Model-based safety analysis of requirement specifications

Abstract

Model-based design primarily aims to establish a communication framework throughout a system’s design. Moreover, models with formal semantics allow verification based on rigorous methods, including the analysis of system safety. However, building formal models is a tedious manual process and cannot be easily applied to real problems.

A key gap that hinders automation of model development is that there is no systematic way to connect system requirements with the activity of model-based design. In this article, we introduce a workflow to tackle this gap and ultimately automate the analysis of system safety using formal methods.

We extend our previous work on boilerplate-based specification of system requirements with ontological semantics towards specifying FDIR (Failure, Detection, Isolation, Recovery) requirements. The workflow is centered around the automated generation of a model skeleton in SLIM, a component-based formal modeling language, from a set of ontology-based requirement specifications. Our approach has been implemented into a dedicated tool, which not only provides visualization of the ontology relations, but also supports traceability of the analysis findings back to the requirements specification. Finally, we provide results on the safety analysis of a real star-tracker system based on a SLIM model derived by minimally changing the auto-generated model skeleton.