Professional Documents
Culture Documents
7tim 2021 3078557
7tim 2021 3078557
Abstract— This article presents a low-cost, open-source, Index Terms— Edge computing, Parallella, phasor measure-
heterogeneous, resource-constrained hardware platform called ment unit (PMU), power quality (PQ), signal compression,
“Parallella” as a measurement device for edge-computing appli- waveform recording.
cations research in smart grid. The unique hardware architec-
ture of the Parallella provides a multitude of edge-computing
I. I NTRODUCTION
resources in the form of a Zynq SoC (dual-core ARM + FPGA)
and a 16-core co-processor called Epiphany. A multifunctional
intelligent electronic device (IED) design is demonstrated to
showcase the capabilities of the platform. A custom I/O board
M OBILE computing architectures have improved the
capabilities of edge devices that are used in smart grid
applications. Multicore processors, graphic processing units
has been developed for the desktop and embedded versions of
Parallella, which can be interfaced with external daughter boards (GPUs), FPGAs, co-processors, and so on are being integrated
and peripherals for measurements. One such daughter board is with mobile-/edge-computing platforms at an ever-increasing
an analog sensing board, which can measure voltages of all the pace. Moreover, in recent years, many of these platforms
three phases and four line currents using a 16-bit synchronous are being made available as a development board to the
ADC set at 32 kHz. The ADC samples are synchronized to the research community, thus enabling unique edge-computing
PPS time clock of a GPS unit for providing global time reference.
These captured seven-channel raw waveform data are sent to a applications. Some popular platforms are NVIDIA-Jetson,
cloud server over a bandwidth-limited communication channel Adapteva Parallella, Zynq Z-board, Intel Agilex, and so on
using a custom anomaly-aware data compression algorithm for industrial/product development. Rasberry-Pi, Beaglebone,
implemented on the ARM. A phasor measurement algorithm and so on are being used for the hobbyists or proof-of-concept
using the Teager energy operator (TEO) is implemented on the developments [1].
field-programmable gate array (FPGA). A parallel power qual-
ity (PQ) measurement algorithm is implemented on the Epiphany. Many measurement platforms with multiple functionali-
The obtained measurements are found to be comparable to a ties, called intelligent electronic devices (IEDs), have been
commercial power analyzer. proposed in the literature for recording and analyzing mea-
sured signals. A microcontroller with two dsPIC33 proces-
Manuscript received January 15, 2021; revised March 26, 2021; accepted sors proposed in [2], one for sampling data at 7.68 kHz
April 21, 2021. Date of publication May 10, 2021; date of current version computations and the other for communication/device control,
June 17, 2021. This work was supported in part by the Bosch Research to satisfy the power quality (PQ) measurement requirements
and Technology Centre, Bengaluru, India, and in part by the Robert Bosch
Centre for Cyber-Physical Systems, Indian Institute of Science at Bengaluru, of the Brazilian power system. A synchronized 10-kHz data
India, through the Project E-Sense: Sensing and Analytics for Energy Aware recorder is discussed in [3] with fast Fourier transform (FFT)
Smart Campus. The Associate Editor coordinating the review process was for feature extraction. The need for low-cost measurement
Dr. Guglielmo Frigo. (Corresponding author: Gurunath Gurrala.)
Ashish Joglekar is with the Robert Bosch Centre for Cyber-Physical devices for research purposes has been recognized in [4]
Systems, Indian Institute of Science at Bengaluru, Bengaluru 560012, India. and developed an OpenPMU, which captures six channels of
Gurunath Gurrala is with the Department of Electrical Engineering, data at 6.4 kHz. In [5], an FPGA-based phasor measurement
Indian Institute of Science at Bengaluru, Bengaluru 560012, India, and
also with the Robert Bosch Centre for Cyber-Physical Systems, Indian unit (PMU) with a sampling rate of 12.8 kHz is proposed.
