一种集成R树、哈希表和B*树的高效轨迹数据索引方法

为兼顾时空索引方法的空间利用率、时间效率和查询种类,提出了一种新的轨迹数据索引方法——HBSTR树。其基本思想是:轨迹采样点以轨迹节点的形式成组集中管理,哈希表用于维护移动目标的最新轨迹节点,轨迹节点满后作为叶节点插入时空R树,另外采用B*树对轨迹节点构建一维索引,既有利于提升索引创建效率,又同时满足时空条件搜索和特定目标轨迹搜索等多种查询类型。为提升时空查询效率,提出了新的时空R树评价指标和节点选择子算法改进时空R树插入算法,同时提出了一种时空R树的数据库存储方案。试验结果表明,HBSTR树在创建效率、查询效率和支持查询类型等方面综合性能优于现有方法,支持大规模实时轨迹数据库的动态更新和高效访问。

An Efficient Trajectory Data Index Integrating R-tree,Hash and B*-tree

To take into account all of efficiency and query capability, this paper presents a new trajectory data index named HBSTR-tree. In HBSTR-tree, trajectory sample points are collectively stored into trajectory nodes sequentially. Hash table is adopted to index the most recent trajectory nodes of mobile targets, and trajectory nodes will not be inserted into spatio-temporal R-tree until full, which can enhance generation performance in this way. Meantime, one-dimensional index of trajectory nodes in the form of B^*-tree is built. Therefore, HBSTR-tree can satisfy both spatio-temporal query and target trajectory query. In order to improve search efficiency, a new criterion for spatio-temporal R-tree and one new node-selection sub-algorithm are put forward, which further optimize insertion algorithm of spatio-temporal R-tree. Furthermore, a database storage scheme for spatio-temporal R-tree is also brought up. Experimental results prove that HBSTR-tree outperforms current methods in several aspects such as generation efficiency, query performance and supported query types, and then supports real-time updates and efficient accesses of huge trajectory database.