Securing Water Distribution Systems: Leveraging Sensor Networks Against Cyber-Physical Attacks Using Advanced Chicken Swarm Optimization

Water distribution systems (WDSs) are essential infrastructures susceptible to cyber-physical attacks (CPAs) and thus require advanced defense mechanisms. This study introduces a novel, scalable approach for real-time CPA detection in WDS using cloud computing and hybrid metaheuristic algorithms. The proposed framework uniquely combines chicken swarm optimization (CSO) and particle swarm optimization (PSO) to effectively select critical features from high-dimensional cyber-physical system (CPS) data, enhancing detection accuracy and system resilience. A robust voting classifier integrates support vector machines (SVMs), decision trees (DTs), random forests (RFs), and extreme gradient boosting (XGBoost) to provide reliable attack detection. Extensive experiments on large-scale WDS datasets [water distribution (WADI), BATtle of the Attack Detection Algorithm (BATADAL)] demonstrate the framework’s scalability and efficacy. For the WADI dataset, the framework achieves perfect accuracy (1.0), high recall (RC) (0.9963), an F1-score (0.9981), a true positive ratio (TPR) of 0.9963, and a false positive ratio (FPR) of 0.002. The BATADAL dataset achieves an F1-score of 0.9888, a precision (PR) of 0.9875, an RC of 0.99, a TPR of 0.99, and an FPR of 0.003. This research advances WDS security by offering a highly accurate and scalable solution for mitigating emerging cyber-physical threats.