中文摘要：

针对无结构化P2P网络的洪泛搜索与随机漫步机制的盲目性，提出了一种利用Hash函数与M—tree技术将文件聚类后，再利用路由表完全分布式存储索引指针的新的路由算法。该算法使每个节点的路由表主要记录拥有各类资源的高能力节点指针，并利用概率统计的方法不断地更新路由表项。当节点收到搜索以后，通过查询路由表，只需一跳就进入能以最大概率回应的节点处查找，并能以较低的网络时延命中多个优质资源副本，达到了高速并行下载的目的。仿真试验和数学分析表明该算法有效地减少了盲目搜索造成的网络流量，提高了查找成功率，并且具有越稀缺的资源越容易找到的特性。

英文摘要：

To improve the current status that the existing blind searching schemes for unstructured P2P networks such as the Flood and the Random Walk incur too much traffic load, the paper presents a novel routing algorithm that can research resources efficiently. The algorithm fully utilize the Hash function, the M-tree technique and the feedbacks of every query to construct clusters of resources and the muting table that maintains pointers of high-capacity nodes. When the peer re- ceives a new query, according to the routing table, it directly forwards the query to nodes with high hit probability. Mul- tiple high-quality replicas of both popular resources and rare resources can be located with the minimum overhead, thus parallel download can be guaranteed. The mathematical analysis and simulation results show that the routing algorithm improves the search efficiency with the high bit rate and the low bandwidth overhead.