Measurement-Based Validation of the 3GPP Spatial Consistency Procedures

To support the design of fifth-generation (5G) beamforming and beamtracking algorithms, a spatially consistent channel propagation model is needed, namely a model whose geometrical properties - angle-of-departure, angle-of-arrival, and delay of its multipath components - evolve smoothly while the receiver is in motion. The 3rd Generation Partnership Project (3GPP) has proposed two spatial consistency (SC) procedures - SC-A and SC-B - as part of its popular TR 38.901 model for 0.5 GHz to 100 GHz channels. In this paper, we perform a comparison of the two procedures with measurements to determine whether the procedures accurately fit our measurements. We use over 700 channel measurements collected with our high-resolution 28 GHz and 60 GHz directional channel sounders, across two indoor and three outdoor environments and for pedestrian and vehicular motion. We show that SC-A - a fully deterministic procedure - provides a much better approximation to the measured channel in all five environments than SC-B - a fully stochastic procedure. We also computed the correlation distance for both procedures - the distance over which they are valid - showing that SC-A's average correlation distance across all environments of 13.4 m is more than twice as long as SC-B's of 6.28 m, and that the channel is spatially consistent for significantly longer outdoors than indoors.