摘要
资源检索是P2P系统研究的热点之一,非结构化P2P资源查找普遍采用泛洪机制。随着查询请求的增加,消息数量呈指数增长,网络拥塞和带宽浪费严重,查询效率得不到保障。针对这一问题,给出了一种基于本地聚类的非结构化P2P资源查找算法。通过对资源特征向量的本地K-means聚类和相似链接的建立,有效地提高了资源检索效率,避免了查询消息的扩散对网络带宽的浪费。实验表明,该方法能有效缩短资源的平均检索长度,提高查找成功率。
Resources search has become a hot research issue in peer-to-peer (P2P) systems. In most unstructured P2P systems with flooding mechanism, with exponential growth of the number of messages, serious network congestion and waste of bandwidth result in low efficiency of resources search. This paper proposed an unstructured P2P resources search algorithm based on local clustering ( LC algorithm) , in which peers performed K-means clustering on local resources separately and then built similar-links between peers that own similar clusters. It avoided the waste of network bandwidth derives from the spread of searching messages and improved the performance of resources search. Experimental results demonstrate that this algorithm effectively shortens the average searching length and gets high success rate.
出处
《计算机应用研究》
CSCD
北大核心
2009年第7期2508-2512,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(60573104)
关键词
对等系统
资源查找
聚类
peer-to-peer(P2P) systems
resources search
clustering