Polarization Sensitive Direction of Arrival Distribution for Radiowave Multipaths: A Comparison of Multiple Geometric Models

In the literature, many geometric models have been derived to characterize the distribution of the direction-of-arrival (DOA) of the multipaths of a signal upon a wireless receiver. Polarization is an essential part of wireless propagation but is often unclear to which particular analytical models work best. This communication performs a comprehensive study of the eleven existing polarization models in propagated signals, using sophisticated mathematical techniques to fit the models with two datasets. Model parameters with the lowest mean squared error are found using both a grid search method and using MATLAB's fmincon constrained optimization algorithm, with the latter yielding parameters with lower mean squared errors for all models with both data sets. The estimates found for the Gaussian model using the constrained optimization algorithm are consistent with previous work, suggesting this method is robust. For both datasets the Gaussian model is outperformed by models with elliptical supports.