摘要： 提出一种提高无结构型对等网络查询效率的机制。在该机制下，节点根据地理位置自动聚类，类之间用Chord方式组合起来，从而减轻了逻辑网络与物理网络拓扑结构的不匹配。为了进一步提高查询效率，引入了一种类间索引技术。该技术使得查询消息不需要遍历所有的类就能获得全局搜索结果。与Gnutella中的洪泛滥查询相比，在TTL=5的情况下，该机制最多能减少超过80%的资源开销，最多可以将响应时间缩短59%。

关键词: 对等网络, 查询机制, 聚类, 索引

Abstract: An efficient unstructured P2P searching scheme is proposed. In the scheme, peers self-organized into clusters based on network locality, and clusters are organized by a Chord graph. The scheme alleviates the topology mismatch problem between the unstructured P2P overlay network and the physical network. In order to further improve the efficiency of the scheme, a cluster index technique is presented. Each cluster has content indices from all peers of some other clusters. When a query is submitted, full search scope can be achieved while part of the clusters are directly probed. Compared with standard flooding used in Gnutella, the new searching scheme with TTL=5 can reduce the average query traffic cost by up to 80% while achieving the same query hits. Additionally, the scheme can shorten the query response time by up to 59%.

Key words: Peer-to-peer, Searching scheme, Cluster, Index

中图分类号: 

• TP393