Abstract: Based on the analysis of current reputation generation algorithms and fuzzy control technology，MSRep which is an approach to generate the reputations of email servers in distributed networks is proposed.MSRep uses the fuzzy controller to control its parameters values，makes the final reputation more in line with the anti-spam requirements.The simulation results demonstrate that，comparing with existing reputation generation algorithms，the reputation obtain by MSRep drops much faster when email server starts to behave maliciously，and the reputation rises more slowly when the email server rebuilds its own reputation.In different email delivering models，the rebuilding index obtain by MSRep is 4.389 times to the TrustGuard algorithm on average，which means it needs higher costs to deliver spam for the malicious nodes.

Key words: distributed networks, reputation generation algorithm, MSRep algorithm, parameters fuzzy controller

摘要： 在分析现有的信誉度生成算法的基础上，结合模糊控制技术设计了一个参数模糊控制器，提出了一种分布式网络中基于模糊控制的邮件服务器信誉度生成算法MSRep。MSRep算法使用参数模糊控制器控制其参数的取值，得到的信誉度更符合信誉度计算原则。仿真结果显示，与现有的信誉度生成算法相比，当邮件服务器开始恶意行为时，MSRep算法计算得到的信誉度下降更快；当邮件服务器重建自身的信誉度时，其信誉度上升则更加缓慢。在不同的发送模型中，MSRep算法计算得到的重建指数平均是TrustGuard算法的4.389倍，恶意节点为发送垃圾邮件所付出的代价更高。

关键词: 分布式网络, 信誉度生成算法, MSRep算法, 参数模糊控制器