Transforming Cyber Defense: AI, Intrusion Detection and the Future

of Security

S. Akilandeswari

, J. Amutha

, S. Sundar

, K. Rahapriya

, T. Sakthivel

and R. Divyabharathi

Department of AI&DS, E.G.S. Pillay Engineering College, Nagapattinam 611002, Tamil Nadu, India

Tech Lead, Ministry of Transportation, U.A.E.

E.G.S. Pillay Engineering College, Nagapattinam 611002, Tamil Nadu, India

Keywords: Intrusion Detection Systems, Cybersecurity, Machine Learning, Deep Learning, Network Security.

Abstract: Thus, Modern Intrusion Detection Systems (IDS) forms an essential part of the critical infrastructures in order

to detect and protect against unwanted malicious actions over networks and hosts. As the cyber threats are

becoming more innovative, advanced deep learning (DL) and machine learning (ML) techniques are widely

used to develop IDS with better performance. This survey focuses on the newest applications and trends in

the field of IDS with respect to current ideas and techniques within the discipline of ML and DL methods

being used along with the challenges they were developed to address and the limitations inherent in their

solutions. In addition, it summarizes the recent techniques, reviews the performance, and indicates the gap for

future research by developing the intelligent and adaptive Intrusion Detection System (IDS).

1 INTRODUCTION

Nishani, L. and Biba, M., 2016. With the rising

numbers and growing sophistication of cyberattacks,

Intrusion Detection Systems became evolved and

improved over time but retained their same core

identity that is still remarkably relevant. Presently, the

deployment of ML and DL provides the basis for all

modern IDS to ensure correct identification of the

irregularities and ascertain the possibility of an attack.

Siddiqui, et al., 2021. This section, both summarizing

current trends and appearing informational on ML

and DL potentiality for identifying advanced attacks

or the trajectory towards constant evolution.

However, this article provides a survey on the main

research publications focusing on the application of

contemporary deep learning-based techniques acting

as intrusion detection system in the fog computing

framework. Some issues regarding cybersecurity

emerge because of this approach, but we can say that

Fog Computing is a decentralized approach that

provides real-time data processing. Goyal, N.,et al,

2021,There are various types of neural networks used

to improve the behavior of IDS and some limitations

about their behavior during the scaling, which are due

to limitation of resources and its capabilities, such as

CNN, RNN, hybrid structures, etcany.

1.1 Deep Learning Paradigms

Goodfellow, I., et al., 2016 Approaches based on deep

learning have achieved impressive results isolating

the intricate patterns present in network traffic, for

example Convolutional Neural Networks (CNNs)

and Recurrent Neural Networks (RNNs). Thakkar,

A. and Lohiya, R., 2021, A few examples of models

built upon RNNs are able to classify malicious

network activity with an accuracy of over 90%.

1.2 Ensemble Learning Strategies

Abdan, M. and Seno, S.A.H., 2022, Ensemble

frameworks which combine different learning models

can improve detection accuracy and robustness.

Xiao, H., et al, 2015 Ensemble methods such as

random forests and boosting algorithms adapt better

to different attack scenarios compared to single

models.

1.3 Optimization Techniques for

Feature Selection

Kanthimathi, S. and Prathuri, J.R., 2020, To improve

this process further, optimization techniques like

genetic algorithm and particle swarm optimization

800

Akilandeswari, S., Amutha, J., Sundar, S., Rahapriya, K., Sakthivel, T. and Divyabharathi, R.

Transforming Cyber Defense: AI, Intrusion Detection and the Future of Security.

DOI: 10.5220/0013890200004919

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 2, pages

800-807

ISBN: 978-989-758-777-1

have been adopted more and more to fine grained the

feature selection process to help to improve the

performance of the IDS focusing on the most

relevant properties Shams, E.A. and Rizaner, A.,

2018.

