一种基于节点全局信任值的BitTorrent系统区分服务机制

BitTorrent系统中节点的搭便车行为严重影响系统的健壮性和稳定性.针对BitTorrent系统中搭便车行为抑制机制存在的不足,提出一种基于节点全局信任值的区分服务机制.在该机制中,Tracker服务器依据节点反馈的本地信任值计算节点全局信任值,依据全局信任值识别贡献节点与搭便车节点,对两类节点采取区分服务策略:为贡献节点与搭便车节点分别返回其同类对等节点,使搭便车节点无法与贡献节点建立连接进而获取文件资源;在贡献节点中又进一步区分高贡献节点和普通贡献节点,引入返回比例因子为其提供差异服务,使高贡献节点获得更快的下载速度.实验表明,该机制能有效隔离并惩罚搭便车节点,奖励高贡献节点.

A Globaltrust-Based Differentiated Service Scheme in BitTorrent

BitTorrent system suffers from free-riding because free-riding behavior has negative effect on the performance of the system. However, existing mechanisms used to counter free-riding in BitTorrent are not effective enough. In this paper, we propose a global-trust-based differentiated service scheme to deal with free-riding problem. In our scheme, tracker serves as a computation agent to calculate peers' global trust values. With these values, tracker identifies free-riders and contributive peers, and then disseminates the information of free-riders to free-riders and the information of contributive peers to contributive peers. Through such isolation, free-riders will have no chance to connect with contributive peers to obtain resources. Moreover, contributive peers are divided into high-contributors and normal-contributors. Tracker provides differentiated service for these peers according to their return ratios, which makes high-contributors achieve faster download bandwidth. Simulations show that the proposed scheme can significantly isolate and penalize free-riders, thus incentivizing peers to donate more upload bandwidth for the system.