A Stochastic Game Theoretic Approach to Attack Prediction and Optimal Active Defense Strategy Decision
Wei Jiang, Zhihong Tian, Hongli Zhang, Xin-fang Song · 2008
This paper presents a stochastic game theoretic approach to analyzing attack prediction and the active defense of computer networks. A Markov chain for privilege (MCP) model to predict attacker's behavior and strategies is proposed. We regard the interactions between an attacker and the defender as a two-player, non-cooperative, zero-sum, finite stochastic game and formulate an attack-defense stochastic game (ADSG) model for the game. An attack strategies prediction and optimal active defense strategy decision algorithm is developed using the ADSG and cost-sensitive model. Optimal defense strategies with minimizing costs are used to defend the attack and harden the network in advance. Finally, a simple example of an attack against a network is modeled and analyzed.