Adaptive sampling for real-time control based on sampling urgency

Abstract

We investigate the problem of optimizing the sampling frequency of a real-time control system under the constraint of available computational resource. We propose the novel notion of sampling urgency to indicate the time-dependent importance of sampling frequency, and establish that system performance is directly related to sampling urgency. We then synthesize an adaptive sampling algorithm that varies the sampling frequency on-line according to sampling urgency and show that this adaptive sampling algorithm leads to significant reduction in computation (without sacrificing performance) in a real-time control process.