Software Engineering Institute

Model-based specification of embedded, critical systems (i.e., systems whose failure is deemed unacceptable) is increasingly becoming the standard of practice. However, analysis of these models can prove challenging when the models grow in size and complexity, which is common as more types and volume of data are loaded into them. One well-studied technique for grappling with complexity in models is slicing, where slices of models are highlighted according to some criterion. In this work, we describe a new software tool: the OSATE Slicer, which extends the concept of slicing to architectural models of embedded, critical systems. It does this by calculating various notions of reachability that can be used to support both manual and automated analyses of system models. We then evaluate the utility of our approach based on several existing analyses and its performance based on a pre-existing corpus of architectural models and find both to be satisfactory.

An archive file accompanies the paper and includes models and documentation to support repeating the performance evaluation.