2 ADVANCEMENTS IN

DETECTION

METHODOLOGIES

2.1 Hybrid Models

Sobehy, A., et al, 2020; Farahani, G., 2021. Hybrid

approaches, integrating supervised and unsupervised

techniques, combine the strengths of anomaly

detection and signature-based methods. Such models

are particularly effective in detecting zero-day attacks

and reducing false positives.

2.2 Proactive Threat Mitigation

Pan, Z., et al., 2020; Scherer, D., et al., 2010,

Proactive approaches leverage predictive analytics to

anticipate and prevent attacks before they occur.

Strategies include employing reinforcement learning

to adapt IDS models dynamically.

2.3 Cloud-Based Intrusion Detection

O'shea, K. and Nash, R., 2015.; Dalal, R., Khari, M.

and Hernandez, M., 2021, The transition to cloud-

centric infrastructures necessitates IDS optimized for

scalability and efficiency in virtualized environments.

Recent advancements emphasize lightweight models

for real-time anomaly detection in high-speed

networks.

3 CHALLENGES AND

RESEARCH GAPS

Despite advancements, IDS development faces

persistent challenges:

• Data Limitations: Access to diverse and

comprehensive datasets remains a critical

bottleneck. Ensuring datasets represent

varied attack scenarios is essential for model

generalization.

• Evolving Threats: The dynamic nature of

cyber threats demands IDS capable of

adapting to novel attack patterns without

manual intervention.

• Interpretability: Deep learning models,

while accurate, often lack transparency.

Enhancing interpretability is vital for

building trust in automated systems. The

comparative analysis of various IDS

approaches is shown in Table 1.

Table 1: Comparative analysis of IDS approaches.

Model

Type

Strengths Weaknesses Applications

Traditional

IDS

Simple,

rule-based

High false

positives,

inflexible

Basic

networks

Deep

Learning

(CNN)

High

accuracy,

pattern

recognition

High

computational

cost

Real-time

traffic

analysis

Deep

Learning

(RNN)

Handles

sequential

data

Slow for large

datasets

Anomaly

detection

Hybrid

Models

Combines

strengths of

multiple

approaches

Complex to

implement,

resource-

intensive

Resource-

constrained

environments

3.1 Key Findings

• Effectiveness of Deep Learning Models:

Studies demonstrate that CNNs and RNNs

significantly improve the accuracy of IDS by

recognizing patterns in large datasets.

• Scalability and Efficiency: Hybrid models

combine cloud and edge processing for

scalability while optimizing resource usage.

• Challenges in Explainability: Capuano, N., et

al., 2022 The fact that deep learning models are

opaque that requires integration of explainable

AI (XAI) to improve trust and usability.

The research underscores the critical role of deep

learning in advancing IDS for fog computing. While

progress has been significant, addressing challenges

like interpretability, data diversity, and ethical

considerations are essential for future developments.

Transforming Cyber Defense: AI, Intrusion Detection and the Future of Security

801

3.2 Detailed Insights from Literature

Review

Table 2: Comparison between traditional and deep

learning-based IDS.

Criteria

Traditional

IDS

DeepLearning-

Based IDS

Adaptability Limited to

redefined rules

Learns and adapts

dynamicall

Accuracy

Moderate,

prone to false

ositives

High, robust

anomaly

detection

Scalability Difficult to

scale

Easily scalable with

cloud and edge

inte

ration

Interpretabili

High (rule-

ased

)

Low (requires

lainableAI

)

The literature reveals the evolution of Intrusion

Detection Systems (IDS) from traditional methods to

sophisticated deep learning models tailored for fog

computing environments. Key highlights include:

• Transition from Rule-Based to Intelligent

Systems: Laqtib, S., et al, 2019; Nweke, et

al., 2018 Traditional IDS relied on

predefined rules, which limited their ability

to adapt to emerging threats. Deep learning

approaches have addressed these limitations

by enabling dynamic anomaly detection.

• Advancements in Neural Architectures:

Bjerrum, E.J. and Threlfall, R., 2017.

Convolutional Neural Networks (CNN)

excel in spatial data analysis, while

Recurrent Neural Networks (RNN) handle

temporal data effectively. Hybrid models

integrating multiple architectures offer

promising results.

