Intrinsic Design Constraints and Tradeoffs of Multivariable Multi-Channel Networked Systems

Description

This project launches a comprehensive investigation into networked control systems, with a particular focus on the constraints (capacity constraint, quantization error, etc) of communication channels on feedback control performance. Our principal objectives are:To obtain analytical solutions and bounds for optimal tracking and regulation performance of multivariable, multi-channel networked systems under information-limited feedback;To develop Bode and Poisson type integral relations for multivariable, multi-channel networked feedback systems;To develop insights and understanding into resource (capacity, power) allocation problems of multi-channel systems;To investigate problems of joint design of controller and communication channel for optimal performance;To develop benchmarks, guidelines, and rules of thumb for the design of multivariable, multi-channel networked control systems.Underlying these objectives is the goal to discover, understand, and quantify the intrinsic constraints, fundamental performance limits, and inherent design limitations and tradeoffs for multivariable, multi-channel networked feedback systems. Built on the PI's expertise and prior success, the project employs a variety of techniques that seek to fuse control theory and information theory.Networked control systems, a setting in which control tasks are executed by exchanging information over wired or wireless communication networks, differ fundamentally from the conventional control systems and present new challenges unexplored in the past. This project tackles one of the most compelling issues facing networked control: the information constraint on control. The proposal outlines a number of key problems, where we believe significant challenges exist and advances can be made. A focused research plan has been delineated, which is supported by concrete, realistic problem formulations and technical approaches. The project will provide benchmarks and guidelines, and will lead to new and better understanding of networked control systems, and potentially to improved design methods and techniques, thus contributing to the advance and application of networked control theories and more generally, to control problems subject to information constraints. The research will also lend insights and thrust to information theoretic problems.