Principal Investigator: Ricardo Calix
Intrusion Detection and Prevention Systems (IDS/IPS) serve a pivotal role in securing computernetworks. Using machine learning for an Intrusion Detection System is important to stop newattacks that do not have known signatures. The further lowering of the barrier to entry formicroprocessor based systems has made it possible to use specialized machine learning coprocessorsto improve analysis performance. This grant project proposes a machine learningapproach on a small, low powered embedded system that uses network based features to predictbetween normal and abnormal network traffic. A hardware based approach using a machinelearning co-processor is compared to a purely software based approach.