On the Use of Adaptive-Density Point Cloud for Site-Specific Ray-Optics Simulations

Multipath channel simulations at millimeter-wave frequencies require more accurate knowledge of the geometry of physical environments than at below-6 GHz frequencies because of the shorter wavelength. The accurate geometry can be obtained as a point cloud through LIDAR (light detection and ranging) scanning of environments. While the use of point clouds for ray-optics simulations of multipath channels has become a popular approach recently, optimized point cloud parameters - for example its density - that allow accurate reproduction of multipath channels are not yet known. Motivated by a conjecture that a same-sized physical object contributes differently to multipath channels when they are near vs far from communicating antennas, we propose to use a point cloud that is denser for environments near antennas and is sparser when further from antennas. Compared to using a uniform-density point cloud with 10 cm resolution on average, the use of such an adaptive-density point cloud in ray tracing simulations shows better reproduction of measured multipath channels, e.g., delay and angular spreads, indicating correctness of our conjecture.