Institute of Science at Bengaluru, Bengaluru 560012, India (e-mail: In [6], a multipurpose board based on Mini-ITX Motherboard
gurunath_gurrala@yahoo.co.in). with peripherals performing PMU, PQ, and signal capture
Puneet Kumar, Francis C. Joseph, and T. S. Kiran are with the Department
of Electrical Engineering, Indian Institute of Science at Bengaluru, Bengaluru at 25.6 kHz is proposed. A 12.8-kHz sampling rate analog-
560012, India. to-digital (A/D) board is proposed in [7], which was inter-
K. R. Sahasranand is with the Department of Electrical Communication faced to notebooks for communicating the data to upstream
Engineering, Indian Institute of Science at Bengaluru, Bengaluru 560012,
India. servers. The notebooks were used for PQ indices computations
Himanshu Tyagi is with the Department of Electrical Communication as well. In [8], a waveform recorder with sampling rate
Engineering, Indian Institute of Science at Bengaluru, Bengaluru 560012, of 7.68 kHz is developed on an FPGA & ARM Cortex
India, and also with the Robert Bosch Centre for Cyber-Physical Systems,
Indian Institute of Science at Bengaluru, Bengaluru 560012, India. M4 board, which uses data compression based on a novelty
Digital Object Identifier 10.1109/TIM.2021.3078557 detection in the waveform using discrete wavelet transform
1557-9662 © 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
9003612 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 70, 2021
to reduce the transmission size. In [9], a PMU is developed sigma–delta ADC at a maximum sampling rate of 64 kHz,
on a BeagleBone Black, which features an ARM core and is developed for Parallella.
two programmable real-time units (PRUs) responsible for A technique to synchronize the ADC’s clock to the GPS
sampling at 12.8 kHz. Usage of either a discrete PLL ASIC time clock PPS reference is also provided [9]. A custom serial
or a software-based PLL with reduced accuracy for the GPS peripheral interface (SPI) ADC driver and a GPS driver to
discipline of the ADC clock was also discussed. The ARM synchronize to GPS time clock reference have been developed
performed frequency and phasor estimation along with data on FPGA.
communication. An FPGA-based PMU sampled at 32 kHz is A custom FPGA binary has been compiled to port the devel-
proposed in [10] using a NovTech IoT Octopus board with an oped custom drivers and algorithms. The developed binary file
iterative Interpolated DFT algorithm for estimation. In [11], exposes the Zynq I/Os without losing the default e-link driver
a review of commercial PMUs, open-source PMUs, and inte- responsible for high-speed communication interface between
grated PMU with digital protective relay functions is provided. the ARM and the Epiphany co-processor. Three applications
In [12], a 50-kHz IED intended for research and prototyping are demonstrated: a custom anomaly-aware data compression
HVDC protection on real-time simulators is discussed. Built algorithm on the ARM to enable the transfer of high-frequency
on ZedBoard with modular peripherals, the IED performs sampled waveform data to a server, a Teager energy operator
16-channel signal sampling on one of its peripheral module (TEO)-based phasor measurement algorithm proposed in [16]
and GPS synchronization on another. A very high-frequency on the FPGA, and a parallel PQ measurement algorithm [17]
sampling rate data capturing device, 215 samples per cycle, on 16-core Epiphany. The accuracy of the raw measurements
is proposed in [13], which requires a powerful desktop for and the PQ measurements is compared with the measurements
real-time signal processing and frequency estimation. obtained from a high-end commercial power analyzer. This is
This article focuses on the development of a measure- a first-of-its-kind effort in the literature, to the best of our
ment platform with edge-computing capabilities for smart knowledge, that explores the applicability of an open-source
grids based on a unique hardware developed by Adapteva, edge-computing platform under 5 W, with three varieties of
advertized as a credit card-sized supercomputer, Parallella. computing modules for a smart grid measurement platform.
