Research On Double Layer Structure Trust Models In Peer-to-Peer Networks

With the rapid development of P2P network technology, its applications in various fields are expanding while the inherent openness and anonymity of P2P networks also provide opportunity for false documents, malicious attacks and other malicious acts; these malicious acts impact the development of P2P networks seriously. Currently, P2P security has become a major obstacle to the application of P2P. Building an effective, reliable and complete P2P trust management mechanisms to address the issue of network security has become an urgent problem.Based on the understanding of P2P technology and the concept of trust, large numbers of relevant trust models are deeply analyzed. Considering the large amount of information flooding with large network bandwidth intensive in distributed trust model, the deficiency of hierarchical trust model inefficient to find the credible value on the structure of the management layer, research the hierarchical structure of the trust model. Hierarchical structure of trust model avoids flooding to search information. But there are still some problems such as:the failure of upper node easily lead to management information lost, the node load is not balanced, and the trust value is one-sided and so on. In this thesis, RTBTTM trust model is proposed, by joining the thread ring and cross-coverage monitoring area, to solve the problem that information lost caused by upper node failure, improving the accuracy of global reputation value at the same time. New algorithm of reputation value is proposed, strengthen the decay characteristics with time of the reputation value, using periodic computer mechanism, can reduce the cost of nodes computing reputation value. In the way of global reputation and local reputation joint computing comprehensive reputation, according to the local time of the transaction constitutes a dynamic adjustment of the density ratio, to further improve the accuracy of reputation values.Finally, performance analysis and simulation results show that RTBTTM model can effectively solve the problem of the management information lost caused by node failure, reduce the load of the root node, improve the accuracy of reputation value, with the ability of resist malicious nodes attack.