This thesis presents a model to predict signal strength for frequencies between 902 MHz and 28 GHz. The model approximates diffraction using the knife-edge concept and equations proposed by Lee (1985). LOS pathways are calculated using the Bresenham algorithm and the corresponding elevations are obtained from a 30m DEM base map. The base map was generated by the procedure outlined in Rose (2001) and includes building elevations. The effect of Fresnel zones on prediction accuracy is considered. The effect of interpolating elevations along the Bresenham line is also considered. An Inverse Distance Weighting algorithm was used to interpolate the elevations.

The accuracy of the model was evaluated using received signal strength data compiled from studies conducted at 902 MHz, 24.12 GHz and 27.525 GHz. In addition to the compiled data, data was also collected for this study at 2.4 GHz. 257 receiver locations were evaluated; 70 samples were Line-of-Sight. The study area incorporates the Virginia Polytechnic Institute and State University campus.

Incorporating Fresnel zones, Interpolating elevations and calculating double blockages do not have an effect on the program's overall ability to predict signal strength. However, for obstructed pathways, it is not adequate to simply use path loss as an estimate of signal strength. Accurate estimates of diffraction gain are crucial for obstructed pathways. In addition, examination of the standard deviation for the data sets indicates that the model is independent of frequency. The average error across the frequencies is positively correlated with frequency, indicating that the model predicts signal strength better for higher frequencies. The smaller wavelengths associated with the higher frequencies require a more directional antenna and are therefore less sensitive to multipath interference. In addition, the smaller wavelengths are less able to diffract around buildings and terrain features.