IAES International Journal of Artificial Intelligence (IJ-AI)

Vol. 13, No. 3, September 2024, pp. 2490~2497

ISSN: 2252-8938, DOI: 10.11591/ijai.v13.i3.pp2490-2497  2490

Learning methodologies towards leveraging security resiliency

in internet-of-things environment

Sowmya Somanath1,2, Usha Banavikal Ajay3

1
Department of Computer Science and Engineering, BMS Institute of Technology and Management, Visvesvaraya Technological
University, Belagavi, India
2
School of Computer Science and Engineering, REVA University, Yelahanka Bangalore, India
3
Department of Information Science and Engineering, BMS Institute of Technology and Management, Visvesvaraya Technological
University, Belagavi, India

Article Info ABSTRACT

Article history: The evolution of artificial intelligence (AI) has faciliated a significant
contribution of machine learning and deep learning in order to improvise the
Received Nov 28, 2023 security features of large internet-of-things (IoT) environment. Since last
Revised Feb 9, 2024 decade there has been different variants of learning-based methodologies
Accepted Mar 2, 2024 towards leveraging security improvements among communication in IoT
devices; however, it is yet to know the strength and weakness of them.
Hence, this paper presents a review of security methodologies adopted in
Keywords: machine learning and deep learning-based techniques in IoT to understand
the degree of resiliency and effectiveness of these techniques. The paper
Deep learning further contributes towards highlighting the current methodologies with
Internet-of-things respect to benefits and limiting factors along with exclusive highlights of
Intrusion detection research trends while the research gap explored assists in offering these
Machine learning insights. The distinct findings of the study assist in paving the work direction
Security in future by harnessing better form of learning scheme.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Sowmya Somanath
School of Computer Science and Engineering, REVA University
Rukmini Knowledge Park, Yelahanka, Kattigenahalli, Bengaluru, Karnataka 560064, India
Email: [email protected]

1. INTRODUCTION
The internet-of-things (IoT) offers a comprehensive deployment of larger number of heterogeneous
connected nodes in order to formulate a well-developed networked system. With a vast number of projected
applications in IoT, the number of evolving applications is still on rise [1]. However, there are some of the
potential issues too associated with it. The first challenge is associated with the weaker’s authentication
which renders the IoT devices exposed to various vulnerable threats and attacks [2]. Different types of IoT
devices also lacks robust and sustainable encryption schemes which let the malicious node to intrude the
network by bypassing the weaker security system and gain an illegitimate access to the sensitive and
confidential information [3]. Apart from this, the processing power is quite limited within an IoT device
which renders inapplicability towards processing high end algorithms towards intrusion detection system. At
present, the manufacturing of the IoT devices is carried out using varied ranges of software and hardware
which has a reported cases of vulnerabilities while the security updates in the form of patches may not be
facilitated by the manufacturers. Further, a distributed nature of IoT makes the nodes working at different
geographical location that are interconnected with different network using different set of protocols, which
cannot be controlled so effectively in case of serious threats [4]. Needless to mention that human factor is
another reason for multiple security breaches in IoT. It is to be noted that there are various study models

Journal homepage: http://ijai.iaescore.com

Int J Artif Intell ISSN: 2252-8938  2491

towards strengthening IoT security in current times using different methodologies; however, the prime
obstactle of successful implementation of same models in practical scenario is less assessed due to
complexity associated with it. IoT system is characterized by high end complexity where incorporating a
potential security solution can be carried out only by expertise in multiple advanced areas of cryptography,
network security, software development, and hardware design. From the perspective of solution towards IoT
security, machine learning approach has proven to its higher degree of effectiveness that emphazied on
algorithms and data using artificial intelligence (AI) for better solutions. Machine learning offers a set of
features which is useful for protecting the IoT devices as well as network from multiple forms of threats [5].
Following are the set of problems associated with implying machine learning for IoT security viz.
i) the explainability and interpretability of machine learning models are quite challenging for large IoT
security, ii) adversarial attacks are still the large level of security problems in machine learning models which
leads to outliers by tampering the data in order to deceive the learning algorithm, iii) resource constraint has
been always a bigger impediment to implement even a strongest encryption algorithm or iterative learning
schemes, iv) machine learning algorithm performance is highly sensitive towards the data quality, v) the
success of accuracy towards detection and mitigation largely depends upon training data, which may not be
facilitated in large IoT environment in security perspective.
For the purpose of realization of above-mentioned problem statement, it is necessary to brief
relevant literatures. Several researchers have presented discussion of solution and issues associated with IoT
threats [6], [7]. Al-Garadi et al. [8] have presented discussion about various security methods lined with both
machine learning and deep learning followed by highlights of comprehensive evaluation of security trends in
[9], [10]. Nguyen et al. [11] have discussed about usage of federated learning approach while discussion of
reinforcement learning-based approach is discussed in [12], [13] towards IoT security. Wu et al. [14] have
presented discussion about AI based security solutions, while Zhou et al. [15] have presented discussion of
methodologies to identify bugs in systems and platforms in IoT.
From the insight of the research-based articles, it is clear that there are various deployment of
learning-based scheme targeting protection of IoT security; however, there is a need to further update the
information with latest studies as well as there is need to explicitely identify the degree of effectiveness in
existing studies. Therefore, the proposed manuscript presents a compact and yet resourceful information
associated with strength and weakness of current significant methodologies of both machine and deep
learning approaches for IoT security. The new value added in this work are i) explicit highlights of
machine/deep learning approaches with respect to specific problems being addressed, advantage, and
limitation, ii) compact snapshot of current research trend, iii) arriving at research gap as the prime
constribution of study, and iv) inferring learning outcomes of proposed study to be helpful for upcoming
research work direction.

