A communication/control co-design paradigm for networked control systems.

摘要

This thesis is dedicated to studying the networked control systems. Different from many existing approaches in the literature, we do not formulate the networked control problems as mere controller design problems assuming that the plant and communication network are given a priori. Instead, we propose a new paradigm called communication/control co-design which provides a novel perspective to re-examine the networked control systems. The essence of this new paradigm is to allow the communication network to be jointly designed with the controller design. Such co- design formulation has tremendous advantages over the controller design formulation, for instance, more design flexibility, less capacity requirement for the communication network, and better control performance, etc.;We explore three different forms of communication/control co-design in this thesis. The first form, called channel/controller co-design, is realized by the twist of channel resource allocation, i.e., assuming the individual channel capacities can be allocated among different channels subject to a total capacity constraint. The second form, called scheduling/control co-design, is put forward to address the scenario when multiple control inputs have to share a small number of communication channels. In such a case, the transmission scheduling of the control inputs becomes an additional design freedom. The third form, called coding/control co-design, is initiated to study the networked control via MIMO communication. The encoder/decoder pair in the MIMO transceiver comes up as an additional design freedom on top of the controller.;Under the communication/control co-design paradigm, three networked stabilization problems are formulated and studied in this thesis, each of which addresses one particular form of communication/control co-design. A fundamental limitation on the total channel capacity required for state feedback stabilization is obtained, which is given by the topological entropy of the open-loop plant. In addition, when MIMO communication is utilized in networked control, a majorization type condition is obtained on the subchannel capacities required for networked stabilizability.