This platform has three compute units: a Zynq SoC (an
industry-proven SoC), housing FPGA + Dual-core ARM II. M ULTIFUNCTIONAL IED H ARDWARE D ESIGN
Cortex-A9 MPCore, clocked at 650 MHz, and a 16-core A 5-W, 18-core credit card-sized microcomputer called
Epiphany co-processor. Parallella is available in three vari- Parallella-Desktop is used as the edge-computing module
ants, namely, server version ($99), desktop version ($159), in this article. The heart of this microcomputer is the
and an embedded version ($249) consuming less than 5 W. Zynq-7010 SoC, which features a seven-series FPGA pro-
This heterogeneous resource-constrained compute architecture grammable logic. The FPGA features 28k logic cells, 2.1 Mb
combines features of CPU, DSP, and ASSP on a single of block RAM, 80 DSP slices, and 150 I/O pins. The
board and makes the Parallella unique. This heterogeneous cores of Epiphany can exchange data with each other via
compute platform has been used as a compute node, accel- a low-latency mesh network on chip [15]. The Epiphany
erator, microserver, and server cluster in research applications architecture allows programming of each core using typical
covering image processing, video filtering, encryption, AI, and C code [18]. An SRAM alongside each core is required to be
machine learning [14]. managed, considering the lack of cache. At 2 W, the Epiphany
Parallella does not come with preprogrammed I/O interface SoC provides promising power efficiency (performance/watt).
options for any external sensor integration. The internal ARM The Epiphany talks with the Zynq over e-Link implemented
ADC is disabled on all versions of the Parallella. The default on the FPGA. The Epiphany e-Link driver can be accessed
FPGA binary file on Parallella does not allow an external sen- from the ARM’s userspace over a memory-mapped AXI inter-
sor interface to the FPGA GPIOs. The co-processor Epiphany face. The device connects to a 10-Mbps Ethernet backhaul.
can only be accessed from the ARM through a proprietary The device also has 1-GB SDRAM and 32 GB of onboard
e-Link protocol module implemented on the FPGA [15]. Any SD card storage.
configuration or modification on Parallella requires modifica- Fig. 1(a) and (b) shows the functional block diagram and the
tion of the FPGA binary file. Modifying the FPGA binary proposed implementation architecture on the Parallella. Three
of Parallella was not attempted before since it might impact voltage measurements for phase voltages, four current mea-
the working of the Epiphany. This limited the application of surements for line currents, and neutral current are provided
Parallella as a sensor node with edge-computing capabilities. using an AFE daughter board. The AFE and Parallella interact
An attempt is made in this article to harness the capabilities via a custom I/O board, which also provides the necessary
of Parallella as a powerful sensor node. hardware interfaces for making peripherals compatible with
A custom I/O board for the Parallella desktop and embedded the Zynq SoC. The developed IED using Parallella and the
versions is developed to expand the parallella capabilities I/O board is shown in Fig. 2(a) and (b), respectively. The IED
for peripheral sensors integration. The developed I/O board provides eight high drive strength GPIOs for the actuation of
provides high-speed, level-shifted GPIO interface to external external circuit breakers or relays. The AFE can accommodate
peripherals. One such peripheral, an analog front-end (AFE) eight differential analog channels for the IED. The clock for
daughterboard for analog measurements, using a synchronous AFE is generated from the GPS reference. The data generated
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
JOGLEKAR et al.: OPEN-SOURCE HETEROGENEOUS CONSTRAINED EDGE-COMPUTING PLATFORM 9003612
Fig. 1. IED design on Parallella. (a) Functional block diagram of the proposed IED. (b) Implementation architecture on Parallella.
Fig. 2. Developed IED using Parallella. (a) Prototype IED. (b) I/O board.
from the ADC move via the I/O board to the FPGA component arbitrator to send sampled data obtained from the FPGA-based
of the Zynq SoC. The data are buffered in a 16-sample ADC driver to the multicore co-processor. The Epiphany is
wide first in–first out (FIFO) buffer from which different used for data crunching applications, such as computation
applications can consume the data. of PQ on the measured signals. For want of space, security
The efficient utilization of the available compute mod- algorithms implemented on Parallella are not discussed, and
ules is an essential step in edge-computing devices, such as the work of [20] can be referred for this.
