- Research Article
- 10.47839/ijc.24.3.4198
Transforming IIoT Security Leveraging Deep Learning and Feature Selection for Superior Intrusion Detection
- Oct 02, 2025
- International Journal of Computing
- Lahcen Idouglid + 2 more +2
The Industrial Internet of Things (IIoT) has revolutionized industrial operations but has also brought forth significant cybersecurity challenges, demanding the development of advanced Intrusion Detection Systems (IDS). This study presents a feature-driven approach to enhance IDS performance in IIoT environments. By utilizing Recursive Feature Elimination (RFE) combined with Mutual Information (MI) for feature selection, we identified the most relevant attributes from the UNSW-NB15 dataset, improving detection accuracy while reducing computational complexity. Several deep learning models, including Convolutional Neural Networks (CNN), Residual Neural Networks (ResNet), Long Short-Term Memory (LSTM), and Bidirectional LSTM (BiLSTM), were evaluated. Among them, BiLSTM delivered the best performance, achieving a recall of 96.96%, an F1-score of 97.06%, and a Matthews Correlation Coefficient (MCC) of 0.93, outperforming other models in detecting complex attack patterns. However, its high computational cost, with training time exceeding 3500 seconds, underscores the need for optimization for real-time deployment. The results highlight the potential of combining feature selection techniques with deep learning models to enhance IDS for IIoT. Future work will focus on optimizing BiLSTM for faster deployment, integrating hybrid models, and testing across diverse datasets to further improve real-time security solutions for IIoT environments.
Read more