Date of Award

Spring 1-1-2015

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Electrical, Computer & Energy Engineering

First Advisor

Jason Marden

Second Advisor

Berouz Touri

Third Advisor

Eric Keller

Abstract

A distributed engineering system requires the design of agent objective functions to control the system sub-entities. These agent functions can cause non-optimal and inefficient system dynamics to occur, and the quality of the design can influence the degree of these undesirable responses. The creation of these agent objective functions is investigated, and an existing method of bounding the induced inefficiencies is expanded to encompass a wider range of games by considering a budget condition. Following, a novel method of designing player objective functions via a convex combination of existing functions is introduced and fully characterized. The significance of these contributions are demonstrated in two theoretic models, and the shortcomings of the expansion are discussed. A final contribution demonstrates that one agent objective function is optimal for some models.

Share

COinS