2. METHODOLOGY
The prime intention of the proposed study is to carry out review of effectivess associated with the
learning-based algorithms and methodologies involved in accomplishing a common goal of optimal IoT
security. For this purpose, this review study has adopted a desk research methodology [16] exhibited in
Figure 1. According to adopted research methodology, the preliminary search was carried out from Google
Scholar as well as reputed technical journal publications. The next step of the adopted method is to perform
an initial filtering on the basis of inclusion and exlusion criteria. The primary inclusion criteria of filtering are
that the published article to mainly posse’s methodological elaboration with results accomplished in order to
understand the effectiveness of approach. The second inclusion criteria also consider the research articles
with technical implementation published between 2013-2023.
The third inclusion criteria are that research articles are only associated with either machine learning
or deep learning as the core implementation. The exclusion criteria are any theoretical papers without results
or implementation model and papers published before 2013. Also, the exclusion criteria include any paper
which has used non-learning-based approach towards improving IoT security as the target of this paper is to
understand the potential of learning-based schemes. The next round of the operation is to perform removal of
the duplicates. The proposed method considers duplicated as i) same author with two extended study models,
ii) usage of exactly similar concept in two different research articles is also considered as duplicates as the
idea is to obtain completely unique study model. The preliminary search yield approximately 31,000 results,
while the initial filtering has resulted in 270 paper. Finally, the removal of the duplicates has resulted in 54
number of research articles which has been reviewed in this manuscript. The learning outcome of the
proposed study further contributes towards exploring research gap, research trend, and exploring distinct
study findings.

Learning methodologies towards leveraging security resiliency in internet-of-things … (Sowmya Somanath)

2492  ISSN: 2252-8938

Preliminary search

Initial Filtering

Inclusion Exclusion
Criteria Criteria

Remove Duplicates

Explore Research Extract Learning Explore Distinct

Trend Outcomes Study Findings

Explore Research Gap

Figure 1. Method adopted in proposed study

3. RESULTS
This section presents discussion about the various learning-based techniques which were utilized for
identifying and preventing various forms of threats and thereby acted as a unique threat mitigation scheme in
IoT environment. It has been also noted that adoption of learning-based schemes is of varied forms where
some of the study models have used individual learning scheme while some of them has used a combined
implementation of learning schemes. Broadly, it is classified into machine learning and deep learning-based
schemes towards threat mitigation in an IoT environment. This section showcases the outcome of these
reviewed approaches.

3.1. Literature survey for machine learning scheme

The study model of Majeed et al. [17] have used naïve Bayes along with blockchain for detecting
the security threats in the data captured by the drone. Adoption of naïve Bayes was also witnessed in work of
Setiadi et al. [18] in order to identify the denial-of-service (DoS) attack in IoT. However, the work carried
out by Jadhav and Pellakuri [19] have developed a detection scheme by integrating naïve Bayes with other
learning approaches e.g., support vector machine (SVM), and K-nearest neighbor (KNN), which are also
another variant of classification algorithms in machine learning.
Existing IoT security approaches have also been reported to use logistic regression (LR).
Korystin et al. [20] have used LR method towards assessing predictive risk associated with the data of
cybersecurity. Li et al. [21] have developed a LR-based scheme in order to identify the anomaly in the
system behaviour in IoT using statistical learning scheme of neural network. Adoption of LR is also reported
in work of Subramanian et al. [22] towards developing a trust model for identifying the malicious node.
Further, clustering approaches has been reported in machine learning approach towards grouping the types of
identified threat. Kammoun et al. [23] have presented a unique clustering approach towards strengthening the
trust system in IoT powered by edge computing. Further adoption of clustering approach is also carried out
by Yadav and Gupta [24] where a reward system has been introduced in order to promote the detection
system. The scheme has also used federated learning for increasing the detection accuracy. The work carried
out by Wu et al. [25] have presented a unique scheme of machine learning integrated with blockchain for
tracing the trust factor. The multiple learning models of random forest, AdaBoost, and decision tree
ensembled with boosting.
Apart from this, SVM is another frequently used classification approach towards addressing security
concerns in IoT. Bagaa et al. [26] have used single class SVM for classifying the threats along with using
network virtualization and software defined network (SDN). Ezhilarasi and Clement [27] have used SVM
along with gated recurrent unit (GRU) in order to difference between malicious and legitimate user in
cognitive radio network, which is another integral part of an IoT system. Ioannou and Vassiliou [28] have
presented a unique mechanism to implement SVM for identifying the normal and abnormal behaviour in IoT
associated with anomaly detection. Existing scheme has also witnessed usage of artificial neural network
(ANN) for improving security performance in IoT. Al-Mohammed et al. [29] have integrated ANN with deep

