A Reputation Aggregation Method Considering File Request Distribution and Peer Behavior in Unstructured P2P Networks

Abstract

In order to establish secure file-sharing in unstructured peer-to-peer networks, reputation aggregation methods are now focused on. In these methods, each peer calculates reputation values which denote the reliability and safety of peers participating in P2P networks. When peers exchange files, they can avoid transacting with the malicious peers by reference to reputation values. Reputation aggregation methods, however, have a problem that file requests concentrate on partial peers with high reputation values. In the existing method, file requests are distributed by making partial peers whose reputation values are high to some extent more likely to be selected temporarily. Although, there are tactical malicious peers providing not only inauthentic but also authentic files in each transaction. And the existing method do not consider the problem that the malicious peers can spread more inauthentic files by providing authentic files intentionally and acquiring high reputation value by the system. Thus in this paper, we propose a reputation aggregation method which decreases the probability of selecting the malicious peers by considering reputation value. Furthermore, the method considers trade-off between the safety of file-sharing and the distribution of file requests. The results of computer simulation demonstrate the distribution of file requests and the advancement of transaction success rate.