Dr. Avimanyu Sahoo, an assistant professor in the Electrical and Computer Engineering department at The University of Alabama in Huntsville (UAH), has secured a $299,969 grant from the National Science Foundation (NSF) to study vulnerabilities in learning-based intelligent cyber-physical systems (CPS) and develop defense strategies.

This research is part of a $600,000 collaboration with Dr. Vignesh Narayan from the Artificial Intelligence Institute at the University of South Carolina and will continue through June 2027.

Cyber-physical systems (CPS), which integrate physical components with intelligent controllers, are essential in fields like smart grids, autonomous vehicles, and robotic networks. These systems use reinforcement learning-based controllers to enhance adaptability and efficiency, mimicking human-like decision-making processes.

However, Dr. Sahoo explained, “The integration of learning components also introduces security challenges, as attackers can manipulate reward signals to divert the controller’s actions.”

This vulnerability could allow adversaries to compromise CPS decision-making, potentially leading to inefficient or harmful system behavior, such as rerouting an autonomous vehicle or disrupting power grids.

The project aims to systematically identify attack vectors and create models to defend against these intrusions. In a practical demonstration, the team will validate its findings in a microgrid environment, and the research will also be presented to high school students through robotics applications, broadening its educational impact.

Jointly funded by NSF’s Secure and Trustworthy Cyberspace (SaTC) program and EPSCoR, this research aligns with increasing global priorities for safeguarding critical infrastructure and advancing secure, resilient technologies in both civilian and military applications.