Flow-based Worm Detection using Correlated Honeypot Logs

Abstract:
Attack detection in high-speed networks is a hot research topic. While the performance of packet oriented signature-based approaches is questionable, flow-based anomaly detection shows high false positive rates. We tried to combine both techniques. In this paper, we study the applicability of flow-based attack detection. We installed a lab environment consisting of a monitoring infrastructure and a well-controlled honeypot. Using correlated honeypot logs and flow signatures, we created a first set of attack pattern. The evaluation of the approach was done within our university network. On the positive side, we were able to prove the successful detection of worm attacks. Problems can occur if incomplete monitoring data is used.