Learning Approaches for Security

and Privacy in Internet of Things

T. Daniya, M. Geetha, Velliangiri Sarveshwaran, and Ch. Madhu Babu

1 Introduction

The Internet of Things (IoT) has enabled everything and anything to be connected
to the internet. By transforming how people use technology, it has caused a multi-
media revolution in the physical world that surrounds it. Supervisory surveillance,
gadgets, commercial vehicles and medical equipment, alarms, fluorescent lamps,
refrigerators, aircraft, pet supplies, smart TVs, and electronic set-tops boxes can
now be connected to the internet through IoT [1]. Industrial applications are becom-
ing more complex and broader as science and technology advance. Many vital pro-
cesses that rely on technology have large-scale control systems in place to keep
them running normally [2]. The Internet of Things (IoT) is a network of uniquely
recognizable, embedded computing equipment that can send and receive data with-
out contact between people-to-people or people-to-computer relations [3].
Many countries have identified cybersecurity as a key national policy concern.
The financial consequences of cybersecurity and cyber warfare are rising as IoT-­
based devices are incorporated into intellectual production and marketing, cities,
advanced transmission networks, intelligent networks, and more elements of

T. Daniya (*)
Department of CSE (AI & ML), GMR Institute of Technology, Rajam, India
e-mail: daniya.t@gmrit.edu.in
M. Geetha
Department of Information Technology, S.A Engineering College, Chennai, Tamilnadu, India
V. Sarveshwaran
Department of Computational Intelligence, SRM Institute of Science and Technology,
Kattankulathur, Chengalpattu, Chennai, Tamilnadu, India
C. M. Babu
Department of CSE, B V Raju Institute of Technology, Narsapur, Telangana, India

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 63

V. K. Gunjan et al. (eds.), Modern Approaches in IoT and Machine Learning
for Cyber Security, Internet of Things, https://doi.org/10.1007/978-3-031-09955-7_4
64 T. Daniya et al.

contemporary life such as banking, finance, independent vehicles, and personal

medicinal devices. More and more cyber-attacks are occurring, and they threaten
IoT devices progressively (such as the Mirai botnet) [4].

Cybersecurity Attacks
Consumer data security breaches, along with major cybersecurity assaults net-
works around the world, have caused IoT users to lose faith in distributors who do
not undertake the precautions necessary to secure its applications [5]. The problem
of IoT networks is cybersecurity, which is a key worldwide problem that necessi-
tates a comprehensive evaluation by either the academic or the manufacturing sec-
tors [6]. Human involvement elevates it to a Cyber-Physical-Human System, also
CPHS, which can be regulated by a human [7].
In CPHS, understanding, validating, and defending such structures are difficult
due to the difficulty of modeling human behavior. Spam and phishing emails,
exploitation of the web address of the network, replication assaults, and denial of
service attacks, in which cyber criminal’s elements, error-prone, and malicious peo-
ple block honest messages from secure broadcast communication, are all threats to
the CPHS [8].
Cyber-Physical System (CPS) can carry out self-confidence, self-adjustments,
and decisions [9]. The way to implement CPS’ environmental tier function is
through machine-to-machine (M2M) communications. In M2M, both wired and
Learning Approaches for Security and Privacy in Internet of Things 65

wireless connections communicate between sensors and smart/mobile devices [10].

