Al-Razi University

information
technology collage
Cybersecurity
Department

IoT Security Challenges and Its Solutions Using Artificial

Intelligence

Under the supervision of Dr. Zaid Al-Marhabi

student preparation
Sadik Ali Sadik
Muhammad Ahmed Ismail Al-Sharabi
Abdul Malik Balghaith Yahya Al-Ahdal
Yaqub Ali Saghir Shalkhuf
Sana’a Yemen 2023/2024

Abstract: The Internet of Things (IoT) is a well-known technology that has a significant
impact on many areas, including connections, work, healthcare, and the economy. IoT has
the potential to improve life in a variety of contexts, from smart cities to classrooms, by
automating tasks, increasing output, and decreasing anxiety. Cyberattacks and threats, on
the other hand, have a significant impact on intelligent IoT applications. Many traditional
techniques for protecting the IoT are now ineffective due to new dangers and
vulnerabilities. To keep their security procedures, IoT systems of the future will need AI-
efficient machine learning and deep learning. The capabilities of artificial intelligence,
particularly machine and deep learning solutions, must be used if the next-generation IoT
system is to have a continuously changing and up-to-date security system. IoT security
intelligence is examined in this paper from every angle available. An innovative method for
protecting IoT devices against a variety of cyberattacks is to use machine learning and
deep learning to gain information from raw data. Finally, we discuss relevant research
issues and potential next steps considering our findings. This article examines how
machine learning and deep learning can be used to detect attack patterns
in unstructured data and safeguard IoT devices

. We discuss the challenges that researchers face, as

well as potential future directions for this research area, considering these findings.
Anyone with an interest in the IoT or cybersecurity can use this website’s content as a
technical resource and reference.

Keywords: internet of things; cyberattacks; anomalies; deep learning;

machine learning; healthcare
1. Introduction unexpected or harmful IoT behaviors and provide a
The Internet of Things (IoT) connects everything in the dynamic, adaptive security solution
modern world and is gaining by leveraging artificial intelligence skills, particularly
traction in business, particularly in healthcare. The IoT machine, and deep learning. To
is one of the most popular new sift through security data in search of novel insights
ideas in recent years. It locates, transmits, and analyzes and trends, machine learning and
data using a network of connected deep learning models commonly use a preset set of
components. In the IoT, “things” are sensors, RFID tags, rules, strategies, or complex transfer
heart rate monitors, and other functions [3]. By recognizing anomalies in the IoT,
smart devices that collect and transmit data. New developed security models might also
devices are added to IoT networks daily. be used to teach robots how to defend themselves
There will be roughly 20.4 billion connected devices in against potential threats or attacks. The
2022, up from 8.4 billion in 2020 [1]. paper’s contributions to the body of literature are
The IoT has an impact on our daily social, commercial, outlined in the following paragraphs.
and economic activities. IoT Data on how these technologies are used in the IoT are
revenue is expected to increase from 892 billion USD in being gathered [4]. Numerous
2018 to more than 4 trillion USD by academic studies have been conducted on IoT security.
2025. This expansion is directly related to the growth For example, some authors have
of the digital economy. The Internet carried out a survey of IoT security vulnerabilities in
which they examine and classify
common security issues relating to the layered design,
of Things has enabled smart meters, remote networking, communication, and
monitoring, process automation, smart homes, management of the IoT [5]. The findings of a second
smart cities, and smart businesses [2]. Current and study on the security of the IoT
future Internet of Things applications were published and produced a list of IoT security
and services have the potential to significantly improve research opportunities and concerns
the ease, speed, and comfort of after taking security in a broad meaning [6]. In addition
customers’ lives [3]. Many cyber threats and attacks, to discussing IoT simulators and
however, are significant impediments models, it summarizes the current state of IoT security
to IoT development. research. The author also gives a
Expansion of IoT networks raises significant issues in quick overview of the principles of IoT security, existing
several areas, including device dangers, potential solutions, and
management, data management, computation, projections for this industry’s future [7]. They
security, and privacy [4]. Several security investigate the problems and state of IoT
flaws discovered may jeopardize the burgeoning IoT. security in their study at the application, network, and
Future IoT applications, such as those perception layers [8].
mentioned above, may fail to fulfill all their promises if The authors consider application domains, security
a dependable framework is not issues, and the process by which
in place as they will be unable to meet individual needs solutions are developed. The authors show how attack
or adhere to social norms. IoT vectors, vulnerabilities, and other
systems are divided into four layers: the application relevant techniques can be used to classify IoT security
layer; the middleware or support layer; issues [9]. The authors overview
the networking and data transmission layer; and the the most recent threats and vulnerabilities related to
perception or sensing layer. There are the IoT by carefully analyzing IoT
many layers to IoT applications, and each needs security research [10]. There have been significant
different technology [2]. At each stage, studies on machine learning, in addition
there are particular security concerns and difficulties. to surveys. We look at IoT security solutions based on
Attacks, including denial-of-service, supervised, unsupervised, and
spoofing, jamming, eavesdropping, data manipulation, reinforcement learning techniques. Their research
and man-in-the-middle, are among primarily focuses on machine-learningbased
the most common IoT risks. authentication, access control, safe offloading, and
Because security threats and attacks are becoming virus detection for IoT data
more numerous and complex, privacy solutions [11]. The authors investigate many
traditional security practices are no longer as effective concerns, including potential attack
as they once were. Future IoT vectors and IoT network security requirements. The
infrastructure requires a security solution that uses use of computers and deep learning to
risk-mitigation technology to reduce secure the IoT is examined [12]. Researchers have
risk. Proponents of the Fourth Industrial Revolution analyzed known and unknowable risks,
argue that artificial intelligence (AI) accessible solutions, and barriers to see how the IoT’s
is critical to the future development of intelligent increasing capabilities affect security
systems. As a result, we can detect and privacy [13].
Understanding the nature of data, the many kinds of IoT network traffic, as part of machine-learning-based
cyber threats, and other pertinent security modeling. This article explores
factors is essential when using machine learning and how ML and DL can be used to uncover attack patterns
deep learning to build data-drivenStudent Name from unstructured data and
protect IoT devices. We address the difficulties
security systems [14]. Regarding connectivity, the IoT
researchers encounter and potential future
controls how and what happens
directions for this study area considering these
when things communicate. This suggests that,
findings. This study analyzes our current
regardless of location, IoT networks are
understanding of AI, focusing on the efficacy of
always available. Networks must remain flexible and
machine-learning- and deep-learningbased
responsive because IoT devices are
IoT security solutions. We introduce a variety of
constantly added and removed [15].
machine learning and deep learning
Because IoT devices are constantly being added and
architectures and techniques and describe how they
withdrawn, the network reconfiguration
can be applied to intelligent security
process must be dynamic and flexible. Ad hoc networks
modelling in order to address the problem of IoT
may rely on nearby devices
security. The abbreviations and their full
for short-range communications [5]. An IoT-enabled
form are shown in Table 1.
device transforms and acts based on
its proximity to the current location [16]. Wireless
Table 1. List of abbreviations.
communication is the primary obstacle
in industrial IoT networks. Highly reliable, low-latency
communication is necessary for
sensitive applications, such as traffic monitoring,
manufacturing on an assembly line, and
medical equipment [17].
An IoT device is a piece of hardware equipped with a
sensor that sends data between
locations through the internet. Because many sensors
are used in a complex system
application, the systems should be set up to use fewer
resources and cost less [15].
There are different techniques of machine learning and
deep learning, such as rulebased 1.1. Research Gap
techniques, the clustering method, optimization of Cyberattacks and threats have a significant impact on
security features, recurrent neural intelligent IoT applications.
network, multi-layer perceptron, and classification and Many traditional techniques for protecting the IoT are
regression techniques, used to protect now ineffective due to new dangers
IoT data. Regression and classification techniques are and vulnerabilities. The capabilities of artificial
well known and frequently used intelligence, particularly machine and deep
in IoT machine security. Predicting the outcome of learning solutions, must be used if the next-generation
discrete values or categories, such as IoT system is to have a continuously
anomaly, average, or attacks, is a standard definition of changing and up-to-date security system. We discussed
classification problems. Clustering how machine learning and deep
algorithms may be very helpful in resolving IoT security learning can be used to detect attack patterns in
issues, such as identifying outliers, unstructured data and safeguard IoT
anomalies, signatures, fraud, and cyberattacks, by devices.
exposing previously hidden patterns
and structures in IoT security data. Rule-based systems
may be essential to IoT security
because they may learn security or policy rules from
data. A well-known machine learning 2. Literature Review
technique called association rule learning looks for The IoT plays an important role in technological
patterns or relationships between the attributes advancement. “IoT” stands for
in a security dataset. This MLP network is used to “Internet of Things”, and the term “Things” refers to
analyze the NSL-KDD dataset’s electrical devices that are linked to
malware, explain IoT parameters, detect malicious the internet. The Fourth Industrial Revolution, also
traffic coming from IoT devices, and known as Industry 4.0, is defined by
create a model for intrusion detection. These enhanced increased automation of traditional industrial and
signature properties may simplify manufacturing processes. The IoT is one
the management of large amounts of IoT security data, of the intelligent technologies being developed for this
such as identifying anomalies in movement [18].
The IoT is a network of objects that may connect to the IoT network have little resources, hackers are drawn to
internet and wireless networks them as easy targets. IoT network
to send and receive data automatically. Different security and privacy issues have been addressed in a
organizations and research groups from variety of ways, the majority of which
distinct viewpoints define the IoT and smart make use of common cryptographic protocols.
environments. The authors claim that RFIDbased However, current solutions cannot address
digital information flows and physical components the security issues that arise with IoT networks and are
make up the IoT [19]. exacerbated by the distinctive
The healthcare industry is quickly adopting the IoT, characteristics of IoT nodes. By implementing machine
which has the potential to improve learning (ML) and deep learning
patient engagement, health, and access to care. IoT into IoT devices and networks, many threats to the
device growth, however, poses security of the Internet of Things (IoT)
significant can be stopped.
security, privacy, and safety hazards to patients and Present IoT opportunities in the healthcare sector are
healthcare workers. Studies promising. It is also well known
on reducing the risks brought on by the IoT in the for its sensing and measuring capabilities, including
healthcare industry are still few and narrowband IoT in low-energy form
Brain
far between. Integrating a secure applications solution (N.B. IoT). It is popular in the healthcare field because
with IoT devices in healthcare of its low energy usage. There are
environments has been the subject of recent research. several ideas for using N.B. IoT in the healthcare
It is crucial to create a specialized IoT business. N.B. IoT is already commonplace
app for health due to the sensitivity of healthcare data and works seamlessly with cellular networks such as
and information [20]. Current IoT LTE. As a result, N.B. IoT has emerged
possibilities for the healthcare sector are promising. as a feasible choice for healthcare-related applications
They are also quite popular because in recent years. The most
of their sensing and measuring capabilities, including critical
narrowband IoT in its low-energy threats to N.B. IoT are security measures and other
system-related issues. If these problems
variant (N.B. IoT). Because of its low energy
and challenges are solved, it has the potential to be
consumption, it is favored in the healthcare
one of the most feasible and popular
industry. Several concepts exist for using N.B. IoT in the
systems for low-power, broad-area healthcare
healthcare industry. It has not been
installations [22]. The IoT risk management
standardized and works flawlessly with cellular systems
model in healthcare is presented in Figure 2.
such as LTE. As a result, N.B. IoT Brain Sci. 2023, 13, x FOR PEER REVIEW 6 of 33
has emerged as a viable option for healthcare-related ways to secure IoT networks using machine learning
applications in recent years. However, techniques, such as supervised learning,
security measures and other system-related difficulties unsupervised learning, and reinforcement learning (RL).
are the most severe dangers to The analysis of data privacy
N.B. IoT. If these concerns and obstacles are addressed focuses on ML-based approaches for authenticating IoT
appropriately, it has the potential devices, controlling access to such
to be one of the most viable and popular solutions for devices, offloading data securely, and identifying viruses.
low-power, wide-area healthcare Future IoT adoption will have
installations [21]. a significant effect on society, business, and the economy.
One of the many difficulties facing the Internet of Because the majority of nodes
Things which connects a wide range in an IoT network have little resources, hackers are drawn
to them as easy targets. IoT
of objects to networks to enable complex and
network security and privacy issues have been addressed in
intelligent applications is protecting user
a variety of ways, the majority
privacy and preventing attacks, including spoofing,
of which make use of common cryptographic protocols.
denial of service (DoS), jamming, and However, current solutions
eavesdropping. The author looks into the flaws in IoT cannot address the security issues that arise with IoT
systems, as well as possible ways networks and are exacerbated by the
to secure IoT networks using machine learning distinctive characteristics of IoT nodes. By implementing
techniques, such as supervised learning, machine learning (ML) and deep
unsupervised learning, and reinforcement learning learning into IoT devices and networks, many threats to the
(RL). The analysis of data privacy security of the Internet of
focuses on ML-based approaches for authenticating IoT Things (IoT) can be stopped.
devices, controlling access to such
devices, offloading data securely, and identifying The IoT is a way to develop intelligent environments,
viruses. Future IoT adoption will have a including smart cities, healthcare
significant effect on society, business, and the systems, and building management systems. This is
economy. Because the majority of nodes in an because of recent improvements. It
also shows how major IoT applications can affect the
economy and the market share they
are projected to control by 2025 [24]. Figure 3 shows
the total number of connected devices
with the IoT.

