Indoor propagation estimation combining statistical models with ray-tracing

In order to improve estimation accuracy of the statistical model, we present a novel method. In this method, we classify NLoS region using number of reflections and diffractions calculated by ray-tracing, and then applies different statistical models to each regions. First, we make statistical models. We classify measurements of NLoS region using the conditions up to 3 times reflections, 1 time diffraction, and others, and develop statistical models of each region. Then we compare fitting errors of the developed statistical models. The fitting error decreased from 4.18 dB to approximately 3 dB after some classification, but the fitting error did not decrease any more. Next, we evaluate the calculation errors using the developed statistical models. The calculation error became minimal of 3.77 dB when we classify NLoS into up to 2 times reflection and others, and then tended to increase when we classify further. From the above results, it is optimal that the NLoS region is classified into 3 regions.