Abstract
Model-Based Design, Systems Engineering ABSTRACT - Automotive mechatronic systems nowadays are facing a big range of challenges. The complexity of those systems is determined by a rising functionality of the single system as well as the steady increase of interactions and interconnections between different systems. On the other hand systems have to be developed faster by using fewer (financial and human) resources for a big variety of product lines, models, and variants.
Software architectures can play an important part in dealing with these topics as they offer the means of achieving all – functional and non-functional – demands and requirements towards a software system.
In this paper we introduce an approach for systematically designing (top-down) or restructuring/ refactoring (bottom-up) software architectures for automotive embedded systems. The process considers all these constraints mentioned above at a time and offers the user different means and tools to design software architectures and monitor their quality objectively and reproducibly. The most relevant of those like objective architecture metrics, model transformations, and a meta model-based architecture depiction including variant handling are presented in detail. By applying this process a goal-oriented as well as incremental and iterative way of evolving/optimizing software architectures is made possible during design or restructuring.
To show explicitly the measurable benefits of the efforts of designing software architectures we picked out a practical example from the driver assistance domain. A historically grown architecture was restructured through several stages. In the end the Dynamic Cruise Control (DCC) function was implemented – one time derived from the old architecture the other time from the new one – as stand-alone solution. The generated results of the resource needs are compared and evaluated.
Keywords: Software Architecture, Automotive Embedded Systems, Driver Assistance