An Approach to Support Variant Management on Safety Analysis using CHESS Error Models

Abstract

Software Product Lines (SPL) and component-based approaches have been largely adopted in the development of safety-critical systems due to benefits such as increased product quality, shorter time-to-market and large-scale reuse. Safety-critical system families built upon SPLs, have to address safety, reliability and availability properties across different system variants. Variations in design and usage context may impact on safety properties such as hazards, their causes, and safety requirements to mitigate hazard effects. Therefore, the adoption of SPLs in the development of critical systems requires the introduction of safety analysis into product line processes. The manual analysis of safety properties across different system variants may be time consuming and error-prone, resulting in project delays and in the increase of development costs. Model-based techniques automate system design and safety analysis and have been recognized by safety standards. AMASS is a model-based platform that supports system design, safety analysis and variability management. In this paper we propose a systematic approach, built upon the AMASS platform, to support the reuse of safety assets and generation certifiable evidence for variant-intensive systems. The approach was evaluated in a realistic variant-intensive automotive braking system and reduced the complexity of performing safety analysis on system variants through the reuse of safety information. It has also automated safety analysis and improved the traceability between development and functional safety assets.