Toward Modeling and Verification of Uncertainty in Cyber-Physical Systems

Abstract

Uncertainty is an intrinsic property of any cyber-physical system(CPS), therefore handling Uncertainty during the operation of a CPS is an essential requirement for a CPS. Design, development, and testing of modern CPS is a rapidly expanding research area. However, uncertainty handling is still relatively unexplored. The first step towards handling Uncertainty is to identify, define, and classify Uncertainty at various levels of a CPS. This review aims to capture the present state-of-the-art of uncertainty modeling in cyber-physical systems. We define appropriate inclusion and exclusion criteria for the studies to be used in this review and extract data from selected studies, and present in this paper. From our selected studies, we find that two significant lines of research exist in this area, and the ultimate objective of both research directions is to develop the uncertainty-aware test-modeling framework. The first research direction from the simula u-test group [1], [6], who first developed a model of Uncertainty based on UML, called the U-Model, then they extended model-based testing (MBT) to the purview of CPS via lengthening the UML profile of Modeling and Analysis of Real-time and Embedded Systems (MARTE) by adding a set of new UML libraries. We focus on exploring the fundamentals of U-Model in this paper. The second approach is to apply methods from Formal Verification of hardware and software systems to CPS [5], [7]. This approach also strives to model and verify the uncertainties of known and unknown types in a CPS. We find the first approach more general and hence more applicable to all kinds of CPS. Finally, as our incremental contribution to this line of research, we extend U-model for certain specific use cases.