• Challenges in Real-World Deployment:

Issues such as computational overhead, lack

of interpretability, and data diversity remain

significant barriers to adoption.The

comparison of Traditional and Deep

Learning-Based IDS is shown in Table 2.

3.3 Methodologies and Key Studies

A variety of methodologies have been employed in

the development of IDS for fog computing. Table 3

highlights some key studies and their contributions.

Table 3: Methodologies, key findings and challenges.

Methodology

Findin

Challenges

CNN-based

IDS

95% accuracy in

anomaly detection

High

computational

cost

Hybrid

CNN-LSTM

Improved

detection of

sequential

atterns

Complex

implementation

Transfer

learning

Enhanced

performance with

less training data

Limited bydata

scarcity

Real-time IDS Significant

reductionin

response time for

DDoS

attacks

Resource

constraints

3.4 Comprehensive Comparison of

AI-Based Intrusion Detection

Systems (IDS)

The comparison between various review articles

based on their focus, approach, and methodology

regarding AI-based Intrusion Detection Systems

(IDS) is shown in Table 4.

The comparative analysis across various

methodologies and systems reveals that while deep

learning-based IDS offer significant advantages in

terms of accuracy and adaptability, they require

substantial computational resources and face

challenges related to interpretability and real-world

deployment. Hybrid models and explainable AI are

promising directions to address these issues.

Table 4: Comparison of AI-based intrusion detection

systems (IDS).

Reference

NIDS

Focused

Appro

ach

Specific

IDS

(

SIDS

)

HybridIDS

Aziz, Et al,

✓

ML,

✓

Ahmad,et al,

✓

ML,

✓

Hodo, E., et al.

✓

✓ ✓

Sultana, Et al.

✓

Moustafa, et

al.,

✓

XAI

✓ ✓

Fejrskov, Et al,

✓

Lunt, T.F.

✓

Axelsson ML

✓

Liao

✓

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

802

3.5 Literature Review on Key Themes

Literature reviews play a fundamental role in the

scenario of academic research, providing a systematic

framework through which a researcher can synthesize

existing knowledge, identify predominant topics and

discover vital gaps to establish a basis for greater

research. In the domain of models for the detection of

efficient intruders that use deep learning techniques

for fog computing, the importance of the exhaustive

review of literature cannot be exaggerated

Poongothai, T. and Duraiswamy, K., 2014, These

revisions serve not only as a complete description of

the knowledge accumulated in the field, but also

function as a critical evaluation tool that encourages

the academic discourse necessary for the progress of

knowledge.

Popli, R., et al., 2021 The proliferation of

computing Nieblahas introduced a myriad of

advances in cloud computing paradigms. The unique

architecture of fog computing, characterized by its

decentralized and distributed nature, introduces

complexities and challenges in cybersecurity, which

requires robust intruder detection mechanisms. This

intersection of deep learning and computing fog

requires a detailed exploration of existing literature to

present not only the technological advances that have

been made but also the critical theoretical and

empirical frameworks applicable within this context.

Hussain, A., et al., 2020; Pilli, E.S., 2018.

Research in intruder detection has gone through

several methodological approaches ranging from

statistical analysis to automatic learning. However,

the advent of deep learning has redefined these

approaches, offering new frames that demonstrate

higher performance by recognizing patterns in large

datasets. A review of the literature that focuses on this

area reveals significant issues, such as the evolution

of intruder detection techniques, from traditional

methods to sophisticated neural networks that focus

in the extraction of characteristics and the

optimization of performance. Studies have illustrated

how CNN and RNN can effectively analyse complex

data flows in real - time fog computer networks,

significantly improving intrusions detection rates

while maintaining the low positive fake relationships.

