STrust:A Trust Model Based On Social-Relation In P2P Network

With the rapid development of network technique and constant extension of its application area, P2P technology gets widely used and developed because of the open and flexible architecture. Applications like file sharing, multimedia transmission, data storage and distributed computing have shown the unique charm of P2P technology. However, natures like open, distributivity, dynamics, anonymity and autonomy have provided convenient conditions for computer virus, Trojan horse and network fraud leading to many secure problems including malicious behavior and inauthentic services which make users lack of security and severely restrict the further development of P2P technology.Considering the above mentioned problems, researchers have introduced the concept of trust model into P2P network by referring interpersonal relationship in our society. Based on a peer’s historical behavior, the peer’s credibility can be judged and used as the reference when other peers are selecting the service provider. The research results prove that trust model can effectively solve the trust issues in P2P network, and that makes trust model one of the most popular subject in network security technology. So far, many researchers have proposed several trust models by analyzing different application scenarios. But all those models suffer from the problem that malicious information cannot be identified accurately and different kinds of attacks cannot be resisted effectively. In this paper, focusing on the inauthentic services and malicious recommendations, we propose a trust model base on the former researches and social-relation named STrust. As per the basic law of people’s contact, this model defines social tightness to describe the network relationship between two peers in P2P network. Our model depicts the influence from social tightness to peer’s providing services and recommendations which can also be considered as the suppression to the unfair feedback from "friend peers" and to the collaborative cheating from malicious peers. Meanwhile, in case the service reputation is replaced by peer reputation, STrust divides the network into different domains in order to describe peer’s behavior more detailedly. Also, for improving the accuracy and reasonability of trust computation, the weight of domain, factor of time and factor of punishment are introduced which can to some extent resist the cheating behavior. Finally, the transaction process is presented based on Chord, the structured P2P network. The simulation results show that STrust is more effective and available on resisting malicious attacks when comparing with EigenTrust and RBTrust.