Network Intrusion Detection in Internet of Things (IoT)

Abstract

The open nature of the internet of things network makes it vulnerable to cyber-attacks. Intrusion detection systems aid in detecting and preventing such attacks. This paper offered a systematic review of studies on intrusion detection in IoT, focusing on metrics, methods, datasets, and attack types. This review used 33 network intrusion detection papers in 31 journals and 2 conference proceedings. The results revealed that the majority of the studies used generated or private datasets. Machine learning (ML)-based methods (85%) were used in the studies, while the rest used statistical methods. Eight categories of metrics were identified as prominent in evaluating IoT performance, and 94.9% of the ML-based methods employed average detection rate. Moreover, over 20 attacks on IoT networks were detected, with denial of service (DoS) and sinkhole being the majority. Based on the review, the future direction of research should focus on using public datasets, machine learning-based methods, and metrics such as resource consumption, energy consumption, and power consumption.