A stochastic defender-attacker-defender model for smart system resilience enhancement against hybrid cyber-physical risks

With the advancement of digital and communication technology, modern critical infrastructure systems, e.g., power grids, tend to be controlled automatically and remotely through cyber systems. Such cyber-physical systems (CPS) promote more efficient operations but induce synergetic physical and cyber vulnerabilities in coupling network structures. These hybrid threats necessitate risk analysis and resilience enhancement of CPS from the cyber-physical perspective. We propose a stochastic defender-attacker-defender model with the consideration of both uncertain physical disruptions and malicious cyberattacks to secure CPS against hybrid risks. The developed model provides decision makers with resilient allocating and operating strategies for distributed energy resources and intelligent firewalls in CPS. The most threatening cyberattack scenario, which targets the availability and integrity of information systems simultaneously, is formulated with the consideration of incomplete knowledge to realistically simulate cyber risks. A column-constraints-generation based adaptive algorithm is developed to address the challenges that come from multilevel stochastic optimization and asymmetric information. Detailed case studies are conducted in the Hong Kong distribution system and Florida power grid with their communication systems to showcase the effectiveness of the proposed approach. The analytical results reveal the interactions and characteristics of concurrent cyber-physical threats and offer valuable managerial insights for decision makers to deploy defense strategies.