A dynamic indexing structure for searching time-series patterns

Abstract

In order to handle time-series patterns efficiently, a database system needs an index mechanism that will help it retrieve patterns quickly according to their temporal and spatial characteristics. Time-series patterns are represented as a sequence of points in multidimensional spaces, thus traditional indexing methods are not well suited to support this requirement. We propose a dynamic index structure called a TIP-index (time-series pattern index) for efficient manipulation of time-series pattern databases. The TIP-index is developed by improving the extended multidimensional dynamic index file (EMDF). We present the structure and give algorithms for searching and inserting in it. We compare the performance of TIP-index with that of EMDF. The results indicate that insertion and search performance is improved, and we conclude that the TIP-index is efficient for applications in time-series patterns.