Spatial Non-Stationary Near-Field Channel Modeling and Validation for Massive MIMO Systems

Massive MIMO is envisioned as a promising technology in 5G and beyond 5G communication. Channel models are of great importance for the development and performance assessment of massive MIMO systems. Since massive MIMO systems are equipped with large-aperture antenna arrays, antenna elements at different spatial positions would observe different channel multipath characteristics, which is so-called spatial nonstationarity (SnS). The SnS property of multipaths has been observed in many reported massive MIMO channel measurements. However, characterization and explanations of SnS have not been adequate in existing statistical channel modeling, and deterministic models (e.g., ray tracing) are difficult to implement due to the high complexity. This article proposes a realistic yet low-complexity SnS channel modeling framework for massive MIMO systems and its validation based on both channel measurements and ray-tracing simulations. In this work, we first perform a 6 GHz-bandwidth millimeter-wave (mmWave) indoor channel measurement campaign with a 0.5 m radius virtual uniform circular array (UCA), where the SnS phenomena are clearly observed. Then, we propose the massive MIMO channel modeling framework that captures the observed SnS property from the physical propagation mechanisms of dominant multipaths in mmWave channels, i.e., blockage, reflection, and diffraction. Compared to traditional stationary channel modeling, only one extra parameter accounting for SnS has been added in the proposed framework, which is desirable for its low-complexity implementation. Finally, the proposed framework is validated with site-specific ray-tracing simulations. The SnS phenomena observed in the measurements are reproduced well in the modeling results according to the proposed framework, and high similarities between the target channels and modeling results are achieved. The proposed framework is valuable for the development of massive MIMO systems since it is realistic, low complexity, and accurate.