Sampling Jitter Compensation in Real-time Control Systems

In control theory, sampling and actuation are generally assumed synchronous and periodic, and a highly deterministic timing of an implementation is assumed [4]. When a control algorithm is executed by a task (or by a set of subtasks) in a multitasking real-time system, those assumptions are not met. This causes control performance degradation and even instability [7]. In addition, the scheduling algorithm may over-constrain the system when trying to fulfil the stringent timing constraints that control theory mandates, resulting in a poor schedulability In this paper, we propose an integrated approach to the design of real-time computer-controlled systems in order to obtain both the best control performance as well as the best system’s schedulability. Real-time scheduling algorithms introduce various forms of jitter to each task instance execution. As a consequence, if those tasks are executing traditional control designs (control tasks), the system performance is not as good as expected. On the other hand, by using an adequate control design strategy that takes into account those jitters, the system performance can improve dramatically. In order to fully use these more suitable control design strategies, the scheduling algorithm may have to provide, at some point, valuable information to each control task instance for compensating the current introduced jitters. At the same time these control design strategies will provide new flexible constraints to each control task instance, allowing a better schedulability of the whole set of tasks (control and non control tasks). In this paper we focus on the sampling jitter problem and show that by using this compensation approach, the control performance degradation due to the sampling jitter introduced by real-time scheduling algorithms is eliminated while guaranteeing stability [8].