Dynamic cross-layer security risk assessment and mitigation for cyber-physical power systems

Cyber-attacks targeting cyber-physical power systems (CPPSs) are increasingly recognized as complex and persistent cyber-to-physical (C2P) security threats, which introduce substantial cross-layer risks to critical power infrastructures. However, existing security frameworks fail to provide a comprehensive approach for risk assessment and mitigation against these ongoing and stealthy cross-layer attacks in CPPSs. This paper presents a cross-layer security risk management method that enables dynamic evaluation of cyber-physical security risks and the formulation of optimal defense strategies to reduce those risks. Specifically, an Extended Hierarchical Bayesian Attack Graph (EHBAG) is introduced to model the C2P attack risk propagation, which can infer the probability of physical-space incidents occurring based on detected attack nodes in the cyber layer. Observation nodes are incorporated into the EHBAG to represent uncertainty in the detected evidence. An attack surface generation algorithm is used to identify the most dangerous set of detected attack nodes within the EHBAG that require immediate attention. Then, a multi-objective security decision-making approach is presented to derive the optimal strategy for defending the highest-value nodes within the attack surface, aiming to reduce the cyber-physical security risks of the system. The proposed approach is implemented and evaluated using a real-world CPPS testbed and the numerical results confirmed its feasibility and effectiveness for risk assessment and mitigation.