Machine Learning-Based Error Correction Codes and Communication Protocols for Power Line Communication: An Overview

Abstract

This study endeavors to investigate the effectiveness of machine learning-based methodologies in enhancing the performance and reliability of Power Line Communication (PLC) systems. PLC systems constitute a critical component within the domains of energy management, monitoring, and automation. The fundamental objective herein is to contribute significantly to the scholarly discourse by conducting a comprehensive review encompassing research investigations and practical applications documented in the extant literature. The primary motivation underpinning this research is predicated upon the necessity for a meticulous evaluation of machine learning techniques that hold the potential to enhance the efficacy and stability of PLC systems. The deployment of these techniques bears the promise of engendering heightened levels of efficiency across the spectrum of energy management, communication, and automation systems. Within this scholarly quest, the study posits a hypothesis: Machine learning-based methodologies possess the capacity to effect marked improvements in the performance and reliability of PLC systems. Methodological scrutiny is executed through a comprehensive evaluation of diverse machine learning techniques, including, but not limited to, deep learning, support vector machines, and random forests, facilitated by a series of empirical experiments and simulations. Empirical findings resoundingly corroborate the proposition, substantiating a significant enhancement in the operational performance of PLC systems when these machine learning methods are judiciously employed. In summation, this study assumes the role of a catalyst in exploring latent, untapped potential inherent within machine learning-based methodologies, customarily calibrated to resonate within the intricate matrix of PLC systems. The zenith of this rigorous investigation stands poised to illuminate the path toward transformative advancements in the domains of energy management, communication, monitoring, and automation systems. The findings contribute significantly to the academic discourse, offering a compass for future research inquiries and practical applications within this burgeoning and dynamic field.

Full text

Received 8 October 2023, accepted 3 November 2023, date of publication 6 November 2023, date of current version 10 November 2023.
Digital Object Identifier 10.1109/ACCESS.2023.3330690
Machine Learning-Based Error Correction Codes
and Communication Protocols for Power Line
Communication: An Overview
TAHIR CETIN AKINCI 1,2, (Senior Member, IEEE),
GOKHAN ERDEMIR 3, (Senior Member, IEEE), A. TARIK ZENGIN 2,
SERHAT SEKER 2, AND ABDOULKADER IBRAHIM IDRISS 4
1WCGEC, University of California at Riverside (UCR), Riverside, CA 92521, USA
2Electrical Engineering Department, Istanbul Technical University (ITU), 344690 İstanbul, Turkey
3Engineering Management and Technology, The University of Tennessee at Chattanooga, Chattanooga, TN 37403, USA
4Department of Electrical and Energy Engineering, Faculty of Engineering, Université de Djibouti, Djibouti City, Djibouti
Corresponding author: Tahir Cetin Akinci (tahircetin.akinci@ucr.edu)
ABSTRACT This study endeavors to investigate the effectiveness of machine learning-based methodologies
in enhancing the performance and reliability of Power Line Communication (PLC) systems. PLC systems
constitute a critical component within the domains of energy management, monitoring, and automation.
The fundamental objective herein is to contribute significantly to the scholarly discourse by conducting a
comprehensive review encompassing research investigations and practical applications documented in the
extant literature. The primary motivation underpinning this research is predicated upon the necessity for a
meticulous evaluation of machine learning techniques that hold the potential to enhance the efficacy and
stability of PLC systems. The deployment of these techniques bears the promise of engendering heightened
levels of efficiency across the spectrum of energy management, communication, and automation systems.
Within this scholarly quest, the study posits a hypothesis: Machine learning-based methodologies possess the
capacity to effect marked improvements in the performance and reliability of PLC systems. Methodological
scrutiny is executed through a comprehensive evaluation of diverse machine learning techniques, including,
but not limited to, deep learning, support vector machines, and random forests, facilitated by a series
of empirical experiments and simulations. Empirical findings resoundingly corroborate the proposition,
substantiating a significant enhancement in the operational performance of PLC systems when these machine
learning methods are judiciously employed. In summation, this study assumes the role of a catalyst in explor-
ing latent, untapped potential inherent within machine learning-based methodologies, customarily calibrated
to resonate within the intricate matrix of PLC systems. The zenith of this rigorous investigation stands
poised to illuminate the path toward transformative advancements in the domains of energy management,
communication, monitoring, and automation systems. The findings contribute significantly to the academic
discourse, offering a compass for future research inquiries and practical applications within this burgeoning
and dynamic field.
INDEX TERMS Power line communication, error correction codes, machine learning, transmission control
protocols, communication protocols, power networks.
I. INTRODUCTION
The Power Line Communication (PLC) technology offers
a reliable and cost-effective communication solution for
The associate editor coordinating the review of this manuscript and
approving it for publication was Oussama Habachi .
various applications such as Smart Buildings, Smart Cities,
and Industry 4.0 [1],[2]. The broadband PLC technol-
ogy development enables higher data rates and supports
more protocols used in Smart Building applications [3],[4].
This technology is especially suitable for continuous power
quality monitoring, electric vehicle charging, and microgrid
124760
2023 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
and distribution generation applications. The deployment of
Broadband Over Power Lines (BPL) networks is expected
to keep growing due to the increasing demand for energy
efficiency and emissions reduction [5],[6].
PLC technology is a method to utilize electrical power
transmission lines as a medium to convey data through a
conductor [7]. This technology is known by various nomen-
clatures, including Power Line Server (PLS), Power Line
Digital Subscriber Line (PDSL), grid communication, power
line telecommunications (PLT), Power Line Network (PLN),
or Broadband Over Power Lines (BPL). Electric power
transmission occurs over high-voltage transmission lines,
is subsequently distributed over medium-voltage lines, and
is ultimately utilized within buildings at lower voltage levels.
PLC systems can seamlessly transition between two distinct
levels, such as the distribution network and plant cabling,
albeit each PLC technology is constrained to a particular
set of cables. To accommodate the creation of expansive
networks, it is possible to amalgamate multiple PLC tech-
nologies, as the transmission of signals can be interrupted by
transformers [8],[9].
Power networks are categorized into three types: DC
sources used in industrial applications such as automotive,
sinusoidal supply used for electrical distribution networks or
domestic applications [10], and expansion units containing
converters and actuators pulse width modulated (PWM) net-
works [11],[12].
PLC technology is often used over sinusoidal and con-
tinuous electrical networks, guaranteeing several hundred
megabit data rates. However, PLC modems cannot operate
in PWM networks due to their wide spectral occupancy [13].
New PLC modems are deployed specifically for PWM net-
works based on a comprehensive review of the inverter
spectrum. These modems achieve reliability and data rate
capacity values. This technology eliminates additional cable
length between the actuator and transducers, resulting in cost
and size advantages.
A. POWER LINE COMMUNICATION
PLC technology enables data communication over the elec-
trical grid [14]. This technology is an alternative solution
to wireless or wired networks, particularly in indoor envi-
ronments. PLC utilizes frequency ranges in the power
grid to facilitate data transmission and can be used for
various applications such as internet connectivity, IPTV,
smart metering devices, smart home systems, and industrial
automation [15],[16],[17].
PLC is widely used in many fields due to its communica-
tion quality, speed, cost, and ease of installation advantages.
However, noise and signal distortions in the power grid
FIGURE 1. Categorization and network visualization of PLC system.
VOLUME 11, 2023 124761

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
FIGURE 2. Essential components of a PLC system [23],[24].
can impact the communication quality of PLC [18],[19].
Different techniques are employed for the development and
improvement of PLC. These techniques include error correc-
tion codes, artificial neural networks, deep learning, support
vector machines, and random forests, among other machine
learning-based methods.
PLC technology may find broader applications in the
future, particularly in areas such as smart cities, intelligent
transportation systems, and renewable energy [20].
In PLC systems, coupling plays a crucial role in PLC, and
it is expected to be more involved in the next generation of
PLC. The performance evaluation of PLC primarily focuses
on how the Connectivity Unit (CU) is designed, interfaced,
and connected to the Power Grid and its performance in
a boisterous environment. In the literature, PLC couplers
are classified based on six criteria: (1) Physical connec-
tion, (2) Voltage level, (3) Voltage type, (4) propagation
mode, (5) frequency band, and (6) number of connec-
tions. Figure 1illustrates the different types of PLC in
each category. The physical link mentioned here involves
integrating, injecting, and extracting the communication
signals to and from the power line. Consequently, to estab-
lish a connection with the line, there are four ways of
physical configuration: antenna, resistive, inductive, and
capacitive [21],[22].
The schematic representation of a PLC network used in an
electrical power system is shown in Figure 2 [23]. The power
transmission line, Transmitter, and Receiver components are
depicted here. PLC is a communication method that utilizes
the existing electrical power infrastructure to transmit data
from the sender to the receiver. The system operates in full
duplex mode and consists of three fundamental parts. A com-
ponent called ‘‘wave-trap’’ or ‘‘line-trap’’ is used in this
system to prevent communication signals from entering the
equipment through the power supply line and to allow signal
division. This component provides high series impedance
to the carriers’ frequencies, comprising different resonant
circuits that block all communication currents while allowing
power frequency to pass. This system is also referred to as a
connection element.
•Terminal Devices: The terminal section encompasses the
essential components of the PLC network. Transceivers
and protective relays are used for initiating and direct-
ing communication. Transceivers facilitate the transfer
of data between the sender and receiver. On the other
hand, protective relays are employed to ensure network
security and provide protection in case of line faults.
•Connection Equipment: The connection section pro-
vides the physical connections for the PLC communica-
tion network. The line adjuster ensures the transmission
of communication signals onto the power line and
adjusts the line parameters accordingly. The connection
capacitor allows for the injection of communication
signals onto the power line and their transmission to
the receiver. Additionally, a combination of components
such as wave traps or line traps aims to block undesired
frequency components on the transmission line, enhanc-
ing communication quality and mitigating disturbances.
•50/60 Hz Power Transmission Line: This section forms
the transmission path for the PLC communication net-
work. Existing 50/60 Hz power transmission lines are
utilized for data transfer across the PLC bandwidth.
Communication signals are transmitted over and deliv-
ered through the power lines within the 50/60 Hz
frequency range. Consequently, data transmission is
achieved without a separate communication infrastruc-
ture [1],[14],[25],[26],[27]. PLC is a highly effective
communication method employed in transformer sta-
tions, leveraging the existing power infrastructure to
reduce costs [28].
•Coupling Capacitor: The coupling capacitor connects
the transmission line and terminal devices, facilitating
the transmission of carrier signals [28]. It is designed
to exhibit high impedance at power frequencies and
low impedance at carrier signal frequencies [29]. These
equipment systems are typically constructed using paper
124762 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
or liquid dielectric materials, primarily for high-voltage
applications. The rating of coupling capacitors varies
based on IEEE guidelines, ranging from 0.004-0.01µF
at 34 kV to 0.0023-0.005µF at 765 kV [30].
•Drain Battery: Figure 2illustrates the role of the dis-
charge coil, which provides high impedance for carrier
frequencies and low impedance for power frequencies.
The limitations of PLC are as follows:
1. Infrastructure Constraints: PLC is constrained by the
characteristics of the existing electrical infrastruc-
ture [31], which introduces various factors that impact
power line channel parameters, including power atten-
uation, noise, impedance, and bandwidth [32],[33].
2. Signal-to-Noise Ratio Requirements: PLC communica-
tion necessitates a high Signal-to-Noise Ratio (SNR)
for effective data transmission [31]. This implies
that a strong signal is required while minimizing
unwanted noise to ensure reliable communication [34].
The design and planning of power lines could be
more suitable for high-frequency signals. In electric
power transmission lines, attenuation, multipath due to
impedance mismatches, and noise are the three most
essential degradation factors that degrade PLC perfor-
mance [35]. Noise in PLC channels is generated by all
electrical devices connected to the Quality, which is a
parameter of the receiver’s noise level and the electrical
signal’s attenuation at different frequencies. The higher
the noise level, the more difficult it is to detect the
received signal. Suppose the signal is attenuated on its
way to the receiver. In that case, it can make the deci-
sion even more difficult as the signal is further obscured
by noise, expressed as the signal-to-noise ratio (SNR)
level. SNR measures how much a signal is degraded by
noise (Equation 1)[35],[36].
SNRdB =10log10 Psignal
Pnoise (1)
3. Unmatched Loads and Variability: The power line net-
work often encounters unmatched loads and exhibits
temporal variations [37]. This can lead to power carrier
attenuation, a significant drawback of PLC, affect-
ing the quality and reliability of the communica-
tion [38],[39].
4. Reflection Losses: Carrier frequencies in PLC systems
experience reflection losses at different points along
the transmission path, such as from the transmitter,
through the coaxial cable, the line adjuster unit, the
coupling capacitor, and the power line. These losses can
result in signal degradation, loss, and distortion during
transmission [37],[38].
5. Security Concerns: PLC faces security challenges as
the power line infrastructure is susceptible to exter-
nal interference and electromagnetic disturbances. This
vulnerability poses potential security risks, requiring
appropriate measures to ensure data confidentiality and
integrity [1].
These limitations underscore the disadvantages [31] and
challenges associated with using PLC as a communication
method over power lines [40],[41].
B. ERROR CORRECTION CODES AN D COMMUN ICATION
Protocols are employed for detecting and correcting errors
during data transmission, with error correction codes serving
as a critical component in this process [42]. Erroneous data
transmission can arise due to channel noise, interference,
parasitic elements, electromagnetic interference, and vari-
ous other environmental factors [43]. Error correction codes
facilitate the rectification or retransmission of erroneous data
packets by appending additional information (parity bits) to
the transmitted data [44],[45]. Fundamental error correction
codes encompass the following:
a. Hamming Codes: Hamming codes provide single-error
correction and double-error detection capabilities.
These codes are used to correct errors at the bit level
during data transmission [46].
b. Reed-Solomon Codes: Reed-Solomon codes are
block-based error correction codes that can correct
multiple-bit errors. They are mainly employed in appli-
cations such as optical and magnetic storage devices,
wireless communication systems, and digital television
broadcasting [47],[48].
c. Turbo Codes and LDPC (Low-Density Parity-Check)
Codes: These codes offer higher error correction
performance, enabling efficient and reliable commu-
nication. They are commonly utilized in space and
satellite communication systems [49],[50].
The classification of forward error codes is given in
Figure 3. Here, under Block Codes and Convolution Codes,
a general category can be made as Hamming, Golay, BCH,
Rs, LDPC, Trellis, and Turbo Codes [22],[51].
FIGURE 3. Classification of error codding [22],[51].
In addition to error correction codes, communication pro-
tocols are crucial in facilitating reliable and efficient data
transmission. These protocols define the rules and procedures
for communication, including data framing, error detec-
tion, flow control, and packet acknowledgment. Examples of
VOLUME 11, 2023 124763

