Getting a Grip on the Ever-Changing Software in Cyber-Physical Systems

Abstract

As industrial cyber-physical systems grow ever more complex, their software grows naturally and changes continuously. In order to make risk-free changes to their software, it is crucial to understand how the system behaves, and how software changes have an impact on system behavior. We propose a generic two-fold approach to infer state machine models capturing system behavior, and to compare these models to determine and visualize the impact of software changes on system behavior, in a way to make them easily understandable for engineers. Our approach has been applied in the industry at ASML to help prevent software regression problems during critical software redesigns. In that, our approach has been shown to reduce risk and to be valuable.