A robust machine-learning-based IoT security system

must consider the IoT cyber
threat environment. Security features must therefore
be designed and refined. A
data-conditionality-reduction technique is essential
because security features and the IoT
Figure 3. Total connected IoT devices and global IoT market data they are associated with have a direct impact on
so far and future prediction [25].
machine-learning-based security
models [40]. “Feature engineering” refers to the
The goal of these smart environments, which
process of developing and refining security
significantly impact business, society,
features. This term is used when discussing the
and the economy, is to offer services based on IoT-
development of security models using
enabled sensor data and clever methods.
machine learning. It may be challenging to accurately
According to Navigant Research, the market for
classify cyber dangers given the
splitting city services will grow froM
potential for irrelevant data in today’s IoT security
93.5 billion USD in 2017 to 225.5 billion USD by 2026.
datasets. If you use this kind of security
Figure 4 shows the economic impact
model to make predictions, you could run into
of IoT applications. The amount of available
problems including high processing costs,
bandwidth, the number of users and smart
excessive variation, the need to build the model, and a
objects in IoT networks, the ability to effectively
lack of generalization. Therefore, if
manage large datasets, and the availability
an IoT security model with high-dimensional datasets
of scalable computing infrastructures, such as the
comprise the right number of security
cloud, are just a few of the factors that
features based on their impact or significance, it might
affect the quality of services offered by IoE applications
be less stressful to design [39].
in creative environments, such as
However, the Internet of Things raises security
intelligent cities [26].
concerns because there are so many
As the digital world expands, both home appliances
devices that can communicate to each other using
and industrial machines are
different protocols. Internet of Things
becoming more intelligent. Security and privacy
devices cannot be made safer because they do not
procedures that are effective in traditional
have enough processing power. This
networks could not be effective in the IoT. IoT
means that the ways we currently protect IoT networks
connections’ versatility causes new security
need to be greatly improved. In
issues. We list a few examples below.
the last few years, security studies have paid a lot more
To enable the creation of reducing applications that
attention to machine learning
could enhance people’s lives, the
models. There may be a need for security for IoT
IoT aims to connect a sizable number of disparate
systems because these devices regularly
devices. IoT devices come in various
produce huge amounts of data that can be used to
sizes and designs, requiring specific hardware and
train machine learning algorithms [40].
software solutions. The IoT connects
New product components are developed using feature
billions of intelligent devices to real-world data in a
selection and principal component
way that has never been carried out
analysis, which together account for the majority of the
before, regarding volume, speed, and organization
significant data. These new brand
[27].
elements could be useful for creating a machine-
The limitations of IoT devices and the dynamic and
learning-based IoT security model [41].
complex nature of the environment
Table 3 shows the dataset used for cybersecurity.
in which they operate exacerbate many of these Table 3. Datasets in the domain of cybersecurity.
concerns beyond the reach of standard Table 3. Datasets in the domain of cybersecurity.
security capabilities, even though many Internet access Datasets References Datasets References
points share the majority of these NSL-KDD [42] Enron Spam [43]
problems [28] (as shown in Table 2). UNSW-NB15 [44] Spam Assassin [45]
DARPA [46] Ling Spam [47]
C.A.I.D.A. [48] D.G.A. [49] making data security crucial in industries such as
ISOT’10 [50] Malware Genome project [51] healthcare. Numerous IoT applications
ISCX’12 [52] Virus Share [53] must make intelligent decisions in real time based on
CTU-13 [54] Virus Total [55]
C.I.C.I.D.S. [56] Comodo [57]
the preferences of the user [62].
Datasets References Datasets References Future-generation wireless networks must be reliable
NSL-KDD [42] Enron Spam [43] and self-sufficient. The individual’s
UNSW-NB15 [44] Spam Assassin [45] use of technology in their daily lives is changing as a
DARPA [46] Ling Spam [47] result of the IoT. Machine
C.A.I.D [48] D.G.A. [49] learning techniques are used by the Internet of Things
ISOT’10 [50] Malware Genome project [51]
to increase the effectiveness and independence
ISCX’12 [52] Virus Share [53]
of the network. Deep learning (DL) is a computationally
CTU-13 [54] Virus Total [55]
C.I.C.I.D.S. [56] Comodo [57] costly and challenging
machine learning (ML) technique. It is difficult to come
As the internet revolution continues, an increasing up with strategies for combining
number of everyday objects and deep learning technologies with IoT infrastructure to
industrial tools begin to function as “smart” devices. enhance the general performance of
Traditional data security and protection IoT applications. A range of methods that achieve a
techniques are unlikely to work on IoT networks. The balance between computing costs and
addition of new services performance are needed for the next generation of IoT
to IoT networks introduces new security flaws. The goal networks [63]. Machine learning
of the IoT is to connect a wide techniques have quickly advanced, and they are
network of various devices so that clamping software presently used in a wide range of academic
can be used to significantly improve advancements [64]. For instance, they are carefully
people’s lives. IoT devices come in a variety of shapes evaluated in a variety of sectors, including
and sizes, and they can perform a the cement business. Although cement enterprises in
wide range of functions, necessitating the use of a wide developing countries make a
range of hardware and software. significant amount of money through the sale of
A network of billions of connected computers makes valuable resources, they still face a number
up the IoT [58]. It also refers to the of difficulties. Optimization in machine learning has
vast amount, rapid rate of change, and organization of grown to be a significant topic of
data derived from the real world. study in recent years. Using the FDH model, the set of
The term “IoT” describes a network of devices capable production possibilities can be built
of two-way data communication. in any way [65]. An innovative three-layer data-mining
As a result, any time and any place can be connected to filtering pre-process for clustering
an IoT network [59]. Theft of techniques has been suggested by experts. It makes
cookies, cross-site scripting, structured query language use of machine learning to increase
injection, session hijacking, and accuracy and filter out irrelevant features and data.
distributed denial of service attacks are all possible on These stages of preparation were
connected IoT devices. DDoS designed to reduce redundant information and
assaults are especially dangerous for large, self- improve precision. Finally, we are aware of
managed IoT networks [1]. IoT devices the top business, best performance model and the
are temporary; thus, network configuration needs to most precise algorithm. The FDH model
be dynamic and flexible. Utilizing consistently performs at the highest possible degree of
nearby devices, ad hoc networks can make efficiency when compared to other
communication over shorter distances easier. suggested models [66]. Out of the three suggested
Proximity is described as how an IoT-enabled object filtering techniques, only the k-means
responds and acts in relation to its algorithm consistently yields the best results. Second
actual surroundings [60]. Networks for industrial IoT and third place, respectively, went to
encounter many difficulties. It is the model’s BCC and CCR. One of the most widespread
critical to have wireless connections that are speedy technologies in modern society is
and reliable. Applications that call the Internet of Things, which has a significant impact
on people’s personal, professional,
for low latency and high reliability connections include
and financial lives. There is a lot of hope that the
tracking, surgical equipment, and
Internet of Things, both now and in the
production on a production line [17]. An IoT device is a
future, will enhance people’s lives in a variety of
piece of hardware with a sensor
environments, from urban infrastructure
that can send information to a remote location over
to classrooms [67]. Automation, consumer comfort,
the Internet. A complicated system
and productivity have all risen as
must be built with the least amount of time, money,
a result of these developments. Yet, threats and
and effort possible because there are so
assaults have a big impact on the way
many sensors involved in its operation [61]. Patient
intelligent Internet of Things applications perform. The
information is sensitive and valuable,
quantity and complexity of threats
to the Internet of Things have increased, and and final outputs in the washbasin while performing
conventional approaches for protecting it complete static analysis. Both the
have not been able to keep up [68]. The security general SmartThings market and our specially created
system of the Internet of Things of the IOTBENCH application corpus were
future must be dynamically updated so it is up to date used to evaluate SAINT. In order to establish the value
for it to operate effectively. Artificial of SAINT and understand how the
intelligence (AI), in particular machine learning and market normally functions, initial research focused on
deep learning techniques, are required the SmartThings sector [71]. The
to make this viable. The author of [69] contrasted second analysis used the IOTBENCH app corpus from
various approaches in order to identify the first one. Our analysis revealed
the most effective one. We showed that this might be that the great majority of currently accessible apps
carried out interactively and how convey sensitive data, and that our
the model could be solved by switching the GDEA dual system is capable of detecting taint sources and sinks.
model to the MOLP. To solve the The outcomes of these tests also
GDEA and identify the MPS within the bounds of each showed that our technology is able to identify the
DMU’s efficiency, one may use this origin and final destination of contamination.
link as the foundation for an interactive MOLP This paper’s main focus is on architectural difficulties
technique. By fusing the STEM and DM because they are the root
methodologies, the GDEA dual model was able to cause of IoT’s poor performance and utility [72]. There
demonstrate the preferences of the DM. are many problems and reasons to
In institutions for stroke care, the max-ordering be worried. Communication, data management, zero-
method was applied to investigate the relationship entropy systems, scalability, massive
between the GDEA dual model and the MOLP [67], data collection, real-time data processing, security and
which is a practical approach privacy, interoperability, a lack of
to securing IoT devices is machine learning. One of the standardization, etc., are just a few of the problems
most advanced AI techniques, that need to be solved. There were
machine learning, performs effectively in massively 20 billion connected things in 2014, and it was
networked environments without anticipated that this number would increase
explicit programming. The system may be trained to to 30 billion by 2020. These connections can be used in
recognize and respond to various countless ways. The devices may
threats using machine learning techniques [13]. In this have features in common, but they are made by
scenario, the majority of attacks different companies and run on different
might be stopped early on. Additionally, it appears that operating systems. Hadoop has trouble dealing with
ML approaches may be useful for data sources that might carry out
spotting new threats and putting strategic defenses in comparable operations but have wildly dissimilar data
place. Machine learning algorithms formats [71]. This lack of consistent
may be employed in the future to create security
standardization is summarized by the phrase “The
standards for IoT devices, making them
Internet of Things May Never Speak a
more dependable and user-friendly than they are now
Single Language”. The lack of standardized protocols is
[25]. IDS’s effectiveness has led to a
now the greatest challenge in the
rise in popularity in recent years. Identification of
path of the Internet of Things, according to a recent
people who do not belong in a particular
survey by Light Reading. This barrier
location is the main purpose of an IDS [70]. Every host
needs to be removed because it prevents the growth of
that tries to join the Internet of
IoT interoperability. Technology
Things without authorization is considered an invader.
progress, data standards, and wireless protocols have
IDS has not been studied enough.
all been covered. Companies regularly
IDS on the IoT uses ML/DL in a variety of ways.
create their own standards, which leads to
Nonetheless, it struggles to deal with
incompatible technology [73]. One of the
difficult problems. In addition, you can only apply these
most important elements affecting people’s daily lives
tactics for select types of blows,
and well-being at work is “worker
and they are not extremely accurate [40]. Right now,
safety”. Studies that have been published in scholarly
one of the biggest problems with
journals have shown that knowing
the Internet of Things is that we do not fully
that they are working in an environment where they
comprehend how apps use data. This study
are less likely to be in an accident
introduces SAINT, a novel static taint analysis tool that
improves employees’ emotions and well-being. It is
locates weak data flows in IoT
crucial that all workplaces have proper
programmers. SAINT transforms the source code of an
safety precautions for their employees and operators,
Internet of Things application into
even though the industrial sector
a lifecycle model. The access points, user inputs,
is the most dangerous for workers. No matter how
events, and actions of the program are
frequent or unusual a job may be, it
represented by this model. We then watch the
must always be protected in order to safeguard the
information flow between sensitive inputs
workers’ health and safety. There are no
published solutions that can also monitor and advise respectively, despite the high values of R2 and Adj. It
people during unusual or dangerous was discovered that the vibration
jobs, even if a range of technologies already meet signal’s R2 value was 0.997%. The findings show that
these needs during “normal” operations the stacked LSTM model outperforms
(e.g., maintenance). The Internet of Things and other other LSTM models in predicting tool wear [76].
real-time applications and services,
3. Methods and Materials
such as video surveillance systems, are growing quickly,
3.2. Exclusion and Inclusion
showing the growing importance
The IoT and machine learning approaches were used as
of technology in our daily lives. The Internet of Things
a keyword string to find
and Industry 4.0 could help identify
publications in databases from the IEEE, Springer,
maintenance problems that have been noticed but not
Scopus, Google Scholar, A.C.M., Science
resolved. Fog devices are now
Direct, andWiley. These works include research on
processing a sizable percentage of IoT application
machine learning categorization, IoT
processing thanks to the development
security, and the integration of health systems. Papers
of fog computing [74]. However, if fog nodes are
that were first chosen for review
underpowered, the device’s reliability
were peer-reviewed before being published. To better
may suffer and IoT apps will not be able to function.
understand how machine learning
Many clear issues with read/write
works and how it might be used to improve IoT
operations and unsafe edge settings must be
security, this research explored publications
addressed. Scalable fault-predictive proactive
that concentrate on machine-learning-based
Brain Sci. 2023, 13, 683 11 of 30
techniques are necessary to improve dependability. approaches. After the initial search, any papers
These algorithms should be capable of found were discarded. We only looked at a few articles
determining whether fog machines are not powered because the review aimed to set
enough to work. The use of a recurrent standards for machine learning research criteria and
neural network to predict proactive problems in fog methodology. The committee did not
devices when there are not enough even read the additional recommendations.
resources is suggested in this research. The method Study Participants
makes use of a new rule-based network
policy for computing, memory, and power, as well as
3.4.1. IoT Attacks on Surface Areas
an entirely theoretical long short-term
We look at several possible attack paths for IoT
memory. An LSTM network is used in the planned CRP
systems and applications in the
to ascertain why the project failed
following sections. There are the following applications
due to a lack of finance. The proposed conceptual
in particular: One of the most
design also includes fault monitors
common entry methods for hackers is through IoT
and failure detectors. They guard against fog nodes
devices. Memory, firmware, physical
failing to provide services to IoT
interfaces, web interfaces, and network resources are
applications. The accuracy of predictions on training
only a few of the IoT systems’ many
data was 95.16 percent and on testing
weak points. Hackers may obtain access through faulty
data, it was 98.69 percent when LSTM and the CRP
parts, vulnerable update systems,
network policy technique were coupled.
and dangerous factory settings, to name a few. IoT
Prior to this, machine learning and deep learning
devices may be attacked through the
techniques were incomparable. This
communication channels they use [77]. The protocols
study uses vibration and acoustic emission sensor data
used by IoT systems may not be
to produce analyzable scalograms.
secure, which would put the plan in danger. IoT devices
To identify whether wavelet functions were useful, we
are vulnerable to network threats,
used the RWE criterion. Further
including spoofing and denial of service. Security flaws
Sin GAN scalograms were produced, and a number of
in web applications and other
picture quality metrics were then
IoT device software could provide unauthorized users
retrieved and used to build feature vectors [75]. The
access to the system. For instance,
experimental data required to train the
hackers might spread malicious firmware upgrades or
LSTM model used to predict tool wear were
steal user credentials using web
insufficient. The feature vector was used to applications [78].
train the bidirectional, stacked, and vanilla LSTM 3.4.2. Architectures and Security Concerns
models. We looked at five performance To highlight the security issues that affect the overall
indicators, including root-mean-square error, mean architecture of the IoT system,
square error, mean absolute error, and we summarize the IoT attack surface parts in this
adjusted root-mean-square error to assess how section. Different IoT concepts have
effectively LSTM models can predict tool been created by several academics and think tanks. A
wear. The MAE, RMSE, and MSE were the lowest, with typical IoT design has three levels:
values of 0.005, 0.016, and 0.0002,
perception, network, and application. However, it turns snooping, allows an attacker to intercept and look
out that the support or middleware through data being exchanged between
layer levels are vital because they must process data two devices [85], as shown in Table 6. A replay attack
and draw wise conclusions [79]. A in an IoT network could be defined
design for the IoT may contain a network layer and a as repeatedly falsifying, changing, or reusing the
support layer depending on its identities of related items. If an attacker
planned use. Many academic studies have also looked has the required time and data encryption keys, they
at how cloud computing might be can execute a timing attack. There are
used for the back-end architecture of the IoT [80]. a lot more ways than just direct node attacks for vital
Figure 7 shows the security challenges information to circulate [86].
of IoT. Table 6. Attacks and countermeasures on physical layer.

