A multilevel reputation system for peer-to-peer networks

Among the five goals of peer-to-peer security, file authenticity and fair trading have attracted the most attention. Reputation and incentive are two traditional systems to deal with file authenticity and fair trading problems, respectively. In this paper we propose a multilevel reputation system which also serves as the base of an incentive system. By introducing the concept of reputation level to a reputation system, we solve the Load Balance problem of peer-to-peer security perfectly. It also helps to prevent the occurrence of Collusion and Cold-Start problems with two proposed reputation level computation schemes, which are Level Up Reputation Computation (LURC) scheme and Level Keeping Reputation Computation (LKRC) scheme. A central reputation level computation and access control enforcement agent called Central Computation and Enforcement Agent (CCEA) is introduced too, which maintains or raises each peer's reputation level according to its periodic contribution to other peers in the network. We also propose a bootstrap process for our multilevel reputation system in a peer-to-peer network that has all the peers in the lowest reputation level at the very beginning.