Janicke, H., et al., 2019., In addition, the

systematic analysis of the methodologies used in

these studies is essential to understand the advances

made in the field. Several researchers have used

various experimental designs, including comparative

analysis of different deep learning models, set

techniques and the merger of automatic learning with

rules -based systems. Mambo, M., et al., 2018. This

methodological diversity articulates the multifaceted

nature of cybersecurity challenges in computing fog

and underlines the need for personalized solutions

that can dynamically respond to evolving threats. An

examination of these methodologies not only informs

researchers about best practices, but also highlights

the deficiencies present in current research, preparing

the scenario for innovative solutions.

Karasfi, B., et al., 2024. Insights derived from the

corpus of studies listed in Table 1 above concerning

deep learning-based intrusion detection in fog

computing highlight the urgent need for an effective

solution. Numerous experiments provide evidence of

how deep -learning methods positively impact the

accuracy and performance characteristics of intrusion

detection systems, as they help to overcome the

aforementioned stagnation issues that have

characterized long-standing traditional techniques.

Moreover, these results shed light on the practical

relevance of such approaches in the real world, e.g.,

taking into account the scalability, adaptability and

efficiency of critical computing resources in fog

environments.

In summary, the exploration of the existing

literature within the scope of intrusions within the

computing environments serves as a critical

component to advance in this field in constant

evolution. This systematic approach not only helps

delineate the existing knowledge, but also facilitates

the identification of gaps that demand more research,

ultimately contributing to the development of

improved intruder’s detection models capable of

safeguarding the integrity of the computer systems of

fog against an increasingly sophisticated landscape of

cyber security threats. The evolution of intrusions

detection systems (ID) was significantly modelled by

the proliferation of different processing

environments, in particular with the advent of fog

technologies. IDs, originally designed to monitor and

analyse network traffic for suspicious activities, have

more and more integrated methodologies advanced to

adapt to the complexities of modern IT

infrastructures. One of the predominant themes in

literature is the transition from traditional ID based on

the network to more dynamic systems and sensitive

to the context capable of operating in distributed

environments.

Min, G., et al., 2017, Another critical theme that

emerges from literature is the challenge of dealing

with large quantities of data generated in the fog

computing scenarios. The researchers identified the

need for scalable architecture that exploit the

distributed nature of the calculation of the fog to

process and analyse the data efficiently. This led to

Transforming Cyber Defense: AI, Intrusion Detection and the Future of Security

803

the proposal of hybrid models that combine the

strengths of the centralized elaboration of the cloud

with the calculation solutions of the edges, allowing

a more robust and resilient IDS framework. These

models often use learning techniques of ensembles,

which aggregate forecasts from various deep learning

models to improve accuracy and reduce false positive

rates, a common problem in many IDS

implementations.

Kumar, V., et al., 2017. In addition, security and

privacy problems in the fog calculation environments

introduce further levels of complexity that are

addressed in existing literature. The dynamic nature

limited to the resources of the nodes of fog requires

the development of light learning models that

maintain performance by minimizing computational

general expenses. Techniques such as the pruning of

the model and the distillation of knowledge have been

designed as a means of optimizing deep learning

algorithms for distribution in bound environments,

ensuring that ID can operate efficiently without

compromising safety measures.

Venkatraman, S., et al., 2019., The integration of

deep learning within IDS is not without critical points

of view. Some scholars underline the interpretation of

profound learning models, raising concerns about the

opaque decision -making processes relating to these

algorithms. This problem has significant implications

for computer security applications, in which the

understanding of the logic underlying the surveys of

the threats is crucial for the response to accidents.

Ranjan, R., et al., 2023.Consequently, literature has

also explored methods to improve interpretation, such

as the integration of AI Techniques of explainable

(XAI), allowing safety analysts to obtain insights on

the underlying processes of detection of anomalies.

In summary, the intersection of intrusions

detection systems, fog and deep learning

methodologies has led to significant progress in the

safety measures within modern processing

environments. Key themes emerging from literature

underline the adaptability of IDS technologies to meet

rapid changes in the generation of data and in the

processing requirements, while facing the intrinsic

challenges posed by scalability, performance and

interpretation., The examination of methodologies

employed in the current research on intrusions has

revealed a significant change in the implementation

of deep learning approaches, particularly in the

context of fog computing environments. Sukarno, P.,

et al., 2024.The need for robust and efficient intrusion

detection systems (IDS) is underlined by the growing

complexity and scale of network architectures that

integrate fog computing, characterized by their

distributed nature and heterogeneous resources.

