Toward Flexible Scheduling of Real-Time Control Tasks: Reviewing Basic Control Models

Abstract

We review state-space control models in order to identify timing properties that can favour flexible scheduling of real-time control tasks. First, from the state-space model of a linear time-invariant discrete-time control system with time delay, we derive a new model that involves computing the control signal with a predicted state vector at the actuation instant. Second, by allowing irregular sampling instants, we show that the new model only forces a single synchronization point (actuation instant) while having a feasible implementation. This augurs better schedulability for a set of real-time control tasks, and can provide robustness against scheduling induced jitters.