Automated Red-Teaming Agent for Resilient Microgrid Design

Red teaming is an adversarial approach used to assess a cyber-physical system's (CPS) defenses in order to identify vulnerability. As the CPS grows in complexity, manually conducting red-teaming grows harder due to the larger attack surface and the need for domain expertise. In this paper, we develop an automated red-teaming agent that can couple with simulation test-beds, which can be used to assess system-wide impacts of cyber-attacks. The proposed agent is adaptive and is able to take the system-state into account to determine the optimal attack action at any given point of time. The efficacy red-teaming agent is demonstrated on a microgrid use-case subjected to Man-In-The-Middle attacks.