P2p Network Credibility Research And Simulation Of The Model

P2P networks, a brand-new Internet technology, is developing rapidly in recentyears.With the expansion of P2P network application， its openness, autonomy,dynamism and anonymity facilitates the propagation of computer viruses, useless dataand fraud. Malicious nodes transmit illegal files and thus damage the system,meanwhile selfish nodes never render but take services the system offers. Abundantresearches have been done in order to address this issue, and the most effectiveapproach is to introduce reputation model.This dissertation goes deeply into the reputation model and its construction in P2Pnetwork. Although existing reputation models offer good mechanisms to deal withmost of these issues, they all fail to address the polarization in network, that is, it isdifficult for new comer that is willing to offer excellent services to obtain a highreputation; meanwhile, the existing designs of incentive punishment mechanisms in thereputation model are so simple that, in general, cannot take effects in incentivepunishment. Aiming at the issues above, this dissertation proposes a reputation modelcombining with incentive punishment mechanism, which is also based on push-pullprotocol and power exponential function. Our model takes the special operations onnewcomers into account, makes it easier for newcomers to establish connections withother nodes by the pull mode. Meanwhile, the reputation accumulation of thosenewcomers willing to offer excellent services are accelerated, by making these nodeseasier to be chosen as interactive nodes with the self-introduce factor. Push-pull modelgreatly reduces the network traffic and the computation of reputation model. Theincentive punishment mechanism based on power exponential function makes thedefined factor the power base and exponent of the power exponential function, mainlyby monotonicity, so as to make the incentive factor and punishment factor changedynamically according to the context. And thus participants are able to obtainincentives when interacting with each other by corporation, and are punished for a longperiod after damage, which makes nodes more willing to share and to interact withothers friendly, and ensures the robustness of the whole network. This dissertation also elaborates the mechanism, working approach and procedure. Simulation demonstratesthe advantages and effectiveness of this model by comparing it with other existingreputation models.