Event-Based Estimation and Control

Researchers: Marcus Thelander AndrénAnton Cervin, Bo Bernhardsson, Kristian Soltesz

Funding: Swedish Research Council

The vast majority of all feedback controllers today are implemented using digital computers, relying on periodic sampling, computation, and actuation. For linear systems, sampled-data control theory provides powerful tools for direct digital design, while implementations of nonlinear control designs tend to rely on discretization combined with fast periodic sampling. In recent years, there has been a growing research interest in event-based control, in particular in connection to distributed and networked control systems. The basic idea is to communicate, compute, or control only when something significant has occurred in the system. The motivation for abandoning the time-triggered paradigm is to better cope with various constraints or bottlenecks in the system, such as sensors with limited resolution, limited communication or computation bandwidth, energy constraints, or constraints on the number of actuations.

During 2016 we have investigated stochastic event-triggered sampling. Using a specific stochastic triggering rule in the sensor node simplifies the estimation problem in the controller, allowing a standard time-varying Kalman filter to be used. We have also developed a simple benchmark for event-based control that is based on LQG-optimal PI(D) control. Using Monte Carlo simulations, the benchmark can be used to compare the performance of continuous-time, sampled-data and various event-based control strategies.