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
popular communication protocols include the Ethernet pro-
tocol, Wi-Fi protocols (e.g., IEEE 802.11 standards), and the
Transmission Control Protocol/Internet Protocol (TCP/IP)
suite [50],[51]. Error correction codes and communication
protocols ensure accurate and robust data transmission in
various applications [51].
Communication protocols facilitate synchronized and
orderly data exchange among participants in data commu-
nication processes. These protocols define communication
processes such as data formatting, timing, error control, and
flow control. The main communication protocols include the
followings:
•TCP (Transmission Control Protocol): TCP is a reliable,
connection-oriented protocol used for data transmis-
sion over the Internet. It provides error control and
flow control mechanisms for ensuring proper deliv-
ery of data packets and retransmission of erroneous
packets [50],[51].
•UDP (User Datagram Protocol): UDP is a fast and
lightweight communication protocol that operates con-
nectionless. Unlike TCP, UDP does not provide error
and flow control mechanisms, making it more prone to
packet loss or errors. This protocol is suitable for real-
time applications, video, and audio streaming [52].
•IP (Internet Protocol): IP is a fundamental network pro-
tocol that enables routing of data packets from source to
destination over a network. It works with TCP and UDP,
operating at the network layer [53].
•ARP (Address Resolution Protocol): ARP is a protocol
that translates IP addresses at the network layer to phys-
ical addresses (MAC addresses) at the data link layer.
This facilitates data transmission between devices on the
network [54].
•HTTP (Hypertext Transfer Protocol): HTTP is an
application layer protocol that transmits web pages
between web browsers and servers. It typically operates
over TCP and can be used securely with the HTTPS
(HTTP Secure) version [55],[56].
•FTP (File Transfer Protocol): FTP is an application layer
protocol used for file transfers. It provides a reliable and
efficient method for uploading and downloading files
between servers [57],[58].
•MQTT (Message Queuing Telemetry Transport): MQTT
is a lightweight and energy-efficient communication
protocol designed specifically for Internet of Things
(IoT) applications. MQTT enables message transmis-
sion with low bandwidth and power consumption
between devices and servers [59],[60].
Error correction codes and communication protocols ensure
data communication’s reliability and performance [61].
A well-designed communication system utilizes effective
error correction codes and protocols to correct errors and
facilitate orderly and synchronized data exchange, leading to
more reliable and efficient communication systems [62],[63].
The error correction analysis graph for a PLC model is
given in Figure 4. Here, the Hamming (7,4) encoding matrix
is used to encode a 4-bit data string, the original data is
represented by input_data, and the encoded data is obtained
by multiplying the data with the encoding matrix and getting
modulo 2.
FIGURE 4. Error correction codes and communication.
The provided illustration serves as a fundamental demon-
stration of error-correcting code implementation. Real-world
error correction codes and communication protocols exhibit
greater intricacy, necessitating meticulous design and opti-
mization tailored to precise requirements and application
contexts.
C. UTILIZATION OF MACHINE LEARNING-BASED
METHODS
Machine learning (ML) is rapidly evolving as a significant
tool in data analysis and pattern recognition [44]. In the field
of communication systems, particularly in areas such as error
correction codes and communication protocols, the utiliza-
tion of machine learning-based methods has led to significant
improvements [58],[59]. This section will discuss the impact
and applications of machine learning-based methods on error
correction codes and communication protocols.
Machine Learning-Based Error Correction Codes: Tra-
ditional error correction codes detect and correct errors
using predefined algorithms and rules. In contrast, machine
learning-based error correction codes aim to detect and cor-
rect errors by learning from a training dataset. This approach
performs better error correction in noisy and complex com-
munication environments [64].
Machine learning-based error correction codes can be
developed using various ML algorithms such as deep learn-
ing, support vector machines, and random forests. These
methods make error correction codes more flexible, scalable,
and adaptable to new communication environments and chal-
lenges [65].
Machine Learning-Based Communication Protocols:
Communication protocols regulate data communication pro-
cesses and operate according to specific standards and rules.
124764 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
Machine learning-based communication protocols enable
more effective and compatible data exchange by analyzing
network traffic and communication processes [66],[67].
Machine learning algorithms can optimize operations in
communication protocols, such as flow control, error control,
timing, and network routing. This allows communication
protocols to respond more swiftly and effectively to changing
network conditions and user demands [68],[69].
The utilization of machine learning-based methods con-
tributes to enhancing reliability and performance in com-
munication systems. Further research and applications in
this field will develop more efficient and reliable com-
munication systems such as energy management, moni-
toring, and automation systems. Moreover, using machine
learning-based methods enables the development of error
correction codes and protocols that can quickly adapt to new
and complex communication environments [69],[70].
Sample Application Areas:
•Wireless communication: Machine learning-based meth-
ods can assist in detecting and correcting erroneous
data packets in wireless communication systems. This
enhances reliability and data transmission speed in wire-
less networks [71].
•Voice and video communication over the Internet: Pro-
tocols developed for real-time voice and video commu-
nication can benefit from machine learning algorithms
to provide more effective error control and flow control.
This ensures a higher-quality and uninterrupted commu-
nication experience [72].
•Internet of Things (IoT) systems: IoT devices typi-
cally operate with low power consumption and low
bandwidth. Machine learning-based error correction
codes and communication protocols can enhance data
transmission reliability and energy efficiency in IoT
systems [73].
•Space and satellite communication systems: Signals in
space and satellite communication systems travel long
distances and encounter various noise and interference.
Machine learning-based methods can help prevent and
correct erroneous data transmission in these systems,
thereby improving communication reliability and effi-
ciency [74].
As a result, using machine learning-based methods leads to
significant advancements and improvements in error correc-
tion codes and communication protocols [75],[76],[77].
These methods enhance communication systems’ reliability,
performance, and energy efficiency, providing a better com-
munication experience in diverse application areas. In the
future, the continued use and development of machine
learning-based methods will contribute to further advance-
ments in communication technologies.
The primary motivation for this research is to explore
the untapped potential of machine learning-based meth-
ods in enhancing the performance and reliability of
Power Line Communication systems. The overarching goal
is to contribute substantively to the scholarly field by
comprehensively reviewing existing research and practical
applications in the literature. By synthesizing this knowledge,
the study aims to pave the way for transformative advance-
ments in energy management, communication, monitoring,
and automation systems. Ultimately, this research aims to
unlock the hidden capabilities of machine learning in PLC
systems and to shape a future characterized by improved
performance and unwavering reliability.
II. MACHINE LEARNING-BASED ERROR CORRECTION
CODES AND COMMUNICATION PROTOCOLS
Machine learning-based error correction codes and commu-
nication protocols are an enhanced version of error coding
methods [71],[76]. These methods utilize machine learning
techniques to detect and correct errors encountered dur-
ing data transmission. These methods are more effective
than traditional error correction coding because they per-
form error detection and correction based on the statistical
properties of the data. Furthermore, machine learning-based
error correction codes and communication protocols allow for
selecting more suitable error correction codes by consider-
ing the characteristics of the transmission channel. Machine
learning-based error correction codes and communication
protocols are commonly employed in environments with
high error rates, such as wireless communication systems.
Data packets can be corrupted or lost in these systems
due to atmospheric conditions or nearby obstacles. In such
cases, machine learning-based error correction codes and
communication protocols ensure secure and accurate data
transmission.
A. DIFFERENC ES BETWEE N TRADITIONAL METHODS AND
MACHINE LEARNING-BASED METHODS POWER LIN E
PLC technology utilizes electrical power lines for data
transmission. PLC systems often face challenges such as
high levels of noise and signal distortion [78]. Therefore,
error correction methods ensure accurate data transmission
and reliability. Traditional error correction methods rely on
pre-defined mathematical rules. Examples of conventional
error correction methods used in PLC systems include dupli-
cation, parity checks, cyclic redundancy checks (CRC), and
Reed-Solomon codes [79]. These methods employ specific
mathematical formulas for error detection and correction.
Machine learning-based error correction methods offer more
effective error correction by automatically determining the
characteristics and patterns of the data. In PLC systems,
machine learning-based methods can detect and modify
errors [51]. These methods perform error correction based
on the statistical properties of the data. For instance, suit-
able error correction codes considering factors like channel
characteristics can be selected using machine learning-based
techniques. Machine learning-based error correction methods
provide more effective error correction than traditional ones.
However, they are more complex and have higher compu-
tational requirements. Additionally, machine learning-based
methods require more data and may have longer training
processes [77].
VOLUME 11, 2023 124765

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
The academic literature has extensively employed PLC
technology with various classification algorithms. Among the
most frequently utilized methods are the following: the Naive
Bayes (NB) algorithm, Linear Discriminant Analysis (LDA),
Logistic Regression (LR), K-Nearest Neighbors (KNN), Sup-
port Vector Machine (SVM), Decision Tree (DT), Random
Forest (RF), Adaptive Boosting (AdaBoost), Extreme Gra-
dient Boosting (XGBoost), and various Rule-based classifi-
cation algorithms. Beyond these, different other algorithms
have also been integrated into practical applications. The
Probability-based Logistic Regression algorithm has been
deployed to estimate typical probabilities. Equation 1herein
presents the mathematical representation of the sigmoid func-
tion, also known as the logistic function (Eq2).
g(z)=1
1+exp(−z)(2)
The Decision Tree (DT) is a widely recognized nonparamet-
ric supervised learning technique employed for classification
and regression purposes. Within the framework of this
algorithm, mathematical expressions for the ‘‘Gini’’ index,
representing Gini impurity (E), and the entropy (Hx) metric
for information gain are provided in Equations 3and 4,
respectively [79].
H(x)= − Xn
i=1p(xi)log2p(xi) (3)
E=1−Xc
i=1p2
i(4)
Within the literature, the customary practice provides an
overview of the typical architectural structure of Machine
Learning (ML) applications utilizing PLCs.
FIGURE 5. The architecture of machine learning-based PLC [79].
This structure typically delineates three fundamental com-
ponents: the database, the testing data, and the model
prediction pertaining to the data received by the PLC
(Figure 5). The classification tree algorithm model is given
in Figure 6.
B. MACHINE LEARNING-BASED ERROR
CORRECTION CODES
PLC systems encounter challenges such as high noise
and signal distortion levels, requiring error correction
FIGURE 6. Classification tree algorithm model [79].
methods to ensure accurate data transmission and reliabil-
ity. Traditional methods are based on pre-defined mathe-
matical rules and employ specific mathematical formulas
for error detection and correction. However, the effec-
tiveness of traditional methods can be limited. Machine
learning-based methods offer more effective error correc-
tion by automatically determining the characteristics and
patterns of the data [77],[78]. These methods perform
error correction based on the statistical properties of the
data. For instance, suitable error correction codes can be
selected considering factors such as channel characteris-
tics. Machine learning-based error correction methods are
more effective than traditional ones [80]. These methods
automatically detect and correct errors, ensuring accurate
and reliable data transmission [81]. However, these meth-
ods are more complex and have higher computational
requirements.
Moreover, they require a greater data volume and may
entail lengthier training processes. Machine learning-based
error correction codes for PLCs can be harnessed to effec-
tively detect and rectify errors within PLC systems [51],[80].
These methodologies can potentially enhance system per-
formance and ameliorate data transmission quality. Machine
learning-based error correction codes for PLCs hold substan-
tial value in critical applications such as data transmission
within intricate systems like smart grids and power distribu-
tion networks [61],[62].
In this study, a feedforward neural network with 10 hid-
den neurons was constructed using a synthetic dataset. The
model was trained on this synthetic dataset to make predic-
tions. By comparing these predictions with the actual data,
we visualize the performance of the trained neural network.
The legend in Figure 7clarifies the distinction between
the actual data points and the model’s predictions, allow-
ing for a comprehensive evaluation of the neural network
architecture.
The performance analysis for PLC based on synthetic data
is presented in Figure 8. The results are categorized into
training, validation, testing, and the best performance out-
comes. The optimal validation performance achieved a value
of 0.55432 at epoch 4.
124766 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
FIGURE 7. Feed-forward neural network architecture.
FIGURE 8. Performance analysis of the PLC model.
The Error-Target outputs plot for the PLC model is
depicted in Figure 9. The graph displays the training,
validation, test, and zero error variations over the training
iterations.
In Figure 10, the training and regression analysis
graph for the PLC model is given. Here, the values are
0.99687 for Training, 0.99115 for Validation, 0.99571 for
Test, and 0.99575 for all.
Data division is randomly selected in the training
algorithm, Training Levenberg-Marquardt, performance
mean squared error and calculation max. The training process
table is given in Table 1.
FIGURE 9. Error-Histogram graph for the PLC model.
FIGURE 10. Training and Regression analysis for the PLC model.
TABLE 1. Training process.
1) DEEP LEARNING-BASED ERROR CORRECTION CODES
Deep learning is a machine learning technique encompass-
ing multi-layered learning processes over artificial neural
networks to extract meaningful features from data [82].
VOLUME 11, 2023 124767

