An EigenTrust-based Hybrid Trust Model in P2P File Sharing Networks

摘要

Reputation systems have been proposed to distinguish malicious peers and ensure the quality of services in P2P networks. How-ever, only relying on global reputation scores to assess the reliability of file providers can weaken the subjective opinions of file requesters, which is vulnerable to some malicious attacks, e.g., camouflage attack. To address this issue, in this paper, we propose a novel hybrid trust model, namely HTM, based on the EigenTrust reputation management system. In HTM, a file requester utilizes a two-phase reference method to take both the global reputation and the direct trust of a file provider into consideration when deciding whether to download from it or not. In particular, in the first phase, a requester selects a provider from the responders based on the roulette wheel selection algorithm; while in the second phase, the requester considers the direct trust on the selected provider with our proposed direct trust evaluation approach. The extended simulation results show that our mechanism can identify and isolate the malicious peers both under individual attack and camouflage attack effectively, thus improving the performance of P2P file sharing networks.