Discrete Markov Chain Model for Reputation Estimation of Unstructure P2p Network

Building trust relationships between peers is an important and difficult part in the security needs of P2P network without a central server. P2P reputation system has been introduced which collects locally generated peer feedbacks and aggregates them to yield global reputation scores. Most P2P applications on the Internet are unstructured, without fast hashing and searching mechanisms, how to perform efficient reputation estimation is a major challenge on unstructured P2P computing. This work thus proposes a twostep reputation estimation approach for the unstructured P2P network. First, a Markov chain model is proposed to determine the reputation value for each one-hop neighbors. A peer’s reputation value (RV) is analyzed from its previous trust manner in this group. The proposed trust model is proven as an ergodic Continuous-Time Markov Chain model. Second, a peer with the highest RV of a group will be selected as the central authentication(CA) server. For increasing reliability, the peer with the second highest RV will be selected as the backup group leaser(BCA) that will take over CA when CA fails. The procedures of the peer’s RV are detailed. Numerical results indicate that the analytical RV of each peer is very close to that of simulation under various situations.