ASIL tailoring on functional safety requirements

Cyber-physical systems like self-driving cars are highly complex and safety-critical. This results in a great number of safety requirements that have different levels of criticality. In automotive, the criticality is categorized in Automotive Safety Integrity Levels (ASIL). As a high ASIL causes high development effort, the goal is to develop most subsystems with lower ASIL requirements. To achieve this ASIL tailoring, subsystems need to be separated or redundantly implemented. These safety measures are usually integrated late in the development process and thus cause costly development iterations. In this paper, we present a systematic, tool-supported ASIL tailoring process for the requirements analysis phase. It is applied on formal safety requirements and automatically generated fault trees for a functional view of the system. The process supports early planning of safety efforts for mixed-criticality systems and avoids costly late development iterations.