In a joint initiative, ICARIS merges PNNL’s data science and grid security strengths with Georgia Tech’s embedded systems security and workforce development to deliver technologies, test beds, and trained talent to safeguard critical infrastructure.

AI translates PLC specifications into temporal logic rules to adapt to shifting configurations and threats, reducing manual abstraction and enabling scalable protection for systems of systems.

Programmable logic controllers drive industrial processes across sectors like energy, banking, transportation, and communications and remain prime targets for cyber attacks, a risk highlighted by incidents such as Stuxnet.

Beyond power grids, the work envisions securing nuclear energy infrastructure and leveraging AI to advance environmental modeling and reactor siting, under DOE leadership and industry partnerships.

Key outputs include the GLIMPSE grid visualization tool and the Physically Aware Cyber Platform, both using machine learning and reinforcement learning to detect, mitigate, and respond to cyber threats in real time.

Plans include Department of Defense collaborations on electronic warfare and communications, signaling potential field deployments of integrated capabilities for warfighter contexts.

Industry engagement is a priority, with tests and potential deployments at Southern Company’s grid test bed to validate real-world applicability and bridge lab research with end users.

The project identifies two PLC attack vectors: adversaries rewriting PLC rules to drive unsafe states and data poisoning that makes unsafe conditions appear safe, causing unstable operation.

ICARIS, a collaboration between PNNL and Georgia Tech, is releasing results from a five-year AI-focused study to protect programmable logic controllers across U.S. critical infrastructure.