A time-frequency detecting method for network traffic anomalies

Traffic anomalies are abnormal changes in network traffic. There are many causes of resulting in traffic anomaly changes, including network fault, abuse, network attack and so on. This paper proposes a novel method for detecting traffic anomalies in a network. Firstly, all Origin-Destination (OD) flows in the network are grouped with accordance with common destination nodes. We then make time-frequency analysis for each group and extract their high frequency component. Based on the extracted high frequency signal, correlation coefficients of each OD flow against the other in a group are computed. Consequently, we can make the exact anomaly detection. Simulation results show that our approach is effective and promising.