Research Of Reputation-based Trust Management In P2P Networks

P2P network is the one of the most popular applications in the Internet. In P2P network, it is not discriminative between service providers and service consumers, that is, all the participation of entities are equal. Each entity can not only provide the service but also use the service. However, because of the anonymous and open characteristics in P2P network, the nodes who interact with each other can not guarantee the reliability of their own behavior, resulting in lack of trust between nodes and it is also difficult to guarantee the quality of resources in the network. How to choose safe and effective resources and reduce network overhead and improve the security and reliability is a key problem.This paper proposes a reputation-based trust model to solve trust issue between nodes in P2P network. The model divides trust into direct trust and indirect trust to calculate trust level and designs the calculation method of the direct trust and indirect trust. Direct trust means two nodes set up relationship of trust through direct interaction, and indirect trust means two nodes set up relationship of trust through other nodes' recommendation. Node can measure the trust according to historical results and other nodes' recommendations and determine trust relationship based on the measure result. The model is adapted to dynamic and open network environments through continuous collecting reputation and updating trust. Furthermore, the paper proposes an ant-based reputation evidence distribution algorithm. This algorithm realizes reputation distribution in P2P network by modeling the ants' behavior of looking for food. It solves the issue that nodes can obtain reputation information safely and efficiently.Theoretical analysis and simulations prove that the trust model proposed in this paper can effectively evaluate the trust level of node and help determine credibility of the service providers, so it can help service requestor to make the right choice and improve the overall quality of network services. The model owns anti-attack capabilities and can also identify and isolate malicious nodes. Ant-based reputation evidence distribution algorithm can choose the best path to obtain reputation information, improve the search efficiency and reduce the load on the system.