Optimal work-conserving scheduler synthesis for real-time sporadic tasks using supervisory control of timed discrete-event systems

Abstract

Real-time scheduling strategies for safety-critical systems are primarily focused on ensuring correctness, both functional and temporal. In order to provide the desired predictability in such systems, it is often advisable that all timing requirements be guaranteed offline, before putting the system into operation. Formal approaches allow for all necessary and sufficiency conditions corresponding to a feasible schedule to be checked in a systematic manner. This enables formal approaches to act as effective mechanisms for providing timing guarantees required by safety-critical systems. In this work, we develop a scheduler synthesis framework for the optimal work-conserving scheduling of dynamically arriving, sporadic tasks using a formal approach known as “supervisory control of timed discrete-event systems” (SCTDES). The synthesis process starts with the construction of a resource-constraint-aware task execution model and a deadline-constraint-aware timing specification model, for each task in the given real-time system. The system model (i.e., composite task execution model) is then derived and transformed to guarantee work-conserving co-execution of tasks. Such a work-conserving approach enables the synthesis of schedules which avoid processor idling in the presence of ready-to-execute tasks. Next, we use the (transformed) system and specification models to obtain a supervisor which can be used to construct an optimal scheduler for the given real-time system. Finally, the applicability of the proposed scheme for real-world scenarios is shown by presenting a case study on an instrument control system (ICS).