Parallella [19]. In this article, FPGA is proposed to be used for
sensing and compute-intensive real-time algorithms. Hardware A. I/O Board
drivers for GPS (UART) and ADC (SPI) are ported on the The e-link Epiphany interface utilizes 48 GPIOs of the
FPGA to meet the communication bandwidth requirements FPGA. Hence, only 24 external GPIOs on the desktop edition
when the data are sampled at 32 kHz (provision is made to use of the Parallella are available for customization. However,
up to 64 kHz depending on the requirement). Fig. 1(b) shows these are not exposed by the Parallella designers for external
various modules added to the FPGA apart from the default use, limiting the usability of Parallella as a sensor node.
e-link interface. Due to the nature of FPGA, all modules A custom I/O board is developed to provide a high-speed,
get executed in parallel. The ARM processor is intended to level-shifted GPIO interface to make the FPGA GPIOs acces-
communicate with the external world, file management on sible for external peripherals, such as analog sensing and
SD card, and security algorithms. The ARM also acts as an environmental sensors. This I/O board is one of the significant
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
9003612 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 70, 2021
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
JOGLEKAR et al.: OPEN-SOURCE HETEROGENEOUS CONSTRAINED EDGE-COMPUTING PLATFORM 9003612
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
9003612 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 70, 2021
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
JOGLEKAR et al.: OPEN-SOURCE HETEROGENEOUS CONSTRAINED EDGE-COMPUTING PLATFORM 9003612
TABLE I
M AXIMUM , M INIMUM , AND AVERAGE CR AND NMSE FOR T WO
D IFFERENT S ETTINGS OF T HRESHOLDS FOR L OSSY C OMPRESSION
AND A NOMALY D ETECTION W ITH n = 16, m = 4, AND BUF = 40
calculated over T blocks and |B| denotes the length of the bit
sequence B. NMSE is considered a standard metric in theory,
but it need not be an appropriate metric for error in power
applications. Indeed, it can be observed that the quality of
recovered waveform is much better than what is reflected by
the NMSE. A schematic summary of the proposed algorithm
is shown in Fig. 7, a preliminary version of the algorithm
appeared in [26], and the adaptation to the IED environment is
described in this work. The first step in the algorithm is to pass
each block of data x (comprising n samples) through a linear a high compression with n = 16, m = 4, and BUF = 40. For
filter. The filter used is an iterated filter where the output the medium case, τ H = 11 for both the voltage and current
y = (y1 , . . . , yn ) is given by y1 = x[1] and y2 = x[2] − x[1] channels and τ B = 7 for voltage channels and τ B = 6 for
and for k = 3, . . . , n, yk = x[k] − 2x[k − 1] + x[k − 2]. the current channels are used. For the high case, τ H = 15
The vector y is converted into a bit block B via a bit-packing and τ B = 14 are used for all the channels. It was observed
procedure in which each sample is stored using bits where that an average CR of around 1.42 for current channels and
is the maximum number of bits needed to represent any 1.64 for voltage channels could be obtained if all data have
sample within the block. The value of (represented using to be retained losslessly. Table I shows that an average CR
log2 r bits) is also included in B. An anomaly detector uses of up to around 45 can be configured for waveform recording
block B and the corresponding to determine whether to label based on the accuracy requirements for the n, m, and BUF.
the block as faulty or normal. If > τ H , a fixed threshold, Figs. 8 and 9 show the heat map of the CR for a snapshot
the block is labeled faulty, normal otherwise. Normal blocks of voltage and current channel for the two settings in the
are set to be stored in a lossy fashion, but not immediately table. Here, a higher CR is attained around the smooth part
since it is necessary to retain a data cycle before and after of the signal and lower around the edges, illustrating adaptive
the fault. Toward this, the bit block stream B1 , B2 , . . . , BT compression. It was observed that the compressed data are
resulting from the input stream x (1) , . . . , x (T ) is passed to an captured without any loss and without overloading the network
FIFO queue of length BUF, which is set to the number of bandwidth. The data collected and the computed data by
cycles intended to be retained around the detected anomaly. the IED are streamed to the local server for storage and
Once the block x (i) is set to be stored in a lossy fashion, visualization. The user can access the streamed timestamped
the bits corresponding to the first sample in x (i) is retained data of the applications via a web browser.
