P2P系统中基于双信任信息的信誉度改进模型

针对P2P网络中诸如诋毁、合谋、欺骗等安全性问题，提出一种新的基于双信任信息的分布式信任模型。区别对待节点提供服务和发送评价的能力，对节点拥有的资源进行分类；通过迭代求解，为每个节点的每类资源分别计算全局服务信任值和回馈信任值；为了防止恶意节点反复实施恶意行为，引入了惩罚机制，激励节点诚信交易。仿真实验表明，该模型能够迅速降低恶意节点的全局信誉值和恶意交易概率，降低了对无过失节点的不公平性。

Improved Trust Model Based on Double Ratings in P2P System

Aiming at the security problems such as slander, collusion and deceptions in P2P networks, a distributed trust model based on double ratings is put forward, which distinguishes the ability of one node to provide services and give ratings for the others, classifying resources of nodes. Resources of nodes are classified. Then, the global service trust value and feedback trust value of each type of resources on a node are computed iteratively. To prevent the repeated bad behaviors of the malicious nodes,the punishment mechanism is introduced into the model to encourage all the nodes trading honestly. The experimental results show that with this model, the global trust value of malicious nodes can be decreased rapidly;the malicious behaviors are restrained effectively; and unfairness to the non-fauh nodes in the network is mitigated.