A Fast Search Algorithm of Anomaly in Data Streams Based on Shifted Wavelet Tree

Anomaly detection is to find aggregate which is different from the most aggregates. Aiming at the limitation of the anomaly detection algorithm of shifted wavelet tree (SWT) in data streams, we propose the improved algorithm which constructs the monotonic search space for binary search after removing the disturbance of bumps to increase the efficiency of detection, and uses the real-time incremental update algorithm for meeting the requirement of the online processing of the data streams. The simulation experiments using two data sets of the Gamma Ray and the Power Quality Disturbance (PQD) verify the high effectiveness and accuracy of our algorithm.