TL;DR
This paper introduces a novel geometry-based stochastic model (GBSM) for Unmanned Aerial Vehicle (UAV) to vehicle (U2V) millimeter-wave (mmWave) communication channels. It accounts for the 3D trajectories of UAVs and vehicles, 3D antenna arrays, and crucially, 3D beamforming, which is essential for overcoming the high path loss in mmWave frequencies. The model offers a good balance of accuracy and complexity by considering the Line-of-Sight (LoS) path, a ground specular path, and the two strongest single-bounce paths.
What Problem Does This Paper Address?
The increasing use of UAVs as flying relays in 5G and beyond communication systems, coupled with the potential of mmWave frequencies for high data rates, presents a unique set of challenges for channel modeling. Traditional channel models, often designed for terrestrial or sub-mmWave frequencies, are insufficient for UAV mmWave communications due to several factors:
- 3D Trajectories: UAVs move in three dimensions, unlike ground vehicles.
- 3D Scattering Environment: The environment around a UAV is inherently 3D.
- mmWave Propagation: mmWave frequencies suffer from high path loss, necessitating the use of directional antennas and beamforming.
- 3D Beamforming: The directional nature of beamforming significantly impacts signal propagation and the effective distribution of scatterers, which traditional models do not adequately capture.
Existing UAV channel models often focus on sub-mmWave bands, are 2D, or do not fully incorporate the complexities of 3D beamforming. This paper aims to bridge this gap by proposing a comprehensive 3D geometry-based beamforming channel model specifically for U2V mmWave communications.
Methods
The authors propose a novel geometry-based stochastic model (GBSM) that incorporates the 3D nature of UAV trajectories, antenna arrays, and beamforming. The model is simplified by considering only the most significant paths:
- Line-of-Sight (LoS) Path: Assumed to always exist.
- Ground Specular (GS) Path: A reflection off the ground.
- Two Strongest Single-Bounce (SB) Paths: The two most dominant single reflections from scatterers.
A key innovation is the hybrid computation method for channel parameters. This method combines:
- Geometry-Based Parameters: Deterministic calculations based on the geometric relationships between the UAV, vehicle, and scatterers. This includes time-variant locations, velocities, and derived angles and delays.
- Data-Based Stochastic Parameters: Random parameters derived from statistical distributions fitted to ray tracing (RT) simulation data. These include angle offsets and delay offsets of rays, and path powers. This approach aims to capture the random scattering effects that are difficult to model purely geometrically.
The model also accounts for the 3D beamforming gain, which varies with the angular deviation of a path from the beam center.
Datasets
The paper utilizes data generated from ray tracing (RT) simulations performed at 28 GHz across four scenarios: urban, forest, hill, and sea. This RT data is used to:
- Analyze the power distribution of different paths (LoS, GS, SB) to justify the selection of the four dominant paths.
- Fit statistical distributions for data-based parameters such as ray delay offsets, azimuth angle offsets, and elevation angle offsets.
The authors also compare their simulated results with measured data from a sub-mmWave channel measurement in [29] to validate the generality and correctness of their proposed model.
Key Insights and Ideas
- Dominance of Few Paths: Analysis of RT data reveals that the LoS path and the three strongest NLoS paths (GS and two SB) account for over 99% of the total path power, justifying the model's simplification.
- Impact of 3D Beamforming: The paper highlights that 3D beamforming is crucial for mmWave U2V links and its gain significantly influences the channel, varying with angular deviation from the beam center.
- Hybrid Parameter Computation: Combining deterministic geometry with data-driven statistics provides a robust approach to modeling complex U2V mmWave channels, balancing accuracy and computational complexity.
- Time-Variant Nature: The model explicitly captures the time-varying nature of channel parameters (delays, powers, angles) due to the dynamic movement of the UAV and vehicle, and the changing beam direction.
- Statistical Properties: The paper demonstrates that the model accurately reproduces key statistical properties like power delay profiles (PDPs), power angle profiles (PAPs), autocorrelation functions (ACFs), and Doppler power spectral densities (DPSDs), matching both theoretical derivations and experimental measurements.
Key Results
- The proposed channel model, considering LoS, GS, and two SB paths, accurately represents the dominant components of U2V mmWave channels.
- Simulations show that the time-variant PDPs and PAPs are consistent with the dynamic nature of the U2V scenario and the beamforming characteristics.
- The simulated ACFs closely match theoretical results at different time instants, validating the model's temporal dynamics.
- The simulated DPSDs illustrate the complex Doppler frequency changes arising from 3D trajectories and beamforming.
- A comparison with measured sub-mmWave channel data shows good agreement, suggesting the model's generality.
Limitations and Future Work
- Beam Tracking Error Ignored: The current model assumes perfect beam tracking, and the angular error of beam tracking is ignored in the simulations.
- Sub-mmWave Measurement Comparison: While the model is for mmWave, validation with measured sub-mmWave data was performed by adjusting parameters. Direct mmWave measurements would provide stronger validation.
The authors suggest that future work will involve performing more extensive channel measurements to further verify the model and optimizing the calculation methods for channel parameters.
Summary updated