4.1.1. Issues with Networking and Data

Communications Layer Security
The main goals of this layer are compatibility, privacy,
and secrecy. At this layer,
it is expected that criminal activities, including
phishing, distributed denial-of-service
Figure 7. Security Challenges of IoT [81,82]. attacks, attacks on data transit, routing attacks,
4. Results identity authentication, and encryption,
4.1. Security Issues in the Perception or Sensing Layer will occur [87]. This layer of the IoT is especially
A conventional IoT design consists of three layers: the vulnerable to phishing attacks, which aim
application layer, the network to obtain sensitive information such as passwords and
layer, and the perception layer [83]. However, the login credentials. When an attacker
support or middleware layer between or unauthorized user gains access to the IoT network
the network and application layers becomes more while IoT apps gather and transfer
important as the significance of data sensitive data, this is characterized as an access attack,
processing and intelligent decision making rises. also known as a continuous advanced
Multiple layers, including a network layer threat. Table 7 shows the attack and countermeasures
and a support layer, may be present in IoT systems. on the data communication layer.
Cloud computing has been used as the Table 7. The attack and countermeasures on data
communication layer.
underlying support layer in numerous studies of IoT
systems.
Various sensors and other devices make up the
perception layer, sometimes called the
sensing layer. This layer’s storage, processing, memory,
and communication capabilities are
limited. The main methods this layer secures in the IoT
network are node authentication,
The most frequent and harmful kinds of network
weak encryption, and access control [84]. Attacks and
attacks are DoS and DDoS attacks.
crimes against the perceiving layer’s
They use up network resources and compromise the
privacy are too common in the real world. One
operation of services. Malicious actors
approach to conduct this is to take control
can also change routing channels’ routes when
of a node. Malicious code usage, data injection, replay
transmitting data by routing attacks, such
assaults, and side-channel attacks as holes and worms [88].
are other techniques. For example, if an attacker takes
over a node, it will stop sending 4.1.2. Security Issues in the Middleware or Support
Layer
Distributed computing solutions have been used to
valid network data and may even stop using the IoT replace centralized cloud environments
security program. It is possible that in a variety of cases, with good results in terms of
the IoT application will not operate as planned if it performance and response time.
receives terrible data or is compromised All sent data should now be checked for accuracy,
by malicious code injection. A technique called concision, and secrecy.
eavesdropping, also called sniffing or When someone inside a network purposefully alters or
steals data or information, this
is known as a malicious inside attack [89]. By inserting
malicious SQL queries into the code,
SQL injection attacks are used to steal data from user
services in the real world. When
damage to one virtual machine spreads to another, this
is a virtualization attack. With the
help of cloud malware injection, a hacker can take over
4.1.3. Application Layer
a cloud service, install malicious
Defining and maintaining IoT applications, including
code, or even create a fake virtual machine. There
their interactions with specific
could be significant consequences if
clients, fall under the scope of the application layer.
attacks are so powerful that cloud infrastructure is
One way to use IoT services is through a
incredibly frustrated [90]. Table 8 shows
user interface. A computer, a smartphone, or any other
the attack and countermeasures on the support layer.
Table 8. The attack and countermeasures on support layer. Internet-enabled smart device could
serve as an interface. The data that the middleware
layer process is used by the application
layer [91]. This holds for a wide range of application
categories, including applications
for smart homes, smart cities, industry, construction,
and health. The security needs of an
application may change depending on how it functions.
When sending information on
climate change forecasts as opposed to when
4.1.3. Application Layer conducting online banking, it is acceptable
Defining and maintaining IoT applications, including to expect a better level of security. The application
their interactions with specific layer must address various security
clients, fall under the scope of the application layer. challenges, such as attacks on access control, malicious
One way to use IoT services is through a code, programming, data leaks,
user interface. A computer, a smartphone, or any other service interruptions, application vulnerabilities, and
Internet-enabled smart device could software flaws [92]. Table 9 shows
serve as an interface. The data that the middleware the attack and countermeasures on the application
layer process is used by the application layer.
layer [91]. This holds for a wide range of application Attacks that interrupt service, commonly referred to as
categories, including applications “Distributed Denial of Service
for smart homes, smart cities, industry, construction, (DoS)” attacks, stop users from using IoT apps by
and health. The security needs of an sending a flood of requests to servers or
application may change depending on how it functions. networks. Threat actors could use sniffer software to
When sending information on monitor data being transmitted by
climate change forecasts as opposed to when IoT apps. Attacks that gain unauthorized access can
conducting online banking, it is acceptable seriously harm a system quickly by
to expect a better level of security. The application preventing users from using IoT-related services and
layer must address various security wiping data [93].
challenges, such as attacks on access control, malicious
Table 9. The attack and countermeasures on the application
code, programming, data leaks, layer
service interruptions, application vulnerabilities, and
software flaws [92]. Table 9 shows
the attack and countermeasures on the application
layer.
Attacks that interrupt service, commonly referred to as
“Distributed Denial of Service
(DoS)” attacks, stop users from using IoT apps by
sending a flood of requests to servers or
networks. Threat actors could use sniffer software to Each layer of an IoT system may be vulnerable to
monitor data being transmitted by different security flaws and attacks,
IoT apps. Attacks that gain unauthorized access can as was already mentioned. Furthermore, there is a
seriously harm a system quickly by severe risk of unknown vulnerabilities.
preventing users from using IoT-related services and One must conduct a thorough investigation to find
wiping data [93]. these hacks. Understanding artificial
intelligence, especially machine learning and deep
learning architectures and techniques, is
an effective way to safeguard the system regarding IoT
security. Figure 8 shows the layers
and function of IoT architecture.
 model and a linear-regression-based strategy for
spotting malicious IoT
malicious nodes [108].
Regression modeling, on the other hand, can be used
to predict attacks or measure the
severity of one. Worms, viruses, and another harmful
software fall under this category [109].
Regression techniques, network packet characteristics,
and quantitative security models
that examine phishing over a specified period are
Figure 8. Layered Architecture [66]. examples of relevant models, as illustrated
in Table 11.
4.2. IoT Security Solutions Based on ML and DL Table 11. Summary of Classification and Regression
IoT devices can use AI technologies, such as Methods.
machine learning and deep learning, to
act correctly after learning from the data they
gather. It is feasible to detect significant
security event trends in IoT data using learning
models, which frequently include rules,
procedures, or complex “transfer functions” [94].
This enables DL and ML, which are entirely
different, to function in real time over
IoT networks. This shows how data-driven IoT
security intelligence models could be created
using ML and DL. IoT security data can be used to iv. Any widely used R.T., such as linear, logistic,
learn new things via classification polynomial, and partial least-squares
and regression analysis, clustering, rule-based regression, can be used to build the quantitative
techniques, feature optimization, and DL security model. For instance, multiple
with ANN, such as the M.N.L.P.N., C.N., and regression analysis can create a correlation between
recurrent networks [95]. The following section human characteristics and how
covers the use of ML and DL to increase the people desire to act in terms of cybersecurity [110].
4.2.2. Clustering Techniques
security of IoT products. A machinelearning-
Clustering is a standard method of unsupervised
based IoT security architecture is shown in F
learning used in machine learning
to analyze IoT security data. It may group or cluster
data points based on similarity or
dissimilarity metrics of security data from IoT devices
from various sources. As a result,
clustering might make finding hidden patterns and
structures in data easier, making it
simpler to spot anomalies or attacks in the IoT. Various
perspectives, such as partitioning,
hierarchies, fuzzy theory, distribution, and grids, can be
used to cluster data. Many wellknown
Figure 9. A machine learning security framework for methods for classifying data include k-means, K-
IoT systems [96]. medoids, and the Gaussian mixture
model [111]. These clustering methods could be used
to fix several IoT issues as illustrated
in Table 12. An example of an algorithm used to profile
The uses of such techniques are presented as follows: unusual IoT device behavior is the
i. The SVM classification approach looks for unusual k-means algorithm, which is one method that can be
behavior in IoT devices and used to find outliers or noisy events
malware on Android to assure the dependability of IoT is a dynamic threshold-based approach. Fuzzy
services [106]. clustering is frequently used to find IoT
ii. Anomalies, denial-of-service assaults, IoT intrusions, intrusions [112].
and irregularities in smart Table 12. Summary of Clustering Techniques.
cities are all detected using the random forest
approach [107].
iii. Two other methods for detecting abnormalities
include a Naive-Bayes-based classification
 with many security attributes evaluated according to
how important or relevant they are
may make it easier to create an IoT security model
[102]. Existing approaches include
the correlation coefficient, the chi-squared test, and
Cybersecurity applications can more effectively find analysis of variance. Techniques for
