Attack-defense game modeling framework from a vulnerability perspective to protect critical infrastructure systems

Critical infrastructure systems (CISs) are increasingly vulnerable to attacks due to their complex interdependencies. To enhance the resilience of CISs against deliberate attacks, we propose a novel Stackelberg attack-defense game model (ADGM) framework based on game theory and network science. In this framework, the functional vulnerability of CISs that accounts for cascading effects is innovatively considered in the construction of the ADGM. Critical target selection and resource allocation problems are integrated into the strategy model, while cumulative prospect theory (CPT) is applied to evaluate payoffs considering the risk attitudes of agents. The particle swarm optimization (PSO) algorithm is utilized to determine the game equilibrium. The results of applying the ADGM to a power grid highlight the necessity to study resource allocation for infrastructure protection from a vulnerability perspective. Identifying critical targets based on their importance, as determined by the vulnerability metric, provides the foundation for players to develop optimal resource allocation strategies. The analysis of optimal strategies under varying levels of resources for both players reveals the importance of striking an investment balance between enhancing component capacity and safeguarding critical components. The ADGM framework proposed in this paper provides valuable decision-making support for the protection of infrastructure systems.