一种基于位向量的位置索引压缩方法*

为了缓解大规模分布式系统中共享资源位置索引信息带来的存储压力，提出一种基于位向量的信息压缩方法。该方法建立在区/站/节点三级分层和区间对等的混合覆盖网结构上，通过在两级代理节点的有序子节点集和一定位数的位向量间建立映射关系，进而把冗余资源的位置索引信息汇聚到一条记录，逆向的位向量查询则可获取副本的分布信息。实验表明，该压缩方法达到了较高的记录压缩比，同时对提高副本定位效率起到了一定的作用。

Novel location-index compression mechanism based on bit vector

To reduce the stress brought by shared resource in large-scale distributed system, a kind of novel location-index compression mechanism based on bit vector is put forward. It is established on hybrid overlay network that is composed with three levels of District/Site/Node and peer-to-peer District level. After mapping sub nodes of District and Site level agent nodes to corresponding bit vector, the location-index of many same replicas can be compressed to one record by bottom-up location-index publish and aggregating. Meanwhile, replicas in system can be located by query on corresponding bit vector. Application and experimental results indicate that this method can achieve high record compression ratio and improve the replica location efficiency.