and others are dropped. This procedure is repeated at a higher
level as well. Namely, if there are m consecutive blocks with B. TEO-Based PMU Algorithm
only the first sample retained, these first samples are passed Fig. 10 shows the block diagram of the TEO-based tech-
through the same linear filter, and the output is used to decide nique proposed in [16] implemented on the FPGA. As shown
whether or not to retain all of the m values or just the first one in the figure, the PMU IP block is fed with a 32-kHz raw
among them. In particular, if the number of bits per sample data stream. The measurement involves four main stages: 1) a
resulting from this filter, > τ B , is a fixed threshold, all m bandpass filter (BPF); 2) TEO; 3) low-pass filter (LPF); and
values are retained. Otherwise, all values, except the first one, 4) moving average filter (MAF). Only a brief explanation of
are dropped. This process of downsampling the data first by n the stages is discussed here due to space constraints. The
and then by m, if needed, results in a hierarchical subsampling. raw data signal is first passed through an adaptive BPF. The
In effect, the algorithm tests whether or not high frequencies adaptive BPF is based on a recursive discrete Fourier trans-
are present and adjusts the sampling frequency accordingly. form (RDFT). It was observed that updating the number of
Reconstruction is performed by using splines to interpolate samples N dynamically is not numerically stable in a recursive
the dropped data points. The IED is connected to measure floating-point implementation [27]. For practical implementa-
the author’s laboratory input power supply. The measured tion, N is fixed and calculated as N = fs / f 0 , where fs and f 0
dataset comprises 5039 files of 1-s duration captured on are sampling and fundamental system frequency, respectively.
January 11, 2020, during daytime. The maximum, minimum, When the previously estimated frequency fˆ[n − 1] drifts from
and average CR and NMSE are calculated for all the 5039 files fundamental ( f 0 ), the phase and amplitude error are calculated
by varying the threshold settings τ H and τ B . Table I provides as discussed in [28], and accordingly, compensation is done
CR and NMSE for two cases corresponding to a medium and for the new sample x(n). Using the calculated amplitude, one
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
9003612 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 70, 2021
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
JOGLEKAR et al.: OPEN-SOURCE HETEROGENEOUS CONSTRAINED EDGE-COMPUTING PLATFORM 9003612
TABLE II
R ESOURCE U TILIZATION FOR PMU A LGORITHM
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
9003612 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 70, 2021
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
JOGLEKAR et al.: OPEN-SOURCE HETEROGENEOUS CONSTRAINED EDGE-COMPUTING PLATFORM 9003612
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.
9003612 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 70, 2021
[11] D. Schofield, F. Gonzalez-Longatt, and D. Bogdanov, “Design and [23] A. E. Emanuel and J. A. Orr, “Current harmonics measurement by
implementation of a low-cost phasor measurement unit: A compre- means of current transformers,” IEEE Trans. Power Del., vol. 22, no. 3,
hensive review,” in Proc. 7th Balkan Conf. Lighting (BalkanLight), pp. 1318–1325, Jul. 2007.
Sep. 2018, pp. 1–6. [24] W. Yao et al., “A novel method for phasor measurement unit sampling
[12] I. Jahn, F. Hohn, G. Chaffey, and S. Norrga, “An open-source time error compensation,” IEEE Trans. Smart Grid, vol. 9, no. 2,
protection IED for research and education in multiterminal HVDC pp. 1063–1072, Mar. 2018.
grids,” IEEE Trans. Power Syst., vol. 35, no. 4, pp. 2949–2958, [25] Y. Kononov, M. Ospishchev, and V. Rudnev, “Current and voltage
Jul. 2020. synchronous measurements using NI cRIO system with sigma-delta
[13] J.-P. Kitzig, S. Schlaghecke, and G. Bumiller, “Power quality mea- ADC NI 92xx and S.E.A GPS modules,” in Proc. 20th IMEKO World
surement system with PMU functionality based on interpolated sam- Symp., 2014, pp. 25–30.
pling,” IEEE Trans. Instrum. Meas., vol. 68, no. 4, pp. 1014–1025, [26] K. R. Sahasranand, B. Rout, A. Joglekar, G. Gurrala, and H. Tyagi,
Apr. 2019. “Fault-aware compression for high sampling rate data acquisition in
[14] Parallella Publications. Accessed: Mar. 20, 2021. [Online]. Available: smart grids,” in Proc. 10th ACM Int. Conf. Future Energy Syst.
https://www.parallella.org/publications/ New York, NY, USA: ACM, Jun. 2019, pp. 422–424.
