A Real-Time Detection Approach to Network Traffic Anomalies in Communication Networks

Abstract

With the advance of new network technologies, new types of applications are quickly arising. Network traffic exponentially is rising. Accordingly, this results in new challenges for network traffic anomaly detections. This paper proposes a new quick detection approach to network traffic anomalies. Firstly, network traffic is regarded as a time series of signals and is constructed into a matrix. Secondly, the principal component decomposition is performed for the matrix. The network traffic is divided into principal and non-principal components. Thirdly, the empirical mode decomposition is carried out for these two components. In this case, a quick anomaly detection algorithm is presented. Simulation results show that our approach is feasible and promising.