Recent literature highlights that traditional intrusion

detection techniques, usually limited by set sets of

predefined rules and inability to adapt to evolving

threats, may be insufficient to meet the demands of

contemporary digital communication infrastructures.

Ai, X., et al., 2023 Moreover, deep learning

methodologies are not only applicable to traditional

network intrusion scenarios, but also were effectively

adapted to improve pedagogical approaches,

particularly in educational environments. Deep

learning models can be used to analyze writing and

understanding, which can be used to detect not only

cyber security threats, but also indicative patterns of

student challenges. This interdisciplinary application

illustrates the movement towards a holistic view of

intrusive detection systems, where the resilience of

educational platforms against potential vulnerabilities

is fundamental.

Tao, X., et al., 2021.In addition to CNNS and

RNNS, a variety of hybrid models have also emerged

that combine the strengths of multiple deep learning

architectures to further improve detection

capabilities. For example, the integration of short-

term memory networks (LSTM) with CNNs has

shown to produce higher results in the recognition of

sequential patterns in the typical data flows of

attempted intrusion. Such innovations emphasize a

methodological trend for the adoption of set learning

techniques that amalgamates various model

predictions to improve general performance and

robustness.

The literature suggests that the complexity of

cyber threats requires a multidisciplinary approach,

drawing from areas such as data science, behavioral

analysis and network theory. The need for

comprehensive literature revisions becomes even

more pronounced in considering the wide range of

emerging challenges - from rapidly changing threat

actors to the wide variety of environments in which

intrusive detection systems are implanted. A strict

analysis of previous research provides a fundamental

context to categorize these challenges and identify

gaps in which more innovation is imperative.

Data errors and complexity in diverse domains

warrant methodological and technological evolution

for effective resolution of intrusions, especially in

adaptive contexts such as those in networks under

consideration of IoT and Cloud computing methods.

As successful initiatives demonstrate, addressing

these modern challenges requires not just an up

technological infrastructure, but also a re-

examination of the geostrategic paradigms that have

historically guided research. These roles highlight the

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

804

importance of ongoing engagement with the existing

literature to shape new questions and research

approaches that address modern cyber security needs.

Last but not least, the review pushes scholars and

practitioners to explore the literature further.

Findings of previous section emphasize how in-depth

revision drives future analyses and increment

success of intrusion detection model. This

amalgamation will help the community go beyond the

limitations of both the frameworks and the emerging

technologies while paving the way for novel

frameworks that would help us in the continuous

evolution of the field and its effective mitigation of

threats. Therefore, the future of intrusion detection

research should be guided by literature reviews that

meet high standards.

4 FUTURE DIRECTIONS

• Explainable AI: Developing interpretable IDS

models will foster collaboration between human

analysts and automated systems, improving

overall threat management.

• Adversarial Robustness: Enhancing resilience

to adversarial attacks is critical. This includes

designing models that can withstand deliberate

perturbations in input data.

• Integration with Emerging Technologies:

Future IDS will benefit from integration with

technologies such as blockchain and quantum

computing for enhanced security and scalability.

5 CONCLUSIONS

Deep learning and machine learning techniques have

merged together to result in improved capabilities of

intrusion detection systems. By addressing current

problems and exploring emerging technologies,

researchers can offer robust defensive capabilities

against the ever-evolving cyber threat landscape and

thus lay the groundwork for next-generation IDS

technology.

Deep learning approaches such as RNN, wrapper-

based feature maps, long short-term memory

(LSTM) are interesting. The goal of these approaches

is to improve the performance and resilience of IDS

against advanced cyber-attacks. In addition,

optimization strategies, on the other hand, integrate

PSO, global optimization algorithms, nature-

inspired optimization, and metaheuristic methods.