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
This method has achieved significant success in various
domains, with error correction codes among them. Deep
learning-based error correction codes are employed in com-
munication systems that contend with high levels of noise
and signal degradation [83],[84]. These techniques enhance
error correction efficiency by automatically discerning data
characteristics and patterns.
Deep learning-based error correction codes primarily
operate on specialized neural networks for encoding and
decoding [84]. These neural networks can be of different
types, such as convolutional neural networks or recurrent
neural networks. These networks extract specific features
from the data and use these features in error detection and
correction. Deep learning-based error correction codes offer
more effective error correction than traditional methods [85].
However, these methods are more complex and have higher
computational requirements. Additionally, they require more
data and may have longer training processes. In conclusion,
deep learning-based error correction codes may have higher
accuracy rates than traditional methods. However, different
neural network models and appropriate hyperparameter tun-
ing may be required depending on the application domain.
2) SUPPORT VECTOR MACHINES-BASED
ERROR CORRECTION CODES
Support Vector Machines (SVM) are machine learning’s
widely used classification and regression methods [86].
SVM can achieve high accuracy rates in linear and non-
linear problems, making SVM-based error correction codes
applicable to error correction processes in PLC systems.
SVM-based error correction codes rely on the assumption
of linear separability in the data [87]. SVM positions data
points in a multi-dimensional space and creates an optimal
separating hyperplane for classification [86]. SVM ensures
maximum margin separation while creating the hyperplane,
allowing for determining boundaries in the widest possible
manner [87],[88]. SVM-based error correction codes yield
good results even in noisy data. SVM can automatically
determine the features of the data, enabling the selection
of the most suitable codes for error correction. SVM-based
error correction codes can reduce the dimensionality of the
data, thereby shortening the processing time [89],[90]. SVM-
based error correction codes also have disadvantages similar
to machine learning-based methods. SVM is a computation-
ally intensive method that may need improvement for large
datasets [91]. Moreover, SVM’s premise of linear separabil-
ity might only sometimes apply to specific problems. Error
correction codes based on SVM can serve as a substitute
for conventional methods used in PLC systems [92]. SVM
can autonomously identify data features and select the most
appropriate error correction codes [93]. Nonetheless, SVM
can be computationally demanding, and the linear separabil-
ity presumption might only sometimes be valid for specific
issues.
In Figure 11, the process of estimating with the SVM is
demonstrated. The code generates a synthetic dataset and
FIGURE 11. Synthetic data and SVM prediction.
constructs and trains the Support Vector Machine (SVM)
model using the fitrsvm function. Subsequently, it plots the
predictions made by the trained SVM model alongside the
actual data points. This visualization clearly represents both
the synthetic data and the model’s estimations.
3) RANDOM FOREST-BASED ERROR CORRECTION CODES
Random Forests are commonly used algorithms in machine
learning [94]. This algorithm combines multiple decision
trees to create a classification or regression model. Ran-
dom Forests perform well on high-dimensional and complex
data. Random Forests can also be used for error correction
codes [95]. In this case, the Random Forest algorithm ana-
lyzes the transmitted data to detect and correct errors [96].
This method yields better results than traditional error correc-
tion methods because the Random Forest algorithm can better
determine relationships and patterns within the data. When
used for error correction codes, the Random Forest algorithm
first creates a model using sample data. This model is then
used to detect errors in the data. Once errors are detected,
the Random Forest algorithm corrects them using a specific
error correction code. Random Forest-based error correction
codes perform well on high-dimensional and noisy data. They
also require less computational power than traditional meth-
ods and have a faster training process. However, Random
Forest-based error correction codes require sufficient sample
data to build the model. Otherwise, the model may need to be
revised or provide misleading results.
Figure 12 illustrates the graph representing the synthetic
data and the predictions made by the random forest algorithm.
Upon examination of the graph, it becomes evident that the
random forest predictions algorithm successfully accurately
approximates the data.
C. MACHINE LEARNING-BASED COMMUN ICATION
PROTOCOLS
PLC is a technology used for data transmission over electri-
cal power lines. PLC systems face challenges such as high
124768 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
FIGURE 12. Synthetic data and random forest predictions.
levels of noise and signal distortion, necessitating specialized
communication protocols to ensure accurate data transmis-
sion and reliability. Traditional communication protocols are
rule-based and may need help to adapt to changing commu-
nication conditions [78].
Machine learning-based communication protocols learn
communication conditions automatically to determine the
most suitable communication strategies [51],[80]. These
protocols analyze data using various machine learning algo-
rithms and select the optimal protocol based on the commu-
nication conditions. This improves communication quality
and enables communication protocols to adapt to changing
conditions [81].
Machine learning-based communication protocols also
ensure data security. Data encryption is crucial for data
privacy and security. Machine learning-based protocols auto-
matically detect security vulnerabilities and implement nec-
essary measures to enhance security [97],[98],[99].
However, machine learning-based communication pro-
tocols typically begin with signal processing and feature
extraction steps to analyze data characteristics and chan-
nel conditions. These steps involve extracting data features,
reducing noise and signal distortion, analyzing frequency
spectra, and feeding this data into machine learning algo-
rithms [100]. Machine learning algorithms, particularly sup-
port vector machines, random forests, deep neural networks,
and various clustering algorithms, can assist in accurate data
classification and error correction processes in the transmis-
sion channel [77]. Machine learning-based communication
protocols can provide higher reliability and efficiency for data
transmission in PLC systems. However, implementing this
technology may present practical challenges due to the need
for high computational power and data resources, especially
for data collection and processing steps.
1) DEEP LEARNING-BASED COMMUNICATION PROTOCOLS
Deep learning-based communication protocols perform error
correction based on the statistical characteristics of data,
distinguishing them from traditional protocols used in PLC
systems [101]. These protocols can analyze data and select
error correction codes suitable for channel characteristics.
Additionally, deep learning algorithms can optimize data
flow and perform data compression operations in power
systems [101],[102].
Deep learning-based communication protocols in PLC sys-
tems can perform higher error correction than traditional
protocols [102]. Moreover, due to their flexible and adapt-
able nature, they can better adapt to changing conditions
in PLC systems. However, deep learning-based commu-
nication protocols are more complex and require higher
computational requirements than traditional protocols [103].
Furthermore, these protocols require more data for train-
ing and have longer training processes. Therefore, selecting
the most suitable communication protocol for PLC sys-
tems is done by considering the system’s characteristics and
requirements [104].
2) SUPPORT VECTOR MACHINES-BASED
COMMUNICATION PROTOCOLS
Support Vector Machines (SVM) based communication pro-
tocols are among the most effective methods for error
correction and signal enhancement in PLC systems. SVM
is a machine learning method used for data classification or
regression analysis. SVM is highly successful in classifying
data linearly or non-linearly, and it is also robust against
noise [77],[78].
In PLC systems, SVM-based communication protocols
are used for data categorization and selection of error
correction codes [89],[90]. Here, SVM analyzes the statis-
tical characteristics of data and determines the appropriate
error correction codes. Additionally, SVM-based protocols
improve signal quality and reduce noise [105].
Compared to other machine learning-based communica-
tion protocols, SVM-based communication protocols require
less computational power and less data [106]. Thus, they
are preferred in scenarios where computational resources are
limited and a smaller amount of data is available.
3) RANDOM FORESTS-BASED COMMUNICATION
PROTOCOLS
Random Forests-based communication protocols are known
for their significant efficacy in guaranteeing precise data
transmission and proficient error detection and correction
within PLC systems [107]. By leveraging the statistical
attributes of the data, Random Forests excel at error correc-
tion and establish diverse decision trees for error detection
purposes. Moreover, they can autonomously identify the dis-
tinctive features of the data [95],[96].
This approach performs better than conventional methods
frequently used in PLC systems regarding error correction.
However, it’s crucial to remember that this protocol requires
more processing power because of its computing needs and
necessitates a more extensive set of data, which lengthens the
training process compared to other protocols.
VOLUME 11, 2023 124769

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
III. EVALUATION OF MACHINE LEARNING-BASED
METHODS IN PLC SYSTEMS
Machine learning-based error correction codes and
communication protocols offer superior efficacy and
efficiency in PLC systems compared to conventional
approaches [108],[109]. Nevertheless, several factors must
be considered when employing and evaluating these tech-
niques. Primarily, the accurate implementation of machine
learning algorithms necessitates ample and representative
datasets [78]. These datasets should encompass various noise
levels, signal distortions, and other variables encountered in
PLC systems. Additionally, machine learning-based methods
rely on the system’s processing power and memory capacity.
The training and implementation of these methods may
require substantial computational resources. Consequently,
carefully evaluating processor capacity and system require-
ments is imperative when selecting the appropriate method.
Machine learning-based error correction codes and com-
munication protocols in PLC systems tend to be more
intricate and place greater demands on processor power than
traditional approaches [110],[111]. However, owing to their
ability to deliver enhanced outcomes and higher data accu-
racy, they are expected to play a pivotal role in numerous
applications within PLC systems.
A. SIMULATION EXPERIMENTS
A simulation experiment uses a dataset obtained from a
PLC system, encompassing data collected under various
speeds, noise levels, and signal distortion conditions [112],
[113],[114]. This dataset can be a training set for diverse
machine learning-based error correction methods. Once the
training set is established, the performance of machine
learning-based error correction methods is assessed using test
data. These test data incorporate measurements from different
speeds, noise levels, and signal distortion conditions. They
evaluate error correction methods’ accuracy, precision, and
performance in specific scenarios.
Simulation experiments can also be carried out to eval-
uate different machine learning-based communication pro-
tocols [115],[116]. In these experiments, datasets and test
data can be gathered explicitly for a given communica-
tion protocol, enabling the assessment of its performance.
Essential features such as transmission rate, latency, and pro-
tocol reliability can be measured during these experiments.
The outcomes of simulation experiments provide insights
into the efficacy of machine learning-based error correction
methods and communication protocols in real-world applica-
tions [117],[118]. Moreover, these experiments facilitate the
comprehension of parameter effects and the optimization of
parameters to achieve optimal results.
1) SIMULATION EXPERIMENTS FOR ERROR
CORRECTION CODES
Simulation experiments are conducted to assess the effective-
ness of machine learning (ML) based error correction codes
in PLC systems. These experiments encompass the following
stages:
Data collection: Data is acquired using systems such as
Advanced Metering Infrastructure (AMI) or Electric Power
Communication Networks (EPIC) [119].
Data processing: The collected data is processed into a
format suitable for error correction algorithms. This involves
preprocessing steps to eliminate noise, extract relevant fea-
tures, and perform feature selection [120].
Algorithm training: ML-based error correction algorithms
are trained using the collected data. The trained model ana-
lyzes the statistical properties of the data to identify and
rectify errors [121].
Performance evaluation: The trained model is tested using
real-time data to evaluate its performance. Performance met-
rics, including accuracy, precision, specificity, false positive
rate, and false negative rate, are employed to assess the
model’s effectiveness [122].
The conducted research reveals that the results of simula-
tion experiments demonstrate that ML-based error correction
algorithms outperform traditional approaches. Additionally,
it is possible to determine the most suitable error correc-
tion algorithm by comparing the performance of various
algorithms.
2) SIMULATION EXPERIMENTS FOR COMMUNICATION
PROTOCOLS
Simulation experiments are conducted to evaluate the per-
formance of communication protocols in PLC systems. The
following simulation experiments are carried out:
Channel Characterization: Communication channels in
PLC systems often operate in noisy environments with poten-
tial signal distortions. Therefore, channel characterization is
crucial. Through simulation experiments, the characteristics
of the communication channel, such as noise level and signal
loss, are determined to assess the performance of communi-
cation protocols [123].
Bit Error Rate (BER) Analysis: Errors may occur during
data transmission in PLC systems, and the rate of these
errors is expressed as the Bit Error Rate (BER). Simulation
experiments enable BER analysis for different communica-
tion protocols. These analyses evaluate the protocols’ error
tolerance and data accuracy [124].
Data Flow Rate Analysis: The data flow rate is a critical cri-
terion in PLC systems. Simulation experiments are conducted
to analyze the data flow rate for different communication
protocols, measuring their data transmission speed and per-
formance [125].
Reliability Analysis: Simulation experiments employ-
ing various communication protocols are conducted to
test data reliability in PLC systems. These analyses
assess the protocols’ data accuracy, reliability, and error
tolerance [126].
Energy Efficiency Analysis: Energy efficiency is a signifi-
cant criterion for PLC systems operating through the power
grid. Simulation experiments are used to analyze the energy
124770 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
efficiency of different communication protocols, determining
their energy consumption and performance [127],[128].
Based on the results of simulation experiments, a compar-
ison can be made among different communication protocols,
leading to the selection of the most suitable protocol. Addi-
tionally, simulation experiments are utilized to enhance the
performance of protocols.
B. REAL-WORLD EXPERI MENTS
Real-world experiments evaluate the efficacy of machine
learning ML-based communication protocols and error cor-
rection codes in PLC systems [129]. These experiments are
typically carried out within the context of energy distribution
systems [130],[131].
Real-world experiments are commonly implemented as
field tests or pilot applications [130],[131]. Multiple PLC
systems are employed to assess the performance of ML-based
communication protocols and error correction codes under
real-world operational conditions [76],[132].
During these experiments, PLC systems are deployed in
actual energy distribution networks, and the real-time data
transmission is meticulously monitored. The experiments
entail observing diverse factors, including noise levels, sig-
nal distortions, and other network-related parameters. The
collected data is subsequently analyzed to evaluate the per-
formance of ML-based communication protocols and error
correction codes [133].
Despite the higher costs and more prolonged duration asso-
ciated with real-world experiments than simulation experi-
ments, the data obtained under real-world conditions tend
to offer greater representativeness than simulation results.
To accurately evaluate the effectiveness of ML-based com-
munication protocols and error correction codes, real-world
experiments are a crucial tool.
1) REAL-WORLD EXPERIMENTS FOR ERROR
CORRECTION CODES
Real-world experiments are typically conducted in industrial
settings to evaluate the performance of PLC systems under
realistic conditions [119]. These experiments involve an ini-
tial assessment of the system’s current state and identifying
performance improvement areas. Subsequently, the system’s
performance is assessed by employing different error correc-
tion codes [120],[121].
Real-world experiments are often conducted as field tests
to measure the system’s performance using various error
correction codes and assess how PLC systems operate in
real-world scenarios [122].
Although real-world experiments are more intricate and
costly than simulation experiments, they allow one to con-
sider factors that cannot be observed in simulations. Real-
world experiments are also instrumental in evaluating the
real-world performance of error correction codes.
C. PERFORMANC E EVALUATION AND COM PARISON
In the realm of ML-based error correction codes and commu-
nication protocols, the research underscores the paramount
significance of performance evaluation and comparison as
pivotal criteria for ascertaining these technologies’ efficacy
and practical utility [134].
Performance evaluation entails the systematic conduction
of tests aimed at gauging the proficiency of an algorithm
against predefined performance metrics [135]. These met-
rics encompass a wide array of parameters, including but
not limited to accuracy, precision, specificity, F1 score,
mean absolute error, and mean squared error, among oth-
ers. It is imperative to acknowledge that the performance
of an algorithm constitutes a variable parameter influenced
by a multitude of factors, such as the nature of the datasets
employed for testing, the chosen performance metrics, and
other pertinent variables.
Conversely, comparison is a pivotal criterion employed to
juxtapose and assess the performance of diverse algorithms.
These comparative analyses entail the utilization of identical
performance metrics in tandem with examining algorithms
using the same datasets. This meticulous approach facil-
itates the discernment of superior or inferior algorithmic
performance under specific contextual circumstances. Ulti-
mately, performance evaluation and comparison, as applied
to ML-based error correction codes and communication pro-
tocols, provide critical insights into how these technologies
will likely perform in authentic, real-world applications.
The significance of performance evaluation and compari-
son is undeniable in the ML-based Error Correction Codes
and Communication Protocols for Power Line Communi-
cation (PLC). However, several complex challenges and
multifaceted considerations come into play, shaping the land-
scape of this research field.
Diverse Data Sets: The diversity of data sets employed
for testing exerts a profound influence on the outcomes of
performance evaluations. Researchers face the crucial task
of meticulously curating representative data sets encompass-
ing many real-world scenarios. This diversity is essential to
ensure the robustness and generalizability of ML algorithms
across various operational contexts.
Performance Metrics Selection: The selection of appro-
priate performance metrics demands thoughtful deliberation.
Different metrics may hold varying degrees of relevance for
specific applications within the PLC system. Researchers
must carefully align their choice of metrics with the pre-
cise objectives and requirements of the PLC system under
scrutiny.
Real-World Variability: Real-world PLC environments are
characterized by their dynamic and unpredictable nature.
Variations in noise levels, interference, network congestion,
and other environmental factors pose significant challenges.
Thorough evaluations of algorithm performance under such
authentic conditions are indispensable to gauge practical
applicability accurately.
Scalability: In the context of large-scale PLC systems, the
scalability of ML-based error correction codes and communi-
cation protocols assumes paramount importance. Researchers
must explore how these technologies fare as the scale of the
VOLUME 11, 2023 124771

