Modeling Embedded Software: State of the Art and Beyond

State-of-the-art software development increasingly relies on describing the system (or software) graphically, abstracting from an actual implementation platform, and supporting to generate an executable system out of the model. Similarly, in electrical engineering often graphically represented models are used to describe the controlled system including its low-level control behavior. Since developing embedded software requires using abstract, functional models of the software as well as incorporating aspects from the implemented algorithmic behavior, a suitable model-based development process must integrate models from both fields. By comparing the results from modeling an automotive embedded controller software module with eight different state-of-the art CASE tools, we show what kind of abstractions, views, and models have proven useful in that domain. Furthermore, we show what analytical and generative development steps are currently supported by CASE tools. Based on those experiences and results from other research on efficiency and effectiveness of software engineering techniques and CASE support, we derive the properties of a model-based development process and illustrate it by example support functionalities