Hyon H. Kwak

Date of Award

3-10-2010

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Electrical and Computer Engineering

First Advisor

Brett Borghetti, PhD

Abstract

Although wireless communication provides connectivity where hardwired links are difficult or impractical, it is still hindered by the environmental conditions where the communicators reside. Signal loss over large distances or because of intervening obstacles can be mitigated by increasing the user's transmission power or adding repeater nodes between the users. Unfortunately, increasing the signal strength strains limited power resources and increases the likelihood of eavesdropping. Stationary repeaters are impractical for highly mobile users in dangerous environments. While mobile relay nodes might be a preferred solution, a centralized control scheme saps bandwidth from important traffic and introduces a single point of failure at the control station. An alternative solution is to create a Mobile Agent Relay Network (MARN). Each autonomous node in the MARN decides where to move to maintain the network connectivity using only locally-available information from onboard sensors and communication with in-range neighbor nodes. This is achieved by borrowing concepts from flocking behaviors that motivates our agents to maintain equal distance between its neighboring nodes. In addition, each agent maintains a filtered list of previously visited locations that provided best connection. This thesis takes the first steps toward realizing a MARN by providing mobile relay agents. Each model-based reflex agent is guided by a modified flocking behavior which considers only trustworthy neighbors and uses a Bayesian model to aggregate observations and shared reputation. The relay agents are able to build a network and maintain connectivity for their users. In this work, MARN agent algorithms are evaluated in a simulated unobstructed environment with stationary users. The system behavior is explored under both benign conditions and with varying numbers of misbehaving nodes.

AFIT Designator

AFIT-GCS-ENG-10-04

DTIC Accession Number

ADA518465

Download

Share

COinS