The IDS Framework for Recognizing Unusual Activity in IoT Networks

Abstract

Protecting against cyberattacks is becoming increasingly crucial as the Internet of Things (IoT) and its resources expand rapidly. The most recent sophisticated attacks leverage characteristics of ordinary traffic to launch novel, covert attacks, while more conventional perimeter-based defences (such as firewalls and authentication) struggle to identify them. The IoT Network Intrusion Dataset (IoTID20) is used to test a lightweight network intrusion detection system designed specifically for the Internet of Things. The approach is based on a workflow for Knowledge Discovery in Databases (KDD) that incorporates thorough data preprocessing. To optimise the trade-off between processing efficiency and efficacy, Artificial Neural Networks (ANNs), K-Nearest Neighbours (KNN), and Random Forests (RFs) are the classifier models created and evaluated. They can also be adjusted for edge implementation. The usual metrics mentioned in the literature are followed while implementing the performance measure. According to the experiment’s findings, RF outperformed the state of the art by 1.2% in accuracy, achieving the best performance in real-time intrusion detection for IoT scenarios.