A game theoretic malicious nodes detection model in MANETs

Network security will face tougher challenges for MANETs due to emerging purposive strategic attacks conducted by smart malicious nodes. All these purposive strategic attacks eventually run out of the throughput which results in network crashes. In this paper, we have proposed a two-level dynamic Bayesian game model to recognize these purposive strategic attacks in MANETs and establish a malicious nodes detection system isolating such attacks. The simulation results have demonstrated that the detection method can effectively enhance cooperation performance and maintain packet forwarding rate under the purposive strategic attacks. Besides, through two-level decision, the system performs well in reducing the missing/false alarming rate, especially suitable for large proportion of malicious nodes in MANETs.