Efficient Sum-of-Sinusoids-Based Spatial Consistency for the 3GPP New-Radio Channel Model

Spatial consistency was proposed in the 3GPP TR 38.901 channel model to ensure that closely spaced mobile terminals have similar channels. Future extensions of this model might incorporate mobility at both ends of the link. This requires that all stochastic processes in the model must be correlated in 3 (single-mobility) and up to 6 spatial dimensions (dual-mobility). Existing filtering methods cannot be used due to the large requirements of memory and computing time. The sum-of-sinusoids model promises to be an efficient solution. To use it in the 3GPP channel model, we extended the existing model to a higher number of spatial dimensions and propose a new method to calculate the sinusoid coefficients in order to control the shape of the autocorrelation function. The proposed method shows good results for 2, 3, and 6 dimensions and achieves a four times better approximation accuracy compared to the existing model. This provides a very efficient implementation of the 3GPP proposal and enables the simulation of many propagation scenarios that were thought to be impossible to realize with geometry-based stochastic channel models.

I. Introduction

Channel models are an essential requirement for the development of new wireless communication protocols and systems. Standardized channel models can be used to evaluate and compare the different proposals against each other. For the upcoming fifth generation (5G) of wireless cellular communications, a new channel model was proposed by the 3rd generation partnership project (3GPP) [1]. An open-source implementation of this model is provided by the quasi deterministic radio channel generator (QuaDRiGa) [2], [3]. An essential new feature that was introduced by 3GPP is spatial consistency which solves one major drawback of previous geometry-based stochastic channel models (GSCMs), namely the lack of realistic correlation in the small-scale-fading (SSF). Without spatial consistency, the positions of individual scattering clusters are generated randomly for each mobile terminal (MT) position. This contradicts the causality and the observations made in channel measurements [4].