UVM Theses and Dissertations
Format:
Print
Author:
Kay, James Michael
Dept./Program:
Electrical Engineering
Year:
2007
Degree:
PhD
Abstract:
Low cost single chip radios enable exciting new applications utilizing large numbers of wireless sensing nodes, including structural health monitoring (e.g., bridges, buildings, aircraft), environmental monitoring and distributed control systems. Many of these applications require reliable operation in harsh or inaccessible environments. One technique to deal with this constraint is to use more than the minimum number of sensors needed to meet the required temporal or spatial resolution allowing the system to operate even as sensors fail throughout the deployment life.
This over-deployment technique also provides flexibility for those systems that require variable spatial or temporal resolution (e.g., a target tracking system that uses one resolution for target acquisition and another for tracking). However, when a system is over-deployed it is important to control the activity of sensors in such a way that energy is not wasted due to acquisition and transmission of redundant information, or due to collisions in the communication channel. In addition to being robust with respect to sensor node loss, these systems will benefit from control techniques that allow the addition of nodes during the system lifetime without explicit reprogramming.
This work develops and analyzes such a technique for controlling the number of active sensors in a wireless sensor network. The technique uses simple finite state automata which are not required to communicate with the other sensing nodes, but only with a central clusterhead. This technique is shown to allow control around the set point with very little variance, robust behavior with respect to sensor loss and addition, and low collision rates at all channel loads.
This over-deployment technique also provides flexibility for those systems that require variable spatial or temporal resolution (e.g., a target tracking system that uses one resolution for target acquisition and another for tracking). However, when a system is over-deployed it is important to control the activity of sensors in such a way that energy is not wasted due to acquisition and transmission of redundant information, or due to collisions in the communication channel. In addition to being robust with respect to sensor node loss, these systems will benefit from control techniques that allow the addition of nodes during the system lifetime without explicit reprogramming.
This work develops and analyzes such a technique for controlling the number of active sensors in a wireless sensor network. The technique uses simple finite state automata which are not required to communicate with the other sensing nodes, but only with a central clusterhead. This technique is shown to allow control around the set point with very little variance, robust behavior with respect to sensor loss and addition, and low collision rates at all channel loads.