A Trust Model To Resist Strategic Attacks For P2P Network

P2P network has a lot of security risks such as management difficulty, malicious attack and virus for its open, anonymous and autonomous characteristics. Lacking of effective trust management system is the cause of security threats in P2P network. The existing trust models mainly focus on how to calculate trust values and predict their future behaviors by nodes'transaction histories. However, nodes are familiar with the trust management system applied in the network. Their attacks are more and more strategic. Consequently, the effectiveness of these trust models is greatly discounted. It is very important to build a trust model which can resist strategic attacks of malicious nodes.This paper presents a method of detecting whether a node's behavior corresponds with the malicious node behavior features. A method of how to calculate the global trust value is proposed. By detecting the feedback sources of the service provider, the node can be judged whether to be honest or not. How to calculate the credibility of feedback source and the evaluation similarity are also proposed in this paper. The algorithm of judging whether a node is a malicious client is presented in detail.The trust model with node detection algorithm is implemented by C programming language. We adopt the False Positive Rate (FPR) and False Negative Rate (FNR) as the evaluation metrics. The results show that the proposed trust model can identify malicious nodes precisely in different circumstances.It is proposed in this paper that malicious node is recognized by checking whether the node's behavior corresponds with the malicious behavior features. The deception of strategic changes can be effectively resisted. The results show that the trust model can cope with the swaying, exaggerating, collude attacks and other strategic attacks. The model can identify malicious nodes correctly and effectively.