High-speed Network Anomaly Information Detection Method with Multiple Time Scale Synchronization

Yu Song

Department of Network Information Management Center, Sichuan University of Science and Engineering, Zigong, 643000, China

Abstract: Aiming at the problem of inaccurate detection of abnormal network information detection methods, a high-speed network anomaly information detection method with multiple time scale synchronization is proposed. The method firstly forms a network traffic time series by collecting the number of IP packets per unit time of the line. Then the Haar wavelet transform is used to decompose the sequence and remove the redundancy to obtain the normal wavelet sequence under the guidance of the "3c" rule of the normal distribution, and it is approximated as Gaussian white noise. Finally, the statistical characteristics of the normal distribution sequence are used to judge the network abnormal information. The results show that compared with the high-speed network anomaly information detection method based on data mining, the anomaly detection accuracy of this method is improved by 9.5%.

Keywords: Multiple time scale; Anomaly information detection; Time series; Haar wavelet transform