect into considered, i.e., share resources
without knowing any information about users’ channels. Though, these schemes are simple in
25
Scheduling Schemes in MmWave
implementation and need no overhead signaling to feedback channel states, they maybe waste
resources to users with bad channel conditions or service only greedy users who required high
data rate at the expense of other users.
On the other side, the channel sensitive scheduling schemes need feedback signaling to know
the associated users’ channels hence using this information on sharing resources to the users.
These schemes are highly recommended in wireless technologies that have highly dynamic
channels such as the case in mmWave communication. Fortunately, during beamforming
training (BT) process in mmWave, users send feedback signal that contains their RX power to
be used in BC sub phase. Hence, this signal can be used as a channel state information that is
needed for scheduling process in channel sensitive scheduling schemes. Thus, feedback signal
in mmWave doesn’t add any overhead packets to the system which means no add delay.
2.3.1 Channel Independent Scheduling Schemes
First-in-First-out (FIFO) and Round robin (RR) scheduling schemes [63, 64] are two examples
of channel independent scheduling where channel is assumed to be time invariant and error-
free. In FIFO system, the first UE associated with AP is the one who is scheduled first by AP.
If this UE requires high data rate, it will be scheduled along too many time slots and wastes the
opportunity from other UEs, i.e., the UEs that will associate with AP later, to be scheduled
early hence they will suffer from delay.
RR scheduling scheme is simple and more efficient than FIFO in terms of delay because it
doesn’t give a priority to UEs based on their arrival time also obtains fairness between UEs.
RR distributes time slots between UEs equally and in ordered manner without considering
channel state information (CSI) of UEs which means that it gives the same preference metric
to all UEs associated to mmWave AP thus it schedules resources fairly. However, RR fails to
provide UEs with high data rate. Also, it wastes resources on UEs with bad channel conditions.
Moreover, mmWave channel is a dynamic channel with blocking probability. Thus, the channel
independent scheme can’t be used in mmWave networks because it will be in efficient in terms
of total system data rate due to variations in channel conditions through time slots. For
comparison and study purpose, we will consider RR as an example of channel independent
schemes that nominated to be used in mmWave communications.
26
MmWave Concurrent Transmission
2.3.2 Channel Sensitive Scheduling Schemes
MmWave channel is highly affected along time and UE locations. Hence, channel sensitive
scheduling schemes [63] is preferable for mmWave networks as they are based on channel
estimation. Some of these schemes aim to maximize the total system data rate, e.g., maximum
rate (MR). While, others target fairness between associated UEs, e.g., Max-min.
In MR scheduling scheme, UEs with good channel conditions that have high signal-to-noise
ratio (SNR) above a certain threshold are scheduled during a certain frame while UEs with bad
channel conditions that have SNR below a certain threshold does not scheduled. That means
AP prefers a group of UEs with good CSI to other UEs which loss the opportunity to be serviced
with high speed data link mmWave network. Although, MR scheme provides the highest
system data rate between all scheduling schemes, it totally ignores obtaining fairness between
UEs. Thus, it is considered to be unfair resource scheduling scheme [63].
While, some schemes obtain optimum fairness between UEs, e.g., max-min scheme. In Max-
min scheme, at each time slot the system schedules the UE that have the bad SNR. This
criterion increases the data rate provided to UEs with bad channel condition hence equalizing
between them at the expense of decreasing the total system rate. Although, this scheme
achieves the optimum fairness between UEs, it obtains the worst system performance in terms
of total system rate because it wastes a lot of time resources on UEs with bad channels [63].
2.4 MmWave Concurrent Transmission
Due to the intensive need for high speed transmission links specially in crowded indoor areas,
e.g., stadiums and airports, the concept of mmWave concurrent links occupy a good position
in researchers' work. Ideally, concurrent transmission guarantees high data rates as the total
rate of system will equal to the sum rate of all links. But this assumption isn't valid in practical
due to the mutual interference generated from concurrent links. To mitigate the mutual
interference between concurrent links, the network needs to select the optimum mmWave
concurrent links that can be established and scheduled at the same time slot [53 - 56].
2.4.1 IEEE 802.11ad Based WLAN with scheduled medium access control
IEEE 802.11 ad standard with scheduled medium access control (MAC) can be applied for
establishing mmWave concurrent transmission [26]. But this conventional scheme can't
27
guarantee selecting the optimal mmWave concurrent links. Because it selects each concurrent
link separately based on the maximum received power criteria without considering the
interference generated from the other mmWave concurrent links that will be connected at the
same time. For example, as shown in Figure 2-5, the ratified IEEE 802.11 ad standard based
WLAN selects , and , to establish mmWave concurrent transmission for UE1 and
UE2, respectively, but these two concurrent links are high interfering links.
On the other side, as shown in Figure 2-6, , and , are low interfering links, hence
using them guarantees higher total system data rate. Thus, if WLAN takes mutual interference
into consideration, these low interfering links can be selected as mmWave concurrent link
hence improving the total system performance. Moreover, using scheduled MAC with RR
scheduling mechanism in each mmWave AP separately to allocate time resources to UEs, can't
guarantee efficient resource allocation hence wasting the opportunity to benefit from all
available capacity. Besides, UEs with highly correlated channel might be selected to connect
with their associated mmWave APs simultaneously which increase the probability of mutual
interference occurrence between concurrent links hence reducing the total system data rate. So,
selecting the optimal mmWave concurrent links is a must in mmWave concurrent transmission
scenarios [53].
MmWave AP 1 MmWave AP 2
High interfering links
UE 1 UE 2
Figure 2-5: Non-optimal establishment of mmWave concurrent links based on IEEE 802.11ad
standard using scheduling MAC.
Low interfering links
UE 1 UE 2
Figure 2-6: Optimal mmWave concurrent links.
28
2.4.2 IEEE 802.11ad Based WLAN with random access scenarios

Applying IEEE 802.11ad standard with random access scenarios using carrier sense multiple
access/collision avoidance (CSMA/CA) [53] can't achieve acceptable performance. Because,
the carrier sense (CS) function might not detect that the medium is busy because of the
predominant nature of directional transmissions and receptions. Moreover, mmWave APs may
start transmitting directional beacon frames for performing the exhaustive search BT
concurrently, or an AP may perform the exhaustive search BT while its nearby APs are in data
transmissions mode, as shown in Figure 2-7, which causes a lot of packet collisions and
interferences especially with the existing data links. Also, as shown in Figure 2-8, high mutual
interferences between the randomly established mmWave links can occur because the UEs are
randomly accessing the medium. This selection of non-optimal mmWave concurrent links
deteriorates the spatial reuse characteristics of mmWave WLANs specially when multiple
number of mmWave operated separately.
Colliding beams
UE 2 UE 1
Exhaustive search BT MmWave data

transmission
Figure 2-7: Establishment of non-optimal mmWave concurrent links based on IEEE 802.11ad using
random access scenarios where Packet collision caused because of un-coordinated mmWave BT.
High interfering links
UE 2 UE 1
MmWave data transmission
Figure 2-8: Establishment of non-optimal mmWave concurrent links based on IEEE 802.11ad using
random access scenarios where Packet collisions caused because of non-optimal user association.
According to that, Concurrent transmission in the mmWave band has several issues. Firstly,
the mmWave antenna array forms a specific beam pattern according to the weight vector
29
calculated based on the measuring channel state information, i.e., AoD and AoA. This
generates high calculation load and large overhead. Secondly, the concurrent links make
beamforming even more difficult because the mutual interference between links is unknown
until the best TX and RX beam patterns for all the concurrent links are established. Finally, no
complete medium access control (MAC) protocol to setup multiple directional communication
links is available. To reduce the complexity of beamforming training and minimize overhead
in CSI, the schemes that don't based on CSI which discussed in the previous section can be
utilized, e.g., numerical search, and Wi-Fi localization-based BT scheme.
However, applying these schemes directly can't guarantee efficient mmWave concurrent
transmission. Because, BT process still needs long setup time specially when pencil beams,
i.e., beams with 5° beamwidth, is used. Also, mutual interference problem still exists as these
schemes didn't take interference generated from other concurrent link into consideration where
in these schemes, mmWave AP considers establishing single communication link. Hence, an
optimum selection of mmWave concurrent links is needed which we will discuss in chapter 4.
30
Chapter 3
Performance Evaluation of Proportional

Fairness Scheduling in MmWave Networks
 Introduction.
 MmWave System Model.
 Proportional Fairness Scheduling in MmWave.
 Simulation and Results.
 Summery.
31
Introduction
Performance Evaluation of Proportional Fairness

Scheduling in MmWave Networks
3.1 Introduction
MmWave has a tenuous channel where it suffers from extremely high propagation and
penetration losses that highly attenuate users’ received power. Thus, mmWave standard, i.e.,
IEEE 802.11 ad standard, proposed directional transmission using antenna beamforming based
on large-scale mmWave antenna arrays to increase the mmWave received (RX) power and
widen the coverage area of the mmWave access point. But, due to the effect of human
shadowing especially in indoor communication areas, the mmWave link becomes highly
susceptible to be blocked which highly degrades the signal strength [65, 66]. This blocking
happens for a short time period and then disappears, i.e., for 0.2 to 1 ms, depending on the speed
of users in indoor areas which makes mmWave channel highly dynamic. This dynamic channel
profoundly effects on the entire system performance of the mmWave networks. To limit the
effect of this blocking, an efficient radio resources scheduling (RRS), i.e., assigning limited
resources to each active user, with considering the changes in mmWave channel is needed.
As it is defined by IEEE 802.11ad standard, mmWave AP has the responsibility of radio

resources management functions such as beamforming, user allocation, handover schemes,
mobility management and RRS. It utilizes hybrid MAC methods to handle these functions. For
MAC layer, IEEE 802.11ad supports contention-based access, i.e., random access using carrier
sense multiple access, scheduled channel time allocation, and dynamic channel time allocation.
The last two methods depended on using time division multiple access (TDMA) to share radio
resources among active mmWave users. Scheduled MAC is more provable than contention-
based scheme because it is more power efficient than random access and eliminate collisions
between active users thus, we will use it.
The final aim from designing a TDMA scheduler is to efficiently distribute time slots between
users that associated with a certain mmWave AP, hence, utilizing an efficient scheduling
scheme along with a robust modulation technique in RRM improves the entire system
32
MmWave System Model
performance. There are several scheduling schemes have been implemented in TDMA systems.
For instance, round robin scheduling scheme is used where time resources shared between
users in an equal manner [64]. Maximum rate (MR) scheduling is used to maximize total
system throughput. Max-min algorithm which achieves optimum fairness between scheduled
user [63]. Also, proportional fairness scheduling (PFS) scheme is used to obtain fairness
between active users without scarifying the total system rate [63].
Although, mmWave standard discussed MAC protocol, it didn’t standardize radio resources
allocation scheme to be implemented in mmWave networks. Thus, the question of how a
certain scheduler performance will be if it is used in a mmWave network with different
conditions and scenarios is needed to be answered. In this chapter performance evaluation of
four common TDMA scheduling schemes, which are RR, MR, Max-min, and PFS, is studied
and presented. This study is performed under different mmWave beamwidths, e.g., 30° , 20° ,
10° and 5° beamwidth. Also, the main parameter that effects on the strength of mmWave
channel that is the blocking probability, is considered with different values, 0.7, 0.5 and 0.3.
Internet MmWave AP
Edge Router WLAN
Gateway Wireless
Router Device
MmWave
Coverage
Figure 3-1: MmWave network architecture.
3.2 MmWave System Model
MmWave, IEEE 802.11 ad, standard for wireless local area network (WLAN) is used to study
the scheduling problem. WLAN usually consists of M mmWave APs and K user equipment
(UEs) where each AP is connected to the network with high speed back-hull links, e.g., fast
Ethernet. Indoor areas with high traffic such as meeting halls and hotel lobbies are a good
example where mmWave APs can be installed. The coverage range of mmWave AP is nearly
33
MmWave System Model
20 m, so we only considered implementing one AP, indicating it by m, in hotel lobby as a case

study assuming there is no interference produced from other APs installed in other parts of hotel.
Figure 3-1 shows this mmWave network architecture. For simplicity, we considered 3D
beamforming only in mmWave AP and considering the UEs antenna models are quasi-omni.
Which means mmWave BT is needed only in transmitter side, i.e., in mmWave AP only.
In IEEE 802.11ad standard, TDMA is considered as a multiple access scheme to handle

