Using Obstacle Analysis to Support SysML-Based Model Testing for Cyber Physical Systems

Cyber-physical systems play a crucial role in various applications, ranging from critical infrastructure control like power grid to the technological revolution of Industry 4.0 aimed to integrate and automate the manufacturing value chain. The Systems Modeling Language (SysML) represents a significant and increasing segment of industrial support for the development of cyber-physical systems partly due to the language's built-in mechanisms for modeling the requirements. In this paper, we leverage goal-oriented obstacle analysis to systematically identify the impediments to the fulfillment of requirements, and further examine several machine learning algorithms' capabilities of classifying these impediments into the components that constitute the cyber physical systems. We then investigate the extent to which a state-of-the-practice SysML tool simulates these obstacles, thereby assessing the risks of system failures at the requirements level. Our work offers concrete insights into model testing which incorporates the deep intertwining of software with hardware in order to improve the robustness of the cyber-physical systems.