helpful information or intelligence embedding information include regularization, Lasso,
in system log data by clustering. Clustering algorithms Ridge, Elastic Net, and tree-based
may be very helpful in resolving IoT feature importance [84]. Using feature selection and
security issues, such as identifying outliers, anomalies, principal component analysis, it is
signatures, fraud, and cyberattacks, possible to create new brand components that explain
by exposing previously hidden patterns and structures the most important data. As part of
in IoT security data [113]. machine-learning-based security modeling, these
4.2.3. Rule-Based Techniques enhanced signature properties may make
Older patterns are less likely to stand out and aid in the it easier to manage large amounts of IoT security data,
identification or prediction of such as identifying anomalies in IoT
IoT security issues than newer unfriendly behavior network traffic [103].
patterns. Selectivity analysis, which 4.2.5. Multi-Layer Perceptron (MLP)
examines current practices, may be more beneficial in Deep learning usually uses the multi-layer MLP, FFAN.
some cases than conventional data The input layer, the hidden
analysis. Another critical goal is to develop a security output layers, and the actual output layer are the three
model for IoT devices that is based layers that make up the traditional
on how recently they have been used. Innovative, M.L.P. design. An AI network links each node in a layer
portable IoT device solutions that take to a specific value in the layer
new data trends into account are required as part of below it. In the end, this number is associated with the
our learning-based research on IoT layer below it. As the model
security [114]. is being built, MLP employs backpropagation to adjust
By creating various links and patterns based on support the internal weight values [117].
and confidence values, rulebased This M.L.P. network is used to analyze the NSL-KDD
procedures are easy to use and complicate the model. dataset’s malware, explain the
The problem might be lessened IoT parameters, detect malicious traffic coming from
with a robust association model. A rule-learning IoT devices, and create a model for
technique that can be used to find intrusion detection [118]. The idea divides network
trustworthy, non-redundant links between ideas is data into secure data and unsecure data.
shown in our earlier work [115]. Policy 4.2.6. Recurrent Neural Network (RNN)
rules in a plan define which network usage is allowed Another variety of artificial neural networks is the
and which is not. Even cyberattacks recurrent neural network. A
with no known vulnerabilities can be stopped by directed graph representing time is constructed from
the connections between the nodes.
security policy monitoring filters and
In the R.N.N. model, neural feed-forward networks are
protections based on rules [116].
used. It looks at its internal
4.2.4. Optimization of Security Features and Principal
state, or memory, to determine how long different
Component Analysis
input sequences last. IoT security,
In the current cyber threat environment, the
natural language processing, and speech recognition
development and optimization of security
can all benefit from the RNN model’s
features are significant barriers to the success of an
capabilities to manage sequential data effectively
ML-based IoT security solution. Security
[119]. IoT devices that are connected
characteristics and IoT data have a direct impact on
provide a lot of sequential data, including information
ML-based security models, necessitating
that changes over time and network
the use of a data-dimensionality-reduction technique.
traffic flows. Recurrent connections in neural networks
“Feature engineering” is the process
can uncover potential defense
of establishing and changing security features or
vulnerabilities when a threat’s communication patterns
variables so that machine-learning-based
change over time. This is because
security models work properly. Today’s IoT security
it has a powerful model for predicting time series
datasets may contain unused or
because of its long short-term Memory,
irrelevant data, making simulation of cyberattacks and
which allows it to remember what it has been told in
other challenges difficult [101]. The
the past. For example, it is possible
forecasting accuracy of a security model can be
to identify and categorize dangerous applications and
harmed by extreme variation, overfitting,
detect intrusions using an L.S.T.M.-
expensive processing, and time-consuming model
model-based recurrent network [120]. It can also be
setup [93]. A high-dimensional dataset
used for further security-related tasks.
The detection and prevention of malware, spoofing, in order to solve the problem of big data [142].
and computer virus attacks across 4.3.5. Security
a wide range of IoT devices can be made using a variety Information security implementation in the IoT is
of deep learning models and challenging. Users communicate
hybrid network models [121]. One type of deep private data to complete tasks. There are various
learning model that could be used to possible opponents for user privacy.
protect IoT devices is a DBN-based security model Therefore, security measures should be implemented
[122]. The authors looked at multiple to safeguard user data and discourage
approaches to in-depth learning. Additionally, they unauthorized access [143].
were referred to as unique features for 4.3.6. Storage
jobs requiring human help and generative for those IoT devices must also be secure to use. Sensors keep an
requiring none. Additionally, hybrid eye on the surroundings and
systems may be used if the data quality calls for it send the information they gather to computers.
[123]. Data-driven security analytics Because there is no encounter measurement,
in the context of the IoT can, therefore, greatly benefit the security of data storage devices cannot be
from the above machine learning guaranteed. As a result, there needs to be a
or deep learning methodologies, way to stop unauthorized access to or monitoring of
4.3.1. Poor Management sensitive data [144].
Systems based on the IoT are having trouble because 4.3.7. Authentication and Authorization
of poor management. The User IDs can be verified using several different
problem is that most of the time, software engineers techniques. The most common approach
try to figure out how to extract useful is to use a login and password, but there are other
data from sensors [138]. They do not care how data are options as well, such as an access
gathered, just that it is. It is easier card, retina scan, voice recognition, or fingerprints.
for attackers to hack a system and steal sensitive user Authorization can also be obtained
data when there is no guarantee. through access control. It is a method of protecting a
Developers must start concentrating on data system by only allowing those who
acquisition as a result [139]. need access to use it. The system has become complex
4.3.2. Naming and Identity Management because it consists of so many nodes
To communicate with other components of a network, and components. The traditional methods of
each component needs to have authentication and permission have failed
its own identity. Therefore, a technique for dynamically in large-scale networks. Although concerns with
identifying each network node authentication and authorization have
with a special identification must exist [140]. When the been researched, they still need to be fixed. To solve
IoT first started, IPv4 was used to these challenges, such an approach is
give each networked device a special identifier. necessary [145].
Because the number of Internet of Things 4.3.8. Secure Network
devices is increasing, IPv6 is used to give each one a Man-in-the-middle and denial-of-service attacks are
distinct name. only two examples of the multiple
4.3.3. Trust Management and Policy ways the transport layer of a network can be used. An
The idea of trust is important and complicated. It is attack that prevents user’s access to
also necessary to have scalability, the targeted system, device, or network resource is
dependability, strength, and availability. It goes above known as a denial-of-service attack [146].
taking safety procedures. IoT A cyberattack known as “man-in-the-middle” occurs
apps ask their users for sensitive information with their when an attacker pretends to be a
permission. Therefore, a privacy third party and transmits and detects messages
guarantee is necessary. User data are protected and between two objectives who believe they
cannot be accessed without permission. are speaking directly to one another. Therefore, a set
Academics have suggested a range of strategies for of protections must be put in place to
improving both trust and privacy in guarantee the security of the network layer [147].
scholarly writings. These strategies for protecting trust Therefore, it is challenging to create new, lightweight
and privacy in IoT applications algorithms or procedures for
have been ineffective. These issues are currently at the IoT devices without first weighing the advantages and
forefront of research on the Internet disadvantages of current teaching
of Things as a result [141]. techniques [148].
4.3.4. Big Data Older patterns are less likely to stand out and aid in the
Currently, billions of devices are connected to the web, identification or prediction of
forming what is known as the IoT security issues than newer unfriendly behavior
IoT. Huge volumes of information are being generated patterns. Selectivity analysis, which
by these devices. IoT struggles with examines current practices, may be more beneficial in
the transmission and processing of massive datasets. some cases than conventional data
Therefore, such a system is essential
analysis [136]. Another critical goal is to develop a other academics and practitioners find and implement
security model for IoT devices that is IoT security solutions in the future.
based on how recently they have been used.
Innovative, portable IoT device solutions that
take new data trends into account are required. As part
of our learning-based research on
IoT security, we examined and evaluated the above
study directions [149]. The security
of the IoT can be improved by including context-aware
computing; “context awareness”
is a term used frequently in IoT computing to describe
a system’s capacity to take in
information about its surroundings and modify its
behavior accordingly [150].
As a result, using chronological, geographical,
individual, dependence, activity, the
relationship between events or exchanges, and other
contextual security data, it is possible
to determine whether suspicious behavior occurs
[151]. For example, a user may be able to
connect to the network in the office but not when
using publicWi-Fi. One area that could
be investigated is how to create IoT security solutions
that work in different contexts and
adapt to them [152].