This in turns allows to make Systems more reliable,

optimize their performance and accurate detection.

Studies of ensemble learning and hybrid models

show that proposed methods are neural networks,

hybrid approaches, and combined classifiers. This not

only improves the performance of IDS but also

addresses real world attacks and increases accuracy,

while reducing false alarms. In addition,

benchmarking and comparison studies provide useful

insights about the benefits and drawbacks of deep

learning models compared to traditional machine

learning models. The importance of cloud-oriented

security solutions speaks to the need for specialized

intrusion detection systems suited for the cloud-

based architectures. Cloud computing networks must

adapt to shifting security threats using scalable,

efficient and secure approaches.

These studies are aimed at protecting wireless

platforms and developing dedicated intrusion

detection systems (IDS) for secure wireless

communications. In conclusion, proactive intrusion

detection research can be referred to as a preventive

approach to detect intrusions beforehand. These

models increase the IDSs’ ability to combat new

threats by focusing on the optimization of search

algorithms and by converging quickly.

REFERENCES

Abdan, M. and Seno, S.A.H., 2022. Machine learning

methods for intrusive detection of wormhole attack in

mobile ad hoc network (MANET). Wireless

Communications and Mobile Computing, 2022(1),

p.2375702.

Ahmad, Z., Shahid Khan, A., Wai Shiang, C., Abdullah, J.

and Ahmad, F., 2021. Network intrusion detection

system: A systematic study of machine learning and

deep learning approaches. Transactions on Emerging

Telecommunications Technologies, 32(1), p.e4150.

Ahmim, A., Maglaras, L., Ferrag, M.A., Derdour, M. and

Janicke, H., 2019, May. A novel hierarchical intrusion

detection system based on decision tree and rules-based

models. In 2019 15th International conference on

distributed computing in sensor systems (DCOSS) (pp.

228-233). IEEE.

Axelsson, S., 2000. Intrusion detection systems: A survey

and taxonomy. Chalmers Univ. Technol., Gothenburg,

Sweden, Tech. Rep. 99-15

Aziz, A.S.A., Sanaa, E.L. and Hassanien, A.E., 2017.

Comparison of classification techniques applied for

network intrusion detection and classification. Journal

of Applied Logic, 24, pp.109-118.

Bjerrum, E.J. and Threlfall, R., 2017. Molecular generation

with recurrent neural networks (RNNs). arXiv preprint

arXiv:1705.04612.

Transforming Cyber Defense: AI, Intrusion Detection and the Future of Security

805

Capuano, N., Fenza, G., Loia, V. and Stanzione, C., 2022.

Explainable artificial intelligence in cybersecurity: A

survey. Ieee Access, 10, pp.93575-93600.

Dalal, R., Khari, M. and Hernandez, M., 2021. Persuasive

simulation of optimized protocol for OppNet. Dynamic

Systems and Applications, 30(5), pp.865-900.

Dwivedi, R., Dave, D., Naik, H., Singhal, S., Omer, R.,

Patel, P., Qian, B., Wen, Z., Shah, T., Morgan, G. and

Ranjan, R., 2023. Explainable AI (XAI): Core ideas,

techniques, and solutions. ACM Computing

Surveys, 55(9), pp.1-33.

Farahani, G., 2021. Black hole attack detection using K‐

nearest neighbor algorithm and reputation calculation

in mobile ad hoc networks. Security and

communication Networks, 2021(1), p.8814141.

Fejrskov, M., Pedersen, J.M. and Vasilomanolakis, E.,

2020, June. Cyber-security research by ISPs: a NetFlow

and DNS anonymization policy. In 2020 International

Conference on Cyber Security and Protection of Digital

Services (Cyber Security) (pp. 1-8). IEEE.

Ghimire, S., Yaseen, Z.M., Farooque, A.A., Deo, R.C.,

Zhang, J. and Tao, X., 2021. Streamflow prediction

using an integrated methodology based on

convolutional neural network and long short-term

memory networks. Scientific Reports, 11(1), p.17497.

