We suggest a systematic modelling method for embedded systems. The goal is to derive models (1) that share the relevant properties with the original system, (2) that are suitable for computer aided analysis, and (3) where the modelling process itself is transparent and efficient, which is necessary to detect modelling errors early and to produce model versions (e.g. for product families). Our aim is to find techniques to enhance the quality of the model and of the informal argument that it accurately represents the system. Our approach is to use joint decomposition of the system model and the correctness property, guided by the structure of the physical environment, following, e.g., engineering blueprints. In this short note we describe our approch to combine Jacksons problem frame approach with a stepwise refinement method to arrive at provably correct designs of embedded systems.
|Name||CTIT Technical Report Series|