5. Conclusions
This research provides a comprehensive review of the
literature on IoT security awareness.
IoT model, IoT-based intelligent environments, and
associated security challenges
are some of the topics highlighted by machine learning
solutions. In this work, we evaluated
the knowledge base on IoT security intelligence. We
investigated the IoT paradigm,
IoT-based smart environments, security issues, and
machine learning solutions to these
problems. Identifying and protecting IoT devices and
systems necessitates a thorough
examination of IoT system architectures, as well as the
cyberattacks that can break them
down layer by layer. We investigated how various
machine learning and deep learning
technologies could be used to improve IoT security. If
IoT security is to be effective, it
must be built on machine learning or deep learning
models that use data attributes. Before
it can assist in making intelligent decisions, the system
must have an effective learning
algorithm based on the IoT security knowledge
acquired and the application for which
it is used. We also talked about potential directions and
approaches for future research
on teaching and learning. Because of these issues, References
there is room for new research in the 1. Hassija, V.; Chamola, V.; Saxena, V.; Jain, D.; Goyal, P.;
field, and this is where the opportunity to develop Sikdar, B. A Survey on IoT Security: Application Areas,
effective strategies for continuously Security Threats,
improving IoT security presents itself. We believe that and Solution Architectures. IEEE Access 2019, 7, 82721–
our research on machine-learningand 82743. [CrossRef]
deep-learning-based security solutions is a step in the 2. ´Slusarczyk, B. Industry 4.0: Are we ready? Pol. J. Manag.
right direction and will help Stud. 2018, 17, 232–248. [CrossRef]
3. Sarker, I.H.; Kayes, A.S.M.; Badsha, S.; Alqahtani, H.; of Things: Challenges, Fundamentals, and Enabling
Watters, P.; Ng, A. Cybersecurity data science: An overview Technologies. IEEE Internet Things J. 2019, 6, 7946–7970.
from machine [CrossRef]
learning perspective. J. Big Data 2020, 7, 41. [CrossRef] 18. Rawat, D.B.; Doku, R.; Garuba, M. Cybersecurity in Big
4. Minerva, R.; Biru, A.; Rotondi, D. Towards a definition of Data Era: From Securing Big Data to Data-Driven Security.
the Internet of Things (IoT). IEEE Internet Initiat. 2015, 1, 1– IEEE Trans.
86. Serv. Comput. 2019, 14, 2055–2072. [CrossRef]
5. Khan, M.A.; Salah, K. IoT security: Review, blockchain 19. Farrokhi, A.; Farahbakhsh, R.; Rezazadeh, J.; Minerva, R.
solutions, and open challenges. Future Gener. Comput. Syst. Application of Internet of Things and artificial intelligence for
2018, 82, smart
395–411. [CrossRef] fitness: A survey. Comput. Netw. 2021, 189, 107859.
6. Minoli, D.; Occhiogrosso, B. Blockchain mechanisms for IoT [CrossRef]
security. IOT 2018, 1, 1–13. [CrossRef] 20. Yahya, F.; Zaki, A.F.A.; Moung, E.G.; Sallehudin, H.; Bakar,
7. Zhang, Z.-K.; Cho, M.C.Y.; Wang, C.-W.; Hsu, C.-W.; Chen, N.A.A.; Utomo, R.G. An IoT-based Coastal Recreational
C.-K.; Shieh, S. IoT security: Ongoing challenges and research Suitability
opportunities. In Proceedings of the 2014 IEEE 7th System using Effective Messaging Protocol. Int. J. Adv.
International Conference on Service-Oriented Computing and Comput. Sci. Appl. 2021, 12, 8. [CrossRef]
Applications, 21. Routray, S.K.; Gopal, D.; Javali, A.; Sahoo, A. Narrowband
Matsue, Japan, 17–19 November 2014; pp. 230–234. IoT (NBIoT) Assisted Smart Grids. In Proceedings of the 2021
8. Noor, M.B.M.; Hassan,W.H. Current research on Internet International
of Things (IoT) security: A survey. Comput. Netw. 2019, 148, Conference on Artificial Intelligence and Smart Systems
283–294. (ICAIS), Coimbatore, India, 25–27March 2021; pp. 1454–
[CrossRef] 1458.
9. Neshenko, N.; Bou-Harb, E.; Crichigno, J.; Kaddoum, G.; 22. Sangra, P.; Rana, B.; Singh, Y. Energy efficiency in IoT-
Ghani, N. Demystifying IoT Security: An Exhaustive Survey on based smart healthcare. In Proceedings of Third International
IoT Conference on
Vulnerabilities and a First Empirical Look on Internet-Scale Computing, Communications, and Cyber-Security; Springer:
IoT Exploitations. IEEE Commun. Surv. Tutor. 2019, 21, 2702– Singapore, 2023; pp. 503–515.
2733. 23. Alshamrani, M. IoT and artificial intelligence
[CrossRef] implementations for remote healthcare monitoring systems:
10. Abbas, S.G.; Hashmat, F.; Shah, G.A. A multi-layer A survey. J. King Saud
industrial-IoT attack taxonomy: Layers, dimensions, Univ. Comput. Inf. Sci. 2021, 34, 4687–4701. [CrossRef]
techniques and 24. Kumar, Y.; Singla, R. Effectiveness of Machine and Deep
application. In Proceedings of the 2020 IEEE 19th Learning in IOT-Enabled Devices for Healthcare System. In
International Conference on Trust, Security and Privacy in Intelligent
Computing and Internet of Things for Healthcare and Industry; Springer:
Communications (TrustCom), Guangzhou, China, 29 Berlin/Heidelberg, Germany, 2022; pp. 1–19. [CrossRef]
December 2020–1 January 2021; pp. 1820–1825. 25. Tahsien, S.M.; Karimipour, H.; Spachos, P. Machine
11. Dalal, K.R. Analysing the role of supervised and learning based solutions for security of Internet of Things
unsupervised machine learning in iot. In Proceedings of the (IoT): A survey.
2020 International J. Netw. Comput. Appl. 2020, 161, 102630. [CrossRef]
Conference on Electronics and Sustainable Communication Brain Sci. 2023, 13, 683 26 of 30
Systems (ICESC), Coimbatore, India, 2–4 July 2020; pp. 75–79. 26. Kaur, J.; Jaskaran; Sindhwani, N.; Anand, R.; Pandey, D.
12. Hussain, F.; Hussain, R.; Hassan, S.A.; Hossain, E. Machine Implementation of IoT in Various Domains, in IoT Based
Learning in IoT Security: Current Solutions and Future Smart Applications;
Challenges. Springer: Berlin/Heidelberg, Germany, 2023; pp. 165–178.
IEEE Commun. Surv. Tutor. 2020, 22, 1686–1721. [CrossRef] [CrossRef]
13. Al-Garadi, M.A.; Mohamed, A.; Al-Ali, A.K.; Du, X.; Ali, I.; 27. Stracener, C.; Samelson, Q.; Mackie, J.; Ihaza, M. The
Guizani, M. A Survey of Machine and Deep Learning Methods Internet of Things Grows Artificial Intelligence and Data
for Sciences. IT Prof.
Internet of Things (IoT) Security. IEEE Commun. Surv. Tutor. 2019, 21, 55–62. [CrossRef]
2018, 22, 1646–1685. [CrossRef] 28. Chen, L.; Hu,W.; Jamieson, K.; Chen, X.; Fang, D.;
14. Zhou, W.; Jia, Y.; Peng, A.; Zhang, Y.; Liu, P. The Effect of Gummeson, J. Pushing the Physical Limits of {IoT} Devices
IoT New Features on Security and Privacy: New Threats, with Programmable
Existing Metasurfaces. In 18th USENIX Symposium on Networked
Solutions, and Challenges Yet to Be Solved. IEEE Internet Systems Design and Implementation (NSDI 21); NSDI: Boston,
Things J. 2018, 6, 1606–1616. [CrossRef] MA, USA,
15. Li, S.; Xu, L.D.; Zhao, S. The internet of things: A survey. 2021; pp. 425–438.
Inf. Syst. Front. 2015, 17, 243–259. [CrossRef] 29. Noura, M.; Atiquzzaman, M.; Gaedke, M. Interoperability
16. Asim, M.; Arif, M.; Rafiq, M. Applications of Internet of in Internet of Things: Taxonomies and Open Challenges. Mob.
Things in university libraries of Pakistan: An empirical Netw.
investigation. Appl. 2018, 24, 796–809. [CrossRef]
J. Acad. Libr. 2022, 48, 102613. [CrossRef] 30. Behrendt, F. Cycling the Smart and Sustainable City:
17. Ma, Z.; Xiao, M.; Xiao, Y.; Pang, Z.; Poor, H.V.; Vucetic, B. Analyzing EC Policy Documents on Internet of Things,
High-Reliability and Low-Latency Wireless Communication for Mobility and
Internet Transport, and Smart Cities. Sustainability 2019, 11, 763.
[CrossRef]
31. Kumar, S.; Tiwari, P.; Zymbler, M. Internet of Things is a 553–558. [CrossRef]
revolutionary approach for future technology enhancement: 43. Fernandes, B.H.P. Smart Bed IoT-Based Wireless Data
A review. Acquisition for Untethered Patients. Ph.D. Thesis,
J. Big Data 2019, 6, 1–21. [CrossRef] Universidade de
32. Li, Y.; Alqahtani, A.; Solaiman, E.; Perera, C.; Jayaraman, Coimbra, Coimbra, Portugal, 2022.
P.P.; Buyya, R.; Morgan, G.; Ranjan, R. IoT-CANE: A unified 44. Janarthanan, T.; Zargari, S. Feature selection in UNSW-
knowledge NB15 and KDDCUP’99 datasets. In Proceedings of the 2017
management system for data-centric Internet of Things IEEE 26th
application systems. J. Parallel Distrib. Comput. 2019, 131, International Symposium on Industrial Electronics (ISIE),
161–172. Edinburgh, UK, 19–21 June 2017; pp. 1881–1886.
[CrossRef] 45. Karn, R.R.; Kudva, P.; Elfadel, I.M. LearningWithout
33. Nawalagatti, A. IoT: A Boon for Advancement of Forgetting: A New Framework for Network Cyber Security
