| Abstract: |
The exponential proliferation of Industrial Internet of Things (IIoT) networks has revolutionized industrial automation while simultaneously introducing critical cybersecurity vulnerabilities that threaten critical infrastructure. This research investigates machine learning-based intrusion detection systems for IIoT environments, addressing the challenges posed by heterogeneous devices, resource constraints, and evolving cyber threats. The study employed deep learning architectures including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, and hybrid models on benchmark datasets Edge-IIoTset and CIC-IoT-DIAD 2024. Through comprehensive experimental analysis, the proposed CNN-LSTM hybrid model achieved 98.7% accuracy, 98.2% precision, 97.9% recall, and 98.1% F1-score in detecting multi-class intrusion scenarios. Results demonstrate superior performance over traditional signature-based approaches in identifying Denial of Service (DoS), Distributed Denial of Service (DDoS), Man-in-the-Middle (MITM), and injection attacks. Statistical analysis reveals significant improvements in detection rates while maintaining minimal false positive rates below 1.5%. The findings indicate that deep learning-based intrusion detection systems effectively enhance IIoT network security by providing real-time threat detection capabilities with high accuracy across diverse attack vectors. |
