Preemption-delay aware self-triggered controller: A scheduler-controller codesign approach

Abstract

Digital control systems traditionally are implemented using real-time periodic tasks, which periodically perform the sampling, computation, and actuation steps. Periods are usually chosen pessimistically too short to guarantee the control system stability and/or control performance. However, most of the times, the control system might be stable, accommodating in an appropriate state, and hence, there is no need to execute some control jobs. Self-triggered controllers have been introduced to address this issue through releasing less jobs, depending on the current state of the system. As a result, fewer jobs are executed during the system life time, leading to less energy consumption in the sensor, processing, actuator, and communicating devices. Stability and control performance of self-triggered controllers are sensitive to I/O delays, namely delays happening between samplings and actuations, which are mostly affected by the interferences between different tasks. This paper presents a scheduler-controller co-design approach by first showing how a predictable preemptive scheduling algorithm can provide an upper-bound on the I/O delay, and then presenting an I/O delay-aware self-triggered controller which can tolerate the bounded delays.