A study on the effect of dimensionality reduction on cyber-attack identification in water storage tank SCADA systems
Explainable AI Within the Digital Transformation and Cyber Physical Systems: XAI Methods and Applications
Attack identification, Cyber-physical systems, Dimensionality reduction, Intrusion detection, SCADA, Water storage tank
Intrusion detection in supervisory control and data acquisition (SCADA) systems can be an effortful task, as the multi-modality of the system often results in high-dimensional streaming data. Such high-dimensional data deteriorates the performance by causing high computational cost, response delay, and low detection accuracy. This study aims to determine the effect of dimensionality reduction on the accuracy of the intrusion detection system (IDS) in SCADA systems. In this regard, a demonstration of the hybrid IDS on a water storage tank cyber-physical system is performed along with a comparative study on 22 advanced dimensionality reduction techniques. The research indicates the enhancement of detection systems accuracy when dimensionality reduction techniques are combined with IDS. The utilized IDS also improves efficiency by reducing the memory usage and using data with better quality, which in turn increases the detection accuracy. The results have been analyzed in terms of F1-score and accuracy w.r.t. two classifiers.
Aljoudi, Ranim; Hallaji, Ehsan; Razavi-Far, Roozbeh; Ahmadi, Majid; and Saif, Mehrdad. (2021). A study on the effect of dimensionality reduction on cyber-attack identification in water storage tank SCADA systems. Explainable AI Within the Digital Transformation and Cyber Physical Systems: XAI Methods and Applications, 171-187.