Reengineering Variants of MATLAB/Simulink Software Systems

Abstract

In a variety of industrial domains, quality and security are paramount factors during software system development. Model-based languages such as MATLAB/Simulink can improve software quality and are used for the development of safety-critical functionality. To comply with changing customer demands, product portfolios oftentimes emerge ad hoc by copying andmodifying existing systems in an undocumented fashion. The proliferation of redundant, almost alike assets adversely affects the quality, maintenance, and evolution of the variant portfolio. To reinstate sustainable development, we describe a holistic approach to migrate the portfolio toward managed reuse by collapsing redundant parts and reengineering specific relations between similar, almost alike system parts. We elaborate on a technique to capture course-grained variability by assessing the portfolio as a whole, which identifies and groups together similar and redundant system parts. Such groups are analyzed further using a fine-grained comparison procedure, which captures their variability by means of common and varying system assets. The result is a variability model with redundant parts collapsed and reusable parts identified. By that, the overall size of the product portfolio is reduced, allowing for better quality assurance and improved maintenance. Furthermore, with knowledge about the relations between variants, affected software systems can be identified across variant boundaries, mitigating security concerns for the entire product portfolio.