Technology. Int. J. Res. Appl. Sci. Eng. Technol. 2022, 10, 652– Threat Detection.
655. [CrossRef] IEEE Access 2021, 9, 137042–137062. [CrossRef]
34. Ja, S.; Dhasb, J.T.M.; Angelc, T.S. Proposed Novel 46. Gharaibeh, M.; Papadopoulos, C. Darpa-2009 Intrusion
Methodology for Automatic Drainage Block Identification in Detection Dataset Report. Tech. Rep. 2014. Available online:
Smart Cities Using https://www.isi.edu/~johnh/PAPERS/Gharaibeh14a.pdf
Internet of Things. In Advances in Parallel Computing (accessed on 31 December 2022).
Algorithms, Tools and Paradigms; IOS Press: Amsterdam, The 47. Ahmed, M.; Byreddy, S.; Nutakki, A.; Sikos, L.F.; Haskell-
Netherlands, 2022. Dowland, P. ECU-IoHT: A dataset for analyzing cyberattacks in
35. Muthulakshmi, S.; Chitra, R. IoT technologies, applications Internet
and challenges, blockchain and its role in IoT: A survey. Int. J. of Health Things. Ad. Hoc. Netw. 2021, 122, 102621.
Internet [CrossRef]
Technol. Secur. Trans. 2022, 12, 321–352. [CrossRef] 48. Moustafa, N. A new distributed architecture for
36. Singh, P.K.; Singh, S.; Usman, H.; Urooj, S. Recent evaluating AI-based security systems at the edge: Network
Advances and Future Trends of IoT-Based Devices. In Energy TON_IoT datasets.
Harvesting; Sustain. Cities Soc. 2021, 72, 102994. [CrossRef]
Chapman and Hall/CRC: Boca Raton, FL, USA, 2022; pp. 179– 49. Kilincer, I.F.; Ertam, F.; Sengur, A. Machine learning
203. methods for cyber security intrusion detection: Datasets and
37. Cˇ olakovic´, A.; Salihovic´, N.; Dželihodžic´, A. Adaptive comparative
Traffic Management Systems Based on the Internet of Things study. Comput. Netw. 2021, 188, 107840. [CrossRef]
(IoT). In 50. Yavanoglu, O.; Aydos, M. A review on cyber security
Proceedings of Advanced Technologies, Systems, and datasets for machine learning algorithms. In Proceedings of
Applications VII: Proceedings of the International Symposium the 2017 IEEE
on Innovative and International Conference on Big Data (Big Data), Boston, MA,
Interdisciplinary Applications of Advanced Technologies (IAT) USA, 11–14 December 2017; pp. 2186–2193.
2022; Springer International Publishing: Cham, Switzerland, 51. Amit, I.; Matherly, J.; Hewlett, W.; Xu, Z.; Meshi, Y.;
2022; Weinberger, Y. Machine learning in cyber-security-problems,
pp. 426–437. challenges and
38. Perwej, Y.; AbouGhaly, M.A.; Kerim, B.; Harb, H.A.M. An data sets. arXiv preprint 2018, arXiv:1812.07858.
Extended Review on Internet of Things (iot) and Its Promising Brain Sci. 2023, 13, 683 27 of 30
Applications; 52. Soheily-Khah, S.; Marteau, P.-F.; Bechet, N. Intrusion
Communications on Applied Electronics: New York, NY, USA, Detection in Network Systems Through Hybrid Supervised
2019; pp. 2394–4714. and Unsupervised
39. Mazhar, T.; Irfan, H.M.; Haq, I.; Ullah, I.; Ashraf, M.; Al Machine Learning Process: A Case Study on the ISCX Dataset.
Shloul, T.; Ghadi, Y.Y.; Imran; Elkamchouchi, D.H. Analysis of In Proceedings of the 2018 1st International Conference on
Challenges Data Intelligence and Security (ICDIS), South Padre Island, TX,
and Solutions of IoT in Smart Grids Using AI and Machine USA, 8–10 April 2018; pp. 219–226. [CrossRef]
Learning Techniques: A Review. Electronics 2023, 12, 242. 53. Apruzzese, G.; Colajanni, M.; Ferretti, L.; Guido, A.;
[CrossRef] Marchetti, M. On the effectiveness of machine and deep
40. Qureshi, A.; Qureshi, M.A.; Haider, H.A.; Khawaja, R. A learning for cyber
review on machine learning techniques for secure IoT security. In Proceedings of the 2018 10th international
networks. In conference on cyber Conflict (CyCon), Tallinn, Estonia, 29
Proceedings of the 2020 IEEE 23rd international multitopic May–1 June 2018;
conference (INMIC), Bahawalpur, Pakistan, 5–7 November pp. 371–390.
2020; 54. Putra, M.A.R.; Hostiadi, D.P.; Ahmad, T. Botnet dataset
pp. 1–6. with simultaneous attack activity. Data Brief 2022, 45,
41. Mazhar, T.; Malik, M.A.; Haq, I.; Rozeela, I.; Ullah, I.; Khan, 108628. [CrossRef]
M.A.; Adhikari, D.; Ben Othman, M.T.; Hamam, H. The Role of 55. Folino, G.; Pisani, F. Evolving meta-ensemble of classifiers
ML, for handling incomplete and unbalanced datasets in the
AI and 5G Technology in Smart Energy and Smart Building cyber security
Management. Electronics 2022, 11, 3960. [CrossRef] domain. Appl. Soft Comput. 2016, 47, 179–190. [CrossRef]
42. Meena, G.; Choudhary, R.R. A review paper on IDS 56. Kasim, Ö. A Robust DNS Flood Attack Detection with a
classification using KDD 99 and NSL KDD dataset in WEKA. In Hybrid Deeper Learning Model. Comput. Electr. Eng. 2022,
Proceedings 100, 107883.
of the 2017 International Conference on Computer, [CrossRef]
Communications and Electronics (Comptelix), Jaipur, India, 1– 57. Kent, A.D. Comprehensive,Multi-Source Cyber-Security
2 July 2017; pp. Events Data Set; Los Alamos National Lab. (LANL): Los
Alamos, NM, USA, 2015.
58. Xiao, L.;Wan, X.; Lu, X.; Zhang, Y.;Wu, D. IoT security 71. Khan, J.; Khan, J.; Ali, F.; Ullah, F.; Bacha, J.; Lee, S.
techniques based on machine learning: How do IoT devices Artificial intelligence and internet of things (AI-IoT)
use AI to technologies in response
enhance security? IEEE Signal Process. Mag. 2018, 35, 41–49. to COVID-19 pandemic: A systematic review. IEEE Access
[CrossRef] 2022, 10, 62613–62660. [CrossRef]
59. Ngo, Q.-D.; Nguyen, H.-T.; Le, V.-H.; Nguyen, D.-H. A 72. Ranaldi, L.; Pucci, G. Knowing Knowledge: Epistemological
survey of IoT malware and detection methods based on static Study of Knowledge in Transformers. Appl. Sci. 2023, 13, 677.
features. [CrossRef]
ICT Express 2020, 6, 280–286. [CrossRef] 73. Hasan, N.; Chen, Z.; Zhao, C.; Zhu, Y.; Liu, C. IoT Botnet
60. Sarker, I.H. Deep Cybersecurity: A Comprehensive Detection framework from Network Behavior based on
Overview from Neural Network and Deep Learning Extreme
Perspective. SN Comput. Learning Machine. In Proceedings of the IEEE INFOCOM
Sci. 2021, 2, 154. [CrossRef] 2022-IEEE Conference on Computer Communications
61. Kumar, K.D.; Sudhakara, M.; Poluru, R.K. Towards the Workshops
integration of blockchain and IoT for security challenges in (INFOCOM WKSHPS), New York, NY, USA, 2–5 May 2022; pp.
IoT: A review. 1–6.
Res. Anthol. Converg. Blockchain Internet Things Secur. 2023, 74. Montanaro, T.; Sergi, I.; Stefanizzi, I.; Landi, L.; Di Donato,
193–209. [CrossRef] L.; Patrono, L. IoT-Aware Architecture to Guarantee Safety of
62. Kumar, K.D.; Venkata Rathnam, T.; Venkata Ramana, R.; Maintenance Operators in Industrial Plants. Appl. Syst. Innov.
Sudhakara, M.; Poluru, R.K. Towards the Integration of 2023, 6, 46. [CrossRef]
Blockchain and 75. Haneef, S.; Venkataraman, N. Proactive Fault Prediction
IoT for Security Challenges in IoT. 2019. Available online: of Fog Devices Using LSTM-CRP Conceptual Framework for
https://www.igi-global.com/chapter/towards-the- IoT
integration-ofblockchain- Applications. Sensors 2023, 23, 2913.
and-iot-for-security-challenges-in-iot/310448 (accessed on 31 76. Shah, M.; Vakharia, V.; Chaudhari, R.; Vora, J.; Pimenov,
December 2022). D.Y.; Giasin, K. Tool wear prediction in face milling of stainless
63. Bin Zikria, Y.; Afzal, M.K.; Kim, S.W.; Marin, A.; Guizani, M. steel
Deep learning for intelligent IoT: Opportunities, challenges using singular generative adversarial network and LSTM deep
and learning models. Int. J. Adv. Manuf. Technol. 2022, 121, 723–
solutions. Comput. Commun. 2020, 164, 50–53. [CrossRef] 736.
64. Mirmozaffari, M.; Yazdani, M.; Boskabadi, A.; Dolatsara, [CrossRef]
H.A.; Kabirifar, K.; Golilarz, N.A. A Novel Machine Learning 77. Mendonça, R.V.; Silva, J.C.; Rosa, R.L.; Saadi, M.;
Approach Rodriguez, D.Z.; Farouk, A. A lightweight intelligent intrusion
Combined with Optimization Models for Eco-Efficiency detection system
Evaluation. Appl. Sci. 2020, 10, 5210. [CrossRef] for industrial internet of things using deep learning
65. Mirmozaffari, M.; Shadkam, E.; Khalili, S.M.; Kabirifar, K.; algorithms. Expert Syst. 2021, 39, e12917. [CrossRef]
Yazdani, R.; Gashteroodkhani, T.A. A novel artificial intelligent 78. Sadhu, P.K.; Yanambaka, V.P.; Abdelgawad, A. Internet of
approach: Comparison of machine learning tools and Things: Security and Solutions Survey. Sensors 2022, 22,
algorithms based on optimization DEA Malmquist 7433. [CrossRef]
productivity index for 79. Tsimenidis, S.; Lagkas, T.; Rantos, K. Deep learning in IoT
eco-efficiency evaluation. Int. J. Energy Sect. Manag. 2021, intrusion detection. J. Netw. Syst. Manag. 2022, 30, 1–40.
