Statistical Techniques for Detecting Traffic Anomalies Through Packet Header Data

This paper proposes a traffic anomaly detector, operated in postmortem and in real-time, by passively monitoring packet headers of traffic. The frequent attacks on network infrastructure, using various forms of denial of service attacks, have led to an increased need for developing techniques for analyzing network traffic. If efficient analysis tools were available, it could become possible to detect the attacks, anomalies and to take action to contain the attacks appropriately before they have had time to propagate across the network.