Confidence-based Trustworthiness of Peers in Fully Distributed P2P Overlay Networks

对等（Peer-Peer，P2P）系统由多个对等进程组成，本质上是完全分布式的。在这里，讨论每个对等点是否可以信任熟人是至关重要的。由于网络中的成员节点上没有集中的协调器，每个节点必须通过与熟人的通信来决定每个熟人的可信度。在传统的信誉概念中，收集目标节点的每一个可信度意见。在这里，可能会包含一些恶意的、过时的观点。在这项研究中，我们讨论了一种新的方式，每个节点只需要一个主观的信任度的目标节点从熟人，该节点可以信任。仿真结果表明，该算法得到的客观可信度比洪泛算法得到的可信度更准确。此外，一个节点可能获得的主观信任的目标对等体不同于客观的。

A peer-peer (P2P) system is composed of peer processes, which is in nature fully distributed. Here, it is critical to discuss if each peer can trust an acquaintance. Since there is no centralized coordinator on member peers in the networks, each peer has to decide on the trustworthiness of each acquaintance by communicating with the acquaintances. In the traditional reputation concept, every trustworthiness opinion on the target peer is collected. Here, some malicious, obsolete opinions might be included. In this research, we discussed a novel way where each peer only takes a subjective trustworthiness on a target peer from acquaintances which the peer can trust. Through the simulation, we showed the objective trustworthiness obtained by the algorithm we discussed is more accurate than the flooding ones. In addition, a peer might obtain the subjective trustworthiness on a target peer different from the objective one.