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
network expands, ensuring that their functionality remains
effective and efficient.
Computational Resources: The computational resource
requirements of ML-based algorithms warrant careful con-
sideration, as they can significantly impact the feasibility of
deployment. Developing algorithms that operate optimally
within specified resource constraints is a coveted objective.
Generalization: Ensuring the robust generalization of
algorithms to diverse PLC infrastructures and deployment
scenarios is of utmost importance. Overfitting to specific data
sets should be vigilantly avoided to guarantee consistent and
reliable performance across diverse operational contexts.
Security: Security concerns come to the forefront with the
increasing integration of ML-based algorithms into PLC sys-
tems. Comprehensive evaluations should encompass facets
of algorithmic security, including vulnerability assessments
pertaining to potential adversarial attacks.
The comprehensive assessment of ML-based Error Cor-
rection Codes and Communication Protocols for Power Line
Communication demands a nuanced, multifaceted approach
to performance evaluation and comparison. Researchers must
navigate the intricate landscape defined by diverse data sets,
metric selection, real-world variability, scalability considera-
tions, computational efficiency, generalization requirements,
and security concerns. Addressing these formidable chal-
lenges is instrumental in advancing and refining ML-based
solutions within the realm of PLC systems.
IV. RELATED WORK ON ML BASED ERROR CORRECTION
CODES AND COMMUNICATION PROTOCOLS
Machine learning-based error correction codes and commu-
nication protocols have become one of the most popular
research topics in recent years [132]. There are numer-
ous studies available in the literature on this subject. Deep
Learning-Based Error Correction for Power-Line Commu-
nications’’: This paper discusses the use of deep learning
methods for error correction coding in PLC systems. The
study combines coding methods used in PLC systems and
deep learning methods to achieve higher accuracy rates [136].
Several important applications [137],[138] have been
developed for accelerating the reinforcement learning-based
channel estimation process in OFDM systems and for analy-
sis, such as machine learning methods, deep learning, robust
meta-learning, performance analysis, low complexity, and
estimation methods [139],[140].
In studies conducted on Machine Learning-Aided Sig-
nal Processing for Power Communications, machine-learning
methods have been used to achieve higher performance in sig-
nal processing operations such as channel estimation, signal
detection, and code decoding [7],[141].
Powerline communication has transcended its conven-
tional applications in the smart grid (SG) and found extensive
use in various domains [142],[142]. Originally employed for
two-way SG communication, advanced metering infrastruc-
ture (AMI) applications, demand response, and power system
control, PLC now serves diverse purposes. These include
broadband Internet applications, consumer home automation,
enabling grid-wide artificial intelligence (AI) applications,
and utilizing PLC modems as sensors for monitoring grid
health, among others. ML-based studies conducted in this
field have investigated the impact oncommunication systems
in narrow and wide-band applications [143],[144]. More-
over, the effectiveness of PLC network devices, commonly
employed in these studies, has been enhanced through the
deterministic data optimization approach [145]. Addition-
ally, the suitability of PLC impedance estimation has been
discussed [146],[147]. The performance of estimators was
thoroughly analyzed and evaluated through numerical assess-
ments [148].
In the challenging power line environment characterized
by significant impulsive noise (IN), it is possible to achieve
more precise channel estimation compared to conventional
strategies. Instead of focusing on suppressing the IN-affected
samples in the received signal, a more effective approach
involves estimating the IN and integrating it into the chan-
nel estimation process [18]. This can be accomplished by
utilizing signal processing schemes proposed in [19],[20],
and [21] to estimate the sparse IN components present in the
received signal. A likelihood function can be formulated to
leverage the estimated IN and observations of the received
signal. As a result, the channel coefficients can be determined
by estimating parameter values that maximize the likelihood
function [148].
A. A MACHINE LEARNING-BASED CHANNE L ESTIMATION
METHOD FOR VEHICULAR COMM UNICATION SYSTEMS
This study deals with the use of machine learning methods in
channel estimation in-vehicle communication systems. The
study proposed a deep learning-based channel estimation
method, and higher accuracy rates were obtained according
to the simulation results compared to other methods. In addi-
tion, studies and applications are carried out for machine
learning-based error correction codes and communication
protocols in wireless networks, image and video compres-
sion, radar, and similar fields. Machine learning-based error
correction codes and communication protocols are widely
used in both academic and industrial applications [149].
Some examples are:
Intelligent energy management: Smart grids use machine
learning-based communication protocols to collect, analyze,
and manage data in electricity transmission and distribution
systems [150].
Automotive industry: In the automotive industry, machine
learning-based error correction codes are used to improve the
accuracy of sensor data in vehicles [151].
Telecommunications: In the telecommunications industry,
machine learning-based error correction codes are used to
correct errors in both communication and PLC lines [152].
Finance: In the financial industry, machine learning-based
communication protocols can help banks interact smarter
with customers and increase the accuracy of financial
data [153].
124772 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
Medicine: In the medical industry, machine learning-based
error correction codes are used to increase the accuracy and
reliability of data in hospital systems. This area is one of the
most important application areas [154].
Agriculture: In the agricultural industry, machine learning-
based communication protocols are used to increase the
accuracy of soil data and agriculture sensor data [155].
In the field of Machine Learning-based PLC systems, the
utilization of the end-to-end learning process is prevalent.
This process facilitates the autoencoder in discovering a
robust representation of the input signal, thereby enabling
the identification of optimal encoding and decoding strate-
gies for stochastic channels. Notably, the autoencoder can
achieve a solution that surpasses the performance of existing
modulation and encoding methods [156],[157]. More-
over, the autoencoder approach operates without making
any assumptions about the channel and, in theory, can
comprehend the dynamics of the PLC channel without
relying on its current manifestation as a linear periodic
time-variable system [66],[156]. To implement this con-
cept, an autoencoder has been developed using TensorFlow
libraries [66],[158] specifically for integration into PLC
systems. A schematic depiction of this implementation can
be found in Figure 13.
FIGURE 13. Semantic description of a PLC system via an autoencoder
model [66].
V. FUTURE DIRECTIONS
Machine learning-based error correction codes and com-
munication protocols are widely used in electric power
transmission systems, which are of paramount importance
in all stages of electrical power transmission, distribution,
and control. These technologies enhance energy efficiency
by providing reliable and high-performance communication,
facilitating system fault detection, and enabling error correc-
tion capabilities.
Current studies demonstrate that machine learning-based
error correction codes offer high accuracy, sensitivity, and
performance levels for PLC systems. Similarly, machine
learning-based communication protocols represent a signifi-
cant advancement in terms of efficiency, speed, and reliability
within PLC systems.
However, challenges persist in this field. A better under-
standing of factors that may impact the performance of
machine learning-based error correction codes and com-
munication protocols in real-world applications is needed.
Concerns about installation and management costs and secu-
rity and data privacy must also be addressed.
Research in this area is expected to focus on integrat-
ing advanced technologies, such as artificial intelligence
and deep learning techniques. Furthermore, developing new
methods by exploring additional application scenarios will
likely render these technologies more suitable for industrial
use.
It is evident that machine learning-driven error correction
codes and communication protocols hold substantial promise
in substantially augmenting the performance of PLC systems.
Furthermore, their adoption is anticipated to experience fur-
ther proliferation in the foreseeable future.
A. POTENTIAL ADVANTAGES AND LIMITATIONS
Machine learning-based error correction codes and commu-
nication protocols are a major advance for PLC systems
and other engineering applications. Potential advantages
include [66],[70]:
Higher efficiency: Machine learning algorithms can
quickly process data to generate error-correcting codes and
communication protocols. This means higher efficiency and
less downtime [159],[160].
Lower cost: Traditional error correction methods and com-
munication protocols rely on manpower and are implemented
manually. Machine learning-based approaches can reduce
these costs and provide a more economical solution.
Higher accuracy: Machine learning algorithms can gener-
ate more accurate error correction codes and communication
protocols. This means fewer bugs and better system perfor-
mance [159],[160]. However, these approaches also have
some limitations:
Data constraints: Machine learning algorithms need suf-
ficient and representative data to achieve high accuracy.
Sometimes, there may not be enough data, or the quality may
be poor.
Model complexity: The complexity of machine learn-
ing models can slow down the learning process or reduce
efficiency [70]. Also, very complex models may become
impractical due to hardware constraints in their application
area [159],[160].
Reliability issues: Machine learning models can be vulner-
able to false data or attacks. Therefore, additional measures
may be required to increase reliability.
Situations requiring human intervention: In some cases,
the results of machine learning models may need human
validation or intervention [70],[161].
Machine learning-based error correction codes and com-
munication protocols may become more common in indus-
trial applications. Advances in data collection technologies
can enable the collection of more and better quality data.
In addition, stronger hardware and software infrastructure can
enable more complex machine learning models to be used.
Although machine learning-based error correction codes
and communication protocols have significant potential to
improve the performance of communication systems, they
VOLUME 11, 2023 124773

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
also come with some limitations. These limitations may
include the need for high computational power, insuffi-
cient data sets, misleading data affecting the accuracy of
algorithms, overfitting problems, model explainability, intel-
ligibility, scalability, and security issues.
However, advanced algorithms and technologies may solve
many of these limitations. For example, algorithms can run
faster by using hardware with higher computing power.
In addition, dataset creation and editing processes can be
made more efficient and effective [14],[15]. More advanced
algorithms and techniques can be developed to detect and
process misleading data. Overfitting problems can be reduced
by developing more complex models and improving model
selection techniques [31],[66]. Model explainability and
intelligibility problems can be addressed by making algo-
rithms more transparent or by using interpretable models.
Scalability issues can be reduced by developing algorithms
that work with larger data sets and using more efficient hard-
ware. Security issues can be resolved with more advanced
encryption and authentication techniques [66],[70].
Future directions in this area could be the development
of more complex and intelligent systems, using more data
sets, using more advanced scalability and security techniques,
and exploring more industrial applications. In addition,
researchers and engineers working in this field should con-
tinue to work on the development of more effective, efficient,
and optimized algorithms for machine learning-based error
correction codes and communication protocols.
B. SUGGESTIONS FOR FUTURE WORK
Several recommendations for prospective research directions
within the realm of machine learning-based error correction
codes and communication protocols are as follows:
Data collection: Extensive data collection and sampling
efforts are imperative to assemble larger, more diversified
datasets. This augmentation of data sources can substantially
enhance the accuracy and efficiency of machine learning
algorithms.
Modeling: Developing more intricate and advanced
machine learning models is crucial for refining error cor-
rection and communication protocols. These models can be
meticulously tailored and optimized to yield superior accu-
racy and efficiency in their outcomes.
Enhanced Classification: Employing superior feature
selection techniques and advanced data preprocessing meth-
ods is essential for elevating the classification accuracy of
machine learning algorithms.
Implementation: Efforts should be directed towards ren-
dering machine learning-based error correction codes and
communication protocols more universally applicable. This
expansion will broaden their utility across a spectrum of
industrial and commercial applications.
Comparative: Comparative evaluations should be under-
taken to discern the performance disparities among diverse
machine learning-based error correction codes and com-
munication protocols. This comparative analysis can help
identify the technologies that offer optimal performance
characteristics.
Security: Further research is required to fortify the security
attributes of machine learning-based error correction codes
and communication protocols, focusing on mitigating vulner-
abilities.
Optimal Design: In-depth research endeavors are neces-
sary to ascertain the optimal design parameters of machine
learning-based error correction codes and communication
protocols, ensuring their maximal effectiveness.
Exploration of Novel Approaches: The exploration of
innovative approaches and techniques within the domain
of machine learning-based error correction codes and com-
munication protocols should be vigorously pursued. This
exploration presents opportunities to surmount the limitations
inherent in existing methodologies and to deliver superior
performance outcomes
VI. DISCUSSION
The originality of this article can be summarized under two
main headings. Firstly, it represents an innovative applica-
tion as a PLC-based machine learning error correction code
algorithm, proposing a novel algorithm. Secondly, it provides
a comprehensive literature overview and aligns with previous
research endeavors.
Unsal and Yalcinoz [164] developed a novel model based
on power line communication (PLC) in their study. Addition-
ally, Huang et al. [165] employed the Q-learning algorithm
to search for the optimal attack sequence against lines in
a dependent power-communication network, leveraging a
model to simulate stepwise faults that occur in the com-
munication network. The system proposed by Hashim and
Al-Mashhadani [166], engages with the IoT system and
directly visualizes real-time data. Efforts have been made
here to facilitate the distribution of the measurement network
over the cloud environment. Practical results indicate that
packet losses in the received data are approximately 0, 1, or
2 characters, and the time difference between the transmitter
and receiver is approximately 5000 milliseconds. In the paper
authored by Omaer et al. [167], a machine learning-based
autonomous fault detection and fault classification system
is proposed, where extreme machine learning is employed
for fault detection and classification. A study conducted by
Ramesh et al. [168] explored the outage probability perfor-
mance of hybrid protocols in power line communication.
In the study conducted by Oliver et al., a deep learn-
ing approach is proposed for detecting trees encroaching on
power and communication lines using street-level images.
This approach further involves performing rapid quantita-
tive and qualitative analyses based on the Grad-CAM++
method [169]. In the research conducted by Topaloglu [170],
a novel approach based on Convolutional Neural Net-
works (CNN) has been developed to classify a specific power
signal according to its relevant power quality condition. Uti-
lizing the Attention Model approach, accuracy and error
values for the Power Quality (PQ) in the electrical power
124774 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
system were obtained, dependent on the absence or scarcity
of direct power disturbances.
Other studies in the literature, as listed in Table 2, pre-
dominantly fall into either classical classification methods or
fault detection methods or are based on Artificial Neural Net-
works (ANN) for fault identification and error code purposes.
Across these studies, Machine Learning (ML) is commonly
employed for data analysis in PLC systems. In this regard, the
proposed method within our research framework effectively
addresses a significant gap within the domain of Machine
Learning-Based Error Correction Codes and Communication
Protocols for Power Line Communication.
Some suggestions for future work in the field of machine
learning-based error correction codes and communication
protocols can be:
Data collection: More data collection and sampling should
be done to create larger and more diverse datasets. This will
help machine learning algorithms become more accurate and
efficient.
Modeling: More sophisticated machine learning models
should be developed for better error correction and communi-
cation protocols. These models can be optimized to produce
more accurate and efficient results.
Classification: Better feature selection and data prepro-
cessing techniques should be used to increase the classifica-
tion accuracy of machine learning algorithms.
Implementation: Machine learning-based error correction
codes and communication protocols should be made more
widely applicable. This will provide a wider range of use in
industrial and commercial applications.
Comparison: Comparisons should be made between dif-
ferent machine learning-based error correction codes and
communication protocols. This will help identify technolo-
gies that provide the best performance.
Security: Machine learning-based error-correcting codes
and communication protocols need further research to further
protect against vulnerabilities.
Optimal design: Machine learning-based error-correcting
codes and communication protocols require further research
to determine their optimal design.
New approaches: New approaches and techniques should
be explored in the field of machine learning-based error cor-
rection codes and communication protocols. This will present
opportunities to overcome the limitations of existing methods
and provide better performance.
This study investigated communication protocols and error
correction codes in Machine Learning PLC [66]. ML tech-
niques can be used in areas such as studies in this field,
energy efficiency optimization, error detection and predic-
tion, and power quality analysis [43],[70]. ML algorithms
can optimize the data transmission of the PLC system,
increase energy efficiency by monitoring and analyzing
energy consumption, predict line faults, and improve power
quality.
ML-based PLC studies are classified in Table 2below with
their references.
TABLE 2. Existing survey and literature review articles with the main
focus highlighted and compared to this paper.
From a practical application perspective, when discussing
potential deployment scenarios of Machine Learning for
Error Correction Codes and Communication Protocols in
Power Line Communication, several approaches can be out-
lined as follows:
Data-Driven Error Correction Codes: Machine learning
can be leveraged to enhance the error correction codes of
PLC systems. More effective codes can be devised using
extensive historical data to address communication errors and
corrections.
Adaptive Communication Protocols: Machine learning can
be employed to create adaptive communication protocols that
continuously monitor and assess network conditions. These
protocols can dynamically adapt to changing noise levels,
traffic patterns, and other factors within the network.
Cybersecurity and Anomaly Detection: Machine learn-
ing can be crucial in bolstering the cybersecurity of power
line communication networks. Techniques such as anomaly
detection and malicious software identification can be applied
to enhance network security.
Spectrum Efficient Utilization: Machine learning can con-
tribute to efficiently utilizing the spectrum. It can assist in
selecting the optimal frequencies and channels for data trans-
mission in the RF spectrum.
Energy Efficiency: Machine learning can contribute to the
development of energy-efficient communication protocols,
reducing the energy consumption of PLC systems.
These scenarios represent practical approaches that can be
employed to evaluate the impact of machine learning on Error
Correction Codes and Communication Protocols in Power
Line Communication. However, it is essential to note that
each scenario has unique challenges and application domains,
and their feasibility should be carefully examined.
VOLUME 11, 2023 124775

