一种基于反馈可信度的分布式P2P信任模型

Robust Feedback Credibility-Based Distributed P2P Trust Model

开放、共享与匿名的Peer-to-Peer(简称P2P)网络已经取得了越来越多的应用,无中心对等的特性也吸引了越来越多的用户.但由于其网络中的节点不受约束,资源的共享是用户自愿的行为,因此节点间的信任很难通过传统的信任机制建立.一种可行的解决方案是借鉴人际网络中的信任关系,建立一种基于信誉的全局信任模型.已有的工作基本建立在信任度高的节点其反馈也更可信这个假设的基础上,将节点的反馈质量简单地等同于服务质量.针对这一问题,提出了一种基于节点反馈可信度的分布式P2P全局信任模型(简称FCTrust),用于量化和评估节点的可信程度,并给出了模型的数学表述和分布式实现方法.分析及仿真实验结果表明,FCTrust较已有的全局信任模型在遏制更广泛类型的恶意节点攻击的有效性、迭代计算的收敛性及消息成本上有较大提高.

The open, sharing and anonymous nature of peer-to-peer （P2P） networks has made it popular in many large-scale distributed applications over the Internet. However, due to the fact that resource-sharing activity of a peer in P2P networks is a volunteer behavior and it is not responsible for its irresponsible bartering history, the trust relationship between participants can not be constructed only on the traditional trust mechanism. A feasible resolution derived from the trust relationship in social networks, is to establish a reputation based global trust model The previous work about the global trust model is mostly based on the assumption that the peer with higher trust value will provide more honest feedbacks, and make the quality of feedback of a peer be approximately equal to that of service of the peer. However, this is not always true. To solve this problem, this paper proposes a robust feedback credibility （FC） based distributed P2P global trust model （FCTrust）, to quantify and evaluate the trustworthiness of participants, and gives the mathematic analyses and distributed implementation method. Theoretical analyses and simulation experiments show that FCTrust has advantages in combating various malicious behaviors such as dishonest feedbacks from malicious peers, the collusion and the strategic attacks to the trust model itself, over the current global trust models, and demonstrates more robustness and effectiveness.