multiuser use case. Where mmWave network frame defined as beacon interval (BI) that shown
in Figure 3-2. Beacon header interval (BHI) initiates BI where it is responsible for exchanging
the management information and network announcements using multiple directionally
transmitted frames. The BHI consists of up to three sub-intervals. First, the Beacon
Transmission Interval (BTI) which used for network announcement and beamforming training
of the mmWave AP’s antenna sectors. Second, the Association Beamforming Training (A-
BFT), which is used by UEs to train their antenna sector for communication with the AP. Third,
during the Announcement Transmission Interval (ATI), the AP exchanges management
information with associated and beam-trained UEs.
The BHI is followed by a data transmission interval (DTI) that contains N time slots where UEs
are scheduled to send their data during these slots. The schedule and medium access
information, which are necessary for UEs, are announced by AP during the BHI.
Figure 3-2: IEEE 802.11ad beacon interval structure.
Regarding to mmWave propagation model, it can be defined as following [28]:
The instantaneous achievable data rate ( ) at user k from mmWave AP m at time slot n, can
be presented as:
( )= log 1+ ( ) , (3.1)
34
MmWave System Model
where and are the bandwidth efficiency and the SNR efficiency, respectively.
indicates to the mmWave bandwidth in Hz, while ( ) indicates to the SNR for user k in
case of connection with AP m by beam ID and can be written as:
( )
( )= , (3.2)
where donates to the noise power. is the received power of user k which is far from
mmWave AP m by distance and connected to it using transmitted beam ID . Hence, this
transmitted power can be calculated by using mmWave pathloss model and written as [28]:
( )[ ]= ( , )+ ( )+ , (3.3)
( )[ ]= ( ) + 10 log + , (3.4)
where indicates the transmitted (TX) power for mmWave in dBm, ( ) refers to the
mmWave average path loss in dB and ( ) is the path loss at a reference distance , we
assume this reference distance equals to 1m. and indicate to the mmWave path loss
exponent and the mmWave shadowing term, respectively. Where ∽ 0, as is the
standard deviation of . ( , ) indicates the 3D beamforming model for mmWave gain of
TX beam in dB. Hence, ( , ) can be written as [28]:
( , )[ ]= [ ] − min [− ( ), + ( ) , Υ], (3.5)
Υ = 12 + [ ], (3.6)
1.6162
[ ] = 20 log , (3.7)
sin
2
where and are the azimuth and the elevation angles, defining the angle of departures (AoDs)
of mmWave channel, while ( ), ( ) are the gains of TX beam in horizontal and vertical
directions, that can be expressed as [28]:
−
( ) = −min 12 ,Υ , (3.8)
35
Proportional Fairness Scheduling in MmWave
−
( ) = −min 12 ,Υ , (3.9)
where and are the centres angles of mmWave AP TX beam, while and
are the beamwidths of mmWave beam in the vertical and the horizontal directions.
3.3 Proportional Fairness Scheduling in MmWave
The wireless network usually goals to different targets, increasing the total system rate or
obtaining fairness among its users or achieving users’ quality of service (QoS). Here, we
interest in obtaining the first two targets. Indeed, these two goals opposite each other where
simpler schedulers that achieves the best system rate obtains the worst fairness and vice versa.
Hence, we propose using PFS as a scheduler scheme in mmWave network because it aims to
maximize the total system data rate with providing acceptable fairness between UEs.
PFS scheme is used for solving an optimization problem which aims to select the UE that can
be scheduled at each time slot to obtain the target cost function that is maximizing the total
system rate over users' average data rate with considering the fairness. This optimization
problem can be formulated as [67]:
max log( ), (3.10)

∈
Subject to ∈
where K is the total UEs that associated with mmWave AP, and k is the index of a certain UE.
While indicates the vectors of all users' achievable data rates.
Here, ( ) = ( ( ): ∈ ) is the indicator vector for the UE assignment with size K at time
slot n. where ( ) = 1, means user k associate with mmWave AP at time slot n, and ( )=
0 elsewhere. Hence, (10) is reformulated as a per time slot UE assignment problem as:
Assignment: max ∑ ∈ log ( ) , (3.11)

( )
S. t. ∑ ∈ ( ) = 1.
36
Proportional Fairness Scheduling in MmWave
where ∑ ∈ ( ) = 1, indicates to only single UE can be associated to mmWave AP at time

slot n. While, ( ) indicates to the average achievable data rate that can be given to UE k till
time slot n, it can be written as [67]:
1 1 (3.12)
( )= 1− ( − 1) + ( ),
where ( ) is the user k instantaneous achievable rate when UE, k associated with AP m at
time slot n.
Following Taylor series expansion:
1
log ( ) ≈ log ( − 1) + ( )− ( − 1) , (3.13)
( − 1)
Thus using (3.12), (3.13) can be expressed as:
1 ( )
log ( ) ≈ log ( − 1) + −1 , (3.14)
( − 1)
( )
Hence, maximizing log ( ) is equivalent to maximize the preference metric ( )
.
Hence, the solution of this optimization problem can be expressed as:
( )
1, = max , (3.15)
( )= ∀ ⊂ ( − 1)
0, ℎ .
The sequence of PFS scheme in mmWave AP can be summarized as following: Firstly, AP

sends a directional beacon interval in all directions at BTI. Then, during A-BFT period, AP
collects the feedback signaling from all UEs. This signal contains the instantaneous CSI of UEs
which reflects to the instantaneous data rate ( ). After that, AP begins to schedule
associated UEs a long all N time slots according to (3.15). At every time slot, AP keep tracking
the average achievable data rate ( ) for each UE k according to (3.12). Hence, mmWave
AP gives a priority to the UE with the best preference metric at time slot n, which satisfies
(3.15) to be scheduled in this time slot. The PFS scheme handle each time slot separately and
then update the system at each time slot.
37
Performance Analysis
3.4 Performance Analysis
The performance evaluation of one of the most efficient channel sensitive scheduling schemes
namely PFS is presented in this numerical analysis. While, RR scheme is considered as a
conventional scheme to compare with due to its simplicity and acceptable performance. From
system point of view, the critical factors to evaluate system performance are the total system
rate which indicates that the system achieves the best benefit from its available radio resources
and the proportion of fairness between system users which indicates to the variety in users’
rates. On the other side, users are interested in their QoS, e.g., delay and packet loss ratio.
In this study, we interested in overall system performance. Hence, we will present total system
data rate and fairness index between users as a performance evaluation while other metrics will
be considered in the future work.
Table 3-1: Study Parameters.

Parameters Value
Room dimensions 20m × 20m × 4m
UEs height from floor 1m
MmWave AP height from floor 4m
Number of mmWave APs 1
TX power of mmWave AP 10 dBm
MmWave Beamwidths 30°, 20° , 15°, 10°
UE antenna model Quasi-omni
MmWave Bandwidth 2.16 GHz
Path loss exponent for mmWave LOS 2.36
MmWave shadowing term 10.4
Bandwidth efficiency 0.6
SNR efficiency 1
Blocking probability 0.7, 0.5, 0.3
3.4.1 Simulation scenario
Table 3-1 shows the main simulation parameters used through this numerical analysis. A
medium size hotel lobby with dimension 20m × 20m × 4m is selected as a target area. Where
a UE height from the floor is 1m. We used one mmWave AP with 10 dBm TX power for
simplification also its coverage range of nearly 20m is acceptable for the aforementioned
38
medium indoor area [51]. The blocking probability assumed to be 0.7, 0.5, and 0.3 for mmWave
line of sight path [65, 66]. MmWave UEs are distributed in target area with the uniform
distribution. Other simulation parameters are defined in Table 3-1. The simulation analysis is
performed with 100,000 Monte Carlo trials.
3.4.2 Results and discussion
Figures 3-3, 4, 5, 6, 7, 8, 9 and 10 show the performance evaluation of the proportional fairness
scheduling and round robin schemes at different number of UEs associated to mmWave AP in
terms of total system data rate in Gbps and Jain’s fairness index between mmWave users. Where
UE association process is assumed to be done separately in AP based on the maximum received
power as it is defined in the standard. MmWave TX beams are adjusted at four different
beamwidths 30° , 20° , 15° and 10° . Also, two mmWave channel models are considered, without
blocking probability and with blocking probability equals to 0.7, 0.5 and 0.3 [65], [66].
Where Jains' fairness index (FI) is evaluated to measure the variance in data rates provided to
UEs in the system, i.e., determine if UEs receiving a fair share of radio resources, which can
be expressed as:
2 K
= K rk 2 (3.16)
k=1 k=1
where rk is the data rate offered to user k a long all time slots N from mmWave AP and can be
written as:
rk = ( ) ( ), (3.17)
Figures 3-3, 4, 5 and 6 prove that PFS scheme obtains total system data rate larger than that
obtained by RR scheme due to the optimization problem that is done to maximize the summation
of average users' rate. Where the changes in mmWave channel due to UEs location and blocking
probability is taken into consideration. In RR scheme, the total system data remains constant
even if the number on UEs increases. While, in PFS scheme, increasing number of UEs
increases the total system data rate. This increment seems to be small in the figures because of
the scale of Gbps.
For instance, in 30° beamwidth with no blocking probability scenario shown in Figure 3-3,
increasing number of UEs from 5 UEs to 10 UEs, increases the total system data rate with 20
39
Mbps. As shown in these figures, decreasing the mmWave beamwidth, i.e., increasing its
directivity hence its gain, decreasing the difference in data rate between PFS and RR schemes.
Because the effect of the distance of UEs according to beam centers becomes small, i.e., there
is a small difference between received powers at UEs.
2.4
RR w/o Blocking
2.2
Total system data rate in Gbps
PFS w/o Blocking

2 RR w Blocking 0.5
1.8 PFS w Blocking 0.5
1.6 RR w Blocking 0.3
RR w Blocking 0.7
1.2
PFS w Blocking 0.7
1
5 10 15 20 25 30
Number of UEs
blocking probability against number of mmWave UEs.
3.5
RR w/o Blocking
3.3
PFS w/o Blocking
3.1

2.1
RR w Blocking 0.7
1.9
PFS w Blocking 0.7
1.7
15 20 5 10
25 30
Number of UEs
40
4.2
RR w/o Blocking
4
PFS w/o Blocking
3.8
3.4
PFS w Blocking 0.5
3.2
RR w Blocking 0.3
3

2.4
PFS w Blocking 0.7
2.2
5 10 15 20 25 30
Number of UEs
5.6
5.4 RR w/o Blocking
5.2 PFS w/o Blocking

5
RR w Blocking 0.5
4.8
4.2
PFS w Blocking 0.3
4
3.6
PFS w Blocking 0.7
3.4
15 20 5
25 10 30
Number of UEs
In Figures 3-3, 4, 5 and 6, considering channel without blocking probability, the rate obtained
by PFS scheme is larger than that obtained by RR scheme with nearly 230 Mbps, 250 Mbps,
190 Mbps and 110 Mbps in case of using 30° , 20° , 15° , and 10° beamwidths, respectively.
While, the blocking probability due to human shadowing increases this difference in data rate
between two schemes. For instance, in probability of 0.3, the difference in data rate obtained
41
using the two schemes is increased to be 280 Mbps, 330 Mbps, 220 Mbps and 190 Mbps in case
of using 30° , 20° , 15° , and 10° beamwidths, respectively. In probability of 0.5, the difference
in data rate obtained using the two schemes is increased to be 460 Mbps, 420 Mbps, 400 Mbps
and 330 Mbps in case of using 30° , 20° , 15° , and 10° beamwidths, respectively. Of course,
increasing the probability of blocking increases this difference as shown in figures.
In blocking probability scenarios where the instantaneous rate of UEs changes with time due to
link blockage with probability of 0.7, 0.5 and 0.3, the highly dynamic mmWave channel has a
high effect on the total system rate in RR scheme comparable to that achieved by PFS scheme.
While, in no blocking scenario the only effect on instantaneous rate is produced due to positions
of UEs, i.e., the distance between UE position and the TX beam centers. Hence, PFS scheme
performance is better than RR in two scenarios because it assigns UE that has the best preference
metric, i.e., best channel condition comparable to its average rate, to the certain time slot at each
slot which means it considered the channel effect during scheduling.
Figures 3-7, 8, 9 and Fig. 10 show the Jain’s fairness index between mmWave UEs. As a result
of a trade-off between total system rate and fairness between UEs, PFS scheme achieves fairness
lower than that achieved by RR scheme. However, this fairness between UEs is acceptable as
the system gains more benefit from its resources in terms of total system data rate. The fairness
between UEs increases with the decrease of mmWave beamwidth because small beamwidth
means the beam gain of UEs are near to each other even if UEs are in different locations, i.e., in
different distances from beam centers. Where, as the beamwidth decreases, the coverage area of
the beam decreases as well.
In Figures 3-7, 8, 9 and 10, considering channel without blocking probability, the fairness
between UEs obtained by PFS (RR) scheme is 0.74 (0.87), 0.82 (0.9), 0.87 (0.92) and 0.92
(0.94) when 5 UEs associated with AP then it decreases with increment of UEs in case of using
30° , 20° , 15° , and 10° beamwidths, respectively. In the blocking probability of 0.5 scenario, the
fairness between UEs obtained by PFS (RR) scheme is 0.56 (0.67), 0.7 (0.8), 0.77 (0.85) and
0.83 (0.88) when 5 UEs associated with AP then it decreases with increment of UEs in case of
using 30° , 20° , 15° , and 10° beamwidths, respectively. As the blocking probability increases,
the fairness between UEs decreases because mmWave channel is being more dynamic and has
a wide variety of values.
42
0.9
RR w/o Blocking

