All ETDs from UAB

Advisory Committee Chair

Thomas Jannett

Advisory Committee Members

Dale Callahan

B Earl Wells

Gary Grimes

Roy Koomulllil

Document Type

Dissertation

Date of Award

2007

Degree Name by School

Doctor of Philosophy (PhD) School of Engineering

Abstract

Management of power and other resources that effect field life is an important consideration in sensor networks. Sensing, data fusion, target tracking, and network resources are some of the factors that have to be effectively managed in order to improve the field life of an underwater distributed sensor network application. A hierarchical sensor network is considered in which the sensors report the range and bearing of the target to the cluster nodes. The cluster nodes then fuse data from the sensor nodes to generate a local estimate of the target position. The master node fuses data from the different cluster nodes to generate a global estimate of the target position. Intelligent agents present on the cluster and master nodes help to manage network resources. This dissertation presents and compares alternative agent paradigms applicable for resource management in the sensor network. The comparisons provide an increased understanding of how the performance of the sensor network changes with the number, functionality, and presence of agents at different levels in the network hierarchy. A modular simulation framework based on object-oriented design was used in performance evaluation and to generate data for a detailed analysis of component interactions. The number of computations, number of communications, tracking performance, and intelligence quotient for different agent paradigms were compared. The simulation framework developed in this dissertation demonstrated the scalability and flexibility of an agent-based system where task-specific agents can be iii added or removed to easily simulate different scenarios. An agent-based scenario in which multiple master nodes were present in the field and another scenario in which master and cluster agents were stacked on the same physical hardware achieved high field-life performance with acceptable tracking errors. Simulation results verify the effectiveness of using agents for resource management in sensor networks. The results comparing the different agent paradigms presented in this dissertation will assist designers of agent-based systems in using agents to their best advantage in scenarios similar to those explored here.

Included in

Engineering Commons

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.