Date of Award

Spring 1984

Document Type

Dissertation

Degree Name

Doctor of Engineering Science in Electrical Engineering

Department

Electrical Engineering

First Advisor

Marshall Chuan Yung Kuo

Second Advisor

Robert De Lucia

Third Advisor

Stanley S. Reisman

Fourth Advisor

Roy A. Plastock

Abstract

This research considers the dynamic routing problem of computer communication networks in the framework of decentralized control theory. The routing dynamics are modeled in terms of a state equation with multiple controllers. Routing, or control of message flow, is formulated as an optimal control problem with multiple decision makers. Each decision maker may have access to different set of information and work cooperatively to optimize a common system performance index.

Necessary and sufficient conditions for optimality are derived for a system with a deterministic and a stochastic traffic patterns under a linear information structure and a quadratic performance index. The resultant control strategies are examined with two extreme information cases: (1) complete information where all state information are available to every local controller through measurement or perfect communication, and (2) partial information where there is no communication among controllers. A three node network is used as a numerical example to interpret the results.

Share

COinS