CyberSecurity Intrusion Detection in Industry 4.0 WSN’s Using ML/DL

Abstract

The increasing deployment of Wireless Sensor Networks (WSNs) in industrial environments exposes critical systems to a variety of cyberattacks. Traditional intrusion detection systems (IDS) often struggle with real-time performance and scalability in resource-constrained environments, limiting their effectiveness in Industry 4.0 applications. This study aims to propose a hybrid machine learning-based IDS that improves detection accuracy, computational efficiency, and adaptability for real-time deployment in industrial WSNs. The proposed system integrates Stacking Classifiers, XGBoost, and Adaboost ensemble learning techniques, designed to enhance attack detection capabilities while minimizing computational overhead. The model was trained and evaluated on a publicly available WSN dataset, employing a 70:30 train-test split and 10-fold cross-validation to ensure robust performance. The system was benchmarked against traditional models like Decision Trees and Random Forests. The proposed model achieved an accuracy of 95.8%, outperforming baseline models (Decision Tree: 84.6%, Random Forest: 90.4%) in terms of recall and F1-score. The system also demonstrated significant computational efficiency, requiring less training time compared to deep learning models while maintaining high detection accuracy. This study presents a novel hybrid IDS solution that balances high accuracy with computational efficiency, making it suitable for real-time intrusion detection in resource-constrained industrial environments. The contributions offer a practical solution for securing WSNs in Industry 4.0, with scalable and adaptable capabilities to detect emerging threats.