Int J Artif Intell, Vol. 13, No. 3, September 2024: 2490-2497

Int J Artif Intell ISSN: 2252-8938  2493

learning technique in order to secure the key distribution process in IoT by detecting the presence of attacker.
Pacheco et al. [30] have also used ANN for identifying intrusion towards the fog nodes in IoT by
investigating the adaptive behaviour of nodes. Further, Sarkar et al. [31] have presented a scheme where
ANN is combinely used with nature-inspired search technique for solving key exchange problem in IoT.
Further, adoption of random neural network is seen in work of Latif et al. [32] in order to predict the
detection of an attacker in industrial IoT use case. Ferrag et al. [33] have used decision tree integrated with
rule-based scheme in order to develop a unique classifier for attack detection in IoT. Adoption of decision
tree approach is also seen in work of Zarzoor et al. [34] where decision tree is integrated with spiking neural
network for minimizing latency while performing predictive attack detection.

3.2. Literature survey for deep learning scheme

Apart from machine-learning based schemes, adoption of deep learning scheme is equivalently on
rise towards improving the security features in IoT. Alasmary et al. [35] have deployed recurrent neural
network (RNN) and long short-term memory (LSTM) for identifying distributed denial-of-service (DDoS).
The study model emphasized more on classification of the DDoS attacks. Further adoption of RNN is
reported in work of Liao et al. [36] towards securing the unreliable storage units in IoT. The model has used
network coding for this purpose. Nearly similar like of research work is also witnessed in the study model of
Ullah and Mahmoud [37] where an anomaly detection model has been developed using RNN. This scheme
has used GRU along with birectional long short-term memory (BiLSTM), and conventional LSTM for this
purpose. The study has been assessed with multiple standard datasets to find its higher accuracy in its threat
detection performance. Zeeshan et al. [38] have developed a scheme to identify both DoS and DDoS attack
on specific dataset of IoT. The scheme has used LSTM for this purpose of detection under varied threat
scheme while the study has selected equal number of data packets from different threat categories in order to
address the overfitting problem.
Adoption of autoencoder towards detection of intrusion is carried out by Alshudukhi et al. [39] with
an idea to select the feature that can contribute towards treat identification. Study towards similar direction is
also carried out by Lee et al. [40] in order to identify the impersonation attack. Further adoption of
autoencoder along with consideration of temporate attribute was reported in work of Salahuddin et al. [41]
where the study model targets to identify the DDoS attack. The study model uses multiple time window
considering temporal-based attributes rowards identifying anomaly behaviour of DDoS attackers. It is also
noted that integration of autoencoder and transfer learning too offers a better attack detection performance as
noted in work of Vu et al. [42]. The predictive model is capable of analyzing both labelled and unlabelled
data acquired from multiple IoT devices using two type soft autoencoders in both supervised and
unsupervised form. Further, the hidden representation of unsupervised autoencoder is subjected to transfer
learning that is further used for detection of attacks.
In the area of deep learning, convolution neural network (CNN) is another frequently deployed
threat detection scheme in IoT. Jeon et al. [43] have developed a security model using CNN for dynamically
investigating propagation of malware in nested environment of cloud. Adoption of CNN is also reported in
work of Li et al. [44] towards identification of malware in IoT. The study model revises CNN scheme by
integrating spatial pyramid pooling with self-attention approach towards classifying different variants of
malwares. Another model developed by More et al. [45] have used CNN with a core idea of this
implementation is to ensure secure transmission of medical images in IoT along with resolving the
computational speed towards processing large number of images. Existing scheme has also witnessed
deployment of optimized version of CNN by adding transfer learning for improving security threat detection
as noted in work of Okey et al. [46]. The model has also implemented quantile transformer for finetuning the
feature vector obtained from images which are then subjected to ensemble algorithms of transfer learning.
Another unique adoption of deep learning towards leveraging IoT security was seen in work carried out by
Taiwo et al. [47]. The study has used CNN for detection of intrusion associated with the automation of home.
Zhang et al. [48] have constructed a security model towards strengthenin the privacy of the data considering
social IoT environment. The study has used CNN integrated with firefly algorithm in order to accomplish the
security features. The firefly algorithm is used for generating adversarial sample while CNN is used for
analyzing similarity of the data. Zhou et al. [49] have used enhanced Bayesian convolution network for
improving the predictive quality of data in wearable IoT.
There are also studies where machine learning and deep learning has been combinely used to
mitigate the potential threat in IoT [50]-[54]. The frequently identified limitation of all the studies is lack of
benchmarking, usage of sophisticated learning approach, lack of extensive analysis to prove, not applicable
for complex intrusion, not applicable for dynamic attackers, and accuracy depends upon dataset size, induces
complexity for large network.

