基于蚁群算法的P2P分发网返源规划  

Seed-returning Planning in P2P Distribution Network Based on Ant Algorithm

作　　者：刘海钦[1] 郑烇[1] 吴刚[1] 朱超[1] 

机构地区：[1]中国科学技术大学网络传播系统与控制联合实验室

出　　处：《计算机仿真》2008年第10期138-141,共4页Computer Simulation

基　　金：国家发改委项目中国下一代互联网(CNGI)示范工程(CNGI-04-15-2A);新世纪优秀人才支持计划资助(NCET-04-0564)

摘　　要：在P2P分发网中,返源策略是影响整个系统吞吐量和带宽利用率的关键技术之一。为了提高节点数据下载阶段的效率及系统总吞吐量,目录服务器需基于先前收集的信息并采用返源策略对系统中各节点之间的连接进行规划。通过描述返源规划问题的模型,提出了一种基于蚁群算法的规划策略来提高互相连接的节点之间数据块的可交换性。仿真结果表明,新算法可以最大化连接节点内容的互补性,继而减少了数据下载时间,提高了系统吞吐量,大大改善了P2P分发网的整体性能。In P2P distribution networks, seed - returning planning policy as one of key factors greatly influences system throughput and the bandwidth utilization. Aiming at enhancing the download efficiency and the system throughput, the directory server needs to schedule the connection process using the formerly collected information. Based on an analytical model, this paper presents a planning policy using the ant colony optimization approach, which takes the overall disjointedness of content among peers into account and therefore makes full use of network bandwidth resources. Experiment demonstrates that the proposed policy can maximize the disjointedness of connected nodes, reduce the average download time, enhance the overall system throughput and greatly improve the performance of P2P distribution network.

关 键 词：内容分发网 对等网络 蚁群算法 返源规划 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]