Optimization of building material properties for accurate indoor ray tracing models

Abstract

A 3D ray-tracing program was divided into a geometric engine and an electromagnetic engine so that the electromagnetic engine could be used in an optimization program. The optimization program computed the 92 dielectric permittivity values of a ray tracing building floor plan to minimize the error between ray tracing and measurement peak powers. The results showed that the optimization is well behaved and can reduce the mismatch on a single measurement to about 1 dB or less in most cases. When several measurements are used, it is seen that the overall error between ray tracing predictions and measurements decreases as more measurements are used in the optimization. They also show that the optimized floor plan produces better ray tracing predictions including at the locations not used in the optimization. The proposed optimization program eliminates the tedious task of the manual tweaking of building material properties and produces an optimized floor plan that can be used by conventional ray tracing programs. It can be also used to discover weaknesses in ray tracing electromagnetic models.