A Trust Model Based On Recommendation And User Behaviors For The P2P Networks

P2P technology allows all network clients could provide the resources, including bandwidth, storage space and computing capabilities. It has been shown unique advantages in collaborative work, distributed information or resource sharing, large-scale parallel computing, etc., and become a new hot spot of Internet. However, P2P technology also has its own problems, such as the complexity of system safety analysis. A significant solution to those problems is to establish a trust model for P2P networks.In this thesis, a useful exploration of the trust model for P2P networks has been studied. By combining recommendation trust and user behaviors trust, a recommended and user actions local trust model (R&UALTrust) for P2P Network was presented. This model considers service satisfaction, service times, recommendation trust factor, influence factors, risk factors and parameters adjustment based on user behaviors.The main research contents of this thesis are as follows: Introduced the status of P2P network technology, and analyzed and summarized its technical features. Compared with the C/S network model, P2P network model has five main advantages:high utilization rate of resources, network scalability, cost-effective, robust, and privacy protection. The deficiencies of P2P networks have also been analyzed.EigenTrust model and its algorithm were studied. And recommended trust model based on EigenTrust model, including direct local trust model (DLTrust) and recommended local trust model (RLTrust) were proposed. Meanwhile, a voting recommended local trust model (VRLTrust) and other improved model added penalty function, the time factor, etc. were introduced.Trust model based on user behavior has been studied, including trust model based on the global user actions and trust model based on the local user actions. Malicious behavior recognition local trust model (MBRLTrust) and behavior similarity recommended local trust model (BSRLTrust) were established.Recommended and user actions local trust model (R&UALTrust) was constructed in this thesis. The model realized the combination of recommendation trust model and user behavior trust model by recommendation factor adjustment after interactions, recommendation factor proactive adjustment and risk factor adjustment based on the recognition of abnormal behaviors.