AdvancingModel-Based Design by Modeling Approximations of Computational Semantics

Over the past decades, engineered systems have increasingly come to rely on embedded computation in order to include advanced and sophisticated features. The unparallelled flexibility of software has been a blessing for implementing functionality with a complexity that could not have been imagined heretofore. One important manifestation of this is in the use of software as the universal system integration mechanism. With the increasing use, however, has come a suite of difficulties in effectively employing software engineering practices because (i) C (the language of choice in embedded software implementation) is very close to the hardware implementation and (ii) software engineering methods typically only consider logical correctness, irrespective of critical characteristics for embedded computation (e.g., response time). To address these problems, Model-Based Design helps raise the level of abstraction while accounting for such critical characteristics. The corresponding models are designed using high-level formalisms such as block diagrams and state transition diagrams whose meaning is particularly intuitive because of their executable nature. The necessity to support increasingly complicated language elements, however, has caused the underlying execution engine to explode in complexity. As a result, the meaning of the high-level formalisms exists almost exclusively by merit of simulation. This paper attempts to present the challenges faced by the current state of Model-Based Design tools and outlines a solution approach by modeling the execution engine.