15, 523–550. [CrossRef] [CrossRef]
66. Sarker, I.H.; Khan, A.I.; Abushark, Y.B.; Alsolami, F. Brain Sci. 2023, 13, 683 28 of 30
Internet of Things (IoT) Security Intelligence: A 80. Alshohoumi, F.; Sarrab, M.; AlHamadani, A.; Al-Abri, D.
Comprehensive Overview, Systematic Review of Existing IoT Architectures Security and
Machine Learning Solutions and Research Directions. Mob. Privacy
Netw. Appl. 2022, 1–17. [CrossRef] Issues and Concerns. Int. J. Adv. Comput. Sci. Appl. 2019, 10,
67. Mirmozaffari, M.; Shadkam, E.; Khalili, S.M.; Yazdani, M. 232–251. [CrossRef]
Developing a Novel Integrated Generalised Data 81. Jha, K.K.; Jha, R.; Jha, A.K.; Hassan, M.A.M.; Yadav, S.K.;
Envelopment Analysis Mahesh, T. A Brief Comparison on Machine Learning
(DEA) to Evaluate Hospitals Providing Stroke Care Services. Algorithms Based
Bioengineering 2021, 8, 207. [CrossRef] on Various Applications: A Comprehensive Survey. In
68. Gupta, C.; Johri, I.; Srinivasan, K.; Hu, Y.-C.; Qaisar, S.M.; Proceedings of the 2021 IEEE International Conference on
Huang, K.-Y. A Systematic Review on Machine Learning and Computation
Deep System and Information Technology for Sustainable Solutions
Learning Models for Electronic Information Security in Mobile (CSITSS), Bangalore, India, 16–18 December 2021; pp. 1–5.
Networks. Sensors 2022, 22, 2017. [CrossRef] 82. Balo, F.; Tor˘ gul, B. Internet of Things: A Survey. Int. J.
69. Koroniotis, N.; Moustafa, N.; Slay, J. A new Intelligent Appl. Math. Electron. Comput. 2016, 104–110. [CrossRef]
Satellite Deep Learning Network Forensic framework for 83. Albalawi, A.M.; Almaiah, M.A. Assessing and reviewing of
smart satellite cyber-security threats, attacks, mitigation techniques in IoT
networks. Comput. Electr. Eng. 2022, 99, 107745. [CrossRef] environment.
70. Thomas, L.; Bhat, S. Machine learning and deep learning J. Theor. Appl. Inf. Technol. 2022, 100, 2988–3011.
techniques for IoT-based intrusion detection systems: A 84. Deep, S.; Zheng, X.; Jolfaei, A.; Yu, D.; Ostovari, P.; Bashir,
literature A.K. A survey of security and privacy issues in the Internet of
review. Int. J. Manag. Technol. Soc. Sci. 2021, 6, 296–314. Things
[CrossRef] from the layered context. Trans. Emerg. Telecommun.
Technol. 2020, 33, e3935. [CrossRef]
85. Navya, P.; Rama, G.S.; Kumar, T.P.; Pasha, S.N.; 99. Aregbesola, M.K.; Griva, I. A Fast Algorithm for Training
Mahender, K. IoT technology: Architecture, stack, security Large Scale Support Vector Machines. J. Comput. Commun.
risks, privacy risks 2022, 10,
and its applications. In Proceedings of AIP Conference 1–15. [CrossRef]
Proceedings; AIP Publishing LLC.: College Park, MD, USA, 100. Agghey, A.Z.; Mwinuka, L.J.; Pandhare, S.M.; Dida, M.A.;
2022; p. 020062. Ndibwile, J.D. Detection of Username Enumeration Attack on
86. Chatterjee, U.; Ray, S. Security Issues on IoT SSH
Communication and Evolving Solutions. In Soft Computing in Protocol: Machine Learning Approach. Symmetry 2021, 13,
Interdisciplinary Sciences; 2192. [CrossRef]
Springer: Berlin/Heidelberg, Germany, 2022; pp. 183–204. 101. Júnior, E.C.; Costa,W.L.; Portela, A.L.C.; Rocha, L.S.;
[CrossRef] Gomes, R.L.; Andrade, R.M.C. Detecting Attacks and Locating
87. Haque, A.K.M.B.; Bhushan, B.; Dhiman, G. Conceptualizing Malicious
smart city applications: Requirements, architecture, security Devices Using Unmanned Air Vehicles and Machine Learning.
issues, J. Internet Serv. Appl. 2022, 13, 11–20. [CrossRef]
and emerging trends. Expert Syst. 2021, 39, e12753. 102. Edemacu, K.; Kim, J.W. Multi-Party Privacy-Preserving
[CrossRef] Logistic Regression with Poor Quality Data Filtering for IoT
88. Jangjou, M.; Sohrabi, M.K. A Comprehensive Survey on Contributors.
Security Challenges in Different Network Layers in Cloud Electronics 2021, 10, 2049. [CrossRef]
Computing. 103. Puthal, D.;Wilson, S.; Nanda, A.; Liu, M.; Swain, S.;
Arch. Comput. Methods Eng. 2022, 29, 3587–3608. [CrossRef] Sahoo, B.P.; Yelamarthi, K.; Pillai, P.; El-Sayed, H.; Prasad, M.
89. Zahran, S.; Elkadi, H.; Helm,W. Fog of Things Framework Decision tree
to Handle Data Streaming Heterogeneity on Internet of based user-centric security solution for critical IoT
Things. In infrastructure. Comput. Electr. Eng. 2022, 99, 107754.
Proceedings of International Conference on Advanced [CrossRef]
Intelligent Systems and Informatics; Springer International 104. Abdaljabar, Z.H.; Ucan, O.N.; Alheeti, K.M.A. An intrusion
Publishing: Cham, detection system for IoT using KNN and decision-tree based
Switzerland, 2022; pp. 653–667. classification. In Proceedings of the 2021 International
90. Rasheed, M.A.; Saleem, J.; Murtaza, H.; Tanweer, H.A.; Conference of Modern Trends in Information and
Rasheed, M.A.; Ahmed, M. A Survey on Fog computing in IoT. Communication
VFAST Technology Industry (MTICTI), Sana’a, Yemen, 4–6 December
Trans. Softw. Eng. 2022, 9, 4. 2021; pp. 1–5.
91. Yassein, M.B.; Shatnawi, M.Q. Application layer protocols 105. Al-Ghaili, A.M.; Kasim, H.; Al-Hada, N.M. A secured data
for the Internet of Things: A survey. In Proceedings of the transform-and-transfer algorithm for energy internet-of-
2016 things
International Conference on Engineering & MIS (ICEMIS), applications. Telkomnika (Telecommun. Comput. Electron.
Agadir, Morocco, 22–24 September 2016; pp. 1–4. Control) 2021, 19, 1872–1883. [CrossRef]
92. Donta, P.K.; Srirama, S.N.; Amgoth, T.; Annavarapu, C.S.R. 106. Menter, Z.; Tee, W.Z.; Dave, R. Application of Machine
Survey on recent advances in IoT application layer protocols Learning-Based Pattern Recognition in IoT Devices. In
and Proceedings of
machine learning scope for research directions. Digit. International Conference on Communication and
Commun. Netw. 2021, 8, 727–744. [CrossRef] Computational Technologies; Springer: Singapore, 2021; pp.
93. Kakkar, L.; Gupta, D.; Saxena, S.; Tanwar, S. IoT 669–689.
architectures and its security: A review. In Proceedings of the 107. Meijin, L.; Zhiyang, F.; Junfeng, W.; Luyu, C.; Qi, Z.; Tao,
Second International Y.; Yinwei, W.; Jiaxuan, G. A Systematic Overview of Android
Conference on Information Management and Machine Malware
Intelligence; Springer: Singapore, 2021; pp. 87–94. Detection. Appl. Artif. Intell. 2021, 36, 2007327. [CrossRef]
94. Ahmad, Z.; Khan, A.S.; Shiang, C.W.; Abdullah, J.; Ahmad, Brain Sci. 2023, 13, 683 29 of 30
F. Network intrusion detection system: A systematic study of 108. Nakip, M.; Gelenbe, E. Botnet attack detection with
machine incremental online learning. In Proceedings of International
learning and deep learning approaches. Trans. Emerg. ISCIS Security
Telecommun. Technol. 2020, 32, e4150. [CrossRef] Workshop; Springer International Publishing: Cham,
95. Sarker, I.H. Deep Learning: A Comprehensive Overview on Switzerland, 2022; pp. 51–60.
Techniques, Taxonomy, Applications and Research Directions. 109. Ravikumar, D. Towards Enhancement of Machine
SN Learning Techniques Using CSE-CIC-IDS2018 Cybersecurity
Comput. Sci. 2021, 2, 420. [CrossRef] Dataset; Rochester
96. Abbas, G.; Mehmood, A.; Carsten, M.; Epiphaniou, G.; Institute of Technology: Rochester, NY, USA, 2021.
Lloret, J. Safety, Security and Privacy in Machine Learning 110. Van der Schyff, K.; Flowerday, S. Mediating effects of
Based Internet information security awareness. Comput. Secur. 2021, 106,
of Things. J. Sens. Actuator Netw. 2022, 11, 38. [CrossRef] 102313. [CrossRef]
97. Han, S.; Mannan, N.; Stein, D.C.; Pattipati, K.R.; Bollas, 111. Sarker, I.H. Machine Learning for Intelligent Data
G.M. Classification and regression models of audio and Analysis and Automation in Cybersecurity: Current and
vibration signals Future Prospects.
for machine state monitoring in precision machining systems. Ann. Data Sci. 2022, 1–26. [CrossRef]
J. Manuf. Syst. 2021, 61, 45–53. [CrossRef] 112. AKTAR, H.; PERKGÖZ, C. Autoencoder Aided Machine
98. Hsieh, S.-C. Prediction of Compressive Strength of Learning Methods for Intrusion Detection Systems. In New
Concrete and Rock Using an Elementary Instance-Based Trends in
Learning Algorithm. Technical, Natural Sciences, Engineering and Health Sciences;
Adv. Civ. Eng. 2021, 2021, 6658932. [CrossRef] Duvar Publishing: Izmir, Türkiye, 2022; p. 101.
113. Giordano, G.; Palomba, F.; Ferrucci, F. On the use of 125. Doshi, R.; Apthorpe, N.; Feamster, N. Machine Learning
artificial intelligence to deal with privacy in IoT systems: A DDoS Detection for Consumer Internet of Things Devices. In
systematic Proceedings
literature review. J. Syst. Softw. 2022, 193, 111475. of the 2018 IEEE Security and Privacy Workshops (SPW), San
