P2p Network Trust Model

In recent years, there has been a great development in P2P network which is one of the most popular applications of Internet. However, the feature of decentralized self-organization as well as the anonymous and open nature of P2P network offers a convenient environment for the spread of viruses and inauthentic files, increasing the distrust among peers and impeding their cooperation and exchange. As a result, the development of P2P network is limited and trust mechanism is one of the effective methods to resolve such problems.After researching into trust mechanisms, trust models and some key technologies on trust model design, the author sets up a trust model based on the direct evaluation of interaction nodes on peers'reputation. This model aims at solving problems existing in current trust models, which are: great message expenses in computing peer's reputation, peers'fake location evaluation, peers'dynamic features and collusion attack etc. The main contributions of this dissertation are as follows:(1) The author provides an algorithm to compute the peers'reputation value without iterative process. This algorithm computes global reputation value through the direct interaction peers'local evaluation weighted its evaluation credibility; therefore it can reduce message expenses in network, improve the model's expansibility and be applied in P2P networks on a large scale.(2) A hash-check-based algorithm on filtering fake evaluation is posed to remove the impact of fake local evaluation on the calculation of node's reputation value.(3) The author believes that local evaluation also has dynamic characteristic. After systematically analyzing those nodes'dynamic activities from the aspects of service and evaluation, the author provides a method of calculating global reputation value and global evaluation credibility, which can control the dynamic characteristic of peers.(4) To solve the problem of collusion attack in trust models, the author proposes an algorithm based on the standard deviation and concentration degree of the location evaluation, which can be helpful to reduce the influence of peers'collusion attack.