Providing Evidence for Correct and Timely Functioning of Software Safety Mechanisms

Abstract

In many application domains, the development of safety-critical systems must follow standards that define process steps and artifacts to establish a comprehensive safety argumentation. Commonly, this involves the identification of hazards and risks as well as the formulation of a safety concept to mitigate these risks. The concept is decomposed into safety requirements, which are finally implemented in hardware and software. All steps must be covered by analyses to ensure that the concept is effective and correctly implemented. This work focuses on timing aspects of the safety concept, i.e., on how it can be ensured that risk mitigation occurs in time. Based on an industrial use case, we show how consistent timing specifications can be derived, decomposed, and implemented in a complete and sound way. The approach extends previous work on contract-based design and investigates on explicating failure modes and fault detection in contract specifications. Finally, we show how model checking can support the verification of safety concepts and their implementation.