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
VII. CONCLUSION
This paper examines the utilization and effectiveness of
machine learning-based methods for error correction codes
and communication protocols in PLC systems. The focus
is on popular machine learning techniques such as deep
learning, support vector machines, and random forests.
Results obtained from simulations and real-world experi-
ments demonstrate that machine learning-based methods can
significantly improve the performance of PLC systems, espe-
cially in challenging communication environments and noisy
channels.
Furthermore, it has emphasized the pivotal role played by
machine learning-based error correction codes and commu-
nication protocols in enhancing the reliability and perfor-
mance of PLC systems. By examining the existing academic
research and industrial applications, it has been observed that
research and applications in this field are rapidly advancing.
Notably, machine learning techniques have been found to
play a critical role in augmenting both the reliability and
effectiveness of PLC systems.
The discussion and future directions section assesses
the potential advantages and challenges of machine
learning-based error correction codes and communication
protocols. Recommendations for future research and poten-
tial application areas in this field are also provided.
Additionally, within the scope of this investigation, a feed-
forward neural network comprising 10 hidden neurons was
constructed and subjected to performance analysis in syn-
thetic data testing. Remarkably, the optimal validation per-
formance ascertained at epoch 4, attained a value of 0.55432.
Furthermore, the Error-Target outputs for the PLC model
were derived to discern variations across iterations for
training, validation, testing, and zero errors, yielding the fol-
lowing values: 0.99687 for Training, 0.99115 for verification,
0.99571 for Testing, and 0.99575 for the aggregate dataset.
In conclusion, this article comprehensively reviews the uti-
lization and effectiveness of machine learning-based methods
for error correction codes and communication protocols in
PLC systems. The findings highlight the potential of machine
learning-based approaches to improve the performance and
reliability of PLC systems. Consequently, further research
and application in this domain will contribute to creating
more efficient and reliable energy management, monitoring,
and automation systems in the future.
REFERENCES
[1] J. P. A. Yaacoub, J. H. Fernandez, H. N. Noura, and A. Chehab, ‘‘Security
of power line communication systems: Issues, limitations and existing
solutions,’’ Comput. Sci. Rev., vol. 39, Feb. 2021, Art. no. 100331, doi:
10.1016/j.cosrev.2020.100331.
[2] P. Mlýnek, M. Rusz, L. Benešl, J. Sláčik, and P. Musil, ‘‘Possibilities
of broadband power line communications for smart home and smart
building applications,’’ Sensors, vol. 21, no. 1, p. 240, Jan. 2021, doi:
10.3390/s21010240.
[3] S. S. B. Sameon, S. Yussof, and B. N. Jørgensen, ‘‘Comparison between
communication technology used in smart building,’’ in Proc. 8th Int.
Conf. Inf. Technol. Multimedia (ICIMU), Selangor, Malaysia, Aug. 2020,
pp. 212–217, doi: 10.1109/ICIMU49871.2020.9243447.
[4] S. Kirmani, A. Mazid, I. A. Khan, and M. Abid, ‘‘A survey on
IoT-enabled smart grids: Technologies, architectures, applications, and
challenges,’’ Sustainability, vol. 15, no. 1, p. 717, Dec. 2022, doi:
10.3390/su15010717.
[5] S. H. Mahmood, A. M. Salih, and M. I. Khalil, ‘‘Broadband services
on power line communication systems: A review,’’ in Proc. 22nd Int.
Conf. Control Syst. Comput. Sci. (CSCS), Bucharest, Romania, May 2019,
pp. 465–470, doi: 10.1109/CSCS.2019.00085.
[6] T. Zhang and W. Liu, ‘‘FFT-based OFDM in broadband-PLC and
narrowband-PLC,’’ in Proc. Int. Conf. Cyber-Enabled Distrib. Com-
put. Knowl. Discovery, Sanya, China, Oct. 2012, pp. 473–478, doi:
10.1109/CyberC.2012.86.
[7] G. Prasad and L. Lampe, ‘‘Full-duplex power line communica-
tions: Design and applications from multimedia to smart grid,’’
IEEE Commun. Mag., vol. 58, no. 2, pp. 106–112, Feb. 2020, doi:
10.1109/MCOM.001.1900519.
[8] E. C. Uwaezuoke and T. G. Swart, ‘‘Network attack analysis of an indoor
power line communication network,’’ in Proc. IEEE Int. Symp. Power
Line Commun. Appl. (ISPLC), Aachen, Germany, Oct. 2021, pp. 96–101,
doi: 10.1109/ISPLC52837.2021.9628606.
[9] M. Singh, A. Amruthakala, M. L. Sudheer, and A. S. Murthy, ‘‘Analysis
of noise in broadband powerline communications (B-PLC) in frequency
range of 150 kHz–30 MHz,’’ in Proc. 4th IEEE Uttar Pradesh Sect. Int.
Conf. Electr., Comput. Electron. (UPCON), Mathura, India, Oct. 2017,
pp. 101–106, doi: 10.1109/UPCON.2017.8251030.
[10] S. T. Y. Alfalahi, A. A. Alkahtani, A. Q. Al-Shetwi, A. S. Al-Ogaili,
A. A. Abbood, M. B. Mansor, and Y. Fazea, ‘‘Supraharmonics
in power grid: Identification, standards, and measurement
techniques,’’ IEEE Access, vol. 9, pp. 103677–103690, 2021, doi:
10.1109/ACCESS.2021.3099013.
[11] D. Joshi, D. Deb, and S. M. Muyeen, ‘‘Comprehensive review on electric
propulsion system of unmanned aerial vehicles,’’ Frontiers Energy Res.,
vol. 10, May 2022, Art. no. 752012, doi: 10.3389/fenrg.2022.752012.
[12] K. Sayed, A. Almutairi, N. Albagami, O. Alrumayh, A. G. Abo-Khalil,
and H. Saleeb, ‘‘A review of DC–AC converters for electric vehi-
cle applications,’’ Energies, vol. 15, no. 3, p. 1241, Feb. 2022, doi:
10.3390/en15031241.
[13] R. Sriharan, S. Dharaganthu, P. A. Francis, and G. Janakiraman,
‘‘A review on power line communication systems,’’ Int. J. Eng. Res.
Technol., vol. 6, pp. 1–5, Apr. 2018, doi: 10.17577/ijertcon062.
[14] L. Lampe and L. T. Berger, ‘‘Power line communications,’’ in Aca-
demic Press Library in Mobile and Wireless Communications. Jan. 2016,
pp. 621–659, doi: 10.1016/b978-0-12-398281-0.00016-8.
[15] P. Rao and B. D. Deebak, ‘‘Role of power line communications in the
smart grid: Applications, challenges, and research initiatives,’’ in Sustain-
able Networks in Smart Grid. New York, NY, USA: Academic, Jan. 2022,
pp. 73–98, doi: 10.1016/b978-0-323-85626-3.00001-6.
[16] H. Habibzadeh, T. Soyata, B. Kantarci, A. Boukerche, and C. Kaptan,
‘‘Sensing, communication and security planes: A new challenge for
a smart city system design,’’ Comput. Netw., vol. 144, pp. 163–200,
Oct. 2018, doi: 10.1016/j.comnet.2018.08.001.
[17] G. van de Kaa, T. Fens, J. Rezaei, D. Kaynak, Z. Hatun, and
A. Tsilimeni-Archangelidi, ‘‘Realizing smart meter connectivity:
Analyzing the competing technologies power line communication,
mobile telephony, and radio frequency using the best worst method,’’
Renew. Sustain. Energy Rev., vol. 103, pp. 320–327, Apr. 2019, doi:
10.1016/j.rser.2018.12.035.
[18] S. Vappangi and V. V. Mani, ‘‘A survey on the integration of visi-
ble light communication with power line communication: Conception,
applications and research challenges,’’ Optik, vol. 266, Sep. 2022,
Art. no. 169582, doi: 10.1016/j.ijleo.2022.169582.
[19] A. S. Menezes, Y. F. Coutinho, and M. V. Ribeiro, ‘‘Hybrid cooperative
spectrum sensing for improving cognitive power line communication
systems,’’ Comput. Electr. Eng., vol. 103, Oct. 2022, Art. no. 108286, doi:
10.1016/j.compeleceng.2022.108286.
[20] X. Deng, L. Wang, J. Gui, P. Jiang, X. Chen, F. Zeng, and S. Wan,
‘‘A review of 6G autonomous intelligent transportation systems: Mecha-
nisms, applications and challenges,’’ J. Syst. Archit., vol. 142, Sep. 2023,
Art. no. 102929, doi: 10.1016/j.sysarc.2023.102929.
[21] K. A. Abdulsalam, J. Adebisi, M. Emezirinwune, and O. Babatunde,
‘‘An overview and multicriteria analysis of communication technologies
for smart grid applications,’’ e-Prime, Adv. Electr. Eng., Electron. Energy,
vol. 3, Mar. 2023, Art. no. 100121, doi: 10.1016/j.prime.2023.100121.
124776 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
[22] R. Kadel, K. Paudel, D. B. Guruge, and S. J. Halder, ‘‘Opportunities
and challenges for error control schemes for wireless sensor networks: A
review,’’ Electronics, vol. 9, no. 3, p. 504, Mar. 2020, doi: 10.3390/elec-
tronics9030504.
[23] HardwareBee. (Oct. 3, 2021). Understanding Power Line Communi-
cation. Accessed: Jun. 2023. [Online]. Available: https://hardwarebee.
com/understanding-power-line-communication/
[24] Electronic Projects for Engineering Students. (May 17, 2020). Power Line
Carrier Communication: Circuit Diagram and Its Working. Accessed:
Jun. 2023. [Online]. Available: https://www.elprocus.com/what-is-
power-line-carrier-communication-working-its-applications/
[25] X. Yang, H. Zhang, Q. Xie, and S. Wang, ‘‘Methods for power line com-
munication carrier equipment identity authentication and wiretapping
equipment locating considering physical layer characteristics,’’ Energy
Rep., vol. 9, pp. 628–637, Apr. 2023, doi: 10.1016/j.egyr.2023.03.011.
[26] Y. Chen, Q. Rao, M. Zhang, J. Xu, P. Zhan, and X. Li, ‘‘Research
on EMI of UHV AC transmission line to power communica-
tion equipment along the line,’’ in Proc. 4th Inf. Commun. Tech-
nol. Conf. (ICTC), Nanjing, China, May 2023, pp. 178–182, doi:
10.1109/ICTC57116.2023.10154686.
[27] C. U. Ndujiuba, S. N. John, and O. Ogunseye, ‘‘Improving data trans-
mission efficiency over power line communication (PLC) system using
OFDM,’’ Int. J. Appl. Eng. Res., vol. 12, no. 5, 2017, Art. no. 705710.
[28] X. Li, Z. Shi, X. Wang, and L. Lv, ‘‘Research on frequency
resource allocation model for power line carrier service of inte-
grated station, equipment and user,’’ in Proc. 7th Asia Conf. Power
Electr. Eng. (ACPEE), Hangzhou, China, Apr. 2022, pp. 803–807, doi:
10.1109/ACPEE53904.2022.9783913.
[29] C. Xu, Z. Xiong, X. Kong, G. Zhao, and S. Yu, ‘‘A packet reception
probability-based reliable routing protocol for 3D VANET,’’ IEEE Wire-
less Commun. Lett., vol. 9, no. 4, pp. 495–498, Apr. 2020.
[30] S. Kasthala and R. Goru, ‘‘A review on power line communi-
cation and its applicability to cable fault monitoring,’’ in Proc.
Int. Conf. Recent Trends Microelectron., Autom., Comput. Commun.
Syst. (ICMACC), Hyderabad, India, Dec. 2022, pp. 212–217, doi:
10.1109/ICMACC54824.2022.10093617.
[31] C. Kotchasarn, ‘‘Performance analysis of broadband power line com-
munications with OFDM transmission,’’ in Proc. 19th Int. Conf. Adv.
Commun. Technol. (ICACT), PyeongChang, South Korea, Feb. 2017,
pp. 316–320, doi: 10.23919/ICACT.2017.7890106.
[32] W. Song, J. Li, D. Duan, J. Li, and H. Liu, ‘‘Optimization of precoders
and rate allocation for rate splitting multiple access over power line
communication channel,’’ Energy Rep., vol. 8, pp. 507–513, Sep. 2022,
doi: 10.1016/j.egyr.2022.03.148.
[33] O. Kolade, A. D. Familua, and L. Cheng, ‘‘Indoor amplify-and-forward
power-line and visible light communication channel model based on a
semi-hidden Markov model,’’ AEU, Int. J. Electron. Commun., vol. 124,
Sep. 2020, Art. no. 153108, doi: 10.1016/j.aeue.2020.153108.
[34] Y. Li, M. Zhang, X. Zhang, G. Li, and J. Lu, ‘‘An adaptive MIMO-OFDM
modulation method for middle voltage power line communication,’’ in
Proc. IEEE 20th Int. Conf. Commun. Technol. (ICCT), Nanning, China,
Oct. 2020, pp. 203–207, doi: 10.1109/ICCT50939.2020.9295905.
[35] S. Schaffenroth, H.-P. Schmidt, and A. Kölpin, ‘‘Mitigation of coloured
impulsive noise in OFDM receiver,’’ in Proc. IEEE Int. Symp. Power Line
Commun. Appl. (ISPLC), Aachen, Germany, Oct. 2021, pp. 43–48, doi:
10.1109/ISPLC52837.2021.9628650.
[36] M. Kashef, M. Abdallah, and N. Al-Dhahir, ‘‘Transmit power optimiza-
tion for a hybrid PLC/VLC/RF communication system,’’ IEEE Trans.
Green Commun. Netw., vol. 2, no. 1, pp. 234–245, Mar. 2018, doi:
10.1109/TGCN.2017.2774104.
[37] Y. Liao, W. Li, K. Wang, J. Guo, Y. Shen, Q. Wang, Y. Zhang, C. Wu,
X. Zhou, T. Guo, and T. W. Kim, ‘‘TENG-inspired LED-in-capacitors
for smart self-powered high-voltage monitoring and high-sensitivity
demodulation of power-line communications,’’ Nano Energy, vol. 102,
Nov. 2022, Art. no. 107698, doi: 10.1016/j.nanoen.2022.107698.
[38] G. Cao, X. Zhang, P. Li, F. Li, Y. Liu, and Z. Wu, ‘‘Load frequency
regulation of multi-area power systems with communication delay via
cascaded improved ADRC,’’ Energy Rep., vol. 9, pp. 983–991, Sep. 2023,
doi: 10.1016/j.egyr.2023.04.379.
[39] E. F. Fuchs and M. A. S. Masoum, ‘‘Impact of poor power quality
on reliability, relaying, and security,’’ in Power Quality in Power Sys-
tems, Electrical Machines, and Power-Electronic Drives. New York,
NY, USA: Academic, Jan. 2023, pp. 805–913, doi: 10.1016/b978-0-12-
817856-0.00009-1.
[40] B. Li, P. Zhang, and Y. Liang, ‘‘Research on power communication
mode under the architecture of mountain microgrid,’’ in Proc. 7th Int.
Conf. Comput. Commun. Syst. (ICCCS), Wuhan, China, Apr. 2022,
pp. 802–806, doi: 10.1109/ICCCS55155.2022.9845901.
[41] F. Fang and Z. Yang, ‘‘Adaptive suppression algorithm of narrow-
band interference for OFDM-based power line communications,’’ in
Proc. 5th Int. Conf. Comput. Sci. Netw. Technol. (ICCSNT), Dec. 2016,
pp. 461–466.
[42] Y. Liang, Z. Shi, X. Wang, and Z. Wu, ‘‘Research on optimization of
modulation and coding mode for power line carrier communication,’’ in
Proc. 4th Int. Conf. Adv. Comput. Technol., Inf. Sci. Commun. (CTISC),
Suzhou, China, Apr. 2022, pp. 1–5, doi: 10.1109/CTISC54888.2022.
9849752.
[43] L. G. de Oliveira, G. R. Colen, M. V. Ribeiro, and A. J. H. Vinck,
‘‘Narrow-band interference error correction in coded OFDM-
based PLC systems,’’ in Proc. Int. Symp. Power Line Commun.
Appl. (ISPLC), Bottrop, Germany, Mar. 2016, pp. 13–18, doi:
10.1109/ISPLC.2016.7476279.
[44] IEEE Standard for low-Frequency (Less Than 500 kHz) Narrowband
Power Line Communications for Smart Grid Applications, IEEE Standard
1901.2, Dec. 2013.
[45] M. Liu, W. Lei, J. Sun, H. Lei, and H. Tang, ‘‘Power and rate control
in wireless communication systems with energy harvesting and rate-
less codes,’’ Phys. Commun., vol. 59, Aug. 2023, Art. no. 102083, doi:
10.1016/j.phycom.2023.102083.
[46] M. A. N. S. Raghavendra and U. S. Acharya, ‘‘Index modulation aided
multi carrier power line communication employing rank codes from
cyclic codes,’’ Phys. Commun., vol. 39, Apr. 2020, Art. no. 100975, doi:
10.1016/j.phycom.2019.100975.
[47] A. Chattopadhyay, K. Sharma, and A. Chandra, ‘‘Error performance of
RS coded binary FSK in PLC channels with Nakagami and impulsive
noise,’’ in Proc. 18th IEEE Int. Symp. Power Line Commun. Appl.,
Glasgow, U.K., Mar. 2014, pp. 184–189, doi: 10.1109/ISPLC.2014.
6812369.
[48] E.-H. Lu, T.-C. Chen, and P.-Y. Lu, ‘‘A new method for
evaluating error magnitudes of Reed–Solomon codes,’’ IEEE
Commun. Lett., vol. 18, no. 2, pp. 340–343, Feb. 2014, doi:
10.1109/LCOMM.2013.122713.132334.
[49] X. Wang, H. Hong, K. Wang, Y. Li, Q. Guo, Y. Zhang, and Q. Yang,
‘‘Performance of LDPC and turbo coded power line communication over
multipath channel and narrowband noise,’’ in Proc. 2nd Asia–Pacific
Conf. Commun. Technol. Comput. Sci. (ACCTCS), Shenyang, China,
Feb. 2022, pp. 263–269, doi: 10.1109/ACCTCS53867.2022.00061.
[50] IEEE Standard for Information Technology—Telecommunications and
Information Exchange Between Systems—Local and Metropolitan Area
Networks—Specific Requirements—Part 11: Wireless LAN Medium
Access Control (MAC) and Physical Layer (PHY) Specifications—
Redline, IEEE Standard 802.11-2020 (Revision of IEEE Standard
802.11-2016), Feb. 2021, pp. 1–7524.
[51] C. Panem, V. Gad, and R. S. Gad, ‘‘Polynomials in error detection and
correction in data communication system,’’ in Coding Theory. London,
U.K.: IntechOpen, Mar. 2020, pp. 29–49, doi: 10.5772/intechopen.86160.
[52] S. Ahmed, D. G. Lee, J. Vong, J. Kurone-Ghariban, C. J. Matchinsky,
T. Kester, M. Cha, S. Dobbs, and Z. Yu, ‘‘Implementation and investiga-
tion of high endurance UAVs in a 4G LTE network for monitoring power
lines,’’ in Proc. Wireless Telecommun. Symp. (WTS), Boston, MA, USA,
Apr. 2023, pp. 1–9, doi: 10.1109/WTS202356685.2023.10131813.
[53] M. Park, G. Jeong, H. Son, and J. Paek, ‘‘Performance of RPL routing
protocol over multihop power line communication network,’’ in Proc.
Int. Conf. Inf. Commun. Technol. Converg. (ICTC), Jeju, South Korea,
Oct. 2020, pp. 1918–1920, doi: 10.1109/ICTC49870.2020.9289175.
[54] A. Dua, V. Jindal, and P. Bedi, ‘‘Covert communication using address res-
olution protocol broadcast request messages,’’ in Proc. 9th Int. Conf. Rel.,
Infocom Technol. Optim. (Trends Future Directions) (ICRITO), Noida,
India, Sep. 2021, pp. 1–6, doi: 10.1109/ICRITO51393.2021.9596480.
[55] N. Wehbe, H. A. Alameddine, M. Pourzandi, E. Bou-Harb, and C. Assi,
‘‘A security assessment of HTTP/2 usage in 5G service-based architec-
ture,’’ IEEE Commun. Mag., vol. 61, no. 1, pp. 48–54, Jan. 2023, doi:
10.1109/MCOM.001.2200183.
[56] M. Selvi, A. Gayathri, S. K. Svn, and A. Kannan, ‘‘Energy effi-
cient and secured MQTT protocol using IoT,’’ Int. J. Innov. Technol.
Exploring Eng., vol. 9, no. 4, pp. 11–14, Feb. 2020, doi: 10.35940/iji-
tee.b6264.029420.
VOLUME 11, 2023 124777