[15] Epiphany Archit. Reference. Accessed: Jan. 10, 2020. [Online]. [27] H. A. Darwish and M. Fikri, “Practical considerations for recursive DFT
Available: https://www.adapteva.com/docs/epiphany_arch_ref.pdf implementation in numerical relays,” IEEE Trans. Power Del., vol. 22,
[16] S. Reza, M. Ciobotaru, and V. G. Agelidis, “Single-phase grid volt- no. 1, pp. 42–49, Jan. 2007.
age frequency estimation using teager energy operator-based tech- [28] M. Wang and Y. Sun, “A practical method to improve phasor and power
nique,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 3, no. 4, measurement accuracy of DFT algorithm,” IEEE Trans. Power Del.,
pp. 1218–1227, Dec. 2015. vol. 21, no. 3, pp. 1054–1062, Jul. 2006.
[17] IEEE Recommended Practice for Monitoring Electric Power Quality, [29] J. F. Kaiser, “On a simple algorithm to calculate the ‘energy’ of a
IEEE Standard 1159-2009 (Revision of IEEE Std 1159-1995), Jun. 2009, signal,” in Proc. Int. Conf. Acoust., Speech, Signal Process., vol. 1, 1990,
pp. 1–94. pp. 381–384.
[18] Epiphany SDK Reference. Accessed: Jan. 10, 2020. [Online]. Available: [30] IEEE Standard for Synchrophasor Measurements for Power Systems—
https://adapteva.com/docs/epiphany_sdk_ref.pdf Amendment 1: Modification of Selected Performance Requirements,
[19] C. Jiang, X. Cheng, H. Gao, X. Zhou, and J. Wan, “Toward compu- IEEE Standard C37.118.1a-2014 (Amendment to IEEE Std C37.118.1-
tation offloading in edge computing: A survey,” IEEE Access, vol. 7, 2011), Apr. 2014, pp. 1–25.
pp. 131543–131558, 2019. [31] Xilinx. Vivado Design Suite User Guide: High-Level Synthesis.
[20] B. P. Robert, P. A. Babu, N. Kashyap, and G. Gurrala, “Practical Accessed: Jan. 10, 2020. [Online]. Available: https://www.xilinx.com/
approaches towards securing edge devices in smart grid,” in Proc. 8th support/documentation/sw_manuals/xilinx2014_1/ug902-vivado-high-
Int. Conf. Power Syst. (ICPS), Dec. 2019, pp. 1–6. level-synthesis.pdf
[21] H. Seljeseth, E. A. Saethre, T. Ohnstad, and I. Lien, “Voltage transformer [32] Xilinx. Vivado Design Suite Tutorial: High-Level Synthesis.
frequency response. Measuring harmonics in norwegian 300 kV and Accessed: Jan. 10, 2020. [Online]. Available: https://www.xilinx.com/
132 kV power systems,” in Proc. 8th Int. Conf. Harmon. Qual. Power., support/documentation/sw_manuals/xilinx2014_1/ug871-vivado-high-
vol. 2, 1998, pp. 820–824. level-synthesis-tutorial.pdf
[22] M. Klatt, J. Meyer, M. Elst, and P. Schegner, “Frequency responses [33] Electromagnetic Compatibility (EMC)—Part 4-30: Testing and Measure-
of MV voltage transformers in the range of 50 Hz to 10 kHz,” in ment Techniques—Power Quality Measurement Methods, document IEC
Proc. 14th Int. Conf. Harmon. Qual. Power (ICHQP), Sep. 2010, 61000 4-30 (2003), 2003.
pp. 1–6. [Online]. Available: http://ieeexplore.ieee.org/document/ [34] A. Grama, V. Kumar, A. Gupta, and G. Karypis, Introduction to Parallel
5625484/ Computing. London, U.K.: Pearson, 2003.
Authorized licensed use limited to: San Francisco State Univ. Downloaded on June 20,2021 at 02:38:57 UTC from IEEE Xplore. Restrictions apply.