基于贝叶斯网络的时间感知的P2P信任管理模型

Time-sensitive P2P trust management model based on Bayesian network

P2P安全是P2P网络中的关键问题,如何选择信任度最高的节点提供资源是目前P2P信任管理的一个热点问题。指出了P2P网络信任弱点,分析了国内外常见的信任模型以及信任度量算法。目前,在各种信任模型中都没有考虑服务节点的离线给资源下载造成的影响,提出了剩余在线时间的概念,并应用到信任模型中。充分考虑节点对资源要求的多样性,有相同需求的节点形成一个簇,簇内节点对同样资源的信任评价角度相似,这样,推荐信任值就得到了保证。提出了系列信任评价指标,并把这些指标用贝叶斯网络形式存储。最后对提出的信任模型在查询周期平台上进行了实验分析,证明该模型具有一定的优越性。

The security is a key problem in P2P network. How to select the most trustable peer as the file provider is a focus problem. Some trust weaknesses are proposed, and some common trust models and trust value algorithms are analyzed as well. Now, affections made by the offline of the file provider have not been considered in most trust models. A new parameter named remain duration time is given into the trust model. The universality of the requirements of the resources are also considered, many peers which have the same or similar requirements form a cluster. The recommendation would be guaranteed because peers in a same cluster have the similar evaluation about the same resource. A series trust evaluation index is proposed, which are stored in the form of Bayesian network. In the end, an experiment is made based on Query Cycle platform to testify the model＇s superiority.