Credibility based feedback for reputation computation in peer-to-peer file sharing network

Abstract

The Peer-to-Peer (P2P) networks are used in a wide range of application domains from file sharing to distributed computing. However, due to open and dynamic and the lack of centralized controlling authority, it provides an idle medium for selfish and malicious peers to exploit the network for their own benefits. This affects the performance, sustainability and future growth of P2P network. In P2P network, peers interact with unknown peers and does not know to what extent they can trust the other peer and the resources provided by them. In this paper, we propose credibility based feedback for reputation computation (CredRep) for a distributed P2P file sharing environment. Due to peer's dual roles in the network, we propose two reputations namely reputation as a requester and reputation as a resource provider. The former is computed using peer's resource contribution and feedback contribution; and the later is computed using feedbacks of other peers on the quality of resources shared by a peer. The reputation as a requester is used to detect and deter free riders and to overcome the feedback sparsity problem. The reputation as a resource provider is used to maintain the quality of resources available in the network. We have evaluated our proposed system by means of a simulation under different network environments. The simulation results show that our system can effectively detect free-riders and increase the probability of peers getting good quality resources.