Variant configuration and management : challenges and opportunities

Abstract

The goal of this thesis is to understand the complexity of configuring and managing variants in Simulink, a model-based design tool that is used in different industries such as automotive, aerospace, and healthcare; and to simplify it by analyzing its capabilities and proposing different solutions using different modeling, workflow and visualization techniques. Models play a major role in system engineering. Model-based development methods have gained widespread adoption in the past two decades. Simulink has been used by thousands of engineers around the world to design and simulate engineering systems. Any improvement in Simulink directly simplifies engineers' workflows and enables them to achieve their tasks more efficiently and effectively. I have studied variability concepts and use cases of configurators in different domains such as (1) personal customization of products (mass customization), (2) choice navigation in retail, and (3) software and model-based design tools. This has enabled me to understand complexities of configurators in different domains, and learn how some of the complexities have been addressed or managed. I conclude that the variant support in Simulink lacks some capabilities to fully support variability for large real-world applications in industry. To improve Simulink variant configuration and management, I propose solutions based on "feature modeling" and "interactive configurator" concepts. Use of "Binary Decision Diagrams" is recommended to achieve higher runtime performance when using interactive configurators for large modular systems and large teams. Also, I propose different visualization techniques to enable model designers to understand variant relationships in the model, check whether they have modeled the variants properly, and help them to further modularize their systems.