Learning methodologies towards leveraging security resiliency in internet-of-things … (Sowmya Somanath)

2494  ISSN: 2252-8938

3.3. Research trend

This part of study will only emphasize towards understanding the trends of adoption of machine
learning and deep learning-based schemes towards effective threat control in IoT. A closer look into the
Table 1 showcases that there are a greater number of research articles published for machine learning-based
approaches total 42,040 publications, in comparison to deep learning-based approaches total 35,585
publications in IoT. These outcomes of publications correspond to research articles published during
2013-2023 only. Table 1 showcases that there are approximately 63 implementation approaches for
classification methods, 20 approaches of regression, 19 approaches for clustering, and approximately 3
approaches for reinforcement based IoT security solution. Number of studies using regression-based machine
learning approaches are still very less to be noticed. Further, it was seen that adoption of supervised deep
learning methods (approximated mean number of publications = 128) are fairly more in contrast to that of
unsupervised deep learning methods (approximated mean number of publications = 6) towards addressing
security issues in IoT as seen in Table 2.

Table 1. Frequently used approaches in machine learning-based IoT security models

Classification No. of publication
Support vector machine 81
Decision tree 79
Random forest 53
K-nearest neighbor 49
Naïve Bayes 56
Regression No. of publication
Logistic regression 56
Lassoregression 1
Support vector regression 4
Clustering No. of publication
K-Means 59
DBScan 9
Agglomerative hierarchical 1
Gaussian mixture 9
Reinforcement learning No. of publication
Q-Learning 7
R-Learning 2
Temporal difference-learning 1

Table 2. Frequently used approaches in deep learning-based IoT security models

Supervised No. of publication Unsupervised No. of publication
Multilayered-perceptron 96 Generative adversarial network 4
CNN 571 Autoencoder 10
RNN 54 Self-organizing map 4
LSTM 43 Restricted Boltzman machine 4
GRU 2 Deep belief network 9
BiLSTM 3

3.4. Research gap

There is no denying the fact that a significant number of research work has been carried out towards
strengthening the IoT security system in modern era where machine learning schemes acts as a significant
contributor. Irrespective of evolving dedicated research contribution, there are various open-ended
shortcoming which are required to be addressed. Following are some of the significant research gap
identified from the proposed review work:
‒ Issues with data acquisition and analysis: majority of the existing approaches using machine learning has
been witnessed to directly subject its algorithm either on publicly available dataset or on its synthetic
data. This task is carried out without any realization that IoT environment is usually massive and
potentially generates unstructured data which further poses potential challenges in analytical operations.
Hence, the higher ranges of acquired accuracy is not rationalized with adopted dataset in existing machine
learning-based techniques on IoT security.
‒ Lack of device constraint modelling: the different ranges of devices used in IoT doesn’t have much
resources nor the processing capability. Apart from this, the processing power is highly limited. Adoption
of deep learning-based scheme as well as iterative machine learning will always demands higher
processing power which cannot be facilitated owing to lack of device constraint modelling. There are few
reported studies where different constraints of device has been considered in implementation.

Int J Artif Intell, Vol. 13, No. 3, September 2024: 2490-2497

Int J Artif Intell ISSN: 2252-8938  2495

‒ Non-inclusion of compatibility factors: there is no doubt that both machine and deep learning-based
approaches towards identifying and mitigating IoT security is quite proven productive. However, there
are various variants of learning schemes within the above two standard mechanism which has different
working principle. For an example, the studies using decision tree approach may yield better result in
normal IoT scenario, however, when the scenario becomes complex with sopshiticated variables, they
suffer from overfitting. At the same time, KNN-based approach may offer better classification accuracy
but they demand higher memory. Similarly, RNN and CNN offers robust anomaly detection with highest
accuracy; however, they have dependency of labelled data as well as specific set of computational
resources in order to carry out training.
‒ Less effective privacy preservation: there are various studies carried out towards privacy preservation in
IoT security; however, such models are designed under restricted research environment. On the other
hand, the aggregation and transmission of data by an IoT device in practical scenario lacks such
consideration on higher scale of deployment. At the sametime, when such higher sensitive information is
exposed to learning algoithm for training purpose, the algorithms itself is at risk. Current, no
benchmarked model is stated to resist learning-based approach to go rogue.
‒ Less emphasis towards computational burden: almost all the studies carried out using learning-based
scheme emphasized towards accomplishing a superior accuracy; however, there is less justification
offered to state the reduced computational effort or burden. Accuracy accomplishment may offer better
security service but practical viability of the study model cannot be offered without benchmarking or
without proving reduction in computational burden over extensive test environment.

