Date of Award

Fall 2015

Project Type


Program or Major

Electrical and Computer Engineering

Degree Name

Master of Science

First Advisor

Nicholas J Kirsch

Second Advisor

Michael J Carter

Third Advisor

Richard A Messner


Broadband internet has grown to become a major part of our daily routines. With this growth increase, those without direct access will not be afforded the same opportunities that come with it. The need for ubiquitous coverage of broadband Internet is clear to provide everyone these opportunities. Rural environments are an area of concern of falling behind the growth as the low population densities make wired broadband solutions cost prohibitive. Wireless options are often the only option for many of these areas; WiFi, cellular, and WiMAX networks are currently used around the world, but with the opening of the unused broadcast television frequencies, deemed TV White Space (TVWS), a new option is hitting the market. This new technology needs to be assessed before it can be seen as a viable solution.

The contribution of this work is two-fold. First, findings from a real, ongoing trial of commercially available TVWS radios in the area surrounding the University of New Hampshire campus are presented. The trial shows that though the radios can provide Internet access to a distance of at least 12.5 km, certain terrain and foliage characteristics of the path can form coverage holes in that region. The second contribution explores the use of empirical path loss models to predict the path loss, and compares the predictions to actual path loss measurements from the TVWS network setup. The Stanford University Interim (SUI) model and a modified version of the Okumura-Hata model provide the lowest root mean squared error (RMSE) for the setup. Additionally, the deterministic Longley-Rice model was explored with the Radio Mobile prediction software. It was determined that without extensively tuning the foliage component of the algorithm, the model could produce significant prediction errors, resulting in a trade-off between low cost, un-tuned predictions, and prediction accuracy.