Reservation-Based State Feedback Scheduler for Hybrid Real-Time Systems

In recent years a new class of hybrid real-time applications has emerged. The peculiarity of these applications is that both the resource requirements and the arrival rates of service requests are not known in advance and each type of job has its own temporal requests. The traditional "open loop" scheduling algorithms perform poorly in those systems because those algorithms are not adjusted base on the system workloads. It has been shown that feedback control is very effective to support the specified performance of those unpredictable dynamic systems. In order to use feedback control scheduling efficiently, it is necessary to have a model that describes the behavior of the system adequately. In this paper, we propose an adaptive reservation based scheduler with a feedback scheme. After describing a precise mathematical model of the scheduler via state-space equations, we design a controller to adjust the resource assignation for different kinds of tasks in order to satisfy their temporal requests. The global stability of our scheduler is discussed as a piece-wise affine system (PWA). Experiments based on VxWorks/Goahead embedded Web server system demonstrate the effectiveness of the proposed scheduling algorithm when there are large bursts of total load in the system.