GeTrust:a Guarantee-based Trust Model In Structured P2P Networks

As a novel overlay network which is set up in the application layer, P2P network is characterized by decentralization, autonomy and anonymity when compared with the traditional C/S network. However, users’unconstrained behavior results in a lack of trust between peers and the P2P network suffers malicious attacks which make the transactions between peers cannot be conducted normally. P2P trust mechanism adopts the way people build trust relationships to build trust model which can collect and quantify the behavioral information of target peer. In this way, the trust model can evaluate the target peer predictively to decide the peer’s credible degree and thus to solve the problem of lack of trust.Among the former researches, trust models based on recommendation are the most commonly used ones, which calculate the target peer’s reputation by using the globally collected recommendations which sent by peers who have transacted with the target peer. Compared to the method based on direct experience, a recommendation-based approach could have a good grasp of the target peer’s behavioral attributes. However, existing recommendation-based trust models suffer from the shortcomings of slow convergence and high complexity of trust computations, as well as huge overhead of network traffic. Also, they usually overlook the difference between the peer’s recommendation credibility and the peer’s global trust, or lack the effective punishment to resist collusion attack, leading to an inaccurate evaluation to the peer.Focusing on the defects exist in the recommendation-based trust models, we propose a guarantee-based trust model in structured P2P network named GeTrust. Referring to the establishment of guarantee relationship in human society, a service peer needs to choose its guarantee peer(s) for the service it is going to provide, and they are both required to pledge reputation mortgages for the service. In order to improve the trust decision, we propose the concepts of guarantee reputation and reputation mortgage which will help to quantify peer’s credible degree and transactional cost more clearly. After the guarantee relationship between service peer and guarantee peer was established, the request peer needs to make evaluations on all the candidates of service peer by referring their service reputations and their guarantee peers’reputations, and selects the one with highest evaluation to be its service provider. After transaction, the archive peers of both service peer and guarantee peer have to collect and verify the feedbacks from request peer and based on the results to refresh the corresponding peers’ reputations. Meanwhile, to enhance GeTrust’s availability and prevent malicious behavior, we also present incentive mechanism and anonymous reputation management strategy. Simulation results show that GeTrust is effective and efficient in terms of improving successful transaction rate, resisting complex attacks and lowering computational complexity.