A Method for Continuous Query Over Data Stream using Wavelet Synopsis

Continuous query is an important aspect for data stream management techniques. The focus is to design one-pass scan algorithm over dataset, maintain an effective synopsis data structures in memory which is far smaller than size of the whole dataset. With this data structure, approximate query result can be finished rapidly. A novel method for continuous query is presented in this paper, which is based on wavelet error tree synopsis. In this method, sliding window model is used, adaptive threshold is selected, and the wavelet coefficients in the sliding window can be incrementally updated. These make the method more efficient in memory and response time. It is suitable for not only streaming data but also large amount of historical data. An experiment using real power load dataset proves effectiveness of this method.