Attack Prediction Model Based On Bayesian Game

Modeling for attack prediction is an important aspect of the research in network security. This paper describes attack prediction model based on static bayesian game (SBGAPM) and dynamic bayesian games (DBGAPM). The SBGAPM model can predict the probability of attacks or defenses that reasonable attacker or defender will take, in order to maximize their payoff. According to the attacker's historical behavior and SBGAPM result, The DBGAPM model reasonably updates the probability of malicious nodes existing in the network by using bayesian law, with which it can predict the probability of attacks or defenses that reasonable attacker or defender will take in the next stage of the game, in order to maximize their payoff.Thus the result can be used to assist security administrators to configure the network system. It may improve the passive detection to the active protection for the defender. This paper also presents the process of experimental and analysis result for validity of the model.In addition, This paper still give an application frame which including DBGAPM and IMS. This application frame can help user build a dynamic and strong prevention system, and the security platform are being controlled and managed. It also indicates security products 's way becomes mixing,working together,managing centrally.