Model-Based Software Engineering and Certification: Some Open Issues

Model-based software engineering methodologies, languages, standards, technologies, and tools are in place since many years. While they proved to be effective in several application sectors, e.g. for embedded systems, empirical studies show that their use in industries pursues a variety of goals and that often they are only partially applied, so it is still not clear to what extent they are actually adopted in the engineering practice and whether they achieve the claimed advantages. Notwithstanding this, model-based techniques are being increasingly advocated for use in critical systems engineering. As critical systems have to undergo certification, the question raised to what extent model-based engineering supports certification. While this is not a new issue, the literature is still at the beginning. We examine key aspects in this respect, and identify some open issues.

I. Introduction

The development of critical systems demands strong efforts, due to rigorous processes and intense activities of verification and validation (V&V) to assure the quality of the product. The approaches of Model-Based Engineering (MBE) and Model-Driven Engineering (MDE) seem a natural way to support the development of the modern complex systems, because they increase the level of abstraction, and enable the automation of development and V&V activities as well [1]. MDE uses models as driving artifacts of the development activities, and derives secondary outputs through (automatic) transformations. MBE is broader in scope, and exploits models to support the engineering tasks [2].¹

In the rest of the paper we will refer generically to model-based engineering techniques, disregarding the distinction between MBE and MDE.