Using Game Theory to Reveal Vulnerability for Complex Networks

This paper proposes a method to identify network vulnerability based on Monte Carlo sampling and game theory. A two-player (attacker vs. immunizer), non-cooperative, constant-sum game model is used to obtain a mixed Nash equilibrium strategy. In this strategy, each node has a probability of being selected by the immunizer. These probabilities reflect the vulnerabilities of network nodes. With the implementation of this mixed strategy, the immunizer will achieve more equilibrium and safe profits. Moreover, this paper finds that the vulnerabilities of nodes in complex networks do not completely depend on static topology characteristics, such as node degree or betweenness.