Moreover, cyber-attacks in networks of the wireless sensor include attacks on a
denial-of-service (DoS), misdirection, selective redirection, sinkhole, Sybil, worm-
hole attacks, and HELLO floods attacks [11].
The probability of severe economic impact from attacks can only be established
through subjective assessment without the relevant threat appraisal process [12].
Cyber risk would contribute to feedback sensors and data structures instantaneously,
which link financial impacts of various IoT vertical systems through collision pat-
tern to crucial foundation. It would help industry and lawmakers to view the issue
and handle the economic burden of IoT-related cyber-attacks [13].
The conceptualization of cyber risk is examined in terms expressed by two cat-
egories of experts: cybersecurity specialists and developers of cybersecurity con-
ceptual frameworks [14]. Both groups attach importance to the postulation of
vulnerability and to their misuse by the assaulter [15]. Hence, the comprehensive
survey has to be discussed for machine learning-based cybersecurity on IoT.
The primary contributions of the article are as follows:
• Firstly, it discusses about the review on various attacks on cybersecurity based on
Internet of things including Denial of Service (DoS) and distributed Denial of
Service (DDoS) attacks, Man-in-the-Middle (MitM) attack, Phishing and spear
Phishing attacks, Drive-by attack, Password attack, SQL injection attack, Cross-­
site scripting (XSS) attack, Eavesdropping attack, Birthday attack, and
Malware attack.
• Secondly, it discusses about the comprehensive survey of security and privacy
authentication of cybersecurity-based machine learning approach on IoT.
The rest of the survey paper is structured as follows: Sect. 2 describes the review
on cybersecurity based on IoT; Sect. 3 describes the comparison on cybersecurity
based on IoT; Sect. 4 describes the future research direction; and finally Sect. 5
concludes the survey paper.

2 Cybersecurity-Based on Machine Learning Algorithms

in Internet of Things

This section, discusses the cyber security-based machine learning on IoT including
various attacks, security, and privacy in cybersecurity.

2.1 Various Attacks in Cybersecurity

Sarker et al. [16] analyzed that the prominence of Internet-of-Things (IoT) and the
huge extension of the communications systems, as well as numerous applications,
have recently paid enormous attention to cybersecurity problems today. Thus, it
66 T. Daniya et al.

became increasingly important to detect different cyber assaults or abnormalities in

the device to build a powerful intrusion detection system that was essential to safety
today. An intelligent intrusion detection system based on data can be used for the
construction of artificial intelligence, in particular machine learning techniques. To
achieve this objective, this chapter presented a machine-learning security model
based on Intrusion Detection Tree (‘IntruDTree’), which first takes into consider-
ation the safety function ranking based on its significance and then builds, based on
the key extracted features, a tree-based generalized intrusion detection model. This
system not only works to predict precision of unidentified test scenarios but also
reduces the computing complexity by decreasing the size of the characteristics.
I.H Sarker et al. [17] presented Cyber Learning, a machine learning model with
a set of the correlated functions and an extensive empirical assessment on the effec-
tiveness of various model machine learning. In cyber learning models, a binary
classification model to predict threats and a multi classification model for a wide
range of cyber-attacks are needed. To create the security model, the 10 popular
methodologies, namely Naive Bayes, Logistic regression, Stochastic gradient
descent, K-nearest Neighbours, Support Vector Machine, Decision Tree, Random
Forest, Adaptive Boosting, eXtreme Gradient Boosting, and Linear discrimination
analysis, are employed. The artificial network-based security model was presented
with a perspective to multiple hidden layers.
Syed et al. [18] analyzed that the usage of appropriate data packets, the imple-
mentation, and the operation of the persistent Internet of Things (IoT) depend delib-
erately. The Message Queuing Telemetry Transport (MQTT) protocol, based on the
publication/subscription landscape, is popular. Cybersecurity threats are expected
to increase with the growing usage by IoT producers of the MQTT protocol. In
particular, IoT may be vulnerable to a Denial of Service (DoS) protocol-based
Application Layer attack that was found to trigger service of interruptions in heri-
tage systems. The application layer DoS attack detection strategy was then pro-
posed for the MQTT protocol and the scheme tested for legit and protocol-compliant
scenarios for the attack DoS. A machine-based learning detection framework devel-
oped for the MQTT protocol was proposed to secure the MQTT message from such
assaults. The influence of these assaults on different MQTT brokers was also dem-
onstrated, and the effectiveness of the introduced strategy for the detection of these
malicious attacks was evaluated.
Damaševičius et al. [19] discussed that Information security is one of the biggest
challenges organizations and institutions face. In recent years, cyber crime has
increased frequency and magnitude with new ways in which information systems
can be stolen, changed and destroyed, and disabled every day. Malware is one of the
types of penetration of confidential data in information systems. An intruder pene-
trates malware into a computer network whereby the data management has sensitive
data information, which is accessed completely or partially. This paper proposed a
group-based malware detection classification methodology. A stacked group of
dense (without a connection) and convolutional neural networks (CNN) performs a
first-phase classification, while a meta-learner performs the final phase
Learning Approaches for Security and Privacy in Internet of Things 67