Jain's Fairness Index
RR w Blocking 0.5
0.7
PFS w Blocking 0.5
RR w Blocking 0.3
0.6
PFS w Blocking 0.3
PFS w Blocking 0.7

0.4
5 10 15 20 25 30
Number of UEs
Figure 3-7: Jain’s fairness index of PFS and RR schemes with 30° beamwidth with/without blocking
probability against number of mmWave UEs.
0.95
RR w/o Blocking
0.9

0.75
PFS w Blocking 0.5
0.7
RR w Blocking 0.3
0.65

0.5
PFS w Blocking 0.7
0.45
5 10 15 20 25 30
Number of UEs
In 30° , 20° , 15° , and 10° beamwidth cases, PFS scheme reduces the fairness between users with
nearly 0.21, 0.11, 0.08 and 0.05 in blocking probability of 0.5 scenario as a trade-off increasing
the total system rate with a factor of 30%, 17.5%, 13% and 7% comparable to RR schemes. That
means the PFS scheme improves the overall system performance with serving the fairness
between UEs in acceptable level and near to the fairness obtained by RR scheme.
43
Summery
0.95
RR w/o Blocking
0.9
PFS w/o Blocking
0.85
RR w Blocking 0.5
0.8
PFS w Blocking 0.5
0.75
RR w Blocking 0.3
0.7
PFS w Blocking 0.3
0.65
RR w Blocking 0.7
0.6
PFS w Blocking 0.7
0.55
5 10 15 20 25 30
Number of UEs
0.95
RR w/o Blocking
PFS w Blocking 0.5

0.8
RR w Blocking 0.3
PFS w Blocking 0.7

0.65
15 5 10
20 25 30
Number of UEs
3.5 Summery
In this chapter, an optimization problem of one of the most efficient channel sensitive schemes
namely proportional fairness scheduling is formulated to maximize total system rate with
obtaining acceptable fairness between UEs. Performance evaluation of PFS scheme is presented
44
Summery
in terms of total system data rate and fairness index between UEs. RR is the simplest scheme
hence it is considered as the conventional scheme to compare with. The comparison between
PFS and RR scheme is performed in different mmWave beamwidth cases which are 30° , 20° ,
15° and 10° beamwidths under two different mmWave channel scenarios, with and without
blocking probability.
Simulation results show that RR scheme obtains better fairness between UEs. While PFS
scheme achieves a balance between total system rate and fairness between UEs. In blocking
probability scenarios, PFS scheme shows an improvement in total system rate comparable to
RR scheme especially in 30° and 20° beamwidth cases. As a conclusion of this study we found
that the blocking probability highly effects on the performance of mmWave channel hence
channel sensitive scheduling schemes is recommended to be used.
This study performed when one mmWave AP implemented in the network. But, if multiple APs
are installed to provide multi Gbps data rate, the interference between mmWave links will highly
degrade the network performance. Thus, the performance of PFS scheme is needed to be
evaluated in this scenario which is the scope of the next chapter.
45
Chapter 4
Joint Proportional Fairness Scheduling for

 Introduction.
 Joint Proportional Fairness Scheduling Optimization Problem.
 Proposed Iterative Search based Joint Proportional Fairness
Scheduling for MmWave Concurrent Transmission.
 Results and discussion.
 Summary.
46
Introduction
Joint Proportional Fairness Scheduling for MmWave

Concurrent Transmission
4.1 Introduction
The intensive need of enormous data rate and low latency connection will be a big issue in the
upcoming years due to the rapidly growth in using IoT applications and high-speed mobile
networks as discussed in Chapter 4. Specially in downlink scenarios, users need massive data
links to cope with their needs of watching high definition videos, performing their work using
cloud services, playing online games, etc. Hence, 5G networks are widely deployed nowadays
specially in indoor areas. 5G network should contain multiple mmWave APs to enable
mmWave concurrent transmission by scheduling multiple mmWave links at the same time slot.
However, these concurrent links suffer from highly mutual interference which incredibly
reduces the total system throughput as discussed in Chapter 2.
To mitigate the mutual interference between concurrent links, network needs to select the
optimum mmWave concurrent links that can be scheduled at the same time slot. Hence, a joint
proportional fairness scheduling (JPFS) optimization problem for mmWave concurrent
transmission is formulated and discussed in this chapter with considering the mutual
interference between concurrent links. This problem aims to select the optimum mmWave
concurrent links that maximizing the total system data rate by jointly performing user
association and time resource allocation.
This JPFS optimization problem is a non-polynomial (NP) time problem particularly when
network consists of multiple number of mmWave APs and multiple number of mmWave UEs.
Hence, finding its optimum solution using exhaustive search (ES), i.e., searching on all
available mmWave APs-UEs patterns in exhaustive manner, is highly complicated. Thus, an
iterative search based JPFS (IS-JPFS) algorithm is proposed to find a sub-optimal solution to
this NP time problem. Where mmWave concurrent links are selected link by link with
considering the already connected links interferences in an iterative manner until achieving a
convergence point instead of searching on all available APs-UEs patterns to select the optimum
one. This scheme highly relaxes the ES-JPFS optimization problem complexity, reduces its
47
The proposed System Model
total setup time and puts a polynomial time solution to it. Besides, it obtains nearly the same
performance that obtained using the ES-JPFS scheme.
4.2 The proposed System Model
The architecture of the proposed network is shown in Figure 4-1. MmWave band used to
provide downlink services. This network consists of M mmWave APs, K users and an access
point controller (APC). Highly traffic indoor areas such as airports, meeting halls, and stadiums
are examples where this network can be installed. High speed back-hull links, e.g., fibre optic,
are utilized to connect APC with all mmWave APs in the network. In this network, APC
utilized to jointly associate and scheduled UEs to mmWave APs by establishing the optimum
(sub-optimal) mmWave concurrent transmission links that can be scheduled simultaneously.
Also, APC is responsible for synchronization between mmWave APs. Besides, APC is
considered as a gateway between a WLAN and the Internet. Moreover, APC has a full
responsibility on the network as it can perform all network functions such as link establishment,
link maintaining, radio resource management and mobility management.
For simplicity, we considered beamforming only in mmWave APs side and considering the
UEs antenna models are quasi-omni. Which means mmWave BT is needed only in transmitter
side, i.e., in mmWave AP only. This mmWave propagation model has been introduced in the
previous chapter. Hence, we will directly describe the JPFS optimization problem.
Select (optimal) sub-optimal

mmWave concurrent links
MmWave AP
APC
mmWave
concurrent links
MmWave beam
coverage area
UE 6 UE 5 UE 4 UE 3 UE 2 UE 1
Figure 4-1: Proposed WLAN Architecture.
48
Joint Proportional Fairness Scheduling Optimization Problem
4.3 Joint Proportional Fairness Scheduling Optimization Problem
In this section the JPFS optimization problem is formulated. The optimization problem aims to
select the optimum APs-UEs configuration pattern that can be scheduled concurrently to obtain
the target cost function that maximizes the total system rate over users' average data rate
and considering the fairness between users. Hence, this problem can be formulated as [68]:
max log( ) (4.1)

∈
Subject to ∈ ,
where K is the total number of UEs that located in the network, and k is the index of a certain
UE. While indicates the vectors of all users' achievable data rates.
The mmWave has a spatial diversity characteristic due to its directional transmission and the
mmWave standard use time division multiple access (TDMA) as a multiple access technique.
Hence, the JPFS optimization problem shall be solved by APC at each time slot n. Thus, it
converted into a per time slot AP-UE assignment problem where ( )=( ( ): ∈
, ∈ ) is the indicator matrix for the AP-UE assignment with size × at time slot n.
( ) = 1, means user k associate with AP m at time slot n, and ( ) = 0 elsewhere.
Assuming that, the number of UEs K is larger than the number of mmWave APs M existence
in WLAN which is a suitable assumption as indoor environments are crowded by users. Hence,
(4.1) can be reformulated as a per time slot AP-UE assignment problem as:
Assignment: max ∑ ∈ log ( ) (4.2)

( )
Subject to ∑ ∈ ( ) = 1,∀ ∈ .
where ∑ ∈ ( ) = 1,∀ ∈ means only single UE can be associated to mmWave AP

m at time slot n. While, ( ) indicates the average achievable data rate that given to user k
up to time slot n which can be written as [68]:
1 1
( )= 1− ( − 1) + ( ), (4.3)
where ( )=∑ ∈ ( ) ( ) is the instantaneous rate that obtained by user k at time

slot n, and ( ) is the user k instantaneous achievable rate when UE, k associated with AP
m at time slot n.
49
Proposed IS based JPFS for MmWave Concurrent Transmission
Following Taylor series expansion:
1
log ( ) ≈ log ( − 1) + ( )− ( − 1) (4.4)
( − 1)
Thus using (4.3), (4.4) can be expressed as:
1 ( )
log ( ) ≈ log ( − 1) + −1 (4.5)
( − 1)
( )
Hence, maximizing log ( ) is equivalent to maximize the preferred metric (
. Hence,
)
the solution of the optimization problem can be expressed as:
⎧ ( )
1, ∀ ∈ = max ,
( )= ∀ ⊂ ( − 1) (4.6)
⎨ ∈
⎩0, ℎ .
where ∀ ⊂ means all users configurations of length M of all network users K. This
solution indicates that the APC selects the user pattern that obtain the maximum of
( )
∑ ∈ ( )
to be scheduled in time slot n, hence for ∀ ∈ the AP-UE assignment
indicator ( ) will be equal to 1. Assuming the network consist of single mmWave AP,
i.e., = 1, (4.6) will become:
( )
1, = max ,
( )= ∀ ⊂ ( − 1) (4.7)
0, ℎ .
4.4 Proposed IS based JPFS for MmWave Concurrent Transmission
The optimization problem that formulated in section 4.3, is a non-polynomial time

