Systematic review of intrusion detection system based on machine learning techniques for internet of things

The Internet of Things (IoT) has become an integral part of modern society, contributing to the intelligent development of various domains. However, the security challenges faced by IoT devices, such as data privacy, network bandwidth constraints, and cyberattacks, require effective solutions. Machine learning (ML)-based intrusion detection systems (IDS) offer a promising approach to enhancing IoT security by analysing real-time data from IoT devices. This paper provides a state-of-the-art analysis of the current state of machine learning techniques applied to IDS in the IoT. A thorough review of approximately 20 high-impact factor journal papers from IEEE, ScienceDirect, and Scopus was conducted. The review examines and summarizes the most employed ML techniques and benchmark datasets, such as BoT-IoT, N-BaIoT, NSL-KDD, and UNSW-NB15, available for IoT IDS and identifies promising avenues for future research. Finally, based on the findings of the review, some important research directions were identified. We highlighted the most efficient machine learning algorithms and feature selection techniques, along with strategies to address issues related to dataset imbalance in benchmarked datasets, with the goal of fostering the development of robust IDS solutions for IoT security.

Actions (login required)

View Item