Goodfellow, I., Bengio, Y. and Courville, A Deep

Learning. Cambridge, MA, USA: MIT Press, 2016.

Hamza, F. and Maria Celestin Vigila, S., 2019. Review of

machine learning-based intrusion detection techniques

for MANETs. In Computing and Network

Sustainability: Proceedings of IRSCNS 2018 (pp. 367-

374). Springer Singapore

Hodo, E., Bellekens, X., Hamilton, A., Tachtatzis, C. and

Atkinson, R., 2017. Shallow and deep networks

intrusion detection system: A taxonomy and

survey. arXiv preprint arXiv:1701.02145.

Ieracitano, C., Adeel, A., Morabito, F.C. and Hussain, A.,

2020. A novel statistical analysis and autoencoder

driven intelligent intrusion detection

approach. Neurocomputing, 387, pp.51-62.

Kanthimathi, S. and Prathuri, J.R., 2020, November.

Classification of misbehaving nodes in MANETS using

machine learning techniques. In 2020 2nd PhD

Colloquium on Ethically Driven Innovation and

Technology for Society (PhD EDITS) (pp. 1-2). IEEE

Kim, S. and Park, K.J., 2021. A survey on machine-learning

based security design for cyber-physical

systems. Applied Sciences, 11(12), p.5458.

Laqtib, S., Yassini, K.E. and Hasnaoui, M.L., 2019,

October. A deep learning method for intrusion

detection systems-based machine learning in MANET.

In Proceedings of the 4th international conference on

smart city applications (pp. 1-8).

Laqtib, S., El Yassini, K. and Hasnaoui, M.L., 2020. A

technical review and comparative analysis of machine

learning techniques for intrusion detection systems in

MANET. International Journal of Electrical and

Computer Engineering, 10(3), p.2701.

Liao, H.J., Lin, C.H.R., Lin, Y.C. and Tung, K.Y., 2013.

Intrusion detection system: A comprehensive

review. Journal of network and computer

applications, 36(1), pp.16-24.

Liu, Y., Fieldsend, J.E. and Min, G., 2017. A framework of

fog computing: Architecture, challenges, and

optimization. IEEE Access, 5, pp.25445-25454.

Lunt, T.F., 1993. A survey of intrusion detection

techniques. Computers & Security, 12(4), pp.405-418.

Mishra, P., Varadharajan, V., Tupakula, U. and Pilli, E.S.,

2018. A detailed investigation and analysis of using

machine learning techniques for intrusion

detection. IEEE communications surveys &

tutorials, 21(1), pp.686-728.

Moustafa, N., Koroniotis, N., Keshk, M., Zomaya, A.Y. and

Tari, Z., 2023. Explainable intrusion detection for cyber

defences in the internet of things: Opportunities and

solutions. IEEE Communications Surveys &

Tutorials, 25(3), pp.1775-1807

Mukherjee, M., Matam, R., Shu, L., Maglaras, L., Ferrag,

M.A., Choudhury, N. and Kumar, V., 2017. Security

and privacy in fog computing: Challenges. IEEE

Access, 5, pp.19293-19304.

Najafli, S., Toroghi Haghighat, A. and Karasfi, B., 2024.

Taxonomy of deep learning-based intrusion detection

system approaches in fog computing: a systematic

review. Knowledge and Information Systems, 66(11),

pp.6527-6560.

Nishani, L. and Biba, M., 2016. Machine learning for

intrusion detection in MANET: a state-of-the-art

survey. Journal of Intelligent Information Systems, 46,

pp.391-407.

Nishani, L. and Biba, M., 2016. Machine learning for

intrusion detection in MANET: a state-of-the-art

survey. Journal of Intelligent Information Systems, 46,

pp.391-407.

Nweke, H.F., Teh, Y.W., Al-Garadi, M.A. and Alo, U.R.,

2018. Deep learning algorithms for human activity

recognition using mobile and wearable sensor

networks: State of the art and research

challenges. Expert Systems with Applications, 105,

pp.233-261.

