Advancing Network Security in Industrial IoT: A Deep Dive into AI-Enabled Intrusion Detection Systems

Abstract

The increasing use of Industrial Internet of Things (IIoT) devices has heightened concerns about cybersecurity threats, particularly botnet attacks. Traditional internet communication methods have consistently faced these challenges, leading to substantial economic losses for numerous manufacturing enterprises. As machine-to-machine communications grow, these attacks are becoming more prevalent. This research addresses the critical need for an AI-powered network intrusion detection system. We conducted an extensive literature review and implemented over 25 advanced Machine Learning (ML) algorithms, including various modifications, to detect botnet attacks on seven IoT devices. The primary objective was to develop robust and accurate models for identifying security threats and allow for a comprehensive performance benchmark for all these models utilizing the same dataset. Our findings revealed that certain models achieved near-perfect performance in detecting botnet attacks, while others were less effective. Our contributions include identifying high-performing ML models for botnet detection and demonstrating their applicability across various IoT devices. Future research should focus on validating these models with new datasets and exploring how the type and function of IoT devices influence detection performance and response time.