TOW:一种新的P2P实时流媒体缓存替换算法  

Time Over Weight:a New Cache Replacement Algorithm for P2P Live Media Streaming

作　　者：胡懋智[1,2] 徐恪[1] 夏树涛[2] 叶明江[1] 

机构地区：[1]清华大学计算机科学与技术系,北京100084 [2]清华大学深圳研究生院网络中心,广东深圳518055

出　　处：《小型微型计算机系统》2009年第8期1484-1489,共6页Journal of Chinese Computer Systems

基　　金：国家"九七三"重点基础研究发展规划基金项目(2003CB314801)资助

摘　　要：随着P2P实时流媒体在网络上的流行,可以预见在未来ISP主干链路的流量中,P2P实时流媒体应用将占有相当大的比例.ISP以前通过部署缓存设备的方法有效缓解了Web和VOD流媒体所产生的流量压力,这种缓存策略也可以应用到P2P流量的管理上.目前很少有直接针对ISP的P2P实时流媒体应用的缓存研究以及它与P2P文件分发应用缓存性能的差别比较.本文研究了各种缓存替换算法以及P2P实时流媒体应用与P2P文件分发应用流量在被缓存效果上差异的原因,提出了一种新的P2P实时流媒体缓存替换算法-时间权算法(TimeOverWeight),分析并用实验证实了它比常用的缓存替换算法具有更高的命中率.With the increasing number of P2P live media streaming users, it is reasonable to foresee that P2P live media streaming media will account for a considerable proportion in future backbone traffic. Caching strategy can be applied to P2P traffic management just as it＇s successfully deployed in Web and VOD media streaming. But few works focus on the P2P live media streaming cache in the view of ISP and directly compare its performance with the P2P file distribution system. The paper studies the various cache replacement algorithm and caching differences between P2P live media streaming and P2P file delivery and proposes a novel cache algorithm for P2P live media streaming called Time over Weight （TOW） and confirms its better performance than popular algorithm such as FIFO, LRU, and LFU.

关 键 词：缓存替换算法 P2P 实时流媒体 命中率 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]