Scalable Industrial Control System Fuzzing Using Explainable AI

ABSTRACT

Learning-based modeling and fuzzing of industrial control systems (ICS) has shown promising results to find ICS attacks without requiring domain-specific expertise. However, ICS fuzzing faces the key challenge of state explosion, where the fuzzing space grows exponentially with ICS size. In this paper, we propose to exploit explainable AI (XAI) to address this challenge. Our results show that XAI accurately explains the ICS model and significantly speeds-up attack fuzzing by 64x.

BIO

Dr. Jingshu Chen currently is assistant professor of Computer Science at Oakland University. Before she joined Oakland University at 2017, she worked at FDA as ORISE researcher and INRIA as postdoc researchers. She earned her Ph.D degree in Computer Science with Dr. Sandeep Kulkarni from Michigan State University at 2013. Her research interests focus on developing novel techniques including formal methods and AI assisted techniques for improving reliability of safety critical systems.