TLDR: This research introduces a novel framework, HMCD, to improve low-altitude wireless networks (LAWNs) in urban areas. It combines collaborative beamforming from UAV swarms with simultaneous transmitting and reflecting reconfigurable intelligent surfaces (STAR-RIS) to overcome signal attenuation from obstacles. The framework optimizes transmission rate and UAV energy consumption using a simulated annealing-based method for STAR-RIS and an enhanced multi-agent deep reinforcement learning for UAVs. Simulations show HMCD outperforms baselines in speed, rate, and energy efficiency, with performance improving as more UAVs and STAR-RIS elements are added.
In our increasingly connected world, low-altitude wireless networks (LAWNs) powered by uncrewed aerial vehicles (UAVs) are becoming crucial for urban communications. These networks offer incredible mobility, flexibility, and broad coverage, making them ideal for extending terrestrial network capabilities, especially in remote areas or during emergencies. However, dense urban environments pose a significant challenge: buildings and other obstructions severely attenuate signals, leading to poor communication quality and interruptions.
To tackle this critical issue, researchers have introduced an innovative approach that combines the strengths of UAVs with a new technology called Simultaneous Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RIS). The core idea is to enhance signal quality and directionality by leveraging collaborative beamforming (CB) from a swarm of UAVs and omnidirectional reconfigurable beamforming (ORB) from STAR-RIS.
The Power of Collaboration: UAV Swarms and STAR-RIS
UAVs, acting as flying base stations, can form a “virtual antenna array” (UVAA). Through collaborative beamforming, these UAVs can jointly transmit signals, steering high-gain beams towards legitimate users while minimizing interference elsewhere. This significantly boosts spectral efficiency and energy utilization without needing hardware changes on the UAVs themselves.
However, even with collaborative beamforming, direct signal paths can still be completely blocked by large urban structures. This is where STAR-RIS comes in. Unlike traditional reflective surfaces that only cover a half-space, STAR-RIS can simultaneously transmit and reflect signals, providing a versatile 360-degree full-space solution. When deployed on building facades, STAR-RIS can redirect signals around obstacles, enabling omnidirectional reconfigurable beamforming (ORB) and greatly enhancing deployment flexibility. The joint use of UVAA and STAR-RIS creates a powerful hybrid active-passive beamforming framework for energy-efficient LAWNs.
Navigating Complexity: The Joint Optimization Challenge
While promising, realizing this combined system faces several hurdles. The environment is highly dynamic, with user positions and channel conditions constantly changing. There’s also an inherent conflict between maximizing the system’s transmission rate and minimizing the energy consumption of the UAV swarm. Furthermore, the diverse devices involved create a high-dimensional, non-convex optimization problem that traditional algorithms struggle with.
To address these challenges, a new framework called Heterogeneous Multi-Agent Collaborative Dynamic (HMCD) optimization has been proposed. This framework aims to maximize the overall system’s transmission rate while minimizing the UAV swarm’s energy consumption, by intelligently optimizing UAV trajectories, excitation current weights, and the STAR-RIS’s reflection and transmission properties.
The HMCD Framework: A Dual-Component Solution
The HMCD framework has two main components:
- STAR-RIS Control: An adaptive temperature-based STAR-RIS optimization (ATSO) strategy, which uses a simulated annealing (SA)-based method to dynamically optimize the reflection and transmission coefficients of the STAR-RIS, thereby enhancing signal propagation.
- UAV Swarm Coordination: An improved multi-agent deep reinforcement learning (MADRL) method for the UVAA. This method incorporates two key enhancements: a self-attention evaluation mechanism to better capture interactions between UAVs, and an adaptive velocity transition mechanism to improve training stability and guide UAVs towards energy-optimal flight patterns.
This integrated approach allows the STAR-RIS to adapt to UAV movements, while UAV strategies adjust to the electromagnetic environment created by the STAR-RIS, forming a continuous, adaptive control loop.
Also Read:
- Securing Low-Altitude Wireless Networks with Collaborative UAV Swarms and Intelligent Reflecting Surfaces
- Optimizing Mobile Edge Computing with Pinching Antenna Systems
Promising Results and Future Outlook
Simulation results demonstrate that the HMCD framework significantly outperforms various baseline methods. It achieves faster convergence speeds, higher average transmission rates, and lower energy consumption. Further analysis reveals that the average transmission rate of the overall system scales positively with both the number of UAVs and the number of STAR-RIS elements. Visualizations of UAV flight trajectories confirm that the HMCD effectively guides UAVs to approach the STAR-RIS while maintaining safe distances from boundaries, ensuring efficient and safe communication path planning in complex urban settings.
This research marks a significant step towards more robust and efficient low-altitude wireless networks. Future work will explore enhancing anti-eavesdropping capabilities, deploying multiple STAR-RIS units, and integrating advanced generative AI models for even more sophisticated resource optimization in next-generation aerial communication networks. You can read the full research paper here.