4. CONCLUSION
This paper has presented discussion about the effectiveness of machine learning as well as deep
learning approaches towards identifying the threats and mitigating them in an IoT environment. Following
are some of the essential findings of the study: i) the security system used by machine learning models are
found to learn towards mapping the input to the output for detecting the anomalies or threatful pattern of the
data. ii) the security solutions introduced by deep learning approaches is mainly witnessed to deploy a neural
network-based approach in order to perform learning and constructing decision towards threat detection. It
was also noted that deep learning models are found to offer an advantage from its counterpartner machine
learning models by making themselves independent of linear models. Various forms of complex operations
can be handled by the deep learning model with an aid of activation function of non-linear forms over every
layer of neural network. iii) both the AI approaches is characterized by dependencies of performing an
iterative and training operation which largely demands either higher central processing unit (CPU) for
machine learning of graphics processing unit (GPU) for deep learning. Unfortunately, such a large demand of
resources for performing analytical operation is not suitable for resource-constraint sensor nodes. Existing
security mechanism has showcased the accomplishment of accuracy, however, there is no much potential
evidence for their viability when exposed to dynamic practical environment. iv) apart from DDoS attack,
which is highly frequent in IoT environment, there are also evolving number of attacks i.e., botnets, malware,
side-channel attack, domain name server (DNS) spoofing, man-in-middle attack, and physical attack, which
has not received much attention in the form of solution. v) majority of the security solution presented in
existing studies are highly specific of attack model, which renders inapplicability of those models when
exposed to different attack environment. As IoT is a large network of heterogeneous IoT devices and
protocols, a security solution much be intelligent enough to understand the vulnerability and resist them.
vi) existing security solutions using learning-based approach has been developed using pre-defined
information of attack strategies which doesn’t work when encountered with dynamic form of attackers.The
future work will be indirection towards improving machine learning scheme for developing better form of
evolving security approaches in IoT. Further, the study can be to optimize the deep learning autoencoders
which can offer better data quality and reliability towards deep learning approach. Hence, a hybrid scheme
can be designed for IoT security.

REFERENCES
[1] M. H. Ali, M. M. Jaber, S. K. Abd, A. Alkhayyat, M. R. Q, and M. H. Ali, “Application of internet of things-based efficient security
solution for industrial,” Production Planning & Control, pp. 1–15, 2023, doi: 10.1080/09537287.2023.2169647.
[2] S. Yempally, S. K. Singh, and V. Sarveshwaran, “A secure and efficient authentication and multimedia data sharing approach in IoT-
healthcare,” Imaging Science Journal, vol. 71, no. 3, pp. 277–298, 2023, doi: 10.1080/13682199.2023.2180140.
[3] H. N. Khan, A. Das, and A. Chaudhuri, “Unique video encryption technique intended for smart city application,” IETE Journal of
Research, vol. 69, no. 9, pp. 5830–5839, 2023, doi: 10.1080/03772063.2023.2190543.
[4] J. Bacquet, R. Riemenschneider, and P. W. -Jensen, “Future trends in IoT,” in Next Generation Internet of Things – Distributed
Intelligence at the Edge and Human-Machine Interactions, New York: River Publishers, 2022, pp. 9–17. doi: 10.1201/9781003338963-2.

Learning methodologies towards leveraging security resiliency in internet-of-things … (Sowmya Somanath)

