Scheduling Adaptive Variable-Rate Tasks using Compositional Real-Time Scheduler for Automotive System

Abstract

Periodic resource model allows integration of multiple real-time applications on a shared hardware with temporal isolation. Meanwhile, recent studies reported that certain automotive application (i.e., engine control) requires its so-called adaptive variable-rate tasks to be executed following the status of its control target (i.e., crankshaft rotation angle). Since such task is not studied in conjunction with periodic resource model, a new method is needed for scheduling them on a periodic resource. For this, this paper proposes a novel method for finding tight periodic resource parameters for given adaptive variable-rate tasks. Our method first reduces each of them into a periodic task where if it is schedulable on a periodic resource, the original task is also guaranteed to be schedulable. Then optimal periodic resource parameters are found considering context switching overheads. To the best of our knowledge, this is the first effort to schedule adaptive variable-rate tasks using periodic resource model.