Abstract
Due to the high-mobility of unmanned aerial vehicles (UAVs), actual UAV channel measurements show that low-altitude air-to-ground (A2G) channels in UAV communications illustrate non-stationary properties. This fact motivates us to develop a non-stationary channel model for UAV-to-ground links. In this paper, we propose a three-dimensional (3D) wideband non-stationary A2G vehicle-to-vehicle (V2V) geometry-based stochastic channel model (GBSM). In the proposed model, both UAVs and ground terminals can be moving, which makes the model more general. In order to mimic the non-stationary channel characteristics, parameters like the number of clusters, power, time delays, angels of departure (AoDs), and angles of arrival (AoAs) are all time-variant. The proposed model combining a line-of-sight (LoS) component, a ground reflection component, a cylinder model, and multiple confocal truncated ellipsoid models has the ability to investigate the impact of UAV heights and transceivers' movements on channel characteristics in diverse environments. Statistical properties like temporal autocorrelation function (ACF), spatial cross-correlation function (CCF), Doppler power spectral density (PSD), and stationary interval of A2G channels are derived and analyzed in detail. In addition, derived root mean square delay spread (RMS-DS), temporal ACF and spatial CCF are validated against channel measurement results. Furthermore, by adjusting channel parameters, the proposed GBSM is sufficiently generic and adaptable to model various UAV-to-ground communication scenarios.