A model-driven methodology to support runtime assurance of open and adaptive systems

Abstract

Cyber-physical systems (CPS) harbor vast economic potential and societal impact, enabling new types of promising applications in different embedded system domains such as automotive, avionics, railway, healthcare, and home automation. These systems perform safety-critical functions, thus in case of failure, they may harm people or lead to the collapse of important infrastructures with catastrophic consequences to industry and/or society. Therefore, CPSs demand the justification of system safety and component reliability. Assurance cases provide an explicit means for justifying/assessing confidence in system dependability with explicit and implicit references to design, safety, and reliability artifacts. The Structured Assurance Case Metamodel (SACM) defines a standardized metamodel for representing structured assurance cases. SACM provides the foundations for model-based system assurance with potential to be applied in emergent open and adaptive CPS domains. Motivation: CPSs are loosely connected and come together as temporary configurations of smaller systems that may dissolve and give place to other configurations. The open and adaptive nature of CPSs, demands a paradigm shift from design-time to runtime system assurance. To ensure the dependability of CPSs, assurance cases are expected to be exchanged, integrated, and verified at runtime. However, enabling the traceability between assurance cases and system design, analysis, and process models, which are part of Executable Digital Dependability Identities (EDDIs) of CPS components, is still a barrier. Objective: This study introduces a novel model-driven methodology to support the specification and synthesis of executable SACM assurance cases from various systems models to demonstrate CPS safety and security at runtime. This work also enhances the SACM standard with pattern extensions that added semantics to the concept of Implementation Constraint to support the specification of executable assurance case pattern (templates) specifications linked to information from design, analysis, and process models (evidence) that constitute the EDDI of a CPS and/or CPS component. This study also provides an assurance case pattern specification and instantiation methodology, comprising a modeling tool for specifying executable assurance case patterns linked to EDDI information, and an instantiation algorithm to support the automatic synthesis of assurance cases from EDDI information. Evaluation: The feasibility of the proposed methodology is evaluated in two illustrative studies from the automotive domain. Results: The effectiveness of the proposed methodology are demonstrated by low instantiation execution time, higher accuracy, and minimal omissions in the generated product assurance cases for the two medium-sized automotive systems.