Proceedings of the International Conference on Computer Safety, Reliability, and Security (SAFECOMP), pp. 298–306
September 2020 · doi: 10.1007/978-3-030-54549-9_20
The Goal Structuring Notation (GSN) is popular among safety engineers for modeling assurance cases. GSN elements are specified using plain natural language text, this giving safety engineers great flexibility to express their arguments. However, pure textual arguments introduce ambiguities and prevent automation. Currently, assurance cases are verified by manual reviews, which are error prone, time consuming, and not adequate for today’s systems complexity and agile development methodologies. In this paper we present our research tool FASTEN.Safe, which extends GSN with a set of higher-level modeling language constructs capturing recurring argumentation patterns and integrating formal system models. This allows automatically checking 1) the intrinsic consistency of assurance models, 2) the consistency of arguments with system models and 3) the verification of safety claims themselves by using external verification tools. FASTEN.Safe is open source and allows experimenting with language abstractions to bridge the world of GSN-based arguments that are common among safety engineers and the world of formal methods that enable automation. Last but not least, we report on the preliminary experience gained with FASTEN.Safe.