classification. A group of five dense neural networks with CNN and ExtraTrees as a
meta-learner is the best performance possible.
Subburaj et al. [20] aimed to investigate and examine various means of protect-
ing websites from SQL attacks through injection. Machine learning algorithms have
been used to detect such attacks in this research project. Machine learning (ML)
algorithms are algorithms that can learn from the given data and deduce interesting
dataset outcomes. The technology was used to detect malicious code through SQL
code as data and ML algorithms. This research project has established a machine
learning model that can predict potential attacks.
Shafiq et al. [21] proposed a hybrid algorithm and a new model architecture to
addressed this issue. The BoT-IoT recognition dataset is first applied, and its 44
effective functions for machine learning algorithms are selected from a variety of
features. Then five efficient machine-learning algorithms are chosen to recognize
malicious and anomalous traffic and also to pick the most commonly used metrics
for ML algorithms. A bijective soft set approach and algorithm are explored to iden-
tify which ML algorithm is successful and should be used to determine IoT anomaly
and traffic intrusion. This has used the proposed algorithm based on a bijective soft
set approach that effectively detects attacks.
Katzir et al. [22] proposed an adverse detection method durability based on a
supervised model of machine learning. Emphasis on multisensory fusion systems
for systematic description of adversarial resilience is given. The Model Robustness
Score (MRB), a measure for assessing the relative resilience of various designs, is
also established, and two new algorithms for the selection of adversary classifiers
are also proposed. The first algorithm only selects features that cannot be realisti-
cally altered by the opponent; the second algorithm controls the durability versus
the precision compromise. Ultimately, the strategy to real-life malware classifica-
tion with a comprehensive, up-to-date body of benign and malware executables was
assessed. The ability to use adversely aware feature selection to create more robust
classifiers has also been shown, and the empirical evidence supporting ensemble
algorithms inherent resilience compared with single model algorithms is provided.

2.2 Security and Privacy in Cybersecurity

Li et al. [28] analyzed that safety approach and smart control system for solar ther-
mal use are of great functional significance in the advancement of Energy Internet
of Things (EIoT), which aims to optimize operational performance and enables
intelligent dynamic change. Computational fluid dynamics modeling was utilized
for the analysis of the solar energy output mechanism for buildings combined with
solar water heating systems. A machine-based approach to predict energy conver-
sion is proposed. This paper proposed a new EIoT platform for the cyber protection
analysis based on machine learning and implements the temperature control plat-
form. The EIoT method showed a high success with the Extreme Gradient Boosting
68 T. Daniya et al.

