Similarity Match in Time Series Streams under Dynamic Time Warping Distance

Time series streams have very wide applications in recent years, more and more scenarios need similarity match in time series streams and it becomes a hot research topic. Due to the fast online arrival and time-critical characteristics, the similarity measure should support handling the variable lengths sequence pairs. Dynamic Time Warping distance suits the request. But it has the disadvantage of high computation time and space cost. We use shortest DTW path algorithm to solve stream monitoring problem. Theoretical analysis shows this method is superior to the Brute Force method in time and space complexity.