Securing IoT Attacks: A Machine Learning Approach for Developing Lightweight Trust-Based Intrusion Detection System

Abstract

The routing process in the Internet of Things (IoT) presents challenges in industrial applications due to its complexity, involving multiple devices, critical decision-making, and accurate data transmission. The complexity further increases with dynamic IoT devices, which creates opportunities for potential intruders to disrupt routing. Traditional security measures are inadequate for IoT devices with limited battery capabilities. Although RPL (Routing Protocol for Low Energy and Lossy Networks) is commonly used for IoT routing, it remains vulnerable to security threats. This study aims to detect and isolate three routing attacks on RPL: Rank, Sybil, and Wormhole. To achieve this, a lightweight trust-based secured routing system is proposed, utilizing machine learning techniques to derive values for devices in new networks, where initial trust values are unavailable. The system demonstrates successful detection and isolation of attacks, achieving an accuracy of 98.59%, precision of 98%, recall of 99%, and f-score of 98%, thereby reinforcing its effectiveness. Attacker nodes are identified and promptly disabled, ensuring a secure routing environment. Validation on a generated dataset further confirms the reliability of the system.