A Robust Distributed Reputation Mechanism for Peer-to-Peer Systems

In P2P content distribution networks, incentive mechanisms are required for sustainable operation of the network. In general, there are two classes of incentive mechanisms, direct and indirect. In direct mechanisms, a content consumer compensates the work of a supplier by providing him some content in return. In indirect mechanisms, the work of a supplier can be rewarded by any peer in the future. A sub-class of indirect incentive mechanisms is based on the concept of reputation. In such a mechanism, based on its past behavior, a peer is assigned a reputation value and based on this value the other peers decide whether to collaborate with him or not. Designing an effective reputation mechanism comes with many challenges such as the accuracy of the reputation values, the security, and the scalability of the mechanism. In this thesis, we investigate these aspects of distributed reputation mechanisms and we incorporate our techniques and algorithms into the BarterCast reputation mechanism, which is an Internet-deployed mechanism in the field of content sharing