Multi-View Design for Cyber-Physical Systems
Hui Zhao, Ludovic Apvrille, Frédéric Mallet · theses.fr (ABES) · 2017
Cyber-Physical Systems (CPS) characterize all these distributed embedded systems that influence so many aspects of people lives without them even being aware of it. They combine small digital systems deeply entangled with physical processes (like factory control systems, smart transportations or smart home systems). As most large scale distributed systems built by putting together subsystems CPS are largely vulnerable to attacks that may impact the safety or physical integrity of people.In Model-Based System Engineering (MBSE), CPS designers bring together the expertises of a wide variety of experts, each of them using different sets of languages and tools. This variety makes it difficult to combine and integrate all these models and artefacts. Looking at this overall process with the goal to improve security is the main scope of our work.We propose a model-based approach to combine heterogeneous models. Rather than trying to build a universal language to combine all the expressiveness of all the sublanguages, we elaborate on subsets of existing languages to keep only what is needed to conduct the required analyses. We use a meta-language to keep the consistency between these subsets.As a case study we start from Capella, an open-source solution used by major integrating companies of critical-systems. Capella covers a wide design flow from functional analysis to component deployment. However, it does not have any specific mechanism for security analysis and would rely on plugins for that purpose. We explore a solution to combine it with another tool, SysML-sec, which is an extension of SysML dedicated to security and safety analysis. We extract only the required subsets to conduct functional analysis with Capella, and security analysis with SysML-sec. We do the same exercise with AADL to do schedulability analysis and show the generality of our process.