Double Wedgie: Efficient Filtering Algorithm for Streaming Time Series

摘要

Filtering streaming time series can benefit monitoring a streaming time series for predefined patterns. Based on the Atomic Wedgie algorithm, we propose a new concept named Double Wedgie and define a tighter lower bound distance on it. The filtering algorithm based on Double Wedgie is more efficient than Atomic Wedgie. Extensive experiments demonstrate that the new algorithm can achieve tremendous improvements in the streaming time series query filtering with guaranteed no false dismissal. Also, the larger difference between the predefined patterns is, the more efficient the algorithm is.