非结构化P2P文件共享网络的应用日益广泛,当前的网络流量主要来自于此类网络中对各类文件资源的搜索。但是目前的搜索机制只适合获取流行资源,存在于个别节点上的稀缺资源很难被搜索到,而用户获取稀缺资源的收益并不小于流行资源,因此高效的搜索命中率将明显提高整个P2P网络的实用性。该文提出一种简单的分布式算法NLIR,使稀缺资源可以像流行资源一样具有较高的搜索命中率。新算法根据度数不同的节点,其处理查询的能力,存储能力,搜索命中率各不相同的特点将稀缺资源的索引副本均匀地扩散到网络中,体现了负载均衡的特性,并且在带宽费用和存储费用都有限的约束条件下,实现了稀缺资源高效的搜索命中率。
Searching for files in popular unstructured P2P file-sharing systems contributes to much of internet traffic. While existing mechanisms can locate popular files easily, studies show they fail to find rare files exiting in the system. In fact the benefit users obtained from rare resources is not less than popular resources. So high searching hit for rare resources will dramatically improve the whole P2P network's practicability. This paper proposes a simple distributed mechanism-NLIR which makes it easy to find rare files as well as popular files. According to peers' heterogeneity, it distributes rare files' index replications uniformly into the networks. Even with constraint conditions that bandwidth and storage is limited, it can achieve high hit-rate for rare resources. Both analysis and simulations demonstrate it imp)roves the performance of the system.