programming (NP-hard) problem. To find the optimum solution to this optimization problem
as given in (4.7), APC needs to exhaustively search on all available APs-UEs configuration
pattern using all available mmWave AP beams to find the optimum pattern maximizes the total
!
system data rate. This solution needs a complexity equals to ( )!
, where B indicates the
number of mmWave AP transmitter (TX) beams.
This solution complexity increases non-linearly with the number of deployed mmWave APs
on the network and the number of UEs desire to connect with a high-speed mmWave
50
Proposed IS based JPFS for MmWave Concurrent Transmission
concurrent links. Therefore, this exhaustive search based JPFS (ES-JPFS) scheme needs a high-
power consumption and a long setup time specially in indoor areas which is the use case under
study in this chapter. To overcome on the complexity issue, an iterative search scheme is
proposed to find the sub-optimum solution of the aforementioned JPFS optimization problem.
This iterative search based JPFS (IS-JPFS) scheme highly relaxes the optimization problem
complexity and provide a polynomial time solution for this problem.
In the proposed IS-JPFS, instead of selecting the optimum APs-UEs pattern directly by
searching on all available patterns, the mmWave concurrent links are selected link by link.
Where APC selects each AP-UE link considering the interference produced by the links that
have been already selected before. This producer is performed iteratively till obtaining a
convergence point. In our proposal, the convergence point is obtained when the achievable rate
obtained by selecting APs-UEs pattern in a certain iteration , is smaller than or equal to
the achievable rate achieved by the APs-UEs pattern that selected in the previous iteration
by a factor of ∆ , i.e., ≤∆ . The threshold ∆ is a design parameter
determined based on the required performance and considering the acceptable complexity from
the IS-JPFS scheme. In this chapter, the threshold ∆ selected to be equal to 0.1 but this
parameter will be put under study in our future work.
The proposed IS-JPFS scheme select the sub-optimum APs-UEs pattern not the optimal one as
in the ES-JPFS and maintains nearly the same performance that can be obtained by using the
NP-hard solution but highly relaxes the system complexity. The proposed IS-JPFS scheme
focuses only on the problem of selecting the sub-optimal APs-UEs pattern without considering
the issue of selecting an efficient beam pattern. Where, there are various efficient mmWave
beamforming training algorithms that proposed in literature as in [31 - 52].
The proposed scheme aims to select the sub-optimal APs-UEs pattern at every time slot that
maximizes the total system data rate and obtaining (4.7). Selecting the sub-optimal APs-UEs
pattern means assigning UE k index and beam identification to the user sets and the
beam sets of all M links at every time slot n. Thus, the proposed IS-JPFS can be explained
as follows: Firstly, and are initialized as = 0⃗ and = 0⃗ with length M. In the
first iteration, the first link is selected by performing BT for single link communication, i.e.,
considering there is no interference will be generated from concurrent links. Thus, [1] and
[1] are assigned by the selected UE and beam that obtained (4.6).
51
Results and discussion
Regarding to the following mth links where 1 < ≤ , the APC performs BT for each link
m with considering the interference produced from the links that have been selected before.
Hence, [ ] and [ ] can be updated with the UE and beam that obtained by (4.6). The
first iteration is ended when APC selects all M concurrent links.
In the next iterations, the selection of each link m considers the interference generated from the
other − 1 links since the user sets and the beam sets have been already determined
before. Hence, in each iterative round, the data rate obtained from the modified selected links
is larger than or equal to that obtained from the previously selected links. So, the procedure of
concurrent links selection is repeated until obtaining a convergence point that is ≤
∆ . Algorithm 1 summarizes the proposed IS-JPFS.
Algorithm 1: Iterative search based JPFS for mmWave concurrent transmission
Input: ∆
Start
1. Initialize the user sets = 0⃗, the beam sets = 0⃗, and = 0.
2. Repeat
3. Assign = .
4. For AP = 1 to M ( is AP number index) do
5. set and according to
( )
6. max ∑ ∈ ( )
,
∀ ⊂
7. end
8. Until ≤∆
Stop
Output: , and
4.5 Results and discussion
In this numerical simulation, the performance of the proposed IS-JPFS scheme for mmWave
concurrent transmission is compared with ES-JPFS and the conventional scheme, i.e., IEEE
802.11ad based WLAN, under different scenarios.
52
Table 4-1: Simulation Parameters.
Parameter Value
Room dimensions 20m × 10m × 4m
UEs height from floor 1m
Number of mmWave APs / UEs 6 / 40
TX power of mmWave AP 10 dBm
MmWave Beamwidth 30°
MmWave Bandwidth 2.16 GHz
Bandwidth efficiency 0.6
SNR efficiency 1
Blocking probability 0.5
AP 2
AP 4
10 m
AP 1 AP 6
AP 3 AP 5
20 m
Figure 4-2: Study area.
Table 4-1 shows the main simulation parameters used in the numerical analysis. Where a
medium conference room, which is shown in Figure 4-2, is selected as a target area with
dimensions 20m × 10m × 4m. Also, the human shadowing blocking probability assumed to
be 0.5 for the mmWave line of sight path [65]. The number of mmWave APs is 6 which is
considered suitable number of APs in such indoor area to provide required data rates to users.
A mmWave beamwidth of 30° is used through this study as an example. In the conventional
scheme, UEs are associated to mmWave APs based on the maximum received power while
53
each mmWave AP scheduled its associated UEs separately based on scheduled MAC as
defined by IEEE 802.11ad standard [26]. On the other side, for the proposed IS-JPFS and ES-
JPFS, mmWave UEs are jointly associated and scheduled to mmWave APs in each time slot.
4.5.1 Performance Metrics
In these numerical simulations, we focus on measuring the total system data rate in Gbps. The
spatial reuse factor is used as an indicator for the obtained rate to the average rate in case of
full isolated links and can be written as [69]:
Sum rate of mmWave concurrent links

= (4.8)
Average rate of isolated links
The large value of means low mutual interference between links and efficient spatial reuse
[69], e.g., if 6 mmWave APs exists in WLAN with no mutual interference between their links,
will equal to 6.
Also, users' fairness index (FI) is evaluated to measure the variance between data rates provided
to different users and can be expressed as:
2 K
= K rk 2 (4.9)
k=1 k=1
where rk is the data rate offered to user k a long all time slots N from all mmWave APs and can
be expressed as:
rk = ( ), (4.10)
Moreover, the total system computational complexity in number of beams switchings is

presented for the three schemes, where the complexity of the conventional IEEE 802.11ad based
WLAN can be expressed as:
. = , (4.11)
where APC synchronize between mmWave APs, so AP begins its BT process on all its beams
when the previous AP ends it BT process.
While the ES-JPFS computational complexity can be written as:
54
!
= (4.12)
( − )!
Where APC searches on all available APs-UEs configuration patterns to find optimum one.
Regarding to the proposed IS-JPFS scheme, the complexity can be defined as:
= , (4.13)
where indicates to the average number of iterations performed by the proposed IS-JPFS
scheme to obtain the convergence point which has been defined in the Section 4.4.
4.5.2 Simulation Results
Figure 4-3, 4 and 5 show the performance of the proposed IS-JPFS, ES-JPFS, and IEEE
802.11ad based WLAN, i.e., the conventional scheme, when different number of mmWave
APs are used for mmWave concurrent transmission at a certain user density which is 1 user/m2.
While, Figure 4-6 presents the comparison between the three scheme in terms of the total
system computational complexity in number of beams switchings.
As shown in Figure 4-3, 4 and 5, the performance achievable by ES-JPFS outperform the
conventional scheme in terms of the total system data rate, spatial reuse factor and user's
fairness index as a result of optimization problem performed where interference between
concurrent links is mitigated. For instance, in case of 6 mmWave APs are used, the proposed
ES-JPFS outperform the conventional scheme by 2.1-time increase in total system date rate, a
2-time increase in spatial reuse factor and nearly a 1.5-time increase in the user's fairness index.
9 Conv.
Total system data rate in Gpbs
IS-JPFS
8
ES-JPFS
7
6
5
4
3
2
1
3 4 2 5 6
Number of used APs
Figure 4-3: The performance of the proposed IS-JPFS, ES-JPFS and IEEE 802.11ad based WLAN
(Conv.) in terms of total system data rate using different number of mmWave APs.
55
5.0
Conv.
4.5 IS-JPFS
4.0 ES-JPFS
Spatial reuse factor

3.5
3.0
2.5
2.0
1.5
1.0
0.5
3 4 25 6
Number of used APs
(Conv.) in terms of spatial reuse factor using different number of mmWave APs.
0.90
0.85
users' fairness index
0.80
0.75
0.70
0.65
0.60
Conv.
0.55 IS-JPFS
ES-JPFS
0.50
3 4 25 6
Number of used APs
(Conv.) in terms of user’s fairness index using different number of mmWave APs.
However, the ES-JPFS significantly increases the total system complexity comparable to the
conventional scheme as shown in Figure 4-6. Hence, the proposed IS-JPFS is proposed to
overcome to this complexity problem with acceptable performance. As shown in Figure 4-6,
the proposed IS-JPFS algorithm highly reduces the total system complexity comparable to ES-
JPFS particularly for large numbers of mmWave concurrent links. Besides, it achieves nearly
the same performance of the ES-JPFS schemes as presented in Figures 4-3, 4 and 5. For
example, in case of 6 mmWave links are used, the reduction of total system complexity of the
proposed IS-JPFS over using the ES-JPFS is more than 99%. In addition, the IS-JPFS scheme
needs only 5-times the complexity of the conventional scheme.
56
1.E+16
Total system computational complexity

1.E+14
1.E+12
1.E+10
1.E+08
Conv.
1.E+06 IS-JPFS
ES-JPFS
1.E+04
1.E+02
1.E+00
2 3 4 5 6
Number of used APs
Figure 4-6: The total system complexity of the proposed IS-JPFS, ES-JPFS and the conventional
scheme when different number of mmWave APs are used.
Figures 4-7, 8 and 9 show the comparison in performance between the proposed IS-JPFS, ES-
JPFS, and the conventional scheme at different user densities [user/m ] in terms of total
system data rate, spatial reuse factor , and users’ fairness index, respectively. In this
simulation, we select only APs 2, 3, 4, and 5 for mmWave concurrent transmission because
these APs are located close to each other hence there will be a high overlapping between their
coverage areas that means high mutual interference between their concurrent links which
clearly tackle the problem under study. Also, we use certain number of UEs which is 40 users
at every simulated point and shrink the area of interest to satisfy the required user density. At
beginning, UEs are distributed in the whole simulated area to obtain a user density of
40/(20m × 10m) = 0.2 user/m . Then, UEs are distributed in area with dimension of
12.5m × 6.25m to obtain a user density nearly equals to 0.5 user/m , and so on till obtaining
a user density of 2 user/m .
As shown in Figures 4-7 and 8, when the user density increased, the total system data rate and
spatial reuse factor are decreased due to the high increase in mutual interference. While, as a
result of interference, the difference between users' data rates is decreased hence the variance
between decreased which improves the fairness between users as presented in Figure 4-9.
57
Total system data rate in Gpbs

7
6
Conv.
5 IS-JPFS
ES-JPFS
4
2
0.5 0.8 1.1 1.4 1.7 0.22
user density [user/m2]
Figure 4-7: Total system data rate of the proposed IS-JPFS, ES-JPFS and the conventional scheme in
different user densities scenarios.
4.0
3.5
Spatial reuse factor
3.0
2.5 Conv.
IS-JPFS
ES-JPFS
2.0
1.5
0.5 0.8 1.1 1.4 1.7 2 0.2
Figure 4-8: Spatial reuse factor of the proposed IS-JPFS, ES-JPFS and the conventional scheme in
0.90
0.85
Users' fairness index
0.80
0.75
Conv.
0.70 IS-JPFS
ES-JPFS
0.65
0.60
0.5 0.8 1.1 1.4 1.7 20.2
Figure 4-9: Users’ fairness index of the proposed IS-JPFS, ES-JPFS and the conventional scheme in
58
Summary
As shown in Figures 4-7 and 8, using the conventional scheme with increasing the user density
from 0.2 to 2 user/m decreases total system data rate by 30%. Besides, a loss of 50% in
spatial reuse factor form ideal value, where = 4, is obtained at user density 0.2 user/m , and
a loss of 61% is obtained at user density 2 user/m . On the other side, using the proposed IS-
JPFS increases total system date rate by 92% and 65% at user densities 0.2 user/m and 2
user/m , respectively. Moreover, the proposed IS-JPFS nearly obtains the ideal value of
spatial reuse factor at user density 0.2 user/m and only a loss of 30% from ideal value at user
density 2 user/m . Also, the proposed IS-JPFS enhanced users' fairness index by 50% in
comparison with IEEE 802.11ad based WLAN as shown in Figure 4-9.
4.6 Summary
In this chapter, an optimization problem for mmWave concurrent transmission in scheduled

MAC scenario is formulated. This problem aims to select the optimum mmWave concurrent
links can be scheduled at the same time. This JPFS optimization problem needs a non-
polynomial time to find its optimal solution using exhaustive search. Hence, an iterative search
based joint proportional fairness scheduling scheme is proposed to highly relax the complexity
of this problem. In which, APC selects a sub-optimal mmWave concurrent links in an iterative
manner, link by link, where each link considers the interference generated from the concurrent
links that have been selected. According to the numerical analysis, the proposed IS-JPFS
scheme improves the system performance compared to IEEE 802.11ad based WLAN in terms
of the total system data rate, spatial reuse factor , and users' fairness index at the expense of
increasing system complexity.
59
Chapter 5
Conclusions and Future Perspectives
 Conclusions.
 Future Perspectives.
60
Conclusions
Conclusion and Future Perspective

