Model-merged development for analysis and design of dependable software systems

Abstract

Current developments of dependable software systems require that developers from different engineering branches perform a cross-checked analysis to evaluate the impact of diverse viewpoints across multiple models. A change on any representation during the development process increases costs and risks since it requires synchronization of models by updating, and then re-verifying/validating each model. This paper presents a novel approach that reduces the co-development effort by enabling system developers who have different viewpoints of the same system architecture, to deal with a single and shared model (no models interfaced with each other as proposed by existing solutions). The modeling technique initially proposed tackles the above challenge by merging the notation of control and software requirements into a single system model. Thus, the impact of changes in the notation of control design can be reflected in the software ones, and vice versa so that the modifications are automatically included when rechecking each model. This paper shows results from an initial merger of models by combining control and software diagrams, a discussion on suitability of modeling technologies, and future research work.