Abstract
Unmanned aerial vehicle (UAV) swarms utilizing collaborative beamforming (CB) in low-altitude wireless networks (LAWN) demonstrate significant potential for enhanced communication range, energy efficiency, and signal directivity through the formation of virtual antenna arrays (VAA). However, environmental disturbances, particularly wind fields, significantly degrade CB performance by introducing positional errors that disrupt beam patterns, thereby compromising transmission reliability. This paper investigates the critical challenge of maintaining CB performance in UAV-based VAAs operating in LAWN under wind field disturbances. We propose a comprehensive framework that models the impact of three distinct wind conditions (constant, shear, and turbulent) on UAV array performance, and formulate a long-term real-time optimization problem to maximize directivity while minimizing maximum sidelobe levels through adaptive excitation current weight adjustments. To address the inherent complexity of this problem, we propose a novel proximal policy optimization algorithm with long short-term memory (LSTM) structure and adaptive learning rate (PPO-LA), which effectively captures temporal patterns in wind field disturbances and enables real-time adaptation without requiring extensive prior training for specific wind conditions. Our simulation results demonstrate that the proposed PPO-LA algorithm successfully recovers degraded CB performance across various wind scenarios, and thus significantly outperforming benchmark algorithms.
1. Introduction
Unmanned aerial vehicle (UAV) swarms are increasingly being deployed as Virtual Antenna Arrays (VAAs) to perform Collaborative Beamforming (CB). This technique allows a group of low-cost UAVs to act as a single, powerful directional antenna, significantly extending the communication range and efficiency of low-altitude wireless networks (LAWN). However, the effectiveness of CB is highly sensitive to the precise positioning of each UAV.
In real-world environments, wind field disturbances—ranging from constant breezes to unpredictable turbulence—displace UAVs from their ideal coordinates. Even minor deviations can lead to significant signal degradation and increased interference (sidelobes). Traditional static algorithms are often too slow or too rigid to handle the non-linear, rapidly changing nature of these wind fields.
2. Modeling Wind Field Disturbances
To address these challenges, the research identifies three primary types of wind disturbances that affect UAV stability:
- Constant Wind: Uniform velocity across the flight area.
- Wind Shear: Velocity that varies with altitude, common in low-altitude layers.
- Turbulent Wind: Random, high-frequency fluctuations modeled using the Dryden turbulence model.
The study quantifies that turbulent wind is significantly more damaging to beamforming than constant or shear wind, highlighting why simple linear correction models usually fail in the field.
3. The PPO-LA Framework
The authors propose PPO-LA, a deep reinforcement learning framework designed to rescue collaborative beamforming in UAV swarms buffeted by unpredictable wind. Instead of attempting to physically counteract the wind (which is energy-intensive), the algorithm adjusts the excitation current weights of each UAV's antenna in real-time to compensate for their actual, drifted positions.
PPO-LA introduces two critical innovations:
- LSTM Integration: By incorporating Long Short-Term Memory (LSTM) layers, the agent can process temporal sequences of wind data. This allows the system to "remember" past patterns and anticipate future positional drifts, enabling proactive adjustments.
- Adaptive Learning Rate (LA): Using logic derived from the Adam optimizer, the algorithm dynamically scales its learning steps based on the "curvature" of the problem's loss surface. This ensures stable convergence even in highly "noisy" turbulent environments.
4. Performance Evaluation and Results
The framework was evaluated using high-fidelity physical simulations of 8-UAV and 16-UAV swarms in a $50 \times 50 \times 50 \text{ m}^3$ airspace.
- Performance Recovery: PPO-LA successfully restored beamforming performance across all wind scenarios, significantly outperforming benchmarks like DDPG, DQN, and standard PPO.
- Directivity Gains: In an 8-UAV swarm, the system maintained a stabilized directivity of 8.25 dB. When scaled to 16 UAVs, it achieved 15 dB, proving the method scales effectively with swarm size.
- Interference Suppression: The algorithm kept maximum sidelobe levels between -1.07 dB and -3.5 dB, ensuring the swarm doesn't become a source of electronic noise for other nearby networks.
- Convergence Speed: Thanks to the adaptive learning rate, PPO-LA reached a stable, high-reward state much faster than traditional DRL methods.
5. Conclusion
The PPO-LA framework provides a robust solution for maintaining reliable communication in UAV-enabled networks subject to environmental stressors. By leveraging the predictive power of LSTMs and the stability of adaptive learning rates, the system can maintain high-quality beamforming without the need for excessive mechanical compensation. Future work will explore the integration of phase synchronization recovery and the optimization of energy consumption for real-time weight adjustments.
Summary updated