WSN中基于MDP与博弈论的入侵检测系统

摘要

针对无线传感器网络（WSNs）中容易遭受多种攻击的问题，提出一种融合马尔可夫决策过程（MDP）和博弈论的WSN入侵检测系统（IDS），称为马尔可夫博弈入侵检测系统（MG-IDS）。MG-IDS采用博弈论和MDP的异常、误用检测技术来确定最佳的防御策略，同时利用MDP和攻击模式挖掘算法，根据攻击记录来预测未来攻击模式。通过仿真实验，比较了MG-IDS、仅博弈论和仅MDP三种方案，在不同攻击频率下，对多类型混合攻击的防御性能进行了比较，实验结果表明，所提出的MG-IDS具有较高的防御成功率。%For the issue that the Wireless Sensor Networks(WSNs)are vulnerable to various attacks, a WSN Intrusion Detection System(IDS)which fuses Markov Decision Process(MDP)and game theory is proposed and named as Markov Game IDS(MG-IDS). MG-IDS determines the optimal defense strategy by using game theory and the anormaly, misuse detection technology of MDP. In the meantime, it forecasts the future attack model on the basis of attack records by using MDP and attack-pattern-mining algorithm. The experimental results show that, compared with the two schemes of game theory and MDP respectively, the proposed MG-IDS has a better performance in defending mixed types attacks with dif-ferent frequency, and has higher defense success rate.