2496  ISSN: 2252-8938

[5] L. Sana, M. M. Nazir, M. Iqbal, L. Hussain, and A. Ali, “Anomaly detection for cyber internet of things attacks: a systematic review,”
Applied Artificial Intelligence, vol. 36, no. 1, Dec. 2022, doi: 10.1080/08839514.2022.2137639.
[6] P. Anand, Y. Singh, A. Selwal, M. Alazab, S. Tanwar, and N. Kumar, “IoT vulnerability assessment for sustainable computing: threats,
current solutions, and open challenges,” IEEE Access, vol. 8, pp. 168825–168853, 2020, doi: 10.1109/ACCESS.2020.3022842.
[7] V. Hassija, V. Chamola, V. Saxena, D. Jain, P. Goyal, and B. Sikdar, “A survey on IoT security: application areas, security threats, and
solution architectures,” IEEE Access, vol. 7, pp. 82721–82743, 2019, doi: 10.1109/access.2019.2924045.
[8] M. A. Al-Garadi, A. Mohamed, A. K. Al-Ali, X. Du, I. Ali, and M. Guizani, “A survey of machine and deep learning methods for internet
of things (IoT) security,” IEEE Communications Surveys & Tutorials, vol. 22, no. 3, pp. 1646–1685, 2020, doi:
10.1109/comst.2020.2988293.
[9] Y. Harbi, Z. Aliouat, A. Refoufi, and S. Harous, “Recent security trends in internet of things: a comprehensive survey,” IEEE Access,
vol. 9, pp. 113292–113314, 2021, doi: 10.1109/ACCESS.2021.3103725.
[10] P. L. S. Jayalaxmi, R. Saha, G. Kumar, M. Conti, and T.-H. Kim, “Machine and deep learning solutions for intrusion detection and
prevention in IoTs: a survey,” IEEE Access, vol. 10, pp. 121173–121192, 2022, doi: 10.1109/ACCESS.2022.3220622.
[11] D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, and H. Vincent Poor, “Federated learning for internet of things: a
comprehensive survey,” IEEE Communications Surveys & Tutorials, vol. 23, no. 3, pp. 1622–1658, 2021, doi:
10.1109/comst.2021.3075439.
[12] A. Uprety and D. B. Rawat, “Reinforcement learning for IoT security: a comprehensive survey,” IEEE Internet of Things Journal, vol. 8,
no. 11, pp. 8693–8706, 2021, doi: 10.1109/JIOT.2020.3040957.
[13] M. Venkatasubramanian, A. H. Lashkari, and S. Hakak, “IoT malware analysis using federated learning: a comprehensive survey,” IEEE
Access, vol. 11, pp. 5004–5018, 2023, doi: 10.1109/ACCESS.2023.3235389.
[14] H. Wu, H. Han, X. Wang, and S. Sun, “Research on artificial intelligence enhancing internet of things security: a survey,” IEEE Access,
vol. 8, pp. 153826–153848, 2020, doi: 10.1109/ACCESS.2020.3018170.
[15] W. Zhou et al., “Reviewing IoT security via logic bugs in IoT platforms and systems,” IEEE Internet of Things Journal, vol. 8, no. 14, pp.
11621–11639, Jul. 2021, doi: 10.1109/JIOT.2021.3059457.
[16] Y. Kunneman, M. A. D. M. -Filho, and J. V. D. Waa, “Data science for service design: an introductory overview of methods and
opportunities,” The Design Journal, vol. 25, no. 2, pp. 186–204, 2022, doi: 10.1080/14606925.2022.2042108.
[17] R. Majeed, N. A. Abdullah, and M. F. Mushtaq, “IoT-based cyber-security of drones using the naïve Bayes algorithm,” International
Journal of Advanced Computer Science and Applications, vol. 12, no. 7, 2021, doi: 10.14569/ijacsa.2021.0120748.
[18] F. F. Setiadi, M. W. A. Kesiman, and K. Y. E. Aryanto, “Detection of dos attacks using naive bayes method based on internet of things
(IoT),” Journal of Physics: Conference Series, vol. 1810, no. 1, 2021, doi: 10.1088/1742-6596/1810/1/012013.
[19] A. D. Jadhav and V. Pellakuri, “Highly accurate and efficient two phase-intrusion detection system (TP-IDS) using distributed processing
of HADOOP and machine learning techniques,” Journal of Big Data, vol. 8, no. 1, 2021, doi: 10.1186/s40537-021-00521-y.
[20] О. Korystin, S. Nataliia, and O. Mitina, “Risk forecasting of data confidentiality breach using linear regression algorithm,” International
Journal of Computer Network and Information Security, vol. 14, no. 4, pp. 1–13, Aug. 2022, doi: 10.5815/ijcnis.2022.04.01.
[21] F. Li, A. Shinde, Y. Shi, J. Ye, X.-Y. Li, and W. Song, “System statistics learning-based IoT security: feasibility and suitability,” IEEE
Internet of Things Journal, vol. 6, no. 4, pp. 6396–6403, 2019, doi: 10.1109/jiot.2019.2897063.
