Research On Trust Models For P2P Networks

The high degree of anonymity, openness, autonomy and some other characteristics of the nodes in P2 P network make its security issues exposed after the technology increasingly widely used, such as malicious viruses, false documents, free-riding and so on. The system is difficult to achieve the maximum shared resource. These security risks greatly limit the further development of P2 P applications. This paper studies in the field of the P2 P trust model based on trust value, the main contributions are as follows:In order to solve the poor adaptability to the node's behaviors in current P2 P trust models and deal with the variety of malicious attacks in the network, a service expectation based P2 P dynamic trust model(SETrust) is proposed. Several parameters are introduced in computing trustworthiness of peers, including capacity evaluation, service expectation, historical trust and so on. Also, these parameters are adjusted in time to reflect the dynamics of the trust environment using time attenuation factor and time decay function. Thus, the trust evaluation is able to simulate the real trust relationship and has better adaptability to the dynamics of trust. Theoretical analysis and simulation show that SETrust has advantages in modeling dynamic trust relationship. It is highly effective in suppression of malicious attacks.In order to improve the detection and withstand capability of current P2 P trust models against the collaborative fraud malicious nodes and deal with the more and more rampant collusion attacks, a reactive mechanism of detection of collusion attacks based P2 P trust model(RDCTrust) is proposed. Several parameters are analyzed to differentiate the collaborative fraud malicious nodes, including feedback credibility of nodes, recommendation comments, and global trust and so on. Theoretical analysis and simulation show that RDCTrust can effectively distinguish between unfair evaluation and group collaboration, and enhance the trust model's ability to resist collusion attacks.