Bloom Filter for Double-Counting Avoidance in Radio Frequency Ray Tracing

Abstract

Brute-force radio-frequency ray tracing scales well on multicore computing architectures as each ray can be traced independently. The discrete nature of rays with no thickness shows a weakness in the aggregation step where nearby rays to the observation point need to be detected and differentiated. The fact that angular defect cannot be distributed evenly for more than 12 rays in space leads to a double-counting (DC) phenomenon, i.e., the radius of a reception sphere cannot exclude all but 1 ray per wavefront. This either leads to significant signal errors or requires the use of space- and time-consuming wavefront differentiation. Bloom filters configured with marginal false-positive rate are proposed here as a replacement of the exact wavefront differentiation, effectively eliminating DC errors. Substantial space gains while computing channel impulse responses are reported. Furthermore, due to the importance of ray-launching template grids for DC avoidance, the analytical angular bounds for the two common icosahedral grids are presented. The frequently referenced approximate DC probability of 20.9% is shown to be highly underestimated when affected by the refraction- and diffraction-induced changes in ray spacing.