Research On Some Key Problems Of Reputation Mechanism In Peer-to-Peer Environments

While P2P technology has been widely deployed and successfully applied in the fields of stream media,file-sharing,content distribution,distributed storage management,distributed computing and so on,there still exists some issues need to be studied.In P2P networks,the dynamic characteristic,openness and autonomy challenge security and fairness.Problems such as Free-Riding,Tragedy of the Commons,and Incredible Resources have emerged.Reputation mechanism is an effective solution to these problems.However,reputation mechanism brings its own security issues and faces the Hot-Spot problem.This dissertation mainly discusses the issues of security and load-balancing in P2P environments and shows our researches on reputation management mechanism, reputation measurement algorithm and security and performance of the mechanism.The simulation results show that the group attacks can be defended against effectively and the resource utilization of P2P networks has been increased.The major topics of this dissertation are as follows:Firstly,the dissertation introduces the development,classification and applications of P2P network and deeply analyzes the main reasons that cause security issues,unfairness,network dynamic characteristic and inhomogeneity of resources utilization.The dissertation also analyzes the existing reputation mechanisms in P2P networks,summarizes the achievements of the related researches and gives the trends of the future research work.Secondly,the dissertation discusses some security issues of reputation mechanism.On the one side,the anonymity of P2P network is discussed and a new model to protect each side of the interactions is proposed.On the other side,the dissertation does some researches on the similarities between P2P networks and social networks.Based on some mature theories in sociology,a new algorithm and a new mechanism are proposed to defend against the group attacks in P2P reputation networks.Thirdly,focusing on the dynamic characteristic and openness of P2P networks, this dissertation proposes a reputation management system to solve the bad influences of Churn and malicious attacks.Finally,the load-balancing problem is discussed.The dissertation shows the requirements of real users who are being engaged in P2P file-sharing systems.In allusion to the problem of Hot-Spot,a reputation model is proposed to promote the resources utilization in P2P file-sharing systems.The main innovations and contributions in this dissertation are as follows:Concerning peers security in reputation mechanism and anonymity of P2P networks,a reputation model(GARM) without CA is proposed.Peers are assigned in some different groups according to the resources they possess and the anonymity between each group is protected.The model realizes the fairness in P2P network.At the same time,it protects service peers and avoids the retaliatory actions between peers.For the group attacks that current P2P networks ignore,a reputation measurement algorithm to defend against these attacks is proposed.Based on the community structure theories in social network,this dissertation analyzes the group structure in the P2P reputation network,and utilizes the reputation value of groups in measuring reputation of peers.On the basis of the algorithm,a group-based reputation mechanism is proposed.This mechanism adapts to dynamic and open P2P environments,and can defend against group attacks without limiting peers joining and leaving.Concerning the dynamic characteristic and openness of P2P network,this dissertation proposes a sustained online duration-based reputation management system(SOD-based system) to reduce the phenomenon of Churn and increases the resourtces utilization in the whole network Rv adding the factor of SOD,the.duration online time of peers is increased.SOD-based system also solves the problems of Cold-Start and Matthew Effect.For the congestion of network and the phenomenon of Hot-Spot in P2P Networks,using economic principles,a downloading requirements preference-based P2P reputation model combining bandwidth commitment with the evaluation of files is proposed.According to the preference of downloading requirements,a server peer allocates the bandwidth commitment dynamically to attract request peers.The simulation results show that this method raises the overall rate of utilization of resources in P2P file-sharing systems and eases the congestion.