O'shea, K. and Nash, R., 2015. An introduction to

convolutional neural networks. arXiv preprint

arXiv:1511.08458

Pan, Z., Wang, Y. and Pan, Y., 2020. A new locally

adaptive k-nearest neighbor algorithm based on

discrimination class. Knowledge-Based Systems, 204,

p.106185.

Pandey, A., Kumar, S., Pattanaik, B. and Pattnaik, M.,

2021. A Survey: Machine Learning Algorithms for

Network Security. SSRN Electron. Journal

Poongothai, T. and Duraiswamy, K., 2014, February.

Intrusion detection in mobile AdHoc networks using

machine learning approach. In International

Conference on Information Communication and

Embedded Systems (ICICES2014) (pp. 1-5). IEEE

Popli, R., Sethi, M., Kansal, I., Garg, A. and Goyal, N.,

2021, August. Machine learning based security

solutions in MANETs: State of the art approaches.

In Journal of physics: conference series (Vol. 1950, No.

1, p. 012070). IOP Publishing

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

806

Popli, R., Sethi, M., Kansal, I., Garg, A. and Goyal, N.,

2021, August. Machine learning based security

solutions in MANETs: State of the art approaches.

In Journal of physics: conference series (Vol. 1950, No.

1, p. 012070). IOP Publishing.

Roman, R., Lopez, J. and Mambo, M., 2018. Mobile edge

computing, fog et al.: A survey and analysis of security

threats and challenges. Future Generation Computer

Systems, 78, pp.680-698.

Scherer, D., Müller, A. and Behnke, S., 2010, September.

Evaluation of pooling operations in convolutional

architectures for object recognition. In International

conference on artificial neural networks (pp. 92-101).

Berlin, Heidelberg: Springer Berlin Heidelberg.

Shams, E.A. and Rizaner, A., 2018. A novel support vector

machine-based intrusion detection system for mobile ad

hoc networks. Wireless Networks, 24, pp.1821-1829

Siddiqui, M.N., Malik, K.R. and Malik, T.S., 2021, May.

Performance analysis of blackhole and wormhole

attack in MANET based IoT. In 2021 International

Conference on Digital Futures and Transformative

Technologies (ICoDT2) (pp. 1-8). IEEE.

Sobehy, A., Renault, É. and Mühlethaler, P., 2020. CSI-

MIMO: K-nearest neighbor applied to indoor

localization. In ICC 2020-2020 IEEE International

Conference on Communications (ICC) (pp. 1-6). IEEE.

Sultana, N., Chilamkurti, N., Peng, W. and Alhadad, R.,

2019. Survey on SDN based network intrusion

detection system using machine learning

approaches. Peer-to-Peer Networking and

Applications, 12(2), pp.493-501.

Thakkar, A. and Lohiya, R., 2021. A review on machine

learning and deep learning perspectives of IDS for IoT:

recent updates, security issues, and

challenges. Archives of Computational Methods in

Engineering, 28(4), pp.3211-3243.

Vinayakumar, R., Alazab, M., Soman, K.P.,

Poornachandran, P., Al-Nemrat, A. and Venkatraman,

S., 2019. Deep learning approach for intelligent

intrusion detection system. IEEE access, 7, pp.41525-

41550.

Wardana, A.A., Kołaczek, G., Warzyński, A. and Sukarno,

P., 2024. Collaborative intrusion detection using

weighted ensemble averaging deep neural network for

coordinated attack detection in heterogeneous

network. International Journal of Information

Security, 23(5), pp.3329-3349.

Weng, C., Chen, C. and Ai, X., 2023. A pedagogical study

on promoting students' deep learning through design-

based learning. International journal of technology and

design education, 33(4), pp.1653-1674.

Xiao, H., Biggio, B., Nelson, B., Xiao, H., Eckert, C. and

Roli, F., 2015. Support vector machines under

adversarial label contamination. Neurocomputing, 160,

pp.53-62.

Transforming Cyber Defense: AI, Intrusion Detection and the Future of Security

807