Security of SCADA Systems Against Cyber-Physical Attacks

SCADA systems have been playing a vital and an increasing role in safety-critical infrastructures of a nation, including transportation systems, electric power grids, gas pipelines, and water networks. The rapid advance of information and communication technology has rendered modern SCADA systems increasingly vulnerable to cyber-physical attacks on both physical and cyber layers. Though approaches focused on information security may provide some elements of security for SCADA systems, their scope may be limited with regards to the defense-in-depth strategy of SCADA systems against malicious attacks. The data-based approach, which does not require the physical and mathematical model of SCADA systems, consists of analyzing hidden patterns in observed data. The detection and isolation tasks can be realized by one-class classification The data-based approach, which does not require the physical and mathematical model of SCADA systems, consists of analyzing hidden patterns in observed data. The detection and isolation tasks can be realized by one-class classification. Mathematically, a semiparametric model consists of three components: a parametric model, a nonparametric model, and random noises. Then, the semiparametric model of the system under both normal operation and abnormal behavior is specified. This semiparametric approach is expected to offer better performance than data-based and model-based approaches.