Principal Investigator: Michael Fowler
Cyber evaluation programs like Common Criteria are focused on testing and certifying products based upon predefined profiles, test plans, and quality assurance guidelines. Testing for unpredictable events, however, is more challenging. To demonstrate confidently that a system can withstand unknown and unpredictable attacks, evaluation laboratories need to move towards intelligent and autonomous testing systems. These systems can adapt and learn from the testing in real-time to stretch and evaluate a much wider range of inputs, sequencing, timing, content, and other parameters. In autonomous testing of avionic systems, a key challenge is determining when the existing data can verify the robustness and resiliency of the system to unknown perturbations, and when more information is needed to elucidate the system and reduce uncertainty to an acceptable level.