(XG Boost) training algorithm following an assessment of a machine-based learn-

ing cybersecurity analysis.
Kumar et al. [29] presented the Trustworthy Privacy-Preserving Secured
Framework (TP2SF) for smart cities. This system consists of three modules: a con-
fidentiality module, a double-level data protection module, and the intrusion detec-
tion module. The blockchain credibility scheme is built in the trustworthiness
module. A blockchain-based enhanced Proof of Work (ePoW) technology and
Principal Component Analysis (PCA) concurrently are implemented to convert data
to a new reduced form in order to avoid interference and toxic attacks on the two-­
level privacy modules. An optimized gradient tree boost method (XGBoost) is used
in the intrusion detection module. Finally, the combined Fog-Cloud infrastructure,
CloudBlock and FogBlock, has introduced the proposed TP2SF platform in the
intelligent city due to its legacy strengths and limitations.
Zhu et al. [30] analyzed that data protection and privacy are becoming more and
more relevant topics in the healthcare industry at present. In introducing digital
health reports based on legislation, restructuring of services, and requiring to share
data between patients, providers and payers, it is emphasized that information secu-
rity is improved. Big information about health care would likely change the out-
comes of the patient, forecast epidemics, gain useful expertise, mitigate illness,
minimize health costs, and improve quality of life research. In this article, the Safety
and Privacy Big Data Analytics Framework for healthcare applications has been
introduced. To maintain a trustworthy Big Data system, there is a need to recognize
the shortcomings of current technologies in future study. In addition, electronic
health records (EHRs) can be shared by different consumers with a view of the qual-
ity of healthcare services. This adds to the major privacy problems to be dealt with
in order to enforce the EHR. It incorporates many technological frameworks and
environmental safeguards and is proven sufficient to address joint risks to network
security appropriately.
Liu et al. [31] analyzed that information can be dynamically interpreted for situ-
ational awareness in power systems with the production and broad implementation
of measurement equipment. However, data spoofing attacks are also jeopardizing
the cybersecurity of power networks. This paper suggested an algorithm based on
the extraction techniques of measurement data source authentication (MDSA)
including ensemble empirical mode decomposition (EEMD), fast Fourier transfor-
mation (FFT), and real-time data classification machine learning. The proposed
algorithm is capable of achieving greater precision MDSA with a shorter data win-
dow from closely positioned synchrophasor sensors.
Lee et al. [32] discussed that when an internet site emerged, a representative
authentication mechanism for the customer was a key issue. However, different
attack methods appeared in this authentication process to steal keyboard password
entries. The keyboard information therefore may not ensure security. A random data
generation data security strategy was developed to detect and mitigate such an
assault. This technique defends keyboard information with dumb keyboard infor-
mation generation when the attacker receives keyboard information. The viability
for the disclosure of keyboard information using the data security technique has
Learning Approaches for Security and Privacy in Internet of Things 69

been shown in this analysis. To illustrate the assault-presented strategy, all the ridic-
ulous keyboard information developed via defensive tool and the user’s actual key-
board information feedback were collected and the threats to cyber protection of
keyboard information were evaluated based on the offensive method of machine
learning. This confirmed that an opponent got 96.2% accurate keyboard data even
though the attack technique, which prevents the keyboard data exposure from
attacking, is being used.
Karimipour et al. [33] analyzed that intelligent grid technology improves electri-
cal grid reliability, safety, and performance. Its strong relationships on digital com-
munication technology moreover create new security flaws, which must be taken
into account for secure and economical power distribution. In this article, a statisti-
cal association between measurements is proposed for an unsupervised detection of
anomalies. The objective is to design an anomaly detector that can discriminate
between a perturbation and an intelligent cyber-attack. It ideals with larger intelli-
gent smart grids. The proposed approach uses symbolic dynamic filters (SDF) as a
functional extraction to minimize computation complexity and to detect causal
interactions between subsystems.
Yin et al. [34] discussed that a cybersecurity solution-based groundbreaking
strategy to the intrusion detection mechanism in which a malicious operation in
Supervision Control and Data Acquisition (SCADA) systems targets the Distributed
Network Protocol (DNP3). Due to the extreme connections between industrial con-
trol systems and the outside Internet world using IoT technologies, information and
communication systems are susceptible to severe and cyber-attacks. In certain cir-
cumstances, cyber-attacks lead to various threats affecting the infrastructure and
continuity of businesses. Owing conventional process systems specificities like
insecure real time protocols, the general-purpose ICT safety mechanisms in SCADA
systems are not completely stable. In this article, a new way of assessing and select-
ing the vulnerability model in various layers using the parsed DNP3 protocol for
machine learning, with additional data containing malware samples, was intro-
duced. In addition, a cyber-attack algorithm was developed that included a method
for classifying and visualizing. In an IoT-based communications environment, the
suggested technique was capable of detecting attacks in real time.
Thakong et al. [35] focused that cybersecurity issues have arisen in many busi-
ness applications in recent times. While researchers previously suggested address-
ing cybersecurity issues, their methodologies frequently replicated training program
to identify datasets of these concerns in non-stationary streaming environments.
The traditional approaches can worsen the adaptive solution to avoid these problems
in dynamic environments. The paper proposed to use the hierarchical structure of
the network to fix these issues in a dynamic environment to create a one-pass throw
away learning environment. In addition, new principles of learning in the form of
recursive functions have been implemented to accelerate the computation time and
to preserve a minimum space complexity for streaming data. In order to minimize
the time of preparation, knowledge gain-based feature selection was also employed.
70 T. Daniya et al.

