Improving the index structure with hierarchical techniques in time-series databases

The R*-tree, as a state-of-the-art spatial index, has already been applied in many fields. But in high-dimensional space its performance is not better than sequential scan algorithm. To solve time-series retrieval problem, we introduce a new index structure—Hierarchical Index Tree. Each time-series of database is divided into several segments with same length, each segment of which corresponds to a cell-tree similar to R*-Tree. Based on the proposed structure, we introduce the corresponding insertion and retrieval algorithm respectively. And we prove no false dismissals for the algorithms. Extensive experiments reveal that Hierarchical Index Tree is superior to the traditional retrieval methods when the dimension of time-series is high.