基于网络延迟的P2P路由算法的研究

近年来,P2P计算应用已经超过Web应用而成为占用互联网带宽最多的网络应用.针对目前P2P系统中采用的随机选择邻居节点的方法会降低路由效率以及增大网络开销方面的问题,在分析Chord方法特点的基础上,提出一种改进的Chord构建算法DeChord.从逻辑上相邻的点在物理上也相邻这一原则出发,DeChord采用Chord数据定位算法;利用全局网络定位系统计算节点坐标并以此为依据计算节点间的物理距离,节点加入时充分考虑节点之间的逻辑距离与物理距离的一致性,系统节点总是选择距离自己物理距离较近的节点作为邻居节点;DeChord算法使得节点的路由表的信息能得到及时的更新.DeChord中的邻居节点选择方式可以降低消息路由过程中每一跳的网络延时,从而降低整个消息路由的开销.模拟实验表明,利用该算法建立的P2P系统能大幅度降低数据定位的延时.

Research of Network-delay-based P2P Routing Algorithm

In recent years, most network bandwidth has been consumed by Peer-to-Peer (P2P)computing applications rather than WEB applications. However, peers in P2P systems choose logical neighbors randomly without any knowledge about underlying physical topology at present, which can decrease greatly the efficiency of message routing and consume more bandwidth. According to the principle that the neighbor peer in logical is also the physical neighbor, an improved Chord construct algorithm DeChord, is proposed. In DeChord, Global Network Position is utilized to compute the coordinate of peers, and then the distance between nodes is computed based on these coordinates. The consistency between logical distance and physical distance is considered when nodes join the system and the closer peers in the physical network are taken as neighbor peers. DeChord makes the information of routing table updated in time. DeChord can decrease the latency of every hop, therefore the latency of total message routing is decreased. Simulation results indicate that the performance of data location in the system that is constructed by DeChord can be significantly improved.