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
[57] G. S. Karthick, ‘‘Energy-aware reliable medium access control proto-
col for energy-efficient and reliable data communication in wireless
sensor networks,’’ Social Netw. Comput. Sci., vol. 4, no. 5, Jun. 2023,
Art. no. 449, doi: 10.1007/s42979-023-01869-z.
[58] A. H. Beshir, S. Negri, X. Wu, X. Liu, L. Wan, G. Spadacini, S. A. Pignari,
and F. Grassi, ‘‘Behavioral model of G3-powerline communication
modems for EMI analysis,’’ Energies, vol. 16, no. 8, p. 3336, Apr. 2023,
doi: 10.3390/en16083336.
[59] M. Brandl and K. Kellner, ‘‘Performance evaluation of power-line com-
munication systems for LIN-bus based data transmission,’’ Electronics,
vol. 10, no. 1, p. 85, Jan. 2021, doi: 10.3390/electronics10010085.
[60] L. Lampe, A. M. Tonello, and T. G. Swart, Power Line Communications:
Principles, Standards and Applications from Multimedia to Smart Grid.
Chichester, U.K.: Wiley, 2016.
[61] H. C. Ferreira, L. Lampe, J. Newbury, and T. G. Swart, Power Line
Communications. Hoboken, NJ, USA: Wiley, 2011.
[62] J. Anatory and N. Theethayi, Broadband Power-Line Communication
Systems: Theory & Applications. Boston, MA, USA: WIT, 2010.
[63] E. Conrad, S. Misenar, and J. Feldman, ‘‘Domain 4: Communica-
tion and network security,’’ in CISSP Study Guide. Amsterdam, The
Netherlands: Elsevier, Jan. 2023, pp. 225–293, doi: 10.1016/b978-0-443-
18734-6.00003-9.
[64] M. Sanz, J. I. Moreno, G. López, J. Matanza, and J. Berrocal, ‘‘Web-
based toolkit for performance simulation and analysis of power line
communication networks,’’ Energies, vol. 14, no. 20, p. 6475, Oct. 2021,
doi: 10.3390/en14206475.
[65] Y. Huo, G. Prasad, L. Lampe, and C. M. V. Leung, ‘‘Smart-grid moni-
toring: Enhanced machine learning for cable diagnostics,’’ in Proc. IEEE
Int. Symp. Power Line Commun. Appl. (ISPLC), Prague, Czech Republic,
Apr. 2019, pp. 1–6, doi: 10.1109/ISPLC.2019.8693287.
[66] A. M. Tonello, N. A. Letizia, D. Righini, and F. Marcuzzi, ‘‘Machine
learning tips and tricks for power line communications,’’ IEEE Access,
vol. 7, pp. 82434–82452, 2019, doi: 10.1109/ACCESS.2019.2923321.
[67] O. Simeone, ‘‘A very brief introduction to machine learning with appli-
cations to communication systems,’’ IEEE Trans. Cogn. Commun. Netw.,
vol. 4, no. 4, pp. 648–664, Dec. 2018.
[68] N. A. Letizia and A. M. Tonello, ‘‘Capacity learning for communica-
tion systems over power lines,’’ in Proc. IEEE Int. Symp. Power Line
Commun. Appl. (ISPLC), Aachen, Germany, Oct. 2021, pp. 55–60, doi:
10.1109/ISPLC52837.2021.9628415.
[69] C. Yang and Z. Zhang, ‘‘Intelligent home system design based on power
line communication,’’ in Proc. 2nd Int. Conf. Telecommun. Commun.
Eng., Nov. 2018, pp. 334–338, doi: 10.1145/3291842.3291852.
[70] M.-S. Koh, ‘‘An end-to-end trainable power line communication system,’’
in Proc. 8th Int. Conf. Mach. Learn. Technol., Mar. 2023, pp. 257–263,
doi: 10.1145/3589883.3589922.
[71] N. Nayakwadi and R. Fatima, ‘‘Machine learning based handover exe-
cution algorithm for heterogeneous wireless networks,’’ in Proc. 5th Int.
Conf. Res. Comput. Intell. Commun. Netw. (ICRCICN), Bengaluru, India,
Nov. 2020, pp. 54–58, doi: 10.1109/ICRCICN50933.2020.9296169.
[72] F. Nisar and S. Baseer, ‘‘A comprehensive survey on mobile communi-
cation generation,’’ in Proc. Int. Conf. Innov. Comput. (ICIC), Lahore,
Pakistan, Nov. 2021, pp. 1–6, doi: 10.1109/ICIC53490.2021.9692972.
[73] P. Liu, J. Luo, H. Liang, and Z. Wen, ‘‘Intelligent monitoring model
of Internet of Things for power transmission corridor status in bad
environment area,’’ in Proc. 13th Int. Conf. Measuring Technol. Mecha-
tronics Autom. (ICMTMA), Beihai, China, Jan. 2021, pp. 188–192, doi:
10.1109/ICMTMA52658.2021.00049.
[74] G. Kumari and C. Selwal, ‘‘System analysis for optical inter-satellite
link with varied parameter and pre-amplification,’’ in Proc. Int. Conf.
Adv. Comput., Commun. Informat. (ICACCI), Jaipur, India, Sep. 2016,
pp. 2157–2161, doi: 10.1109/ICACCI.2016.7732371.
[75] M. S. Ibrahim, W. Dong, and Q. Yang, ‘‘Machine learning driven
smart electric power systems: Current trends and new perspec-
tives,’’ Appl. Energy, vol. 272, Aug. 2020, Art. no. 115237, doi:
10.1016/j.apenergy.2020.115237.
[76] S. A. A. Kazmi and M. S. Iqbal, ‘‘Machine learning assisted delay mini-
mization in full duplex energy constrained cooperative communication
network,’’ Ad Hoc Netw., vol. 149, May 2023, Art. no. 103208, doi:
10.1016/j.adhoc.2023.103208.
[77] I. H. Sarker, ‘‘Machine learning: Algorithms, real-world applications and
research directions,’’ Social Netw. Comput. Sci., vol. 2, no. 3, pp. 1–21,
Mar. 2021, doi: 10.1007/s42979-021-00592-x.
[78] Y. Zhang, T. Huang, and E. F. Bompard, ‘‘Big data analytics in smart
grids: A review,’’ Energy Informat., vol. 1, no. 1, Aug. 2018, Art. no. 8,
doi: 10.1186/s42162-018-0007-5.
[79] C. Liu, Z. H. Rather, Z. Chen, and C. L. Bak, ‘‘An overview of decision
tree applied to power systems,’’ Int. J. Smart Grid Clean Energy, vol. 2,
no. 3, pp. 413–419, 2013, doi: 10.12720/sgce.2.3.413-419.
[80] M. Bonavita and P. Laloyaux, ‘‘Machine learning for model error infer-
ence and correction,’’ J. Adv. Model. Earth Syst., vol. 12, no. 12,
Nov. 2020, Art. no. e2020MS002232, doi: 10.1029/2020ms002232.
[81] I. H. Sarker, ‘‘Deep learning: A comprehensive overview on techniques,
taxonomy, applications and research directions,’’ Social Netw. Comput.
Sci., vol. 2, no. 6, Aug. 2021, Art. no. 420, doi: 10.1007/s42979-021-
00815-1.
[82] N. Jariwala, C. L. Putta, K. Gatade, M. Umarji, S. N. R. Rahman,
D. M. Pawde, A. Sree, A. S. Kamble, A. Goswami, P. Chakraborty, and
T. Shunmugaperumal, ‘‘Intriguing of pharmaceutical product develop-
ment processes with the help of artificial intelligence and deep/machine
learning or artificial neural network,’’ J. Drug Del. Sci. Technol., vol. 87,
Sep. 2023, Art. no. 104751, doi: 10.1016/j.jddst.2023.104751.
[83] H. Chang, X. Liang, H. Li, J. Shen, X. Gu, and L. Zhang, ‘‘Deep
learning-based bitstream error correction for CSI feedback,’’ IEEE Wire-
less Commun. Lett., vol. 10, no. 12, pp. 2828–2832, Dec. 2021, doi:
10.1109/LWC.2021.3118923.
[84] R. Zhang, G. Li, S. Bu, G. Kuang, W. He, Y. Zhu, and S. Aziz, ‘‘A hybrid
deep learning model with error correction for photovoltaic power fore-
casting,’’ Frontiers Energy Res., vol. 10, Aug. 2022, Art. no. 948308, doi:
10.3389/fenrg.2022.948308.
[85] L. D. Colomer, M. Skotiniotis, and R. Muñoz-Tapia, ‘‘Reinforcement
learning for optimal error correction of toric codes,’’ Phys.
Lett. A, vol. 384, no. 17, Jun. 2020, Art. no. 126353, doi:
10.1016/j.physleta.2020.126353.
[86] R. Gholami and N. Fakhari, ‘‘Support vector machine: Principles,
parameters, and applications,’’ in Handbook of Neural Computation.
New York, NY, USA: Academic, 2017, pp. 515–535, doi: 10.1016/b978-
0-12-811318-9.00027-2.
[87] S. Ghosh, A. Dasgupta, and A. Swetapadma, ‘‘A study on support vector
machine based linear and non-linear pattern classification,’’ in Proc. Int.
Conf. Intell. Sustain. Syst. (ICISS), Palladam, India, Feb. 2019, pp. 24–28,
doi: 10.1109/ISS1.2019.8908018.
[88] S. Huang, J. Huang, Y. Ou, W. Ruan, J. Lin, X. Peng, and X. Wang,
‘‘Transmission line faults classification based on alienation coefficients
of current and voltage waveform and SVM,’’ in Proc. 5th Asia Conf.
Power Electr. Eng. (ACPEE), Chengdu, China, Jun. 2020, pp. 60–64, doi:
10.1109/ACPEE48638.2020.9136270.
[89] J. Xie, A. P. S. Meliopoulos, and B. Xie, ‘‘Transmission line fault classi-
fication based on dynamic state estimation and support vector machine,’’
in Proc. North Amer. Power Symp. (NAPS), Fargo, ND, USA, Sep. 2018,
pp. 1–5, doi: 10.1109/NAPS.2018.8600658.
[90] A. S. Nagane, C. H. Patil, and S. M. Mali, ‘‘Classification of
brahmi script characters using HOG features and multiclass error-
correcting output codes (ECOC) model containing SVM binary learn-
ers,’’ in Proc. Int. Conf. Intell. Innov. Technol. Comput., Electr.
Electron. (IITCEE), Bengaluru, India, Jan. 2023, pp. 448–451, doi:
10.1109/IITCEE57236.2023.10091084.
[91] C. Shu, C. Yang, and P. An, ‘‘An online SVM based VVC intra fast par-
tition algorithm with pre-scene-cut detection,’’ in Proc. IEEE Int. Symp.
Circuits Syst. (ISCAS), Austin, TX, USA, May 2022, pp. 3033–3037, doi:
10.1109/ISCAS48785.2022.9937410.
[92] P. Santos, L. Villa, A. Reñones, A. Bustillo, and J. Maudes, ‘‘An SVM-
based solution for fault detection in wind turbines,’’ Sensors, vol. 15,
no. 3, pp. 5627–5648, Mar. 2015, doi: 10.3390/s150305627.
[93] M. Awad and R. Khanna, ‘‘Support vector machines for classification,’’
in Efficient Learning Machines. Amsterdam, The Netherlands: Elsevier,
2015, pp. 39–66, doi: 10.1007/978-1-4302-5990-9_3.
[94] Y. Shen, J. Ruijsch, M. Lu, E. H. Sutanudjaja, and D. Karssenberg,
‘‘Random forests-based error-correction of streamflow from a large-
scale hydrological model: Using model state variables to estimate error
terms,’’ Comput. Geosci., vol. 159, Feb. 2022, Art. no. 105019, doi:
10.1016/j.cageo.2021.105019.
[95] X. Chen and H. Ishwaran, ‘‘Random forests for genomic data
analysis,’’ Genomics, vol. 99, no. 6, pp. 323–329, Jun. 2012, doi:
10.1016/j.ygeno.2012.04.003.
[96] G. Biau, ‘‘Analysis of a random forests model,’’ J. Mach. Learn. Res.,
vol. 13, pp. 1063–1095, Apr. 2012.
124778 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
[97] H. Taherdoost, ‘‘Blockchain and machine learning: A critical review
on security,’’ Information, vol. 14, no. 5, p. 295, May 2023, doi:
10.3390/info14050295.
[98] A. Gerodimos, L. Maglaras, M. A. Ferrag, N. Ayres, and I. Kantzavelou,
‘‘IoT: Communication protocols and security threats,’’ Internet
Things Cyber-Phys. Syst., vol. 3, pp. 1–13, Jan. 2023, doi:
10.1016/j.iotcps.2022.12.003.
[99] M. I. Tariq, N. A. Memon, S. Ahmed, S. Tayyaba, M. T. Mushtaq,
N. A. Mian, M. Imran, and M. W. Ashraf, ‘‘A review of deep learning
security and privacy defensive techniques,’’ Mobile Inf. Syst., vol. 2020,
pp. 1–18, Apr. 2020, doi: 10.1155/2020/6535834.
[100] C. Wen, D. Dematties, and S.-L. Zhang, ‘‘A guide to signal process-
ing algorithms for nanopore sensors,’’ ACS Sensors, vol. 6, no. 10,
pp. 3536–3555, Oct. 2021, doi: 10.1021/acssensors.1c01618.
[101] R. Maskeli ¯
unas, R. Pomarnacki, V. K. Huynh, R. Damaševičius, and
D. Plonis, ‘‘Power line monitoring through data integrity analysis with
Q-learning based data analysis network,’’ Remote Sens., vol. 15, no. 1,
p. 194, Dec. 2022, doi: 10.3390/rs15010194.
[102] X. Huang, T. Hu, C. Ye, G. Xu, X. Wang, and L. Chen, ‘‘Electric
load data compression and classification based on deep stacked auto-
encoders,’’ Energies, vol. 12, no. 4, p. 653, Feb. 2019, doi: 10.3390/
en12040653.
[103] N. Rajkumar and E. Kannan, ‘‘Deep learning-based key transmission
(DLKT) protocol for secured group communication in cloud,’’ Soft Com-
put., vol. 25, no. 16, pp. 10709–10721, Jul. 2021, doi: 10.1007/s00500-
021-05959-z.
[104] A Complete Overview of IoT Communication Protocols | Expan-
ice. Accessed: Jul. 2023. [Online]. Available: https://expanice.com/
article/iot-communication-protocols-comparison
[105] H. Lin, W.-Y. Shin, and J. Joung, ‘‘Support vector machine-based trans-
mit antenna allocation for multiuser communication systems,’’ Entropy,
vol. 21, no. 5, p. 471, May 2019, doi: 10.3390/e21050471.
[106] R. K. Satyanarayana and K. Selvakumar, ‘‘Bi-linear mapping integrated
machine learning based authentication routing protocol for improv-
ing quality of service in vehicular ad-hoc network,’’ e-Prime, Adv.
Electr. Eng., Electron. Energy, vol. 4, Jun. 2023, Art. no. 100145, doi:
10.1016/j.prime.2023.100145.
[107] S. D. D. Anton, S. Sinha, and H. D. Schotten, ‘‘Anomaly-based intru-
sion detection in industrial data with SVM and random forests,’’ 2019,
arXiv:1907.10374. Accessed: Jul. 29, 2023.
[108] M. D. Lal and R. Varadarajan, ‘‘A review of machine learning approaches
in synchrophasor technology,’’ IEEE Access, vol. 11, pp. 33520–33541,
2023, doi: 10.1109/ACCESS.2023.3263547.
[109] A. Alayil, P. Sarkar, D. Bose, and C. K. Chanda, ‘‘Prediction of power
outage during cyclone using machine learning,’’ in Proc. IEEE Cal-
cutta Conf. (CALCON), Kolkata, India, Dec. 2022, pp. 255–260, doi:
10.1109/CALCON56258.2022.10060493.
[110] X. Tang, P. Reviriego, W. Tang, D. G. M. Mitchell, F. Lombardi, and
S. Liu, ‘‘Joint learning and channel coding for error-tolerant IoT systems
based on machine learning,’’ IEEE Trans. Artif. Intell., early access,
Jan. 10, 2023, doi: 10.1109/TAI.2023.3235778.
[111] S. Liu, P. Reviriego, X. Tang, W. Tang, and F. Lombardi, ‘‘Result-
based re-computation for error-tolerant classification by a support vector
machine,’’ IEEE Trans. Artif. Intell., vol. 1, no. 1, pp. 62–73, Aug. 2020,
doi: 10.1109/TAI.2020.3028321.
[112] P. Mlynek, J. Misurec, P. Silhavy, R. Fujdiak, J. Slacik, and Z. Hasirci,
‘‘Simulation of achievable data rates of broadband power line communi-
cation for smart metering,’’ Appl. Sci., vol. 9, no. 8, p. 1527, Apr. 2019,
doi: 10.3390/app9081527.
[113] D. Liang, S. Ge, H. Guo, Y. Wang, Z. Liang, and C. Chen, ‘‘Mon-
itoring power line faults using impedance estimation algorithms in
power line communication equipment,’’ in Proc. 10th Int. Conf. Power
Energy Syst. (ICPES), Chengdu, China, Dec. 2020, pp. 404–408, doi:
10.1109/ICPES51309.2020.9349656.
[114] Z. Shi, J. Lu, and B. Li, ‘‘Research on power line carrier com-
munication channel modeling for power distribution and consump-
tion IoT,’’ in Proc. 3rd Int. Conf. Comput. Vis., Image Deep
Learn. Int. Conf. Comput. Eng. Appl. (CVIDL ICCEA), Changchun,
China, May 2022, pp. 322–325, doi: 10.1109/CVIDLICCEA56201.2022.
9824830.
[115] J. P. A. León, L. J. de la Cruz Llopis, and F. J. Rico-Novella, ‘‘A machine
learning based distributed congestion control protocol for multi-hop wire-
less networks,’’ Comput. Netw., vol. 231, Jul. 2023, Art. no. 109813, doi:
10.1016/j.comnet.2023.109813.
[116] J. Chen, H. Chen, and Z. Li, ‘‘A double serial concatenated code
using CRC-aided error correction for highly reliable communica-
tion,’’ Comput. Netw., vol. 216, Oct. 2022, Art. no. 109260, doi:
10.1016/j.comnet.2022.109260.
[117] P. Ledesma, D. Gotti, and H. Amaris, ‘‘Co-simulation platform for
interconnected power systems and communication networks based on
PSS/E and OMNeT++,’’ Comput. Electr. Eng., vol. 101, Jul. 2022,
Art. no. 108092, doi: 10.1016/j.compeleceng.2022.108092.
[118] E. A. A. Alaoui, S. C. K. Tekouabou, Y. Maleh, and A. Nayyar,
‘‘Towards to intelligent routing for DTN protocols using machine learn-
ing techniques,’’ Simul. Model. Pract. Theory, vol. 117, May 2022,
Art. no. 102475, doi: 10.1016/j.simpat.2021.102475.
[119] M. A. Kippke, ‘‘Technologies for data collection in power systems,’’
in Encyclopedia of Electrical and Electronic Power Engineering.
Amsterdam, The Netherlands: Elsevier, Jan. 2023, pp. 320–326, doi:
10.1016/b978-0-12-821204-2.00063-5.
[120] Y. Li, T. Bao, Z. Chen, Z. Gao, X. Shu, and K. Zhang, ‘‘A missing
sensor measurement data reconstruction framework powered by multi-
task Gaussian process regression for dam structural health monitoring
systems,’’ Measurement, vol. 186, Dec. 2021, Art. no. 110085, doi:
10.1016/j.measurement.2021.110085.
[121] D. DeBonis, T. Estrada, R. E. Grant, K. T. Pedretti, J. H. Laros, III,
and D. Arnold, ‘‘APE: Metrics for understanding application perfor-
mance efficiency under power caps,’’ Sustain. Comput., Informat.
Syst., vol. 34, Apr. 2022, Art. no. 100702, doi: 10.1016/j.suscom.2022.
100702.
[122] H. Sheng, H. Y. Zhang, F. Yang, C. H. Li, and J. Wang, ‘‘A CSMA/CA
based MAC protocol for hybrid power-line/visible-light communication
networks: Design and analysis,’’ Digit. Commun. Netw., Oct. 2022, doi:
10.1016/j.dcan.2022.09.019.
[123] M. Wolkerstorfer, B. Schweighofer, H. Wegleiter, D. Statovci,
H. Schwaiger, and W. Lackner, ‘‘Measurement and simulation framework
for throughput evaluation of narrowband power line communication
links in low-voltage grids,’’ J. Netw. Comput. Appl., vol. 59, pp. 285–300,
Jan. 2016, doi: 10.1016/j.jnca.2015.05.022.
[124] F. Krajcovic, M. Krajmer, M. Lipovsky, M. Majchrak, and
P. Podhoransky, ‘‘Investigation of harmful interference to
power line communication bit rate,’’ in Proc. 19th Int. Conf.
Radioelektronika, Bratislava, Slovakia, Apr. 2009, pp. 51–55, doi:
10.1109/RADIOELEK.2009.5158723.
[125] B. Yue, Z. Wei, C. Zheng, Y. Ding, B. Li, D. Li, X. Liang, and
X. Zhai, ‘‘Power consumption prediction of variable refrigerant flow
system through data-physics hybrid approach: An online prediction test
in office building,’’ Energy, vol. 278, Sep. 2023, Art. no. 127826, doi:
10.1016/j.energy.2023.127826.
[126] C. Huo, J. Yuan, G. Song, and Z. Shi, ‘‘Node reliability based multi-path
routing algorithm of high-speed power line communication network,’’
in Proc. IEEE 4th Int. Conf. Cloud Comput. Big Data Anal. (ICC-
CBDA), Chengdu, China, Apr. 2019, pp. 570–573, doi: 10.1109/ICC-
CBDA.2019.8725798.
[127] A. H. K. S. A. Saidi, S. A. Hussain, S. M. Hussain, A. V. Singh, and
A. Rana, ‘‘Smart water meter using power line communication (PLC)
approach for measurements of accurate water consumption and billing
process,’’ in Proc. 8th Int. Conf. Rel., Infocom Technol. Optim. (Trends
Future Directions) (ICRITO), Noida, India, Jun. 2020, pp. 1119–1122,
doi: 10.1109/ICRITO48877.2020.9197956.
[128] H. Chiroma, P. Nickolas, N. Faruk, E. Alozie, I.-F.-Y. Olayinka,
K. S. Adewole, A. Abdulkarim, A. A. Oloyede, O. A. Sowande, S. Garba,
A. D. Usman, L. S. Taura, and Y. A. Adediran, ‘‘Large scale survey for
radio propagation in developing machine learning model for path losses
in communication systems,’’ Sci. Afr., vol. 19, Mar. 2023, Art. no. e01550,
doi: 10.1016/j.sciaf.2023.e01550.
[129] S. Singh, A. Trivedi, and D. Saxena, ‘‘Unsupervised LoS/NLoS identi-
fication in mmWave communication using two-stage machine learning
framework,’’ Phys. Commun., vol. 59, May 2023, Art. no. 102118, doi:
10.1016/j.phycom.2023.102118.
[130] A. Kumbhar, P. G. Dhawale, S. Kumbhar, U. Patil, and P. Magdum,
‘‘A comprehensive review: Machine learning and its application in inte-
grated power system,’’ Energy Rep., vol. 7, pp. 5467–5474, Nov. 2021,
doi: 10.1016/j.egyr.2021.08.133.
[131] M. Farhoumandi, Q. Zhou, and M. Shahidehpour, ‘‘A review of machine
learning applications in IoT-integrated modern power systems,’’ Electr.
J., vol. 34, no. 1, Jan. 2021, Art. no. 106879, doi: 10.1016/j.tej.2020.
106879.
VOLUME 11, 2023 124779