5.1 Conclusions
The persistent demand of massive data rate directs both research and industry eras to work hard
to enable 5G and beyond networks. These networks aim to service the IoT and tactile internet
applications which become rapidly increasing recently. Where, these applications require
massive data rate communication links and ultra-low latency with high availability, spectral
efficiency, energy efficiency, user experience, reliability and security. Enabling these networks
can be done by immigration to unlicensed wideband technologies, i.e., mmWave band.
However, mmWave suffers from technical issues. It needs complex BT process to associate
with user which also reflects on its performance in concurrent transmission scenarios. Besides,
efficient scheduling scheme in mmWave network is needed to handle blocking probability and
changes in dynamic mmWave channel.
This thesis focus on the problems that appear in downlink scenario only, which are the
performance of scheduling schemes in mmWave communication and mmWave concurrent
transmission while other problems such as efficient beamforming training, scheduling schemes
performance in mobility scenarios will be considered in our future work. Thus, the contribution
of this thesis can be summarized as follows:
Firstly, an optimization problem of one of the most efficient channel sensitive schemes namely
proportional fairness scheduling is formulated to maximize total system rate with obtaining
acceptable fairness between UEs. Performance evaluation of PFS scheme is presented in terms
of total system data rate and fairness index between UEs. A comparison between PFS and RR
schemes is performed in different mmWave beamwidth cases which are 30° , 20° , 15° and 10°
beamwidths under two different mmWave channel scenarios, with and without blocking
probability. Simulation results show that RR scheme obtains better fairness between UEs.
While PFS scheme achieves a balance between total system rate and fairness between UEs. In
blocking probability scenarios, PFS scheme shows an improvement in total system rate
comparable to RR scheme especially in 30° and 20° beamwidth cases. As a conclusion of this
study we found that the blocking probability highly effects on the performance of mmWave
61
Future Perspectives
channel hence channel sensitive scheduling schemes is recommended to be used. This study
performed when one mmWave AP implemented in the network. But, if multiple APs are
installed to provide multi Gbps data rate, the interference between mmWave links will highly
degrade the network performance. Thus, the performance of PFS scheme is needed to be
evaluated in this scenario which considered in the following.
Secondly, the optimization problem for mmWave concurrent transmission in scheduled MAC
scenario is formulated to select the optimum mmWave concurrent links that can be scheduled
at the same time. However, this JPFS optimization problem needs a non-polynomial time to
find its optimal solution using exhaustive search scheme. Hence, an iterative search based joint
proportional fairness scheduling scheme is proposed to highly relax the complexity of this
problem. In which, APC selects a sub-optimal mmWave concurrent links in an iterative
manner, link by link, where each link considers the interference generated from the concurrent
links that have been selected. According to numerical analysis that are performed, the proposed
IS-JPFS scheme improves the system performance compared to IEEE 802.11ad based WLAN
in terms of the total system data rate, spatial reuse factor , and users' fairness index at the
expense of acceptable increase in system complexity.
5.2 Future Perspectives
The future work can include other scenarios and use cases in mmWave communication. The
scenarios can be categorized to two parts: the first one is distributed case, i.e., each AP works
separately without knowing any information about other APs exist in the network as chapter 3
considered, and the second case is centralized under the control of APC as chapter4. Thus, the
future work would be:
1) In distributed case: considering mobility scenario, scheduling schemes is needed to
modify with considering the effect of user’s movement where UEs that move with high
speed comparable to coverage area of mmWave beam shall have a low priority to be
scheduled while UEs with average and low speed must have a high priority to be
scheduled in certain sub frame (beacon interval). Also, schemes that reduce the
complexity of beamforming training can be proposed with considering mobility
management.
2) In multiple APs scenario, i.e., mmWave concurrent transmission, conflict beams
causing interference to the existing mmWave links can be localized and eliminated
62
Future Perspectives
from the BT process allowing the establishment of low interference links. Moreover,
optimum (sub-optimal) concurrent links can be selected by searching on a group of TX
beams in each mmWave AP instead of searching over all AP beams hence reducing the
complexity and setup time of concurrent transmission. Several other mmWave
communication issues can be effectively addressed using the proposed network.
63
REFERENCES
REFERENCES
[1] J. Lin, W. Yu, N. Zhang, X. Yang, H. Zhang and W. Zhao, "A Survey on Internet of Things:
Architecture, Enabling Technologies, Security and Privacy, and Applications," in IEEE Internet of
Things Journal, vol. 4, no. 5, pp. 1125-1142, Oct. 2017.
[2] A. Zanella, N. Bui, A. Castellani, L. Vangelista and M. Zorzi, "Internet of Things for Smart Cities,"
in IEEE Internet of Things Journal, vol. 1, no. 1, pp. 22-32, Feb. 2014.
[3] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari and M. Ayyash, "Internet of Things: A
Survey on Enabling Technologies, Protocols, and Applications," in IEEE Communications Surveys
& Tutorials, vol. 17, no. 4, pp. 2347-2376, Fourthquarter 2015.
[4] NOKIA. (3/03/2017). Available: http://info.networks.nokia.com/Indoor-Deployment-Strategies-
White-Paper-LP.html
[5] Cisco, “Cisco visual networking index: global mobile data traffic forecast update, 2016-2021,”
Mar. 2017 [online]. Available: https://www.cisco.com/c/en/us/solutions/collateral/service-
provider/visual-networking-index-vni/mobile-white-paper-c11-520862.html
[6] J. G. Andrews et al., "What Will 5G Be?," in IEEE Journal on Selected Areas in Communications,
vol. 32, no. 6, pp. 1065-1082, June 2014.
[7] A. Osseiran et al., "Scenarios for 5G mobile and wireless communications: the vision of the METIS
project," in IEEE Communications Magazine, vol. 52, no. 5, pp. 26-35, May 2014.
[8] C. Wang et al., "Cellular architecture and key technologies for 5G wireless communication
networks," in IEEE Communications Magazine, vol. 52, no. 2, pp. 122-130, February 2014.
[9] F. Boccardi, R. W. Heath, A. Lozano, T. L. Marzetta and P. Popovski, "Five disruptive technology
directions for 5G," in IEEE Communications Magazine, vol. 52, no. 2, pp. 74-80, February 2014.
[10] N Bhushan, et al., "Network densification: the dominant theme for wireless evolution into 5G," in
IEEE Communication Magazine, Vol. 52, no. 2, pp. 82-89, February 2014.
[11] N. Zhang, N. Cheng, A. T. Gamage, K. Zhang, J. W. Mark and X. Shen, "Cloud assisted HetNets
toward 5G wireless networks," in IEEE Communications Magazine, vol. 53, no. 6, pp. 59-65, June
2015.
[12] C. Yang, J. Li, M. Guizani, A. Anpalagan and M. Elkashlan, "Advanced spectrum sharing in 5G
cognitive heterogeneous networks," in IEEE Wireless Communications, vol. 23, no. 2, pp. 94-101,
April 2016.
[13] T. E. Bogale and L. B. Le, "Massive MIMO and mmWave for 5G Wireless HetNet: Potential
Benefits and Challenges," in IEEE Vehicular Technology Magazine, vol. 11, no. 1, pp. 64-75,
March 2016.
[14] T.L. Marzetta, E.G. Larsson, H. Yang, H.Q. Ngo, Fundamentals of Massive MIMO (Cambridge
University Press, 2016)
[15] S. Lien, K. Chen, Y. Liang and Y. Lin, "Cognitive radio resource management for future cellular
networks," in IEEE Wireless Communications, vol. 21, no. 1, pp. 70-79, February 2014.
[16] T. S. Rappaport, S. Sun, R. Mayzus, H. Zhao, Y. Azar, K. Wang, et al., "Millimeter wave mobile
communications for 5G cellular: It will work!," IEEE access, vol. 1, pp. 335-349, 2013.
[17] M. Xiao et al., "Millimeter wave communications for future mobile networks," in IEEE J. Sel.
Areas Commun., vol. 35, no. 9, pp. 1909-1935, Sept. 2017.
[18] K. Sakaguchi, G. K. Tran, H. Shimodaira, S. Nanba, T. Sakurai, K. Takinami, et al., "Millimeter-
wave evolution for 5G cellular networks," IEICE Transactions on Communications, vol. 98, pp.
388-402, 2015.
[19] K. Sakaguchi and et al. “Millimeter-wave wireless LAN and its extension toward 5G,” in IEICE
Trans. on Commun. vol. E98-B, no.10, pp. 1932-1948 Oct. 2015.
[20] L. Feng, R. Q. Hu, J. Wang, P. Xu and Y. Qian, "Applying VLC in 5G Networks: Architectures
and Key Technologies," in IEEE Network, vol. 30, no. 6, pp. 77-83, November-December 2016.
[21] S. Wu, H. Wang and C. Youn, "Visible light communications for 5G wireless networking systems:
from fixed to mobile communications," in IEEE Network, vol. 28, no. 6, pp. 41-45, Nov.-Dec. 2014.
64
REFERENCES
[22] M. Ayyash et al., "Coexistence of WiFi and LiFi toward 5G: concepts, opportunities, and
challenges," in IEEE Communications Magazine, vol. 54, no. 2, pp. 64-71, February 2016.
[23] M. Simsek, A. Aijaz, M. Dohler, J. Sachs, and G. Fettweis, "5G-enabled tactile internet," IEEE
Journal on Selected Areas in Communications, vol. 34, pp. 460-473, 2016.
[24] G. P. Fettweis, "The tactile internet: Applications and challenges," IEEE Vehicular Technology
Magazine, vol. 9, pp. 64-70, 2014.
[25] M. Maier, M. Chowdhury, B. P. Rimal, and D. P. Van, "The tactile internet: vision, recent progress,
and open challenges," IEEE Communications Magazine, vol. 54, pp. 138-145, 2016.
[26] IEEE 802.11ad standard: “Enhancements for Very High Throughput in the 60 GHz Band," in IEEE
Std 802.11ad-2012 (Amendment to IEEE Std 802.11-2012, as amended by IEEE Std 802.11ae-
2012 and IEEE Std 802.11aa-2012) , vol., no., pp.1-628, 28 Dec. 2012 [Online]
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6392842&isnumber=6392841
[27] T. S. Rappaport, F. Gutierrez, E. Ben-Dor, J. N. Murdock, Y. Qiao, and J. I. Tamir, "Broadband
millimeter-wave propagation measurements and models using adaptive-beam antennas for outdoor
urban cellular communications," IEEE transactions on antennas and propagation, vol. 61, pp. 1850-
1859, 2013.
[28] “FP7-ICT-608637 MiWEBA Project Deliverable D5.1 channel modeling and characterization,”
June 2014. [Online]: http://www.miweba.eu/wp-content/uploads/2014/07/MiWEBA D5.1
v1.01.pdf.
[29] F. Sohrabi and W. Yu, "Hybrid Digital and Analog Beamforming Design for Large-Scale Antenna
Arrays," in IEEE Journal of Selected Topics in Signal Processing, vol. 10, no. 3, pp. 501-513, April
2016.
[30] J. Song, J. Choi and D. J. Love, "Codebook design for hybrid beamforming in millimeter wave
systems," 2015 IEEE International Conference on Communications (ICC), London, 2015, pp.
1298-1303.
[31] A. Alkhateeb, O. El Ayach, G. Leus and R. W. Heath, "Channel Estimation and Hybrid Precoding
for Millimeter Wave Cellular Systems," in IEEE Journal of Selected Topics in Signal Processing,
vol. 8, no. 5, pp. 831-846, Oct. 2014.
[32] L. Zhao, D. W. K. Ng and J. Yuan, "Multi-User Precoding and Channel Estimation for Hybrid
Millimeter Wave Systems," in IEEE Journal on Selected Areas in Communications, vol. 35, no. 7,
pp. 1576-1590, July 2017.
[33] S. Payami, M. Shariat, M. Ghoraishi and M. Dianati, "Effective RF codebook design and channel
estimation for millimeter wave communication systems," 2015 IEEE International Conference on
Communication Workshop (ICCW), London, 2015, pp. 1226-1231.
[34] L. Zhou and Y. Ohashi, “Efficient codebook-based MIMO beamforming for millimeter-wave
WLANs,” in Proc. IEEE PIMRC, Sep. 2012, pp. 1885–1889.
[35] J. Wang et al., “Beam codebook based beamforming protocol for multi-Gbps millimeter-wave
WPAN systems,” IEEE J. Sel. Areas Commun., vol. 27, no. 8, pp. 1390–1399, Oct. 2009.
[36] H. H. Lee and Y. C. Ko, “Low complexity codebook-based beamforming for MIMO-OFDM
systems in millimeter-wave WPAN,” IEEE Trans. Wireless Commun., vol. 10, no. 11, pp. 3607–
3612, Nov. 2011.
[37] Z. Xiao, T. He, P. Xia and X. Xia, "Hierarchical Codebook Design for Beamforming Training in
Millimeter-Wave Communication," in IEEE Transactions on Wireless Communications, vol. 15,
no. 5, pp. 3380-3392, May 2016.
[38] B. Li, Z. Zhou, H. Zhang, and A. Nallanathan, “Efficient beamforming training for 60-GHz
millimeter-wave communications: A novel numerical optimization framework,” IEEE Trans. On
Vehi. Techn., vol. 63, no. 2, pp. 703-717, Feb. 2014.
[39] B. Li, Z. Zhou, W. Zou, X. Sun, and G. Du, “On the efficient beamforming training for 60 GHz
wireless personal area networks,” IEEE Trans. Wireless Commun., vol. 12, no. 2, pp. 504–515,
Feb. 2013.
65
REFERENCES
[40] X. Gao, L. Dai, C. Yuen, and Z. Wang: ‘Turbo-like beamforming based on tabu search algorithm
for millimeter-wave massive MIMO systems’, IEEE Trans. On Vehi. Techn., 2016, 65 (7), pp.
5731–5737.
[41] Y. Tsang, A. Poon, and S. Addepalli: ‘Coding the beams: Improving beamforming training in
mmwave communication system’. Proc. IEEE Global Tel. Conf. (GLOBECOM), Houston, 2011,
pp. 1–6.
[42] L. Filippini, V. Sciancalepore, F. Devoti, A. Capone, “Fast Cell Discovery in mm-wave 5G
Networks with Context Information,” Networking and Internet Architecture, arXiv:1705.07291,
May 2017.
[43] P. Kela et al., "Location Based Beamforming in 5G Ultra-Dense Networks," 2016 IEEE 84th
Vehicular Technology Conference (VTC-Fall), Montreal, QC, 2016, pp. 1-7.
[44] Rezagah, Roya E., et al. "Cell discovery in 5G HetNets using locationbased cell selection."
Standards for Communications and Networking (CSCN), 2015 IEEE Conference on. IEEE, 2015.
[45] W. D. Abbas and M. Zorzi, “Context Information Based Initial Cell Search for Millimeter Wave
5G Cellular Networks,” in European Conference on Networks and Communications, Oulo, Finland,
June 2016.
[46] Peng, Hailan, Kazuya Moriwaki, and Yasuhiro Suegara. "Macro-Controlled Beam Database-
Based Beamforming Protocol for LTE-WiGig Aggregation in Millimeter-Wave Heterogeneous
Networks." 2016 IEEE 83rd Vehicular Technology Conference (VTC Spring). IEEE, 2016.
[47] E. M. Mohamed, K. Sakaguchi, and S. Sampei, “Millimeter wave beamforming based on WiFi
fingerprinting in indoor environment,” in Proc. 2015 IEEE ICC, 2015, pp. 1155-1160.
[48] A. S. Mubark, E. M. Mohamed, and H. Esmaiel “Millimeter wave beamforming training, discovey
and association using Wi-Fi positioning in outdoor urban environment,” in Proc. of IEEE ICM,
Dec. 2016.
[49] A. Abdelreheem, E. M. Mohamed and H. Esmaiel: ‘Millimeter wave location-based beamforming
using compressive sensing’. Proc. IEEE ICM, Egypt, 2016, pp. 213-216.
[50] Abdelreheem, E. M. Mohamed and H. Esmaiel: ‘Location-Based Millimeter Wave Multi-Level
Beamforming Using Compressive Sensing’, IEEE Commun. Letters, 2018, 22 (1), pp. 185-188.
[51] Nor, A.M. & Mohamed, E.M. “Li-Fi Positioning for Efficient MmWave Beamforming Training
in Indoor Environment”. Mobile Netw Appl (2018). https://doi.org/10.1007/s11036-018-1154-4.
[52] A. M. Nor and E. M. Mohamed, "Millimeter Wave Beamforming Training Based on Li-Fi
Localization in Indoor Environment," GLOBECOM 2017 - 2017 IEEE Global Communications
Conference, Singapore, 2017, pp. 1-6.
[53] E. M. Mohamed, K. Sakaguchi and S. Sampei, "Wi-Fi Coordinated WiGig Concurrent
Transmissions in Random Access Scenarios," in IEEE Transactions on Vehicular Technology, vol.
66, no. 11, pp. 10357-10371, Nov. 2017.
[54] J. Qiao, L. X. Cai, X. S. Shen and J. W. Mark, "Enabling Multi-Hop Concurrent Transmissions in
60 GHz Wireless Personal Area Networks," in IEEE Transactions on Wireless Communications,
vol. 10, no. 11, pp. 3824-3833, November 2011.
[55] J. Qiao, X. Shen, J. W. Mark and Y. He, "MAC-Layer Concurrent Beamforming Protocol for
Indoor Millimeter-Wave Networks," in IEEE Transactions on Vehicular Technology, vol. 64, no.
1, pp. 327-338, Jan. 2015.
[56] J. Qiao, L. X. Cai, X. Shen and J. W. Mark, "STDMA-based scheduling algorithm for concurrent
transmissions in directional millimeter wave networks," 2012 IEEE International Conference on
Communications (ICC), Ottawa, ON, 2012, pp. 5221-5225.
[57] Ziyu Guo, Xiaodong Wang, Wei Heng: ‘Millimeter-Wave Channel Estimation Based on 2-D
Beamspace MUSIC Method’, IEEE Trans. On Wireless Commun., 2017, 16 (8), pp. 5384-5394.
[58] Sungwoo Park, et al: ‘Exploiting Spatial Channel Covariance for Hybrid Precoding in Massive
MIMO Systems’, IEEE Trans. On Signal Processing, 65 (14), pp. 3818-3832.
[59] Shu Sun, Theodore S. Rappaport: ‘Millimeter Wave MIMO channel estimation based on adaptive
compressed sensing’. Proc. IEEE Int. Conf. on Commun. Workshops (ICC Workshops), Paris,
2017, pp. 47-53.
66
REFERENCES
[60] Abdelreheem, E. M. Mohamed and H. Esmaiel, “Adaptive Location-Based Millimeter Wave

