A critical study on the importance of feature selection for diagnosing cyber-attacks in water critical infrastructures
Explainable AI Within the Digital Transformation and Cyber Physical Systems: XAI Methods and Applications
Attack identification, Cyber-physical systems, Feature selection, Intrusion detection, SCADA, Water storage tank
Supervisory control and data acquisition (SCADA) systems are often imperiled by cyber-attacks. Such threats can be detected using an intrusion detection system (IDS). However, the performance and efficiency of IDS can be affected by a number of factors such as the quality of data, high dimensionality of the data, and computational cost. Feature selection techniques can eliminate most of these challenges by eliminating the redundant and non-informative features, which in turn increases the detection accuracy. This chapter studies the importance of feature selection on the performance of IDS. To do this, a multi-modular IDS is designed that is connected to the SCADA system of a water storage tank. A comparative study is also performed by employing twelve advanced feature selection techniques. The most reliable feature selection algorithms, crucial features, and the significance of the resulted improvements are then reported for the selected case study in terms of accuracy and F1-score.
Hallaji, Ehsan; Aljoudi, Ranim; Razavi-Far, Roozbeh; Ahmadi, Majid; and Saif, Mehrdad. (2021). A critical study on the importance of feature selection for diagnosing cyber-attacks in water critical infrastructures. Explainable AI Within the Digital Transformation and Cyber Physical Systems: XAI Methods and Applications, 153-169.