Fast Subsequence Matching in Time-series Database

This paper proposes a new algorithm of fast subsequence matching in time-series database. Dramatically differing from FRM algorithm, we develop an index structure using VA-file, which maps data points directly into the index structure. We also propose a new algorithm for Range Query on this index structure. We implement our method and carry out experiments in different ways on three kinds of time-series data. The result shows that our algorithm accelerated the search time significantly.