Peer-to-peer Streaming Network Trust And Reputation Mechanism Research

In nowadays, streaming applications have huge amounts of resources and canprovide rich audio and video streaming services for the overwhelming majority of users.However, abundant malicious users and selfishness hinder the development of thesystem and also threat the safety of users’ using streaming services. Trust and reputationmechanisms through user behavior’s historical records to evaluate and classify users toprevent malicious and selfish users have bad effect on the system and other users, and toencourage legitimate users become the participants or the providers of system services.Current trust and reputation model, however, usually only consider the user’scontribution behaviors to the system, and ignore many factors such as the relationshipbetween users, the resource features, network connections, lead these models difficult todeploy in the actual system. Therefore, based on the streaming application’s resourcedistribution characteristics, a robust trust and reputation model is proposed. On thisbasis, the stability and accuracy of the model are analyzed. Finally the validity of themodel is verified in the experimental environment. The main work includes thefollowing parts:(1) A feedback correlation based distributed reputation model is put forward. Themodel adopts the hierarchical network structure, based on the current usernodes’ providing historical evaluation calculates the reputation value,combined with user’s huddling character proposes group reputation to calculatethe global reputation. In addition, the model also introduces correlationfiltering function to inhibit users’ bad behavior and dishonest users, so as toimprove the capability of the model against security threats and to enhance therobustness of the reputation model.(2) An incentive mechanism based on the rewards and punishment is proposed.System divides users into different trust grades, to provide differentiatedservices. During the system strategy design, trust level based provider selectionand utility based service providing strategies are proposed. According todifferent user behavior to reward positive and punish selfish user. (3) System security threat models are established. Design the strategic alteringbehavior, dishonest feedback behavior, studies the user security problemsunder the attack scenarios. Using PeerSim as the simulation tool to realize thevalidity verifying of the reputation and incentive models. The results show thatreputation model has good robustness to the dishonest feedback behavior, caneffectively detect out strategic altering behavior, and can provide securedifferentiated service for different types of users.(4) The deployment of trust and reputation model in streaming system is designed.The model’s deployment method provides a reference for this kind of solutiondeployment in peer-to-peer system. Test and analysis of the prototype systemshows that trust and reputation mechanism works effectively in the actualsystem to improve the security of the system.