Advancing IoT Cybersecurity through AI and ML: A Comparative Study on Intrusion Detection and Privacy Protection
Michael Oghale Ighofiomoni *
Department of Computer Engineering and Systems Engineering, Southern Delta University, Ozoro, Nigeria.
Joel Segun Ojerinde
Department of Electrical and Electronics Engineering, University of Ilorin, Nigeria.
Olaniyan Iqmat
Department of Mathematics and Computer Science, Faculty of Basic and Applied Sciences, Institution, Elizade University, Nigeria.
Peter Paul Issah
Department of Computer Science, Kwame Nkrumah University of Science and Technology, Ghana.
Adepitan Oluwatosin Similoluwa
Department of Science Education (Mathematics), University of Ilorin, Nigeria.
Oyenuga Oluwaseun
Department of Electronics and Computer Engineering, Lagos State University, Nigeria.
Emmanuel Tomisin David
Department of Electrical Electronic Engineering, Olabisi Onabanjo University, Nigeria.
Adedotun Idowu
Department of Computer Engineering, Ladoke Akintola University of Technology, Nigeria.
Idowu Moyinoluwa Emmanuel
Department of Computer Science, Ladoke Akintola University of Technology, Nigeria.
Confidence Adimchi Chinonyerem
Department of Accountancy, Abia State Polytechnic, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
The wide use of Internet of Things (IoT) devices in residences and industries has brought unexpected ease, but concurrently, unprecedented new privacy attacks and cybersecurity threats. Classical security measures lag in tackling the dynamic and complex nature of IoT ecosystems due to limited resources and device variety. This study examines the use of Artificial Intelligence (AI) and Machine Learning (ML) methods to enhance the security posture of IoT ecosystems, specifically to counter data breaches and protect user privacy. Publicly available datasets, the TON_IoT and CICIDS2018 datasets, were used to benchmark the performance of several machine learning models, such as Random Forest (RF), Support Vector Machine (SVM), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM) networks. The models were trained and tested on classifying and labelling cyberattacks such as DoS attacks, reconnaissance, and data exfiltration attempts in IoT network traffic and telemetry logs. The findings indicate that CNN recorded the best detection accuracy (94.3% on TON_IoT and 96.2% on CICIDS2018) and performed better than traditional algorithms, whereas Random Forest recorded the best compromise between performance and computational cost and was thus appropriate for real-time use. The research affirms that intrusion detection in IoT networks can be dramatically enhanced through AI/ML methods and that model choice must be determined on the basis of deployment factors like available computational resources, as well as whether real-time processing is required.
Keywords: Artificial intelligence, internet of things, intrusion detection system, long short-term memory, data breach prevention, privacy preservation