Control Oriented Direct Feedback Scheduling

The performance of multitasking control systems with resource constraints depends not only on controller design, but also on efficient scheduling of the shared computing resources. To support codesign of feedback control and real-time scheduling under overload conditions, a direct feedback scheduling approach is proposed. The goal is to optimize the overall system performance through dynamically allocating limited CPU time based on control requirements. A control application oriented task model is suggested, providing an interface between control performance and task scheduling decision. Based on this model, a novel scheduling algorithm intended for control tasks is presented. The overload condition is effectively handled using a job skipping scheme. Preliminary simulations exhibit the benefits of the proposed approach, with comparison against traditional control systems design methodologies.