Beamforming Using Compressive Sensing Based Channel Estimation” in IET, vol 13, no. 9, pp.
1287-1296, June 2019.
[61] J. García-Rois et al., "On the Analysis of Scheduling in Dynamic Duplex Multihop mmWave
Cellular Systems," in IEEE Transactions on Wireless Communications, vol. 14, no. 11, pp. 6028-
6042, Nov. 2015.
[62] Y. Li, E. Pateromichelakis, N. Vucic, J. Luo, W. Xu and G. Caire, "Radio Resource Management
Considerations for 5G Millimeter Wave Backhaul and Access Networks," in IEEE
Communications Magazine, vol. 55, no. 6, pp. 86-92, June 2017.
[63] Yaxin Cao; Li, V.O.K.; "Scheduling algorithms in broadband wireless networks," Proceedings of
the IEEE, vol.89, no.1, pp.76-87, Jan 2001.
[64] Hahne, E.L.; "Round-robin scheduling for max-min fairness in data networks," Selected Areas in
Communications, IEEE Journal on vol.9, no.7, pp.1024-1039, Sep 1991.
[65] M. Gapeyenko et al., "Analysis of human-body blockage in urban millimeter-wave cellular
communications," 2016 IEEE International Conference on Communications (ICC), Kuala Lumpur,
2016, pp. 1-7.
[66] A. M. Nor, H. Esmaiel and O. A. Omer, "Performance Evaluation of Proportional Fairness
Scheduling in MmWave Network," 2019 International Conference on Computer and Information
Sciences (ICCIS), Sakaka, Saudi Arabia, 2019, pp. 1-6.
[67] H. Kim, K. Kim, Y. Han, and S. Yun, “A proportional fair scheduling for multicarrier transmission
systems,” IEEE Commun. Letters, vol. 9, no. 3, pp. 210-2012, Mar. 2005.
[68] K. Son, S. Chong, and G. Veciana, “Dynamic association for load balancing and interference
avoidance in multi-cell networks,” IEEE Trans. On Wireless Commun., vol. 8, no. 7, pp. 3566-
3576, Jul. 2009.
[69] S. Sur, V. Venkateswaran, X. Zhang, P. Ramanathan, “60 GHz Indoor networking through flexible
beams: a link-level profiling,” in Proc. ACM SIGMETRICS, 2015.
67
List of Publications
List of Publications
[1] A. M. Nor, H. Esmaiel and O. A. Omer, "Performance Evaluation of Proportional Fairness
Scheduling in MmWave Network," 2019 International Conference on Computer and Information
Sciences (ICCIS), Sakaka, Saudi Arabia, 2019, pp. 1-6.
68
‫اﻟﻤﻠﺨﺺ اﻟﻌﺮﺑﻰ‬
‫اﻟﺒﯿﺎﻧﺎت ﻓﻰ اﻟﻨﻈﺎم ﻣﻊ ﺗﺤﻘﯿﻖ ﺟﺰء ﻣﻘﺒﻮل ﻣﻦ اﻟﻌﺪل ﻓﻰ ﻣﻌﺪل ﻧﻘﻞ اﻟﺒﯿﺎﻧﺎت ﺑﯿﻦ اﻟﻤﺴﺘﺨﺪﻣﯿﻦ‪ .‬ﺛﻢ ﯾﺘﻢ اﻗﺘﺮاح‬
‫طﺮﯾﻘﺔ ﺑﺤﺚ ﻣﺘﺘﺎﺑﻊ ﻹﯾﺠﺎد اﻟﺤﻞ اﻟﻤﻨﺎﺳﺐ ﻟﮭﺬه اﻟﻤﺸﻜﻠﺔ ﺑﺸﻜﻞ أﻗﻞ ﺗﻌﻘﯿﺪا ﻣﻘﺎرﻧﺔ ﺑﻄﺮﯾﻘﺔ اﻟﺒﺤﺚ ﻓﻰ ﺟﻤﯿﻊ‬
‫اﻟﺤﺎﻻت اﻟﻤﻤﻜﻨﺔ‪ .‬وﯾﺘﻢ اﻟﺘﺤﻘﻖ ﻣﻦ ھﺬا اﻟﺘﻘﻠﯿﻞ ﻓﻰ اﻟﺘﻌﻘﯿﺪ ﻣﻦ ﺧﻼل اﻟﺘﺤﻠﯿﻞ اﻟﻌﺪدى‪.‬‬
‫ﺗﺘﻜﻮن ھﺬه اﻟﺮﺳﺎﻟﺔ ﻣﻦ ﺧﻤﺲ ﻓﺼﻮل ﻛﺎﻻﺗﻰ‪:‬‬
‫‪ ‬اﻟﻔﺼﻞ اﻷول ‪ :‬ﯾﺴﺘﻌﺮض ھﺬا اﻟﻔﺼﻞ ﻣﻘﺪﻣﺔ ﻋﺎﻣﺔ ﻋﻦ اﻟﺘﻄﻮر اﻟﻤﺘﺰاﯾﺪ ﻓﻰ اﺳﺘﺨﺪاﻣﺎت ﻧﻘﻞ اﻟﺒﯿﺎﻧﺎت‬
‫واﻟﺘﻰ ﻗﺎﻣﺖ ﻋﻠﻰ أﺳﺎﺳﮭﺎ ﺷﺒﻜﺎت اﻟﺠﯿﻞ اﻟﺨﺎﻣﺲ‪ .‬ﻛﻤﺎ ﯾﺘﻀﻤﻦ اﻻﺗﺠﺎھﺎت اﻟﻤﻤﻜﻨﺔ ﻟﺘﻔﻌﯿﻞ ﺗﻠﻚ‬
‫اﻟﺸﺒﻜﺎت وﻣﻦ أھﻤﮭﺎ اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﻣﻊ ﻋﺮض ﻧﺒﺬة ﻣﺨﺘﺼﺮة ﻋﻨﮭﺎ وﻋﻦ أھﻢ ﺧﺼﺎﺋﺼﮭﺎ‬
‫واﻟﻤﺸﺎﻛﻞ اﻟﺘﻰ ﺗﻮاﺟﮭﺎ‪ .‬ﻛﻤﺎ ﯾﻮﺿﺢ ھﺬا اﻟﻔﺼﻞ اﻟﺪواﻓﻊ اﻟﺘﻰ ﺑﻨﯿﺖ ﻋﻠﯿﮭﺎ اﻟﺮﺳﺎﻟﺔ واﻷھﺪاف اﻟﺘﻰ‬
‫ﺗﺴﻌﻰ ﻟﺘﻘﺪﯾﻤﮭﺎ‪.‬‬
‫‪ ‬اﻟﻔﺼﻞ اﻟﺜﺎﻧﻰ‪ :‬ﯾﺘﻀﻤﻦ ھﺬا اﻟﻔﺼﻞ ﺗﻮﺿﯿﺢ أﻛﺜﺮ ﻟﻠﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﻣﻊ ﻋﺮض ﻣﻔﺼﻞ ﻋﻦ‬
‫ﺧﺼﺎﺋﺼﮭﺎ ‪.‬ﺛﻢ ﯾﺴﺘﻌﺮض ﺑﺘﻔﺼﯿﻞ أﻛﺒﺮ اﻟﻤﺸﺎﻛﻞ اﻟﺘﻰ ﺗﻮاﺟﮫ اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ‪ .‬ﺑﺎﻹﺿﺎﻓﺔ اﻟﻰ‬
‫ذﻟﻚ ﯾﺤﺘﻮى ﻋﻠﻰ أھﻢ اﻟﺪراﺳﺎت اﻟﺴﺎﺑﻘﺔ اﻟﺘﻰ ﺳﺎھﻤﺖ ﻓﻰ ﺣﻞ ھﺬه اﻟﻤﺸﺎﻛﻞ وﯾﺴﺘﻌﺮض ﻣﻤﯿﺰات‬
‫وﻋﯿﻮب ﻛﻞ ﺣﻞ ﻣﻦ ھﺬه اﻟﺤﻠﻮل‪.‬‬
‫‪ ‬اﻟﻔﺼﻞ اﻟﺜﺎﻟﺚ‪ :‬ﯾﺤﺘﻮى ھﺬه اﻟﻔﺼﻞ ﻋﻠﻰ دراﺳﺔ ﻋﻦ طﺮق اﻟﺠﺪوﻟﺔ اﻟﻤﻤﻜﻦ ﺗﻄﺒﯿﻘﮭﺎ ﻓﻰ ﺷﺒﻜﺎت‬
‫اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﻓﻰ ﺣﺎﻟﺔ رﻏﺒﺔ أﻛﺜﺮ ﻣﻦ ﻣﺴﺘﺨﺪم اﻻﺗﺼﺎل ﻣﻦ ﺧﻼﻟﮭﺎ‪ .‬ﻣﻊ اﻗﺘﺮاح ﺗﻄﺒﯿﻖ‬
‫ﺧﻮارزﻣﯿﺔ )‪ (proportional fairness scheduling‬ﻟﺘﺤﺴﯿﻦ أداء اﻟﻨﻈﺎم ﺣﯿﺚ أﻧﮫ ﯾﻌﻈﻢ ﻣﻌﺪل‬
‫ﻧﻘﻞ اﻟﺒﯿﺎﻧﺎت اﻟﻜﻠﻰ ﻣﻊ اﻟﻤﺤﺎﻓﻈﺔ ﻋﻠﻰ اﻟﻌﺪل ﻓﻰ ﺗﻮزﯾﻊ ﻣﺼﺎدر اﻟﻨﻈﺎم‪.‬‬
‫‪ ‬اﻟﻔﺼﻞ اﻟﺮاﺑﻊ‪ :‬ﯾﺤﺘﻮى ھﺬا اﻟﻔﺼﻞ ﻋﻠﻰ ﻣﻘﺘﺮح ﻟﺤﻞ ﻣﺸﻜﻠﺔ اﻟﺘﺪاﺧﻞ ﺑﯿﻦ وﺻﻼت اﻟﻤﻮﺟﺎت‬
‫اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﺣﯿﺚ ﺗﺤﺪث ھﺬه اﻟﻤﺸﻜﻠﺔ ﻓﻰ ﺣﺎﻟﺔ ﺗﻮاﺟﺪ أﻛﺜﺮ ﻣﻦ ﻣﺮﺳﻞ ﻓﻰ ﻧﻔﺲ اﻟﻨﻄﺎق‪ .‬وھﺬا اﻟﺤﻞ‬
‫ﯾﮭﺪف إﻟﻰ ﺗﺤﻘﯿﻖ أﻓﻀﻞ اﺳﺘﻔﺎدة ﻣﻦ اﻟﺸﺒﻜﺔ ﻋﻦ طﺮﯾﻖ وﺻﻞ وﺟﺪوﻟﺔ اﻷﺟﮭﺰة اﻟﻤﺴﺘﻘﺒﻠﺔ ﻣﻊ‬
‫اﻟﻤﺮﺳﻠﯿﻦ ﻓﻰ ﻧﻔﺲ اﻟﻮﻗﺖ‪ .‬وإﯾﻀﺎ ﯾﺘﻢ ﺗﻘﺪﯾﻢ ﺧﻮارزﻣﯿﺔ اﻟﺒﺤﺚ اﻟﻤﺘﺘﺎﺑﻊ ﻟﺘﻘﻠﯿﻞ اﻟﺘﻌﻘﯿﺪ ﻓﻰ اﺧﺘﯿﺎر‬
‫اﻟﻮﺻﻼت ﻣﻘﺎرﻧﺔ ﺑﻄﺮﯾﻘﺔ اﻟﺒﺤﺚ اﻟﻜﻠﯿﺔ‪.‬‬
‫اﻟﻔﺼﻞ اﻟﺨﺎﻣﺲ‪ :‬وﻓﯿﮫ ﺗﻢ ﻋﻤﻞ ﻣﻠﺨﺺ ﻹﺳﮭﺎﻣﺎت اﻟﺮﺳﺎﻟﺔ واﻟﻤﻮﺿﻮﻋﺎت اﻟﻤﻘﺘﺮﺣﺔ ﻟﻨﻘﺎط ﺑﺤﺜﯿﺔ‬ ‫‪‬‬
‫ﯾﻤﻜﻦ اﻟﻌﻤﻞ ﻋﻠﯿﮭﺎ ﻓﻲ اﻟﻤﺴﺘﻘﺒﻞ‪.‬‬
‫‪70‬‬
‫إن ﺗﻄﺒﯿﻘﺎت اﻧﺘﺮﻧﺖ اﻷﺷﯿﺎء و اﻻﻧﺘﺮﻧﺖ اﻟﻤﻠﻤﻮس أﺻﺒﺤﺖ ﻣﻨﺘﺸﺮة ﺑﺸﻜﻞ ﻛﺒﯿﺮ ﻣﺆﺧﺮا‪ .‬و أﺻﺒﺤﺖ ﺷﺒﻜﺎت‬
‫اﻻﺗﺼﺎﻻت ﻋﺎﻟﯿﺔ اﻟﺴﺮﻋﺔ ﻣﻄﻠﺐ ﻣﻠﺢ ﻓﻰ ھﺬه اﻷﯾﺎم ﺧﺼﻮﺻﺎ ﺑﺴﺒﺐ اﻻﺣﺘﯿﺎج اﻟﻜﺒﯿﺮ ﻟﻨﻘﻞ اﻟﺒﯿﺎﻧﺎت‪ .‬ﻟﺬا وﺟﮫ‬
‫اﻟﻌﻠﻤﺎء اھﺘﻤﺎﻣﮭﻢ إﻟﻰ ﺷﺒﻜﺎت اﻟﺠﯿﻞ اﻟﺨﺎﻣﺲ ﻟﻜﻰ ﯾﻮاﻛﺒﻮا ھﺬه اﻷﺣﺘﯿﺎﺟﺎت‪ .‬ﺣﯿﺚ ﺗﻌﺘﺒﺮ اﻟﮭﺠﺮة إﻟﻰ ﺣﯿﺰ‬
‫اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﻣﻦ اﻷﺗﺠﺎھﺎت اﻟﻤﺮﺷﺤﺔ ﻟﺘﻔﻌﯿﻞ ﺷﺒﻜﺎت اﻟﺠﯿﻞ اﻟﺨﺎﻣﺲ وذﻟﻚ ﻹﻧﮭﺎ ﺗﺘﻤﯿﺰ ﺑﺎﻣﺘﻼﻛﮭﺎ ﺣﯿﺰ‬
‫ﻣﺘﺴﻊ ﻣﻦ اﻟﻄﯿﻒ اﻟﻤﻮﺟﻰ اﻟﻐﯿﺮ ﻣﺴﺘﻐﻞ‪ .‬وﻣﻊ ذﻟﻚ‪ ,‬ﺗﻮاﺟﮫ اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ اﻟﻌﺪﯾﺪ ﻣﻦ اﻟﺘﺤﺪﯾﺎت اﻟﺘﻘﻨﯿﺔ‬
‫اﻟﺘﻰ ﺗﻌﯿﻖ ﻋﻤﻠﮭﺎ ﺣﯿﺚ إﻧﮭﺎ ﺗﻀﻌﻒ ﺑﺸﺪة ﻓﻰ اﻟﻮﺳﻂ اﻟﻤﺤﯿﻂ ﺑﺴﺒﺐ ﺿﻌﻔﮭﺎ أﺛﻨﺎء اﻧﺘﺸﺎرھﺎ أو اﺧﺘﺮاﻗﮭﺎ‬
‫ﻟﻠﺤﻮاﺟﺰ‪ .‬إﯾﻀﺎ ﺗﻮاﺟﮫ اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﻣﺸﻜﻠﺔ ﺗﺄﺛﺮھﺎ اﻟﻜﺒﯿﺮ ﻧﺘﯿﺠﺔ اﻋﺘﺮاض اﻟﺒﺸﺮ ﻻﻧﺘﺸﺎر ﻣﻮﺟﺎﺗﮭﺎ‪.‬‬
‫ﻟﺬا ﻋﺮف ﻣﻌﯿﺎر )‪ (IEEE 802.11 ad‬اﻻﻧﺘﺸﺎر اﻟﻤﻮﺟﮫ ﻟﻠﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﺑﺎﺳﺘﺨﺪام زواﯾﺎ اﻟﮭﻮاﺋﻰ ﺗﺤﺖ‬
‫ﻣﺴﻤﻰ ﻋﻤﻠﯿﺔ )‪ (beamforming training‬وذﻟﻚ ﻟﻠﺘﻐﻠﺐ ﻋﻠﻰ ﻣﺸﺎﻛﻞ اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ وزﯾﺎدة ﻗﻮة‬
‫ﺷﻌﺎﻋﮭﺎ‪ .‬وﻣﻊ ذﻟﻚ‪ ,‬ﻓﺈن ھﺬه اﻟﻌﻤﻠﯿﺔ ﻣﻌﻘﺪة وﺗﺤﺘﺎج وﻗﺖ طﻮﯾﻞ ﻻﻛﺘﺸﺎف اﻟﺸﻌﺎع اﻟﺬى ﯾﺤﻘﻖ أﻋﻠﻰ ﻗﯿﻤﺔ طﺎﻗﺔ‬
‫ﻣﺴﺘﻘﺒﻠﺔ‪ .‬وﻗﺪ اﻗﺘﺮح ﻣﻌﯿﺎر )‪ (IEEE 802.11 ad‬اﺳﺘﺨﺪام اﻟﺘﺤﻜﻢ ﻓﻰ وﺳﻂ اﻻﺗﺼﺎل اﻟﻤﺠﺪول اﻟﻤﻌﺘﻤﺪ ﻋﻠﻰ‬
‫ﺗﻘﺴﯿﻢ اﻟﻔﺘﺮة اﻟﺰﻣﻨﯿﺔ )اﻟﻮﻗﺖ( ﻟﻜﻰ ﯾﻮزع اﻟﻤﺼﺎدر اﻟﻤﺘﺎﺣﺔ ﻋﻠﻰ ﻣﺴﺘﺨﺪﻣﻰ اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ اﻟﻤﺘﺼﻠﯿﻦ‬
‫ﺑﺎﻟﺸﺒﻜﺔ وﻟﻜﻨﮫ ﻟﻢ ﯾﻌﺮف طﺮﯾﻘﺔ ﻣﻌﯿﻨﺔ ﻟﺬﻟﻚ‪ .‬وﻣﻦ ﺛﻢ‪ ,‬ﻓﺈن ﺗﻘﯿﯿﻢ اﻷداء ﻟﻄﺮق اﻟﺠﺪوﻟﺔ اﻟﻤﺨﺘﻠﻔﺔ ﺷﻰء ﺿﺮورى‬
‫ﻗﺒﻞ اﺧﺘﯿﺎر طﺮﯾﻘﺔ ﻣﻨﮭﻢ ﻻﺳﺘﺨﺪاﻣﮭﺎ ﻓﻰ ﺷﺒﻜﺎت اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ‪ .‬ﺑﺎﻹﺿﺎﻓﺔ إﻟﻰ ذﻟﻚ‪ ,‬ﻓﻰ ﺣﺎﻻت ﻧﻘﻞ‬
‫اﻟﺒﯿﺎﻧﺎت ﻷﻛﺜﺮ ﻣﻦ ﻣﺴﺘﺨﺪم ﻋﻦ طﺮﯾﻖ أﻛﺜﺮ ﻣﻦ ﻣﺮﺳﻞ ﯾﻌﺪ اﺧﺘﯿﺎر وﺻﻼت اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ اﻟﺘﻰ ﺗﻌﻈﻢ‬
‫ﻗﯿﻤﺔ ﻣﻌﺪل ﻧﻘﻞ اﻟﺒﯿﺎﻧﺎت ﻟﻠﻨﻈﺎم ﻣﺸﻜﻠﺔ ﻛﺒﯿﺮة اﺧﺮى‪ .‬وذﻟﻚ ﻹن ﻋﻤﻠﯿﺔ )‪ (beamforming training‬ﺗﺘﻢ‬
‫دون أى ﻣﻌﺮﻓﺔ ﻋﻦ اﻟﺘﺪاﺧﻞ اﻟﺤﺎدث ﺑﯿﻦ وﺻﻼت اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ وﻣﻦ ﺛﻢ ﺧﺴﺎرة ﻓﺮص اﺧﺘﯿﺎر‬
‫وﺻﻼت ﻟﮭﺎ ﻗﯿﻢ ﻗﻠﯿﻠﺔ ﻣﻦ اﻟﺘﺪاﺧﻞ ﻣﻊ ﺑﻌﻀﮭﺎ واﻟﺘﻰ ﺗﻮﻓﺮ ﻣﻌﺪل ﻋﺎﻟﻰ ﻣﻦ ﻧﻘﻞ اﻟﺒﯿﺎﻧﺎت ﻟﻠﻤﺴﺘﺨﺪﻣﯿﻦ‪ .‬ﯾﺘﻢ‬
‫ﻋﺮض ﻣﻮﺿﻮﻋﯿﻦ ﻓﻰ ھﺬه اﻟﺮﺳﺎﻟﺔ‪ ,‬أوﻟﮭﻤﺎ ھﻮ دراﺳﺔ ﺗﻘﯿﯿﻢ اﻷداء ﻷﻧﻈﻤﺔ ﺟﺪوﻟﺔ ﻣﺨﺘﻠﻔﺔ ﻓﻰ ﺷﺒﻜﺎت اﻟﻤﻮﺟﺎت‬
‫اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ وﻣﻦ أﻣﺜﻠﺘﮭﺎ )‪ (round robin‬و)‪ (proportional fairness scheduling‬وذﻟﻚ ﺗﺤﺖ‬
‫ظﺮوف ﻣﺨﺘﻠﻔﺔ أى ﺑﺎﺳﺘﺨﺪام اﺗﺴﺎﻋﺎت ﻣﺨﺘﻠﻔﺔ ﻷﺷﻌﺔ اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ واﺳﺘﺨﺪام ﻗﯿﻢ ﻣﺨﺘﻠﻔﺔ ﻣﻦ اﺣﺘﻤﺎل‬
‫ﺣﺪوث اﻟﻌﻮاﺋﻖ اﻟﺒﺸﺮﯾﺔ‪ .‬وﺗﺤﺴﺐ ھﺬه اﻟﺪراﺳﺔ ﻣﻘﯿﺎﺳﯿﻦ ﻣﮭﻤﯿﻦ ھﻤﺎ اﻟﻤﻌﺪل اﻟﻜﻠﻰ ﻟﻨﻘﻞ اﻟﺒﯿﺎﻧﺎت ﻓﻰ اﻟﺸﺒﻜﺔ‬
‫وﻣﺪى اﻟﻌﺪل ﻓﻰ ﻣﻌﺪل ﻧﻘﻞ اﻟﺒﯿﺎﻧﺎت ﺑﯿﻦ اﻟﻤﺴﺘﺨﺪﻣﯿﻦ‪ .‬أﻣﺎ اﻟﻤﻮﺿﻮع اﻟﺜﺎﻧﻰ ﻓﻰ اﻟﺮﺳﺎﻟﺔ ﯾﻨﺎﻗﺶ ﻣﺸﻜﻠﺔ اﺧﺘﯿﺎر‬
‫ﻋﺪة وﺻﻼت ﻟﻠﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﻓﻰ ﺣﺎﻻت ﻧﻘﻞ اﻟﺒﯿﺎﻧﺎت ﻷﻛﺜﺮ ﻣﻦ ﻣﺴﺘﺨﺪم ﻋﻦ طﺮﯾﻖ أﻛﺜﺮ ﻣﻦ وﺻﻠﺔ‬
‫ﻣﻦ وﺻﻼت اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﯿﻤﺘﺮﯾﺔ ﻓﻰ ﻧﻔﺲ اﻟﻮﻗﺖ ﻓﻰ اﻟﺸﺒﻜﺎت اﻟﻼﺳﻠﻜﯿﺔ ﺻﻐﯿﺮة اﻟﺘﻐﻄﯿﺔ‪ .‬ﺣﯿﺚ ﯾﺘﻢ ﺗﻌﯿﯿﻦ‬
‫وﺻﯿﺎﻏﺔ ﻣﺸﻜﻠﺔ )‪ (joint proportional fairness scheduling‬اﻟﺘﻰ ﺗﻌﻈﻢ ﻣﻦ اﻟﻤﻌﺪل اﻟﻜﻠﻰ ﻟﻨﻘﻞ‬
‫‪69‬‬
Scanned by CamScanner
‫ﻛﻠﯿـﺔ اﻟﮭﻨﺪﺳـﺔ‬ ‫ﺟﺎﻣـﻌﺔ اﺳـﻮان‬
‫ﺗﻘﯿﯿﻢ أداء اﻟﻤﻮﺟﺎت اﻟﻤﻠﻠﻲ ﻣﺘﺮﯾﺔ ﻓﻰ ﺷﺒﻜﺎت اﻟﺠﯿﻞ‬

