Evaluating a formal scenario-based method for the requirements analysis in automotive software engineering

Abstract

Automotive software systems often consist of multiple reactive components that must satisfy complex and safety-critical requirements. In automotive projects, the requirements are usually documented informally and are reviewed manually; this regularly causes inconsistencies to remain hidden until the integration phase, where their repair requires costly iterations. We therefore seek methods for the early automated requirement analysis and evaluated the scenario-based specification approach based on LSCs/MSDs; it promises to support an incremental and precise specification of requirements, and offers automated analysis through scenario execution and formal realizability checking. In a case study, we used ScenarioTools to model and analyze the requirements of a software to control a high-voltage coupling for electric vehicles. Our example contained 36 requirements and assumptions that we could successfully formalize, and we could successfully find specification defects by automated realizability checking. In this paper, we report on lessons learned, tool and method extensions we have introduced, and open challenges.