Statistical Techniques for Detecting
Internet Traffic Anomalies
Research Office (ARO),
Computing and Information Sciences.
Paschalidis, Boston University.
This short-term project aims at further developing statistical anomaly
detection approaches that can be used to monitor the aggregate Internet
traffic associated with a group of computer systems. It will built on
recent work by the PI on random and Markovian models for characterizing
typical traffic distributions and detecting deviations from them. In
addition, it will develop an approach for combining inputs from several
traffic anomaly detectors to detect spatial anomalies that could
identify a more global and coordinated attack campaign.