Low Complexity Optimal Policies for Networked Control Systems | Manali Dutta | IISc

Описание к видео Low Complexity Optimal Policies for Networked Control Systems | Manali Dutta | IISc

Title: Low Complexity Optimal Policies for Networked Control Systems

Speaker(s): Manali Dutta, PhD Student, IISc.

Time: 5:00 PM - 6:00 PM (IST)

Date: 03/09/2024

Venue: MP 20 and Online on Zoom

Abstract: We design optimal scheduling policies for wireless networked control system (WNCS) composed of plants, sensors, controllers, and actuators which are connected via unreliable wireless channels. Sensor observes the plant, and encodes these observations into data packets before transmitting them to the controller via wireless communication channel. The controller estimates the plant state, and generates controls based on the information received from the sensor. It then transmits the control packets to the actuator via an unreliable communication channel. The actuator, in turn, applies these controls to the plant. We derive low complexity optimal policies for the following cases (i) when the communication channel is partially observed, (ii) when the objective function is a risk-sensitive cost criterion, meaning in addition to penalizing the mean cost, we also penalize its higher order moments, and (iii) controller is half-duplex. Specifically, we show that for (i) the sensor attempts transmission only when the current belief state of the channel exceeds a certain threshold, for (ii) the sensor transmits only when the magnitude of the current error (different between current and a priori estimate of plant state) exceeds a certain threshold, and for (iii) in case there is a data packet available with the controller, the sensor activates controller--actuator channel only when the magnitude of the current plant state exceeds a certain threshold.

Bio: Manali Dutta is a PhD student at Indian Institute of Science, Bangalore, having joined the program in 2021. Her research interests lie in the domain of networked control systems, stochastic control, and reinforcement learning. Prior to her PhD studies, she obtained her M.Tech degree from the Indian Institute of Technology, Kharagpur, in 2021.

Комментарии

Информация по комментариям в разработке