NIS06-5: Monitoring Network Data to Detect IP Security Violations

We propose a method of detecting security violations by monitoring IP address space usage at per interface level. Abnormal increases in the unique number of destination IP addresses seen over some fixed time intervals usually indicate scanning activities. An increase in the number of unique source IP addresses could indicate address spoofing. The data used are flow level data collected over some peering links. We discuss in this paper the method developed to monitor the use of IP address space to alert for suspicious activities, such as scanning and possible IP address spoofing. We illustrate the idea of using IP graph to visually display the IP address space usage, and discuss the statistical approaches that can be applied for anomaly detection. To demonstrate the effectiveness of the method, we show how some scans are detected.