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
[132] P. Sanjeevikumar, T. Samavat, M. A. Nasab, M. Zand, and M. Khoobani,
‘‘Machine learning-based hybrid demand-side controller for renewable
energy management,’’ in Sustainable Developments by Artificial Intel-
ligence and Machine Learning for Renewable Energies. Amsterdam,
The Netherlands: Elsevier, Jan. 2022, pp. 291–307, doi: 10.1016/b978-
0-323-91228-0.00003-3.
[133] T. Kaneko, R. Wada, M. Ozaki, and T. Inoue, ‘‘Hybrid physics-based
and machine learning model with interpretability and uncertainty for real-
time estimation of unmeasurable parts,’’ Ocean Eng., vol. 284, Sep. 2023,
Art. no. 115267, doi: 10.1016/j.oceaneng.2023.115267.
[134] N. A. Azhar, N. A. M. Radzi, K. H. M. Azmi, F. S. Samidi, and
A. M. Zainal, ‘‘Criteria selection using machine learning (ML) for com-
munication technology solution of electrical distribution substations,’’
Appl. Sci., vol. 12, no. 8, p. 3878, Apr. 2022, doi: 10.3390/app12083878.
[135] L. Pereira and N. Nunes, ‘‘An empirical exploration of performance
metrics for event detection algorithms in non-intrusive load monitor-
ing,’’ Sustain. Cities Soc., vol. 62, Nov. 2020, Art. no. 102399, doi:
10.1016/j.scs.2020.102399.
[136] S.-M. Tseng, W.-C. Hsu, and D.-F. Tseng, ‘‘Deeplearning based decoding
for polar codes in Markov Gaussian memory impulse noise channels,’’
Wireless Pers. Commun., vol. 122, no. 1, pp. 737–753, Aug. 2021, doi:
10.1007/s11277-021-08923-0.
[137] K. Mei, J. Liu, X. Zhang, K. Cao, N. Rajatheva, and J. Wei, ‘‘A low
complexity learning-based channel estimation for OFDM systems with
online training,’’ IEEE Trans. Commun., vol. 69, no. 10, pp. 6722–6733,
Oct. 2021, doi: 10.1109/TCOMM.2021.3095198.
[138] H. Mao, H. Lu, Y. Lu, and D. Zhu, ‘‘RoemNet: Robust meta learn-
ing based channel estimation in OFDM systems,’’ in Proc. IEEE
Int. Conf. Commun. (ICC), Shanghai, China, May 2019, pp. 1–6, doi:
10.1109/ICC.2019.8761319.
[139] K. Mei, J. Liu, X. Zhang, N. Rajatheva, and J. Wei, ‘‘Perfor-
mance analysis on machine learning-based channel estimation,’’ IEEE
Trans. Commun., vol. 69, no. 8, pp. 5183–5193, Aug. 2021, doi:
10.1109/TCOMM.2021.3083597.
[140] J. Liu, K. Mei, X. Zhang, D. Ma, and J. Wei, ‘‘Online extreme learning
machine-based channel estimation and equalization for OFDM systems,’’
IEEE Commun. Lett., vol. 23, no. 7, pp. 1276–1279, Jul. 2019, doi:
10.1109/LCOMM.2019.2916797.
[141] J. H. Fernandez, A. Omri, and R. D. Pietro, ‘‘Power grid surveillance:
Topology change detection system using power line communications,’’
Int. J. Electr. Power Energy Syst., vol. 145, Feb. 2023, Art. no. 108634,
doi: 10.1016/j.ijepes.2022.108634.
[142] C. Tang, Z. Chang, H. Liang, L. Zhang, and B. Pang, ‘‘Feature
extraction method of HPLC communication signal based on genetic
algorithm,’’ IET Commun., vol. 17, no. 13, pp. 1553–1561, Jun. 2023,
doi: 10.1049/cmu2.12641.
[143] E. Oyekanlu and J. Uddin, ‘‘Random forest-based ensemble machine
learning data-optimization approach for smart grid impedance prediction
in the powerline narrowband frequency band,’’ in Deterministic Artificial
Intelligence. IntechOpen, May 2020, doi: 10.5772/intechopen.91837.
[144] J. Slacik, P. Mlynek, M. Rusz, P. Musil, L. Benesl, and M. Ptacek, ‘‘Broad-
band power line communication for integration of energy sensors within
a smart city ecosystem,’’ Sensors, vol. 21, no. 10, p. 3402, May 2021, doi:
10.3390/s21103402.
[145] T. Mazhar, H. M. Irfan, I. Haq, I. Ullah, M. Ashraf, T. A. Shloul,
Y. Y. Ghadi, Imran, and D. H. Elkamchouchi, ‘‘Analysis of challenges
and solutions of IoT in smart grids using AI and machine learning
techniques: A review,’’ Electronics, vol. 12, no. 1, p. 242, Jan. 2023, doi:
10.3390/electronics12010242.
[146] F. Marcuzzi and A. M. Tonello, ‘‘Topology-based machine learning: Pre-
dicting power line communication quality in smart grids,’’ IEEE Access,
vol. 11, pp. 24851–24862, 2023, doi: 10.1109/ACCESS.2023.3245361.
[147] A. Dogan and D. C. Dogan, ‘‘A review on machine learning models in
forecasting of virtual power plant uncertainties,’’ Arch. Comput. Methods
Eng., vol. 30, no. 3, pp. 2081–2103, Nov. 2022, doi: 10.1007/s11831-022-
09860-2.
[148] D. Shrestha, X. Mestre, and M. Payaró, ‘‘On channel estimation
for power line communication systems in the presence of impulsive
noise,’’ Comput. Electr. Eng., vol. 72, pp. 406–419, Nov. 2018, doi:
10.1016/j.compeleceng.2018.10.006.
[149] I. Shakeel, ‘‘Machine learning based featureless signalling,’’ in Proc.
IEEE Mil. Commun. Conf. (MILCOM), Los Angeles, CA, USA,
Oct. 2018, pp. 1–9, doi: 10.1109/MILCOM.2018.8599694.
[150] T. Ç. Akinci and A. A. Martinez-Morales, ‘‘Cognitive based electric
power management system,’’ Balkan J. Electr. Comput. Eng., vol. 10,
no. 1, pp. 85–90, Jan. 2022, doi: 10.17694/bajece.1060998.
[151] H. Soy and İ. Toy, ‘‘Design and implementation of smart pressure sen-
sor for automotive applications,’’ Measurement, vol. 176, May 2021,
Art. no. 109184, doi: 10.1016/j.measurement.2021.109184.
[152] Y. M. Chung, ‘‘Performance comparisons of broadband power line com-
munication technologies,’’ Appl. Sci., vol. 10, no. 9, p. 3306, May 2020,
doi: 10.3390/app10093306.
[153] M. Leo, S. Sharma, and K. Maddulety, ‘‘Machine learning in banking risk
management: A literature review,’’ Risks, vol. 7, no. 1, p. 29, Mar. 2019,
doi: 10.3390/risks7010029.
[154] O. Ali, W. Abdelbaki, A. Shrestha, E. Elbasi, M. A. A. Alryalat, and
Y. K. Dwivedi, ‘‘A systematic literature review of artificial intelligence
in the healthcare sector: Benefits, challenges, methodologies, and func-
tionalities,’’ J. Innov. Knowl., vol. 8, no. 1, Jan. 2023, Art. no. 100333,
doi: 10.1016/j.jik.2023.100333.
[155] K. Liakos, P. Busato, D. Moshou, S. Pearson, and D. Bochtis, ‘‘Machine
learning in agriculture: A review,’’ Sensors, vol. 18, no. 8, p. 2674,
Aug. 2018, doi: 10.3390/s18082674.
[156] Z. Peiling and Z. Hongxin, ‘‘Measurement and modeling of indoor
low voltage power distribution line channel,’’ in Proc. 6th Asia–Pacific
Conf. Environ. Electromagn. (CEEM), Shanghai, China, Nov. 2012,
pp. 396–398, doi: 10.1109/CEEM.2012.6410652.
[157] M. Z. Nisar, M. S. Ibrahim, M. Usman, and J.-A. Lee, ‘‘A lightweight
deep learning model for automatic modulation classification using resid-
ual learning and squeeze–excitation blocks,’’ Appl. Sci., vol. 13, no. 8,
p. 5145, Apr. 2023, doi: 10.3390/app13085145.
[158] M. Abadi et al., ‘‘TensorFlow: Large-scale machine learning on hetero-
geneous distributed systems,’’ 2016, arXiv:1603.04467.
[159] H. Asghari, R. E. Newman, and H. A. Latchman, ‘‘Bandwidth-efficient
forward-error-correction-coding for high speed powerline communica-
tions,’’ in Proc. IEEE Int. Symp. Power Line Commun. Appl., Jan. 2006,
pp. 355–360, doi: 10.1109/ISPLC.2006.247488.
[160] Microcontrollers Lab. (Oct. 8, 2019). Introduction to Power Line
Communication. Accessed: Jul. 2023. [Online]. Available: https://
microcontrollerslab.com/introduction-to-power-line-communication/
[161] A. O. Aderibole, E. K. Saathoff, K. J. Kircher, S. B. Leeb, and
L. K. Norford, ‘‘Power line communication for low-bandwidth control
and sensing,’’ IEEE Trans. Power Del., vol. 37, no. 3, pp. 2172–2181,
Jun. 2022, doi: 10.1109/TPWRD.2021.3106585.
[162] G. López, J. Matanza, D. De La Vega, M. Castro, A. Arrinda,
J. I. Moreno, and A. Sendin, ‘‘The role of power line commu-
nications in the smart grid revisited: Applications, challenges, and
research initiatives,’’ IEEE Access, vol. 7, pp. 117346–117368, 2019, doi:
10.1109/ACCESS.2019.2928391.
[163] L. Lampe, A. M. Tonello, and T. Swart, Power Line Communica-
tions: Principles, Standards and Applications From Multimedia to
Smart Grid, 2nd ed. Hoboken, NJ, USA: Wiley, 2016, doi: 10.1002/
9781118676684.
[164] D. B. Unsal and T. Yalcinoz, ‘‘Applications of new power line
communication model for smart grids,’’ Int. J. Comput. Electr.
Eng., vol. 7, no. 3, pp. 168–178, 2015, doi: 10.17706/ijcee.2015.7.3.
168-178.
[165] W. Huang, T. Zhang, and X. Yao, ‘‘Optimization for sequential commu-
nication line attack in interdependent power-communication network,’’
Phys. A, Stat. Mech. Appl., vol. 592, Apr. 2022, Art. no. 126837, doi:
10.1016/j.physa.2021.126837.
[166] S. M. Hashim and I. B. Al-Mashhadani, ‘‘Adaptation of powerline
communications-based smart metering deployments with IoT cloud plat-
form,’’ Indonesian J. Electr. Eng. Comput. Sci., vol. 29, no. 2, p. 825,
Feb. 2023, doi: 10.11591/ijeecs.v29.i2.pp825-837.
[167] M. O. F. Goni, M. Nahiduzzaman, M. S. Anower, M. M. Rahman,
M. R. Islam, M. Ahsan, J. Haider, and M. Shahjalal, ‘‘Fast and accu-
rate fault detection and classification in transmission lines using
extreme learning machine,’’ e-Prime, Adv. Electr. Eng., Electron.
Energy, vol. 3, Mar. 2023, Art. no. 100107, doi: 10.1016/j.prime.2023.
100107.
[168] R. Ramesh, S. Gurugopinath, and S. Muhaidat, ‘‘Hybrid protocols for
relay-assisted non-orthogonal multiple access in power line communica-
tions,’’ IEEE Open J. Commun. Soc., vol. 4, pp. 1813–1825, 2023, doi:
10.1109/OJCOMS.2023.3289185.
124780 VOLUME 11, 2023

