An MDE approach for the design of platform-aware controls in performance-sensitive applications

Model-Based Design is widely adopted in control domains for the early validation of systems properties using simulation or formal verification and the possibility of automatic generation of code. Most tools used in the industrial practice allow for the representation of the controller functionality abstracted from the implementation details. These models may be inaccurate in those cases in which computation and communication delays affect the performance of the controls. To address this problem, we propose a Model-Driven approach in which a Simulink functional model of controls is matched to a model of the execution platform through a mapping model, representing the implementation as a set of tasks and messages. The platform and the implementation are modeled in SysM-L/MARTE and are used to automatically generate a new Simulink model with an additional set of blocks representing the execution time of the tasks running under the control of a selected scheduler. Acceleo and QVTo model-to-text and model-to-model standard transformation languages are used to automatically generate the intermediate models, the task and scheduler blocks.