A History And Honest Recommendation Based Trust Model In Peer-to-Peer Networks

Nowadays Peer-to-Peer (P2P) network has attracted more research and attentions, it has been applied in several fields and has great development in the future. Due to the P2P network's opening and anonymity, there are more and more misbehaviors of malicious peers, and the performance of network is severely destroyed. So how to recognize those malicious peers, and how to restrain the misbehaviors of dishonest peers have become more and more important in trust mechanism.Based on the typical resolve of"trust mechanism"both at home and abroad, this paper mainly studies the computation of peer's trustworthiness in the distributed environment and provides a"trust mechanism", including the evaluation of Peer's behavior, the method of trust data management, and the strategies of security. In addition, we also offer a series of simulations and analysis for the model.This trust model introduces several parameters to simulate the behaviors of peers in the networks, and estimate peers'possible behaviors in the future by computing general trustworthiness. We present a history and honest recommendation-based algorithm of trust computation, with dynamic weight between direct transaction histories and indirect honest recommendations from other peers. We distinguish the transaction and recommendation reputation definitely, and we can compute the peer's recommendation reputation through the deviation of feedbacks. This method avoids iterative computation, can be easily implemented. To most important, it can distinguish dishonest feedbacks and resist collusive attacks.We realize the distributed implementation of this trust model by Chord. It can guarantee the security of placement and access of trust data, and it increased the anonymity and security of the transactions in the P2P networks. We also analysis the potential threats in the existed networks, and bring forward some security strategies for identity imitation, malicious imputations and collusive cheatings. Besides, we expand the computation of trustworthiness.At last, we compare the improved computation of trustworthiness with some other recommendation-based computations. The simulation results prove that our model have better performance in evaluation accuracy, security and implementation.