[22] A. K. Subramanian, A. Samanta, S. Manickam, A. Kumar, S. Shiaeles, and A. Mahendran, “Linear regression trust management system
for IoT systems,” Cybernetics and Information Technologies, vol. 21, no. 4, pp. 15–27, 2021, doi: 10.2478/cait-2021-0040.
[23] N. Kammoun, R. Abassi, and S. Guemara, “Towards a new clustering algorithm based on trust management and edge computing for IoT,”
2019 15th International Wireless Communications & Mobile Computing Conference (IWCMC). IEEE, 2019. doi:
10.1109/iwcmc.2019.8766492.
[24] K. Yadav and B. B. Gupta, “Clustering based rewarding algorithm to detect adversaries in federated machine learning based IoT
environment,” in 2021 IEEE International Conference on Consumer Electronics (ICCE), IEEE, Jan. 2021, pp. 1–6. doi:
10.1109/ICCE50685.2021.9427586.
[25] Y. Wu, X. Jin, H. Yang, L. Tu, Y. Ye, and S. Li, “Blockchain-based internet of things: machine learning tea sensing trusted traceability
system,” Journal of Sensors, vol. 2022, pp. 1–16, 2022, doi: 10.1155/2022/8618230.
[26] M. Bagaa, T. Taleb, J. B. Bernabe, and A. Skarmeta, “A machine learning security framework for IoT systems,” IEEE Access, vol. 8, pp.
114066–114077, 2020, doi: 10.1109/ACCESS.2020.2996214.
[27] I. E. Ezhilarasi and J. C. Clement, “GRU-SVM based threat detection in cognitive radio network,” Sensors, vol. 23, no. 3, Jan. 2023, doi:
10.3390/s23031326.
[28] C. Ioannou and V. Vassiliou, “Network attack classification in IoT using support vector machines,” Journal of Sensor and Actuator
Networks, vol. 10, no. 3, 2021, doi: 10.3390/jsan10030058.
[29] H. A. Al-Mohammed et al., “Machine learning techniques for detecting attackers during quantum key distribution in IoT networks with
application to railway scenarios,” IEEE Access, vol. 9, pp. 136994–137004, 2021, doi: 10.1109/ACCESS.2021.3117405.
[30] J. Pacheco, V. H. Benitez, L. C. F. -Herran, and P. Satam, “Artificial neural networks-based intrusion detection system for internet of
things fog nodes,” IEEE Access, vol. 8, pp. 73907–73918, 2020, doi: 10.1109/ACCESS.2020.2988055.
[31] A. Sarkar, M. M. Singh, M. Z. Khan, and O. H. Alhazmi, “Nature-inspired gravitational search-guided artificial neural key exchange for
IoT security enhancement,” IEEE Access, vol. 9, pp. 76780–76795, 2021, doi: 10.1109/ACCESS.2021.3082262.
[32] S. Latif, Z. Zou, Z. Idrees, and J. Ahmad, “A novel attack detection scheme for the industrial internet of things using a lightweight random
neural network,” IEEE Access, vol. 8, pp. 89337–89350, 2020, doi: 10.1109/ACCESS.2020.2994079.
[33] M. A. Ferrag, L. Maglaras, A. Ahmim, M. Derdour, and H. Janicke, “RDTIDS: rules and decision tree-based intrusion detection system
for internet-of-things networks,” Future Internet, vol. 12, no. 3, 2020, doi: 10.3390/fi12030044.
[34] A. R. Zarzoor, N. A. S. Al-Jamali, and D. A. A. Qader, “Intrusion detection method for internet of things based on the spiking neural
network and decision tree method,” International Journal of Electrical and Computer Engineering (IJECE), vol. 13, no. 2, pp. 2278–
2288, 2023, doi: 10.11591/ijece.v13i2.pp2278-2288.
[35] F. Alasmary, S. Alraddadi, S. Al-Ahmadi, and J. Al-Muhtadi, “ShieldRNN: a distributed flow-based DDoS detection solution for IoT
using sequence majority voting,” IEEE Access, vol. 10, pp. 88263–88275, 2022, doi: 10.1109/ACCESS.2022.3200477.
[36] C. -H. Liao, H. -H. Shuai, and L.-C. Wang, “RNN-assisted network coding for secure heterogeneous internet of things with unreliable
storage,” IEEE Internet of Things Journal, vol. 6, no. 5, pp. 7608–7622, 2019, doi: 10.1109/jiot.2019.2902376.
[37] I. Ullah and Q. H. Mahmoud, “Design and development of RNN anomaly detection model for IoT networks,” IEEE Access, vol. 10, pp.
62722–62750, 2022, doi: 10.1109/ACCESS.2022.3176317.
[38] M. Zeeshan et al., “Protocol-based deep intrusion detection for DoS and DDoS attacks using UNSW-NB15 and Bot-IoT data-sets,” IEEE
Access, vol. 10, pp. 2269–2283, 2022, doi: 10.1109/ACCESS.2021.3137201.
[39] A. F. Alshudukhi, S. A. Jabbar, and B. Alshaikhdeeb, “A feature selection method based on auto-encoder for internet of things intrusion

