SERA – an AI driven re-programmable simulation environment for explainable resiliency analysis in cyber-physical systems

Abstract

Testbeds are a practical way to perform security exercises on cyberphysical systems (CPS) to understand vulnerabilities and the progression/ impact of cyber-attacks on CPS safety. Most CPS testbeds use a hardware-in-the-loop architecture. However, it limits the testbed's ability to easily reconfigure with new features, new capabilities and new vulnerabilities. Moreover, cyber attacks recreated within the testbed can potentially damage the hardware that the cyber attack targets. These factors suggest a software-based testbed for CPS cyber security and resiliency research. In this paper, we present SERA, an AI-driven, software-based simulation environment for resiliency analysis in CPS. It is specifically geared towards explainable resiliency analysis using the Resiliency Graph model proposed by Bashir et al. [3] and is fully reconfigurable and extendable for experimentation on different types of CPSes.