Open Real-time Systems: From Modeling to Implementation

Abstract

Correct and efficient implementation of open real-time systems is still a costly and error-prone process. An open real-time system interacts with its execution environment while meeting tim-ing constraints. Rigorous design methodologies for these systems should be based on models able to express the interactions with environment and the timing constraints they satisfy. We present a general model-based implementation method of open real-time systems based on the use of two models: (i) an abstract model representing the interactions between the environment and the application and its timing behavior without considering any execution platform (based on an abstract notion of time); (ii) a physical model representing the behavior of the abstract model running on a given platform, that is, it takes into account execution times. We define an Execution Engine that performs the online computation of schedules for a given application so as to meet its timing constraints. Furthermore, it can detect time-safety viola-tions and stop the execution if the execution times are not compatible with the timing con-straints of the model. We implemented the Execution Engine for BIP programs and validated our method for a module of an autonomous rover—Antenna of the Dala robot.