Abstract: Currently there exists slander, exaggeration, cooperative cheating, strategic deception and other malicious transactions problems in P2P network. The rapid increasing proportion of malicious nodes results in rapid decline of system trading?success rate. This paper proposes a trust model which has a detection mechanism and good convergence. Simultaneously, it makes use of the service widespread degree evaluation mechanism, the periodic strategy similarity detection mechanism and the evaluation similarity detection mechanism. They efficiently suppress?bad service of malicious nodes. By comparison with current several typical global trust models, simulation results show that this model has improved greatly in the transaction success rate.

Key words: dynamic trust model, peer-to-peer network, detection

摘要： P2P网络目前存在诋毁、夸大、协同作弊、策略性欺骗等恶意交易问题，而且恶意节点比例增加导致系统交易成功率下降速度过快。提出了一种具有检测机制的动态信任模型，该模型具有良好的收敛性。在模型中引入了服务广泛度评价机制、周期性策略相似度检测机制和评价相似度检测机制，较好地抑制了恶意节点的不良服务。通过与目前几个典型的全局信任模型的对比仿真实验，表明该模型在交易成功率方面有较大提高。

关键词: 动态信任模型, 对等网络, 检测