THE DEVELOPMENT OF INTERNET TRAFFIC MODELS TO PREDICT NETWORK BEHAVIOR AND DETECT MALICIOUS CYBER ACTIVITY

Abstract

In collaboration with the company NIKSUN located in Princeton, NJ, models have been developed for performing k-step ahead predictions of utilization rates associated with first flow and then packet utilization. Using both the models developed and the invariants discovered during the modeling process, methodologies for performing anomaly detection are proposed. Further suggestions for research are also proposed.