The integration of computation, communication and sensing technologies into our transportation, power grid, healthcare, and manufacturing systems presents unprecedented challenges in ensuring the security and safety of these systems. Cyber attacks on such systems cross from the cyber realm into the physical world, and we must deal with new attack trends that may cause vehicles to veer off the road, manipulate devices responsible for power generation, distribution and consumption, and exploit robotic/drone systems for malicious activities. This project develops methodologies that ensure the resiliency of such systems through developing defense strategies in response to the attack strategies available to the adversary.
To ensure the resiliency of Cyber-Physical Systems (CPS) against cyber attacks, this project explores the use of a game-theoretic framework to orchestrate the CPS with the proper combination and configuration of action blocks. This project posits that action blocks can be used to ensure the integrity of a system's control and measurement signals by checking the signals' values and manipulating these values in manners devised to detect attacks. This exploratory effort will develop initial game theoretic models and accompanying reinforcement learning techniques to investigate possible defense strategies that rely on mixtures of action blocks, focusing on the generalizability, implementability, and scalability of the approach. This project lays the ground work for a more extensive open library of action blocks in which CPS developers may not only contribute blocks but also invoke the resulting methods to assemble potent defense strategies for their systems.
|