Monitoring capabilities of schedulers in model-driven development of real-time systems

摘要

Model-driven development has the potential to reduce the design complexity of real-time embedded systems by increasing the abstraction level, enabling analysis at earlier phases of development, and automatic generation of code from the models. In this context, capabilities of schedulers as part of the underlying platform play an important role. They can affect the complexity of code generators and how the model is implemented on the platform. Also, the way a scheduler monitors the timing behaviors of tasks and schedules them can facilitate the extraction of runtime information. This information can then be used as feedback to the original model in order to identify parts of the model that may need to be re-designed and modified. This is especially important in order to achieve round-trip support for model-driven development of real-time systems. In this paper, we describe our work in providing such monitoring features by introducing a second layer scheduler on top of the OSE real-time operating system's scheduler. The goal is to extend the monitoring capabilities of the scheduler without modifying the kernel. The approach can also contribute to the predictability of applications by bringing more awareness to the scheduler about the type of real-time tasks (i.e., periodic, sporadic, and aperiodic) that are to be scheduled and the information that should be monitored and logged for each type.