The scale self-adjusting multi-resolution network traffic anomaly detection

Identifying network traffic anomalies accurately and rapidly is very critical to efficient operation of computer network. In this paper, to improve the existing anomaly detection, we propose a novel multi-resolution network traffic anomaly detection approach based on S transform with self-adjusting scale. By introducing S transform, we can decompose network traffic signal into a group of different frequency sub-bands according to the traffic signalpsilas characteristics. By means of self-adjusting reconstruction of the signal from different frequency sub-bands, our method is able to confirm the anomaly characteristics and enhances the reliability of detection. By means of self-adaptive window selection, we are able to determine the length of detection window according to the spectrum characteristics of the corresponding signal. The simulation results prove that the method can detect the network traffic anomaly efficiently and rapidly, and excels the existing multi-resolution anomaly detection methods.