Research Of The Trust Prediction Mechanism Based On Game Theory In P2P Networks

P2P networks has been developed quickly in recent years, along with its applications are expanding; it is bringing more convenience which is increasing every year for users. P2P networks has a unique operation, it do not dependent on any server and provides a completely free interactive environment for users. Each node in the networks has equal and double identities that is the service and be serviced. P2P networks brings a new network model for people, but also because of its shorter development time and related technology has not been a more comprehensive study. Therefore, as the time it provides more services, P2P network security is more obviously.The scope of P2P networks is expanding, but no mechanism to manage it, the nodes through multiple transactions, its attitude become more rational. The selfish nodes become habitual betrayal in order to get maximization benefit, it's free riding on other nodes, collaborative cheating others, a lot of nodes'malignant behaviors which make a waste of network resources, and make full of false information in network environment. But the malignant behaviors of the nodes have not been constrained control and affect the interact mind of other nodes, then much more malignant nodes exist in the network, so the efficiency of network operation is getting lower and lower. Based the problem what has been analysis above, it is more important to build a reasonable P2P trust mechanism to punish malicious nodes, in order to maintain an active role in the network. This article is based on this point, based on building a trust relationship for the nodes, so from the micro it design a trust mechanism to constraints nodes which choose betrayal, then from the macro to manage the running status of network.The paper compares and research more mature P2P trust models, it can be found that many models are only based on calculate the credibility of nodes, but based on the nodes'trust or betrayal behavior, the models can not reward or punish; The existing models generally take a different approach to solve the problems which have appeared, but can not predict the future state of the network to take more timely measures. For the reasons what we have talked above, the paper based on build the game framework between the nodes, design a game mechanism which has a predictive function, finally, devises the game process between the mechanism and the nodes. From the details to the whole way, the predicted mechanism promotes nodes and nodes, nodes and networks to form more positive relationships. Before and after the reward or punishment strategies execute, the trends of the nodes'trust game are compared and simulated, furthermore, the predict ability of the mechanism is also simulated. With the corresponding strategies, it can be got the changes of a node status transfer, and gained more effective interval of rewards and penalties, through a series of experiments, finally it can be proved that the effectiveness of forecasting mechanism, the rules of the mechanism is reasonable.