Graph-Based Interdependent Cyber-Physical Risk Analysis of Power Distribution Networks

This paper proposes a graph-based model for interdependent cyber-physical risk analysis of power distribution networks. The proposed methodology evaluates device criticality in the interdependent cyber and physical networks that comprise modern power distribution networks. A network traversal algorithm is developed to identify latent interdependent support and dependency structures as network subgraphs. Identifying these subgraphs provides a context-aware approach to model feasible cyber-attack vectors, define metrics for device criticality, and measure overall network resilience to cyber-attacks. The proposed methodology is demonstrated in a cyber-attack detection study. The results indicate that the proposed context-aware criticality metrics, based on the dependency subgraph framework, better identify the risks of cascading dependencies in power distribution networks. Further, several characteristics of the power distribution network such as interdependencies, device vulnerabilities, and attack target criticalities detailed in the proposed methodology have significant impacts on the performance of attack detection strategies.