Date of Award

Summer 8-19-2022

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

Advisor

Ali Abedi

Second Committee Member

Walter Rawle

Third Committee Member

Rick Eason

Additional Committee Members

Rosemary L. Smith

Vincent Caccese

Abstract

This thesis utilizes wireless sensor network systems to learn of changes in wireless network performance and environment, establishing power efficient systems that are low cost and are able to perform large scale monitoring. The proposed system was built at the University of Maine’s Wireless Sensor Networks (WiSe-Net) laboratory in collaboration with University of New Hampshire and University of Vermont researchers. The system was configured to perform soil moisture measurement with provision to include other sensor types at later stages in collaboration with Alabama A & M University. In the research associated with this thesis, a general relay energy assisted scenario is considered, where a transmitter is powered by an energy source through both direct and relay links. An energy efficient scheduling method is proposed for the system model to determine whether to transmit data or stay silent based on the stored energy level and channel state. An analytical expression has been derived to approximate outage probability of the system in terms of energy and data thresholds. In addition, we propose a model for evaluating the outage probability of a solar powered base station, equipped with a selected photo voltaic panel size and battery configuration. The energy harvesting environment location has been selected as the state of Maine, during a variety of weather conditions, considering base station loading during different days of the week. Simulation results shows the required photo-voltaic panel size and number of batteries for specific tolerable outage probability of the system. The fundamental contribution of this work is in development of hardware and software based on new methodologies to optimize network longevity using AI/ML. One of the most important metrics to define longevity and reliability is the outage probability of a network. We have derived equations for the outage probability, based upon power configuration panel size, battery capacity and the environmental factors, meteorological and diurnal. This will impact the observed cost function which is outage probability. The system models proposed in this thesis result in much more energy efficient systems with less outage probabilities compared to the current systems.

Download

Files over 10MB may be slow to open. For best results, right-click and select "save as..."

Share