‫اﻟﺨﺎﻣﺲ‬
‫رﺳﺎﻟﺔ ﻋﻠﻤﯿﺔ ﻣﻘﺪﻣﺔ اﻟﻰ ﻗﺴﻢ اﻟﮭﻨﺪﺳﺔ اﻟﻜﮭﺮﺑﺎﺋﯿﺔ ﺑﻜﻠﯿﺔ اﻟﮭﻨﺪﺳﺔ‬
‫ﺟﺎﻣﻌﺔ أﺳﻮان‬
‫ﻛﺎﺳﺘﯿﻔﺎء ﺟﺰﺋﻲ ﻟﻤﺘﻄﻠﺒﺎت اﻟﺤﺼﻮل ﻋﻠﻰ درﺟﺔ ﻣﺎﺟﺴﺘﯿﺮ اﻟﻌﻠﻮم‬
‫ﻓﻲ‬
‫اﻟﮭﻨﺪﺳﺔ اﻟﻜﮭﺮﺑﯿﺔ‬
‫ﻣـﻘﺪﻣـﺔ ﻣﻦ‬
‫اﻟﻤﮭﻨﺪس‪ /‬أﺣﻤﺪ ﻣﺤﻤﺪ ﻧﻮر اﻟﺪﯾﻦ اﻟﺰﻗﻠﻮﺑﻰ‬

