Unmanned aerial vehicles (UAVs) are increasingly deployed in ultra-wideband (UWB) environments to support applications that demand high data rates and reliable connectivity. However, effective communication between UAVs and ground terminals is challenged by six-dimensional (6D) posture-induced signal variation, non-stationary channel behavior, and dynamic resource constraints. Existing models are limited in addressing 6D posture-aware fading, frequency non-stationarity, and spatiotemporal variability in multipath propagation, which are critical in UAV-to-ground (U2G) communication in UWB environments. This study introduces a 6D UAV posture based non-stationary U2G channel model that incorporates UWB signal features, 6D UAV posture dynamics, and intelligent reflecting surface (IRS) beam control. A graph neural network (GNN) is developed to learn spatiotemporal correlations among UAV 6D movement, IRS states, and frequency-selective channel behavior. The model is initialized with ray-tracing (RT) data and supports resource allocation (RA) by adjusting UAV transmit power and IRS reflection coefficients. Performance is evaluated using key statistical metrics, including space–time–frequency correlation functions (STF-CFs), power delay profiles (PDP), root-mean-square delay spread (RMS-DS), and Doppler spectral density (DPSD). Results show that the proposed framework effectively captures 6D posture-aware fading and frequency non-stationarity across wide bandwidths, achieving an 11.6% reduction in path-loss RMSE, 8.2% improvement in RMS-DS fidelity, and 23.4% increase in sum-rate compared with conventional baselines.
Ahmad et al. (Sat,) studied this question.