3 Comparison for Cybersecurity Based on Machine

Learning in Internet of Things

In this segment, comparison table for various attacks, security, and privacy authen-
tication in Cybersecurity-based machine learning approaches can be depicted below:

3.1 Various Attacks in Cybersecurity

S. No Author Methodology Advantage Disadvantage

1. Anthi Adversarial machine It is more robust This method may not
et al. learning using Jacobian-­ be sufficient also
[23] based Saliency Map attack difficult
2. Bilen Support vector machine and Faster and more Technical details can
et al. Linear regression effective to detect crime be easier to steal
[24] and criminals
3. Singh Variational mode Better efficiency and Cyber-attack
et al. decomposition (VMD) and computational resiliency cannot be
[25] Decision Tree (DT) Processing time can be enhanced
algorithm achieved
4. Tariq Genetic Algorithm (GA) Accuracy is high and Payload fluctuation
et al. with Reinforcement effectively detects the will be high
[26] Learning (RL) attack
5. Ghiasi Hilbert-Huang Transform Enhance information Cyber-attack details
et al. Methodology with exchange privacy, more for unofficial users
[27] Blockchain-based Ledger precision and robust are not accessible
Technology

3.2 Cybersecurity and Privacy Authentication in IoT

S. No Author Methodology Advantage Disadvantage

1. Kalinin Neural network approach High classification New features to
et al. [36] accuracy when datasets cannot be
working with big added.
data
2. Serrano Blockchain random neural Cybersecurity Block chain with
et al. [37] network for cybersecurity resilience increased, another neural
applications in holistic decentralized the network cannot be
digital and physical user access and validated
cybersecurity users and connectivity
channel authentication
methods
(continued)
Learning Approaches for Security and Privacy in Internet of Things 71

S. No Author Methodology Advantage Disadvantage

3. Zhang Mobile edge computing It achieves both Optimum perturbation
et al. [38] enabled federated learning efficiency and strength for private
framework (FedMEC) privacy data cannot be
explored
4. Latif et al. Block chain-based Reduces Great potential of
[39] architecture computational block chain
complexity and technology for smart
enhances security industry cannot be
achieved
5. Abramson A trust framework based on Performed privacy Messaging function
et al. [40] decentralized identifiers and preserving workflow with separate structure
verifiable credentials for for ML
ML communication cannot
be focused

