A Systematic Review of Intrusion and Anomaly Detection Techniques in IoT
A Systematic Review of Intrusion and Anomaly Detection Techniques in IoT
Seyed Amin Heydari1 Hossein Izadi2 Mohammad Ghanbri3
1) Department of Computer Engineering, National University of Skills (Shahrekord), Chaharmahal and Bakhtiari, Iran
2) B.Sc. Student in Computer Engineering, National University of Skills (Shahrekord), Chaharmahal and Bakhtiari, Iran
3) B.Sc. Student in Computer Engineering, National University of Skills (Shahrekord), Chaharmahal and Bakhtiari, Iran
IOI Code
:
AAHD-ABHZBE
Publication :
National Conference on Information Technology, Nanotechnology, Artificial Intelligence and Technological Futures Studies(itnaf.ir)
Abstract :
The growing adoption of Internet of Things, IoT, devices has introduced significant security challenges, necessitating robust intrusion detection systems tailored to resource-constrained, heterogeneous environments. This survey provides a comprehensive review of contemporary intrusion and anomaly detection techniques for IoT systems. We systematically examine three core pillars: first, feature engineering approaches, from classical statistical methods to advanced selection algorithms; second, machine learning paradigms, including supervised, unsupervised, and ensemble methods; and third, advanced deep learning architectures such as convolutional neural networks, recurrent neural networks, autoencoders, and recent Transformer-based large language models. Furthermore, we address critical deployment challenges including resource constraints, adversarial attacks, dataset limitations, and privacy preservation via federated learning. By synthesizing findings from key studies published primarily between 2020 and 2026, while acknowledging foundational literature, this paper offers researchers a structured taxonomy of current techniques and identifies open research gaps toward secure, scalable IoT intrusion detection.
Keywords : internet of things intrusion detection anomaly detection feature engineering machine learning deep learning