A Multirate Filterbank Approach Toward Joint Controller and Coder/Decoder Design of Wireless Networked Control Systems

This proposal initiates a comprehensive research program into the study of networked control systems, with a particular orientation toward wireless networked systems. Our principal goal is to investigate the fundamental design constraints and performance limits of wireless networked feedback control, and the design of communication components such as coder/decoder to circumvent the constraints and thus to improve performance. We employ the benchmark performance index of tracking as our performance measure, and attempt to accomplish the following technical objectives:To quantify the fundamental constraints and limits imposed by essential wireless characteristics such as packet loss, random delay, fading and multipath propagation processes on control design, and to develop computational solutions and analytical bounds for optimal tracking over feedback channels with these characteristics;To develop a mean-square small-gain approach for the analysis and design of networked feedback systems with random logarithmically quantized and random multiplicative channels, which are known to capture a variety of wireless phenomena and provide a unified framework addressing packet loss, random delay, fading and multipath propagation;To investigate how coder/decoder may be designed to improve tracking performance, and in particular how controller and coder/decoder may be jointly designed to optimize tracking performance;To develop a multirate filterbank small-gain approach, and lifting techniques for the analysis and design of wireless networked feedback systems.Underlying these objectives is the central task that seeks to investigate how wireless networks may result in design constraints and hence degrade control performance in ways unfound previously, and to find ways to design coder/decoder to circumvent the constraints and improve the performance. We propose a research plan that addresses and incorporates key characteristics of wireless communication channels, and employs key wireless technologies such as multirate filterbank coders/decoders. We develop a mean-square small-gain approach particularly relevant to these problems, which draws upon theories and techniques of robust control, multirate signal processing, and optimal functional approximation. The project seeks to advance significantly the current state of networked control research, and is believed to hold considerable academic interest and practical importance. Indeed, as wireless communication technology is continuing to grow at a breathtaking speed, control over wireless networks promises to have an enormous scope of applications. The project seeks to provide answers to key and fundamental issues facing the design of wireless networked control systems, and thus has the potential to impact a variety of these areas in both theoretical and applied research.