Reputation Research Based On The Cluster In Peer-to-PEER

P2P network, because of its unique advantage, is gradually becoming important part of the internet. But with the extensive application of P2P technology, has seriously exposed the problem of security and selfish:(1) Security problem. Because P2P networks lack monitor from the third-party, and also don't have the corresponding security mechanism, in addition to the feature of its'openness and anonymous, so it is vulnerable to various security threats, such as providing false files, virus, etc.(2) Selfish problem. Selfish is that between the nodes in P2P network don't actively participate in cooperation. Every node tries best to get the resources but not to share their service with other nodes, so it leads to the present free-riding, load balance, public tragedy and so on.The problem of security and selfish has seriously restrained the further development of P2P network. Therefore, if we want to improve the present state of P2P network, we must manage the nodes' autonomous behaviors, and encourage them the active cooperation. On basis of the backgroup, the papar is doing the research.On the basis of the present reputation mechanism, I am doing in-depth analysis and research, and propose "reputation mechanism based on cluster" (BCR), which is clustering according to geographical location, and taking cluster as basic management unit; Based on the heterogeneity of nodes (such as reputation, the capability of process, average on-line time of history, and so), we divide the node of the cluster inner node into two type of common node and service node. BCR uses strict protocols for specifying the behavior of the nodes and the strict reputation evaluation mechanism for protecting the security of transaction. In order to further promote more and active cooperation between the nodes, the paper posed the concept of reputation rating services that is referred when between the nodes transact each other, the better the reputation is, the more the chance and priority is. In view of the fluctuations of P2P network, this paper considers such as on-line time, network bandwidth, node performance to enhance the stability of the system. To reduce the costs in the course of clustering, the paper improved the management of cluster.At last, by the simulation experiment, the proposed mechanism in the paper is compare to the present typical mechanism in aspects of the rate of the effective load, cluster growth, and network flow. Experiment results show that BCR in aspect of security, selfish, stabilization, and network cost have achieved the better effect.