Introduction to Two P2P Network Based Reputation Systems

Peer-to-peer (P2P) network is an important information sharing architecture. The original design of P2P networks assumes peers in the networks altruistically contribute their resources to others. However, most peers are unwilling to cooperate with each other without incentives and it leads to a so called "free-riding" behavior. Issues such as "hotspot" and "tragedy of the digital commons" are coming along with serious free-riding problems. Eventually, the prevalence of free riders will downgrade the performance of entire system and make it vulnerable. One way to encourage resource sharing is through service differentiation based on peer\´s contributions to the communities. This paper presents two reputation systems for P2P networks. One of them uses fuzzy logic method which uses requester\´s reputation and provider\´s inbound bandwidth as input information to create incentives for sharing and to avoid over-loading problems for primary file providers. The other one utilizes the clustering effect of virtual communities for reputation management. Reputation sharing in systems is implemented by decentralized selective polling which determines a peer\´s global contribution. The use of selective polling can significantly decrease overheads for reputation sharing. I illustrate the structures and algorithms for these two methods separately and I try to compare system performances for both methods.