摘要： 在时空数据库中，频繁更新会导致TPR树更新与查询性能下降。针对该问题，提出MAH_TPR索引方法，分别对预处理过程、索引结构及更新算法进行优化。在构建索引及更新操作时，通过使用空间聚类来减少节点间空间区域的交叠几率。引入基于磁盘的Hash辅助存储结构，在直接访问叶节点的基础上进一步减少磁盘I/O的操作。引入基于内存的移动对象辅助存储结构，用于存储发出频繁更新请求，以避免主索引结构节点的合并和分裂。实验结果表明，MAH_TPR索引方法的查询性能优于HTPR方法和LGU方法，更新性能优于HTPR索引方法。

关键词: 频繁更新, 空间聚类, MAH_TPR索引构建, MAH_TPR索引更新, 移动对象, Hash辅助存储结构

Abstract: The MAH_TPR indexing method is proposed which aims to solve the problem of decreased update performance and query performance because of frequent updates in spatial-temporal database. This method is optimized by prepared processing, indexing structure and update algorithm. Overlapping probability among the spatial areas of nodes is significantly reduced by using the spatial clustering in structuring indexing and updating. The leaf nodes can be accessed directly and further the disk I/O operation is decreased by introducing a disk-based hash auxiliary structure. Node merging and splitting in main indexing structure are avoided by employing a memory-based auxiliary storage structure which is used to store the moving objects that have frequent update requests. Experiments in update and query performances of the method are studied. The results show that the MAH_TPR indexing method has a better query performance than HTPR indexing method and LGU indexing method. Its update performance is better than that of HTPR indexing method.

Key words: frequent update, spatial clustering, MAH_TPR index constructing, MAH_TPR index updating, moving object, Hash auxiliary storage structure

中图分类号: 

• TP311.13