[CrossRef] Francisco, CA, USA, 24 May 2018; pp. 29–35. [CrossRef]
114. Kallitsis, M.; Prajapati, R.; Honavar, V.;Wu, D.; Yen, J. 126. Zhang, C.; Sun, X.; Xia,W.; Zhang, J.; Zhu, H.;Wang, X.
Detecting and Interpreting Changes in Scanning Behavior in Deep Learning Based Double-Contention Random Access for
Large Network Massive
Telescopes. IEEE Trans. Inf. Forensics Secur. 2022, 17, 3611– Machine-Type Communications. IEEE Trans. Wirel. Commun.
3625. [CrossRef] 2022, 22, 1794–1807. [CrossRef]
115. Masum, M.H.R. IT-Security Challenges for IoT 127. Anidu, A.; Obuzor, Z. Evaluation of Machine Learning
Infrastructures; Frankfurt University of Applied Sciences: Algorithms on Internet of Things (IoT) Malware Opcodes.
Frankfurt am Main, Handb. Big
Germany, 2022. Data Anal. Forensics 2022, 177–191. [CrossRef]
116. Gordaliza, P.M. Computer-Aided Assessment of 128. Zhou, W.; Yu, B. A cloud-assisted malware detection and
Tuberculosis with Radiological Imaging: From Rule-Based suppression framework for wireless multimedia system in IoT
Methods to Deep based on
Learning. Ph.D. Thesis, Universidad Carlos III de Madrid, dynamic differential game. China Commun. 2018, 15, 209–
Madrid, Spain, 2022. 223. [CrossRef]
117. Elghamrawy, S.M.; Lotfy, M.O.; Elawady, Y.H. An 129. Chauhan, J.; Seneviratne, S.; Hu, Y.; Misra, A.;
Intrusion Detection Model Based on Deep Learning and Seneviratne, A.; Lee, Y. Breathing-Based Authentication on
Multi-Layer Perceptron Resource-Constrained
in the Internet of Things (IoT) Network. In International IoT Devices using Recurrent Neural Networks. Computer
Conference on Advanced Machine Learning Technologies and 2018, 51, 60–67. [CrossRef]
Applications; 130. Ismaeel, H.; Elmedany, W. Anomaly-based detection
Springer: Berlin/Heidelberg, Germany, 2022; pp. 34–46. Technique using Deep Learning for Internet of Things: A
[CrossRef] Survey. In
118. Uhricek, D.; Hynek, K.; Cejka, T.; Kolar, D. BOTA: Proceedings of the 2022 International Conference on
Explainable IoT Malware Detection in Large Networks; IEEE: Innovation and Intelligence for Informatics, Computing, and
New York, NY, USA, Technologies
2022; p. 1. [CrossRef] (3ICT), Sakheer, Bahrain, 20–21 November 2022; pp. 278–
119. Madhu, B.; Chari, M.V.G.; Vankdothu, R.; Silivery, A.K.; 284.
Aerranagula, V. Intrusion detection models for IOT networks 131. An, N.; Duff, A.; Naik, G.; Faloutsos, M.; Weber, S.;
via deep Mancoridis, S. Behavioral anomaly detection of malware on
learning approaches. Meas. Sens. 2022, 100641. [CrossRef] home routers. In
120. Al-Shareeda, M.A.; Manickam, S.; Saare, M.A. DDoS Proceedings of the 2017 12th International Conference on
attacks detection using machine learning and deep learning Malicious and Unwanted Software (MALWARE), Fajardo, PR,
techniques: USA,
Analysis and comparison. Bull. Electr. Eng. Inform. 2023, 12, 11–14 October 2017; pp. 47–54.
930–939. [CrossRef] 132. da Costa, K.A.; Papa, J.P.; Lisboa, C.O.; Munoz, R.; de
121. Gopal, S.B.; Poongodi, C.; Nanthiya, D.; Kirubakaran, T.; Albuquerque, V.H.C. Internet of Things: A survey on machine
Logeshwar, D.; Saravanan, B.K. Autoencoder based learningbased
Architecture for intrusion detection approaches. Comput. Netw. 2019, 151,
Mitigating Phishing URL attack in the Internet of Things (IoT) 147–157. [CrossRef]
Using Deep Neural Networks. In 2022 6th International 133. Wahi, V.; Yadav, S.; Thenuia, Y.; Chauhan, A. Anomaly
Conference Based Intrusion Detection for IoT. In Proceedings of the 2022
on Devices, Circuits and Systems (ICDCS); IEEE: New York, NY, 3rd
USA, 2022; pp. 427–431. [CrossRef] International Conference for Emerging Technology (INCET),
122. Bhattacharya, S.; Ghorai, S.; Khan, A.K. Systematic Study Belgaum, India, 27–29 May 2022; pp. 1–7.
of Detection Mechanism for Network Intrusion in Cloud, Fog, Brain Sci. 2023, 13, 683 30 of 30
and 134. Yang, X.; Peng, G.; Zhang, D.; Lv, Y. An Enhanced
Internet of Things Using Deep Learning. In Human-Centric Intrusion Detection System for IoT Networks Based on Deep
Smart Computing; Springer: Berlin/Heidelberg, Germany, Learning and
2023; Knowledge Graph. Secur. Commun. Netw. 2022, 2022,
pp. 31–43. [CrossRef] 4748528. [CrossRef]
123. Saheed, Y.K.; Baba, U.A.; Orje-Ishegh, T.; Longe, O.B. An 135. Thavamani, S.; Mahesh, D.; Sinthuja, U.; Manoharan, G.
Efficient Machine Learning and Deep Belief Network Models Crucial attacks in internet of things via artificial intelligence
for Wireless Intrusion Detection System. 2022. Available techniques:
online: The security survey. In AIP Conference Proceedings; AIP
https://www.researchgate.net/publication/364203267_ Publishing LLC.: College Park, MD, USA, 2022. [CrossRef]
An_Efficient_Machine_Learning_and_Deep_Belief_Network_ 136. Khan, R.; Yang, Q.; Ullah, I.; Rehman, A.U.; Bin Tufail, A.;
Models_for_Wireless_Intrusion_Detection_System (accessed Noor, A.; Rehman, A.; Cengiz, K. 3D convolutional neural
on networks
31 December 2022). based automatic modulation classification in the presence of
124. Deng, L.; Li, D.; Yao, X.; Wang, H. Retraction Note to: channel noise. IET Commun. 2021, 16, 497–509. [CrossRef]
Mobile network intrusion detection for IoT system based on 137. Haji, S.H.; Ameen, S.Y. Attack and Anomaly Detection in
transfer IoT Networks using Machine Learning Techniques: A Review.
learning algorithm. Clust. Comput. 2021, 24, 589. [CrossRef] Asian J.
Res. Comput. Sci. 2021, 9, 30–46. [CrossRef]
138. Abideen, Z.U.; Mazhar, T.; Razzaq, A.; Haq, I.; Ullah, I.; IPv6 Extension for Next-Generation IoT Networks. IEEE
Alasmary, H.; Mohamed, H.G. Analysis of Enrollment Criteria Internet Things J. 2021, 10, 2095–2103. [CrossRef]
in
Secondary Schools Using Machine Learning and Data Mining
Approach. Electronics 2023, 12, 694. [CrossRef]
139. Khan, W.U.; Imtiaz, N.; Ullah, I. Joint optimization of
NOMA-enabled backscatter communications for beyond 5G
IoT networks.
Internet Technol. Lett. 2020, 4, e265. [CrossRef]
140. Esmalifalak, M.; Liu, L.; Nguyen, N.; Zheng, R.; Han, Z.
Detecting Stealthy False Data Injection Using Machine
Learning in Smart
Grid. IEEE Syst. J. 2017, 11, 1644–1652. [CrossRef]
141. Pourghebleh, B.;Wakil, K.; Navimipour, N.J. A
Comprehensive Study on the Trust Management Techniques
in the Internet of
Things. IEEE Internet Things J. 2019, 6, 9326–9337. [CrossRef]
142. Hajjaji, Y.; Boulila, W.; Farah, I.R.; Romdhani, I.; Hussain,
A. Big data and IoT-based applications in smart
environments:
A systematic review. Comput. Sci. Rev. 2021, 39, 100318.
[CrossRef]
143. Stergiou, C.; Psannis, K.E.; Gupta, B.B.; Ishibashi, Y.
Security, privacy & efficiency of sustainable Cloud Computing
for Big Data &
IoT. Sustain. Comput. Inform. Syst. 2018, 19, 174–184.
[CrossRef]
144. Chui, K.T.; Liu, R.W.; Lytras, M.D.; Zhao, M. Big data and
IoT solution for patient behaviour monitoring. Behav. Inf.
Technol. 2019,
38, 940–949. [CrossRef]
145. Mkrttchian, V.; Gamidullaeva, L.; Finogeev, A.;
Chernyshenko, S.; Chernyshenko, V.; Amirov, D.; Potapova, I.
Big data and
Internet of Things (IoT) technologies’ influence on higher
education: Current state and future prospects. Int. J. Web-
Based Learn.
Teach. Technol. (IJWLTT) 2021, 16, 137–157. [CrossRef]
146. Khalil, H.; Rahman, S.U.; Ullah, I.; Khan, I.; Alghadhban,
A.J.; Al-Adhaileh, M.H.; Ali, G.; ElAffendi, M. A UAV-Swarm-
Communication Model Using a Machine-Learning Approach
for Search-and-Rescue Applications. Drones 2022, 6, 372.
[CrossRef]
147. Dehghantanha, A.; Choo, K.-K.R. Handbook of Big Data
and IoT Security; Springer: Berlin/Heidelberg, Germany, 2019.
148. Dwivedi, S.K.; Roy, P.; Karda, C.; Agrawal, S.; Amin, R.
Blockchain-Based Internet of Things and Industrial IoT: A
Comprehensive
Survey. Secur. Commun. Netw. 2021, 2021, 7142048.
[CrossRef]
149. Yousafzai, B.K.; Khan, S.A.; Rahman, T.; Khan, I.; Ullah, I.;
Rehman, A.U.; Baz, M.; Hamam, H.; Cheikhrouhou, O.
Student-
Performulator: Student Academic Performance Using Hybrid
Deep Neural Network. Sustainability 2021, 13, 9775.
[CrossRef]
150. Raj, A.; Shetty, S.D. IoT Eco-system, Layered
Architectures, Security and Advancing Technologies: A
Comprehensive Survey.
Wirel. Pers. Commun. 2021, 122, 1481–1517. [CrossRef]
151. Gupta, D.; Juneja, S.; Nauman, A.; Hamid, Y.; Ullah, I.;
Kim, T.; Tag eldin, E.M.; Ghamry, N.A. Energy Saving
Implementation in
Hydraulic Press Using Industrial Internet of Things (IIoT).
Electronics 2022, 11, 4061.
152. Hussain, A.; Nazir, S.; Khan, F.; Nkenyereye, L.; Ullah, A.;
Khan, S.; Verma, S. Kavita A Resource-Efficient Hybrid Proxy
Mobile