Int J Artif Intell, Vol. 13, No. 3, September 2024: 2490-2497

Int J Artif Intell ISSN: 2252-8938  2497

detection,” International Journal of Electrical and Computer Engineering (IJECE), vol. 12, no. 3, pp. 3265–3275, 2022, doi:
10.11591/ijece.v12i3.pp3265-3275.
[40] S. J. Lee et al., “IMPACT: impersonation attack detection via edge computing using deep autoencoder and feature abstraction,” IEEE
Access, vol. 8, pp. 65520–65529, 2020, doi: 10.1109/ACCESS.2020.2985089.
[41] M. A. Salahuddin, V. Pourahmadi, H. A. Alameddine, M. F. Bari, and R. Boutaba, “Chronos: DDoS attack detection using time-based
autoencoder,” IEEE Transactions on Network and Service Management, vol. 19, no. 1, pp. 627–641, 2022, doi:
10.1109/TNSM.2021.3088326.
[42] L. Vu, Q. U. Nguyen, D. N. Nguyen, D. T. Hoang, and E. Dutkiewicz, “Deep transfer learning for IoT attack detection,” IEEE Access, vol.
8, pp. 107335–107344, 2020, doi: 10.1109/ACCESS.2020.3000476.
[43] J. Jeon, J. H. Park, and Y.-S. Jeong, “Dynamic analysis for IoT malware detection with convolution neural network model,” IEEE Access,
vol. 8, pp. 96899–96911, 2020, doi: 10.1109/ACCESS.2020.2995887.
[44] Q. Li, J. Mi, W. Li, J. Wang, and M. Cheng, “CNN-based malware variants detection method for internet of things,” IEEE Internet of
Things Journal, vol. 8, no. 23, pp. 16946–16962, 2021, doi: 10.1109/jiot.2021.3075694.
[45] S. More et al., “Security assured CNN-based model for reconstruction of medical images on the internet of healthcare things,” IEEE
Access, vol. 8, pp. 126333–126346, 2020, doi: 10.1109/access.2020.3006346.
[46] O. D. Okey, D. C. Melgarejo, M. Saadi, R. L. Rosa, J. H. Kleinschmidt, and D. Z. Rodriguez, “Transfer learning approach to IDS on cloud
IoT devices using optimized CNN,” IEEE Access, vol. 11, pp. 1023–1038, 2023, doi: 10.1109/access.2022.3233775.
[47] O. Taiwo, A. E. Ezugwu, O. N. Oyelade, and M. S. Almutairi, “Enhanced intelligent smart home control and security system based on
deep learning model,” Wireless Communications and Mobile Computing, vol. 2022, pp. 1–22, 2022, doi: 10.1155/2022/9307961.
[48] P. Zhang, Y. Wang, N. Kumar, C. Jiang, and G. Shi, “A security- and privacy-preserving approach based on data disturbance for
collaborative edge computing in social IoT systems,” IEEE Transactions on Computational Social Systems, vol. 9, no. 1, pp. 97–108, Feb.
2022, doi: 10.1109/TCSS.2021.3092746.
[49] Z. Zhou, H. Yu, and H. Shi, “Human activity recognition based on improved Bayesian convolution network to analyze health care data
using wearable IoT device,” IEEE Access, vol. 8, pp. 86411–86418, 2020, doi: 10.1109/ACCESS.2020.2992584.
[50] N. M. Y. -Naula, C. V. -Rosales, and J. A. P. -Diaz, “SDN-based architecture for transport and application layer DDoS attack detection by
using machine and deep learning,” IEEE Access, vol. 9, pp. 108495–108512, 2021, doi: 10.1109/ACCESS.2021.3101650.
[51] M. Savic et al., “Deep learning anomaly detection for cellular IoT with applications in smart logistics,” IEEE Access, vol. 9, pp. 59406–
59419, 2021, doi: 10.1109/access.2021.3072916.
[52] P. Sudhakaran, C. Malathy, T. H. Vardhan, and T. Sainadh, “Detection of malware from IoT devices using deep learning techniques,”
Journal of Physics: Conference Series, vol. 1818, no. 1, 2021, doi: 10.1088/1742-6596/1818/1/012219.
[53] M. -Q. Tran et al., “Reliable deep learning and IoT-based monitoring system for secure computer numerical control machines against
cyber-attacks with experimental verification,” IEEE Access, vol. 10, pp. 23186–23197, 2022, doi: 10.1109/access.2022.3153471.
[54] X. Zhou, W. Liang, W. Li, K. Yan, S. Shimizu, and K. I. K. Wang, “Hierarchical adversarial attacks against graph-neural-network-based
IoT network intrusion detection system,” IEEE Internet of Things Journal, vol. 9, no. 12, pp. 9310–9319, Jun. 2022, doi:
10.1109/JIOT.2021.3130434.

BIOGRAPHIES OF AUTHORS

Sowmya Somanath received the B.Eng. degree in Computer Science and

Engineering from VTU University in 2008, M.Tech. degree in Computer Science and
Engineering from VTU university in 2015, and currently pursuing Ph.D. under Visvesvaraya
Technological University Belgavi, in security in IoT using machine learning domain.
Currently, she is an assistant professor at the Department of Computer Science and
Engineering, REVA University. She is a member of International Association of Engineers
(IAENG). Her research interests include internet of things, network security, and machine
learning. She can be contacted at email: [email protected].

Dr. Usha Banavikal Ajay is a dedicated motivational individual committed to

maximizing learning opportunities in diverse academic settings using consistent and organized
practices. Energetic and ambitious professional with 18 years of experience in teaching. She
has completed her research in the area of information security. She Obtained her Ph.D. from
Visvesvaraya Technological University, Belagavi, Karnataka, India in the year 2016. She has
published more than 35 research papers in reputed international journal and conferences where
some of them are Scopus indexed and also has good impact factors. Currently, she is working
as professor in the Department of Information Science and Engineering, BMS Institute of
Technology and Management, Bengaluru. Her research interests include IoT, information
security, cyber security, data mining, artificial intelligence, and machine learning. She can be
contacted at email: [email protected].

Learning methodologies towards leveraging security resiliency in internet-of-things … (Sowmya Somanath)