查看文章

Peer-to-peer (P2P) file sharing networks are widely used by Internet users but due to their anonymity and self-organizational nature, they are often targeted by malicious users. Therefore, reputation systems are used to distinguish between good and malicious peers. Many reputation systems propose the use of the Bayesian model and it has proved to be effective for reputation calculation in a small, fully distributed system. This paper focuses on the distributed rank-based reputation system (DRank), one of the latest systems applying the Bayesian model in P2P streaming. DRank developers evaluated it on a small network with a maximum 200 nodes. In this paper, DRank is examined in a P2P file sharing system and on larger networks. This examination showed a drop in reputation propagation and a rise in the number of polluted files downloaded. Therefore, we propose a number of optimizations to DRank, such as …