Estimation of simplified reflection coefficients for improved modeling of urban multipath

An algorithm to estimate reflection coefficient properties of materials based on code multipath errors is developed and tested in two urban environments. The algorithm extends an existing model that integrates 3D structure models, electromagnetic ray-tracing algorithms, antenna pattern measurements, and receiver tracking loop models to simulate code multipath errors. Instead of using the Fresnel equations to compute reflection coefficients, we utilize a simplified, fixed reflection coefficient model with no incident/reflection angle dependence. In previous model validation work these coefficients were adjusted manually to fit to experimental data. Herein, a minimum least-squares (LS) filter is used to improve upon initial reflection coefficient estimates by comparing experimental and simulated multipath errors. The modeled environments are the University of Colorado (CU) Engineering Center rooftop and the campus of the University of New South Wales (UNSW). For both, simulations that mimic open-air experiments are run to provide information on multipath ray paths, geometric delays, interaction types, and the receiving antenna response. The LS filter then adjusts the reflection and phase coefficients for materials that produce reflections in the model. Short segments of experimental data with distinct multipath signatures are used to update reflection coefficients for the concrete walls of the CU rooftop and concrete sidewalks at UNSW, with estimated values of 0.42 and 0.48, respectively. In addition, a hill covered with foliage is estimated to have a reflection coefficient of 0.1.