Performance evaluation of a ray tracing based microcellular coverage prediction tool

Based on measurements in five urban environments in three different cities, the accuracy of a ray tracing based model is characterized. First, the conventional characterization using the mean and standard deviation of the prediction error is used. Then, a figure of merit defined as the percentage of correct prediction with respect to a coverage threshold is computed. Finally, an example is given to illustrate how prediction results could be displayed to provide some insights into the accuracy of the model. For comparison purpose, results from power law curve fits are also analyzed. It is concluded that the standard deviation is a fair characterization of the accuracy of a model as long as its value is small (<-7 dB). For larger value, the figure of merit can be more useful. However, further work is needed to find reliable characteristics which could provide almost as much information as the visual inspection of the prediction results overlaid on the measurements