A 2007 study by the National Academy of Sciences provides strong motivation to explore the use of dependability cases as a means to address verification, and ultimately, certification, of highly complex systems. Kestrel Technology, LLC, is developing a prototype extensible workbench to develop, maintain, and analyze safety cases – a specialized form of dependability cases. The CertWare workbench contributes several core modules supporting safety case models, and extends these with a service-based APIs for plugging new capabilities into the workbench for processing these models.
In addition to typical features such as multi-user safety case model editing, change tracking, and templates, the CertWare workbench emphasizes support for analysis of safety case model structure and content. The APIs for external tool integration provide support for a variety of analyses of the safety case models. Model transformations enable syntactic and semantic analysis. Probabilistic reasoning provides support for uncertain evidence and computations of confidence levels. Logical reasoning provides support for answering queries on the model, determining the impact of negative evidence or inconsistent arguments, and incorporating unknown data. Historical reasoning and trend monitoring supports project management for case production, including assessing the cost and schedule impact of revised requirements or rework.
Initial funding for CertWare development was provided by NASA Langley Research Center Contract NNL10AA08C.