Distributed controller synthesis and decision making
Yadav, Vikas (2007) Distributed controller synthesis and decision making. PhD thesis, Iowa State University.
This is the latest version of this eprint.
Full text available as:
Distributed controller design and distributed decision making have been hot topics of investigation in the last few years. New technologies have led to systems where it is critical to identify architectures that distribute the controller effort over sub-controllers to respect the information flow and/or resource constraints. The communication uncertainty between sub-controllers partly governs the optimality of the architecture of the controller. The related synthesis methodology for optimal distributed controller has to address internal stability concerns and has to incorporate the effect of communication uncertainty into the performance metric. In the first part of this thesis, a methodology is developed to address the concerns of sub-controller communication uncertainty. It is demonstrated that different canonical architectures of a centralized design result in appreciably different performance. Methods to identify architectures of information flow where the optimal performance problem is convex are developed. In addition, synthesis methods to incorporate robustness measures with respect to model uncertainty of the communication channel are obtained for the associated distributed architectures. These methods are further refined for specific structures of information flow in the system. In the second part of this thesis, issues in distributed decision making in a large network of nodes are discussed, in particular a distributed averaging consensus protocol is considered which converges asymptotically. However, each node individually never comes to know of the occurrence of convergence, and thus it keeps running required computation and communication throughout its life. This is not desired, as in most of the networks the power of each node is a very limited resource. This thesis provides a distributed algorithm through which each node can distributively detect when the convergence has occurred within a given error margin. This distributed detection takes finite time and happens simultaneously.
Available Versions of this Item
Archive Staff Only: edit this record