Date of Award

Summer 2007

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering

First Advisor

Yaz, Edwin E.

Second Advisor

Josse, Fabien J.

Third Advisor

Johnson, Michael T.

Abstract

A sensor network is usually deployed to collaboratively obtain and process information from the physical environment using a large number of sensors. Advances in wireless communications and electronics have brought significant development of low power, multifunctional sensors. These sensors are designed to be small and energy efficient and be able to collect, transmit and communicate information with each other. Due to usually the wireless nature of this kind of network and possible high traffic density issues even in wired networks, the communication within the network is sometimes not reliable, and therefore, the estimation and control in sensor network must be dynamically adaptive to this stochastic property. In this work, some linear, estimation and control issues are addressed for sensor networks with probability of failure for sensors/actuators or probability of data loss. Previous results on all sensors/actuators have the same failure rate are extended to multiple rates. New estimation algorithms under two different sets of assumptions for designing linear minimum variance estimators are developed to address the data loss and sensor failure issues arising from unreliable communication links between sensor nodes. Then the linear quadratic optimal control problem is studied for multiple actuators which may fail randomly with a certain failure rate and a method is proposed to determine the mean square stabilizability of a system that is monitored and controlled by a sensor network.

Share

COinS

Restricted Access Item

Having trouble?