Attack–Defense Game in Interdependent Networks: A Functional Perspective

Abstract

The normal operation of critical infrastructure is related to the stability of society and a country’s development. Based on game theory and network science, this paper analyzes the attack–defense issues of critical infrastructure, proposing a network attack–defense game model based on virtual flow. Five coupling methods are established according to the complex interdependent characteristics and important evaluation indicators (the degree value and betweenness value) in the network. Considering the network’s information transmission, redistribution, and cascading failures, a cascading failure model based on betweenness virtual flow is established. A two-player game is adopted, in which the attacker and the defender act simultaneously. The scale-free, small-world, and random networks are related using these five coupling methods. Simulation and theoretical analyses are carried out on the interdependent network game scenario under different coupling modes. The game equilibrium under different network coupling conditions and coupling methods is simulated, and the preferences of the game participants are determined. In addition, a sensitivity analysis of the game parameters is conducted, facilitating the identification of certain effective strategies that can support decision-makers. Critical infrastructure networks play an important role in maintaining social stability. Based on game theory and network science, this paper analyzed the attack–defense problems of critical infrastructures and proposed an attacker–defender game model based on the betweenness virtual flow. The attack–defense game in the interdependent network is analyzed in terms of coupling modes, network types, and tolerance coefficient. A sensitivity analysis of the game parameters is also conducted. The results show that the interdependent networks under random coupling and reverse coupling strategies are more vulnerable to attack. In addition, the cost-sensitive parameter is a key factor affecting the game equilibrium. Even if the two subnetworks in the interdependent network are the same, the payoffs and the robustness will be different when the order of removing nodes is different. This paper provides decision support for actual infrastructure protection.