一种对等网络中动态随机中继节点路由算法

Dynamic Stochastic Relay Selection Algorithm in Peer-to-Peer Networks

提出了一种对等网络中继节点选择的随机路由算法,该算法应用随机规划框架,通过权衡本地路径真实时延以及非本地网络路径的时延统计分布,从而选择端到端期望时延最短的中继节点完成流量传输。随机中继路由算法可分布式实现,通过相邻节点动态更新路由的统计测量信息,相比于经典的静态路由算法能够获得更低的时延性能。为了更好地测量非本地网络覆盖路径的统计时延分布,路由算法拟合覆盖链路上的历史时延测量数据,并通过仿真实验表明,基于本算法建立的中继单路径/多路径可有效减少端到端路径时延和丢包率。

A stochastic routing algorithm for selecting appropriate relay nodes in peer-to-peer networks was proposed. This algorithm was constructed using a stochastic programming framework by leveraging the actual delay of local links and the statistical delay distributions of non-local overlay links. In order to approximate the statistical delay distribution of non-local overlay links, the historical delay values of each link were utilized to approximate the link delay distribution. This algorithm was fully distributed and the stochastic link measurement data could be updated between neighboring nodes at a longer time granularity to reduce routing overhead. This algorithm and two other traditional algorithms were evaluated to find single and multiple overlay paths via relays between two end-hosts using simulation experiments. The experiment results demonstrate that this algorithm may achieve significant stochastic gain in terms of shorter delay and smaller packet loss than two routing algorithms including the deterministic shortest path algorithm and the minimum hop routing algorithm.