4 Future Research Direction for Cybersecurity

• Future work can evaluate IntruDTree model efficacy through accumulating huge
amounts of datasets with larger sizes of IoT protection and by measuring their
effectiveness in the area of cybersecurity at application level.
• The IoT device can accumulate latest security details with larger sizes and con-
struct a safe system handled by data through learning technology in future.
• As future work, the intention is to study an authentic MQTT dataset to evaluate
its characteristics and how these can help to unmark legitimate and malicious
MQTT traffic.
• In order to offer valuable knowledge to malware analysts, the use of explanatory
artificial intelligence (XAI) strategy to decipher the results of deep learning mod-
els for malware detection would be implemented in future. It also plans to inves-
tigate the architectures of group learning and to do more tests on broader malware
databases. This seeks to enhance the classification potential and precision of the
ensemble learning algorithm by refining and validating the model design for
many malware datasets in future work.
• For future work, a prototype that examines the conceptual safety requirements of
the overall strategy will expand the proposed structure. In addition, the TP2SF
architecture will be studied to integrate deep learning and edge computing in
smart cities. This will take into account a blockchain network-allowed edge and
deep learning in the intelligent city. This incorporation will help to improve the
safety and privacy demands of developing intelligent cities.
• Moreover, interactive and distributed machine-based learning systems are not
investigated. On distributed and heterogeneous devices, IoT needs ML-based
solutions. Such strategies must work together and do not rely on centrally trained
results. Federated learning will in this respect be seen as a basis for such methods.
72 T. Daniya et al.

• This will in the future strengthen the mathematical integration to describe vari-
ous machine learning tasks and unsupervised machine learning cases in a cor-
rect way.
• The strategy to distributed multisensory SIEMs is to be tested in the future.
These systems use a wide range of independent sensors that improve detection
capacity. Today’s major challenge among modern SIEMs is how to decrease
false alerts so that safety professionals use these systems more efficiently. The
forensic analysis showed signs of the attack, which were initially ignored due to
the large number of false alarms in virtually every public-ally documented attack
on a larger organization. Thus, the purpose of this paradigm is to test strategy and
to verify that an intruder is unable to increase the number of proven false warns
intentionally.
• In the future, a 6G device is designed to increase scalability and reliability, very
low latency and low energy use. However, there may be restrictions on current
blockchain consensus algorithms about scalability, durability, latency, and
energy consumption. Implementing blockchains to achieve key design features,
such as fault tolerance, safety, low latency, and decentralization at the same time
creates significant challenges to scalability and confidence, one of 6G systems’
main objectives. In order to profit from the deployment of blockchain in 6G sys-
tems, a new consensus algorithm, which offers increased reliability and scalabil-
ity while offering tradeoffs between intrusion detection, protection and delay,
should be suggested.

5 Conclusion

In some domains like IoT, cybersecurity has historically been one of the key issues
of ICT and is even more important. This paper surveyed the state of the art of the
existing researches in Cybersecurity-based machine learning approaches on Internet
of things. It analyzed the various attacks in cybersecurity including malware attack,
cross-site scripting attack, DoS attack, etc. Also, this article surveyed the security
and privacy authentication in cybersecurity. Then the comparison table was ana-
lyzed with advantages and disadvantages. Finally, the future research directions
were discussed in IoT-based on cybersecurity. As a result, cybersecurity will prog-
ress and become more critical as the intelligent world proliferates on machine learn-
ing approaches in IoT.

References

1. K. Kimani, V. Oduol, K. Langat, Cyber security challenges for IoT-based smart grid networks.
Int. J. Crit. Infrastruct. Prot. 25, 36–49 (2019)
2. M.M. Rana, R. Bo, IoT-based cyber-physical communication architecture: Challenges and
research directions. IET Cyber-Physical Systems: Theory & Applications 5(1), 25–30 (2020)
Learning Approaches for Security and Privacy in Internet of Things 73