T. C. Akinci et al.: ML-Based Error Correction Codes and Communication Protocols for PLC
[169] A. A. Oliveira, M. S. Buckeridge, and R. Hirata, ‘‘Detecting tree and wire
entanglements with deep learning,’’ Trees, vol. 37, no. 1, pp. 147–159,
May 2022, doi: 10.1007/s00468-022-02305-0.
[170] I. Topaloglu, ‘‘Deep learning based a new approach for power quality
disturbances classification in power transmission system,’’ J. Electr. Eng.
Technol., vol. 18, no. 1, pp. 77–88, Jan. 2023, doi: 10.1007/s42835-022-
01177-1.
TAHIR CETIN AKINCI (Senior Member, IEEE)
received the bachelor’s degree in electrical
engineering, in 2000, and the master’s and
Ph.D. degrees, in 2005 and 2010, respectively.
From 2003 to 2010, he was a Research Assis-
tant with Marmara University, Istanbul, Turkey.
He is currently a Full Professor with the Electri-
cal Engineering Department, Istanbul Technical
University (ITU), in 2020. He was the Vice Dean
with the Graduate School, from 2020 to 2021,
and the Electrical and Electronic Engineering Faculty, from 2020 to 2021.
He assumed the role of a Visiting Scholar with the University of California
at Riverside (UCR). His current research interests include artificial neural
networks, deep learning, machine learning, cognitive systems, signal pro-
cessing, power systems, power line communication, and data analysis.
GOKHAN ERDEMIR (Senior Member, IEEE)
received the B.Sc., M.Sc., and Ph.D. degrees from
Marmara University, Turkey. He was a Research
Scholar with the Robotics and Automation Lab-
oratory, Michigan State University, East Lans-
ing, MI, USA, and the Health Management and
Research Center, University of Michigan, Ann
Arbor, MI, USA. He is currently an Associate
Professor with the Engineering Management and
Technology Department, The University of Ten-
nessee at Chattanooga (UTC). His current research interests include control
theory, robotics, industrial automation, AGVs, and engineering education.
A. TARIK ZENGIN received the B.S. degree in
electrical and electronics engineering from Ege
University, Turkey, in 2007, and the M.E. and
Ph.D. degrees from the Department of Computer
Science and Electrical Engineering, Kumamoto
University, Japan, in 2010 and 2013, respectively.
Currently, he is an Assistant Professor with Istan-
bul Technical University. His current research
interests include autonomous systems and control
theory.
SERHAT SEKER received the degree from the
Electrical Engineering Department, Istanbul Tech-
nical University (ITU), and the master’s and
Ph.D. degrees from the Electrical Engineering
Division, Science and Technology Institute, ITU.
He studied the Ph.D. thesis with the Energy
Research Centre of the Netherlands (ECN) and
worked on signal analysis techniques. He was
an Assistant Professor and an Associate Profes-
sor with ITU, in 1995 and 1996. He worked in
industrial signal processing with the Maintenance and Reliability Centre,
The University of Tennessee, Knoxville, TN, USA, in 1997. He was
the Vice Dean with the Electrical and Electronic Engineering Faculty,
from 2001 to 2004, and the Department Head of the Electrical Engineering,
from 2004 to 2007. He was also the Dean of the Faculty of Electrical and
Electronics, from 2013 to 2020.
ABDOULKADER IBRAHIM IDRISS received
the Ph.D. degree in photonic engineering from
Université de Franche-Comté, Besançon, France,
with specialization in optical nano-antennas for
the inspection of photonic structures. He was the
Director of the Logistic and Transport Centre
(Centre of Excellence), financed by the World
Bank, from 2019 to December 2021. He is cur-
rently a Professor with the Department of Electri-
cal Engineering. He is the Dean of the Faculty of
Engineering. He is also an Assistant Professor with the Faculty of Engineer-
ing, Université de Djibouti, Djibouti. His current research interests include
materials, photonic, and nanomaterials for renewable energy. He is a Guest
Editor for the Special Issue Big Data in Renewable Energy of Renewable and
Sustainable Energy Reviews (Elsevier).
VOLUME 11, 2023 124781