Machine Learning–Based Intrusion Detection Framework for Industrial Internet of Things Environments

Abstract

The Industrial Internet of Things has transformed manufacturing and critical infrastructure, yet introduces significant cybersecurity vulnerabilities. This research proposes a comprehensive machine learning-based intrusion detection framework tailored for IIoT environments. The framework integrates multiple ML algorithms including Random Forest, Support Vector Machines, and deep learning architectures to detect diverse cyberattacks. Using the Edge-IIoTset benchmark dataset containing 2.2 million instances across 15 attack categories, the proposed framework achieved 99.60% detection accuracy with minimal false positive rates. The study hypothesizes that hybrid ML models combining Convolutional Neural Networks with ensemble methods outperform traditional signature-based detection systems. Experimental validation demonstrates the framework's capability to identify DDoS attacks, Man-in-the-Middle intrusions, injection attacks, and malware threats in real-time. Results confirm superior performance across multiple IIoT protocols including MQTT, Modbus TCP/IP, and HTTP, establishing the framework's effectiveness for protecting Industry 4.0 infrastructures against evolving cyber threats.