3. S. Tanwar, S. Tyagi, S. Kumar, The role of internet of things and smart grid for the devel-
opment of a smart city, in Intelligent Communication and Computational Technologies,
(Springer, Singapore, 2018), pp. 23–33
4. P. Radanliev, D. De Roure, S. Cannady, R.M. Montalvo, R. Nicolescu, M. Huth, Economic
impact of IoT cyber risk-analysing past and present to predict the future developments in IoT
risk analysis and IoT cyber insurance (2018)
5. A. Jain, T. Singh, Security challenges and solutions of IoT ecosystem, in Information and
Communication Technology for Sustainable Development, (Springer, Singapore, 2020),
pp. 259–270
6. S. Tweneboah-Koduah, K.E. Skouby, R. Tadayoni, Cyber security threats to IoT applications
and service domains. Wirel. Pers. Commun. 95(1), 169–185 (2017)
7. S.K. Sowe, E. Simmon, K. Zettsu, F. de Vaulx, I. Bojanova, Cyber-physical-human systems:
Putting people in the loop. IT Prof 18(1), 10–13 (2016)
8. S.A. Kumar, B. Bhargava, R. Macêdo, G. Mani, Securing iot-based cyber-physical human sys-
tems against collaborative attacks, in 2017 IEEE International Congress on Internet of Things
(ICIOT), (IEEE, 2017, June), pp. 9–16
9. Y. Jiang, S. Yin, O. Kaynak, Data-driven monitoring and safety control of industrial cyber-­
physical systems: Basics and beyond. IEEE Access 6, 47374–47384 (2018)
10. K.M. Renuka, S. Kumari, D. Zhao, L. Li, Design of a secure password-based authentica-
tion scheme for M2M networks in IoT enabled cyber-physical systems. IEEE Access 7,
51014–51027 (2019)
11. B. Ali, A.I. Awad, Cyber and physical security vulnerability assessment for IoT-based smart
homes. Sensors 18(3), 817 (2018)
12. P. Radanliev, D.C. De Roure, R. Nicolescu, M. Huth, R.M. Montalvo, S. Cannady, P. Burnap,
Future developments in cyber risk assessment for the internet of things. Comput. Ind. 102,
14–22 (2018)
13. P. Radanliev, D. De Roure, S. Cannady, R. Mantilla Montalvo, R. Nicolescu, M. Huth,
Analysing IoT cyber risk for estimating IoT cyber insurance, in Living in the Internet of Things:
Cybersecurity of the IoT-2018. IET Conference Proceedings, (The Institution of Engineering
and Technology, London, 2018)
14. S. Velliangiri, K.K. Kasaraneni, Machine Learning and Deep Learning in cyber security for
IoT, in ICDSMLA 2019, (Springer, Singapore, 2020), pp. 975–981
15. K. Kandasamy, S. Srinivas, K. Achuthan, V.P. Rangan, IoT cyber risk: A holistic analysis of
cyber risk assessment frameworks, risk vectors, and risk ranking process. EURASIP J. Inf.
Secur. 2020, 1–18 (2020)
16. I.H. Sarker, Y.B. Abushark, F. Alsolami, A.I. Khan, Intrudtree: A machine learning based
cyber security intrusion detection model. Symmetry 12(5), 754 (2020)
17. I.H. Sarker, CyberLearning: Effectiveness analysis of machine learning security modeling
to detect cyber-anomalies and multi-attacks. Internet of Things 14, 100393 (2021)
18. N.F. Syed, Z. Baig, A. Ibrahim, C. Valli, Denial of service attack detection through machine
learning for the IoT. J. Inf. Telecommun 4(4), 482–503 (2020)
19. R. Damaševičius, A. Venčkauskas, J. Toldinas, Š. Grigaliūnas, Ensemble-based classifica-
tion using neural networks and machine learning models for windows PE malware detection.
Electronics 10(4), 485 (2021)
20. V.H. Subburaj, B.A. Pham, Cybersecurity: Using Machine Learning Algorithms to Detect
SQL Injection Attacks (2020)
21. M. Shafiq, Z. Tian, Y. Sun, X. Du, M. Guizani, Selection of effective machine learning algo-
rithm and Bot-IoT attacks traffic identification for internet of things in smart city. Futur. Gener.
Comput. Syst. 107, 433–442 (2020)
22. Z. Katzir, Y. Elovici, Quantifying the resilience of machine learning classifiers used for cyber
security. Expert Syst. Appl. 92, 419–429 (2018)
23. E. Anthi, L. Williams, M. Rhode, P. Burnap, A. Wedgbury, Adversarial attacks on machine
learning cybersecurity defences in industrial control systems. J. Inf. Secur. Appl. 58,
102717 (2021)
74 T. Daniya et al.