‫ﺑﻜﺎﻟﻮرﯾﻮﯾـﺲ اﻟﮭﻨـﺪﺳـﺔ اﻟﻜﮭﺮﺑﯿـﺔ – ﻛﻠﯿـﺔ اﻟﮭﻨﺪﺳـﺔ ‪ -‬ﺟﺎﻣﻌـﺔ أﺳـﻮان – ‪٢٠١٦‬‬
‫‪٢٠١٩‬‬
‫‪View publication stats‬‬

Performance Evaluation of MmWave in 5G Networks

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

Performance Evaluation of MmWave in 5G Networks

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Performance Evaluation of MmWave in 5G Networks

Thesis · September 2019

The user has requested enhancement of the downloaded file.

Performance Evaluation of MmWave in 5G

A Thesis Submitted to the Faculty of Engineering

In Partial Fulfillment of the Requirements for the Degree of

Ahmed Mohammed Nor

JOINT PROPORTIONAL FAIRNESS SCHEDULING FOR MMWAVE CONCURRENT

HCF Hybrid Coordination Function.

M Number of mmWave APs.

N Number of time slots.

The instantaneous achievable data rate at user k from mmWave AP m at time

The bandwidth efficiency.

The SNR efficiency.

BW MmWave bandwidth in Hz.

( ) The SNR for user k in case of connect with AP m by beam ID .

The received power of user k which is far from mmWave AP m by distance

The noise power.

The transmitted power for mmWave AP in dBm.

( ) The mmWave average path loss in dB.

( ) The path loss at a reference distance .

The mmWave path loss exponent.

The mmWave shadowing term.

The standard deviation of .

( , ) The 3D beamforming gain for mmWave TX beam in dB.

The azimuth angle.

The elevation angle.

( ) The gains of TX beam in horizontal direction.

( ) The gains of TX beam in vertical direction.

The center angle of mmWave AP TX beam in horizontal direction.

The center angles of mmWave AP TX beam in vertical direction.

The beamwidths of mmWave beam in the horizontal directions.

The beamwidths of mmWave beam in the vertical directions.

users' average data rate.

The vectors of all users' achievable data rates.

B The number of mmWave AP TX beams.

Spatial reuse factor.

. The complexity of the conventional IEEE 802.11ad based WLAN.

The complexity of ES-JPFS scheme.

The complexity of IS-JPFS scheme.

The average number of iterations performed by the proposed IS-JPFS scheme

Connected IoT Devices in

MmWave Band Light Fidelity Band

1.1.1 MmWave Technology

User MmWave coverage area

Figure 1-4: MmWave AP antenna beamforming.

1.1.1.1 Directional communication

Beamforming process needs an estimation of mmWave channel to efficiently determine the

Motivated by the fact that mmWave communication is almost a line-of-sight (LOS)

location-based BT is mainly depending on the accuracy of the used localization technology. In

1.1.1.2 Scheduling Schemes in MmWave

1.1.1.3 MmWave Concurrent transmission

MmWave AP 1 MmWave concurrent links MmWave AP 2

Figure 1-5: MmWave concurrent transmission.

1.3 Thesis objectives

The contributions of this thesis are:

 Firstly, the performance evaluation of radio resource scheduling schemes in mmWave

1.4 Thesis Outline

In Chapter 3, the performance evaluation of radio resource scheduling schemes in mmWave is

 Millimeter Wave Technology.

2.2 Millimeter Wave Technology

2.2.1 MmWave physical layer