24. A. Bilen, A.B. Özer, Cyber-attack method and perpetrator prediction using machine learning
algorithms. PeerJ Comput. Sci 7, e475 (2021)
25. V.K. Singh, M. Govindarasu, A cyber-physical anomaly detection for wide-area protection
using machine learning. IEEE Transactions on Smart Grid. PP, 1 (2021)
26. I. Tariq, M.A. Sindhu, R.A. Abbasi, A.S. Khattak, O. Maqbool, G.F. Siddiqui, Resolving
cross-site scripting attacks through genetic algorithm and reinforcement learning. Expert Syst.
Appl. 168, 114386 (2021)
27. M. Ghiasi, M. Dehghani, T. Niknam, A. Kavousi-Fard, P. Siano, H.H. Alhelou, Cyber-attack
detection and cyber-security enhancement in smart DC-microgrid based on blockchain tech-
nology and Hilbert Huang transform. IEEE Access 9, 29429–29440 (2021)
28. Q. Li, L. Zhang, R. Zhou, Y. Xia, W. Gao, Y. Tai, Machine learning-based stealing attack of
the temperature monitoring system for the energy Internet of Things. Secur. Commun. Netw
2021, 1 (2021)
29. P. Kumar, G.P. Gupta, R. Tripathi, TP2SF: A trustworthy privacy-preserving secured frame-
work for sustainable smart cities by leveraging blockchain and machine learning. J. Syst.
Archit. 115, 101954 (2021)
30. S. Zhu, V. Saravanan, B. Muthu, Achieving data security and privacy across healthcare applica-
tions using cyber security mechanisms. Electron. Libr. 38, 979 (2020)
31. S. Liu, S. You, H. Yin, Z. Lin, Y. Liu, W. Yao, L. Sundaresh, Model-free data authentication for
cyber security in power systems. IEEE Trans. Smart Grid 11(5), 4565–4568 (2020)
32. S. Velliangiri, An enhanced multimedia video surveillance security using wavelet encryption
framework. J. Mobile Multimed 15(3), 239–254 (2020)
33. H. Karimipour, A. Dehghantanha, R.M. Parizi, K.K.R. Choo, H. Leung, A deep and scalable
unsupervised machine learning system for cyber-attack detection in large-scale smart grids.
IEEE Access 7, 80778–80788 (2019)
34. S. Velliangira, J. Premalata, A novel forgery detection in image frames of the videos using
enhanced convolutional neural network in face images. Comput. Model. Eng. Sci 125(2),
625–645 (2020)
35. M. Thakong, S. Phimoltares, S. Jaiyen, C. Lursinsap, One-pass-throw-away learning for
cybersecurity in streaming non-stationary environments by dynamic stratum network. PLoS
One 13(9), e0202937 (2018)
36. M. Kalinin, V. Krundyshev, P. Zegzhda, Cybersecurity risk assessment in Smart City infra-
structures. Mach. Des. 9(4), 78 (2021)
37. W. Serrano, The blockchain random neural network for cybersecure IoT and 5G infrastructure
in smart cities. J. Netw. Comput. Appl. 175, 102909 (2021)
38. J. Zhang, Y. Zhao, J. Wang, B. Chen, FedMEC: Improving efficiency of differentially private
federated learning via mobile edge computing. Mobile Netw. Appl 25(6), 2421–2433 (2020)
39. S. Latif, Z. Idrees, J. Ahmad, L. Zheng, Z. Zou, A blockchain-based architecture for secure and
trustworthy operations in the industrial Internet of Things. J. Ind. Inf. Integr. 21, 100190 (2021)
40. W. Abramson, A.J. Hall, P. Papadopoulos, N. Pitropakis, W.J. Buchanan, A distributed trust
framework for privacy-preserving machine learning, in International Conference on Trust and
Privacy in Digital Business, (Springer, Cham, 2020, Sept), pp. 205–220