Introduction
As wireless networks evolve toward 6G, achieving ubiquitous coverage remains a primary challenge. Reconfigurable Intelligent Surfaces (RIS) have emerged as a promising solution, acting as "intelligent mirrors" to redirect signals into dead zones. A more recent advancement, Simultaneously Transmitting and Reflecting RIS (STAR-RIS), extends this capability by allowing signals to both pass through and bounce off the surface simultaneously, potentially offering 360-degree coverage.
Mounting these surfaces on Unmanned Aerial Vehicles (UAVs) provides significant deployment flexibility but introduces complex 3D geometric considerations. This study provides a rigorous head-to-head performance comparison between aerial RIS and STAR-RIS, accounting for realistic factors such as deployment altitude, surface orientation, and directional antenna patterns.
System Architecture and Methodology
The researchers developed a comprehensive 3D system model featuring a multi-antenna base station (BS), a drone-mounted surface (either RIS or STAR-RIS), and multiple ground users. Unlike simplified models, this framework incorporates directional radiation patterns and Rician fading to accurately reflect how the physical orientation of the drone impacts signal gain.
To ensure a fair comparison, the study formulated a sum-rate maximization problem, optimizing both the base station's beamforming and the surface's phase shifts and amplitudes. Due to the non-convex nature of the optimization problem, a sophisticated iterative algorithm was employed, combining Weighted Minimum Mean Square Error (WMMSE), Block Coordinate Descent (BCD), and the Penalty Dual Decomposition (PDD) method. Simulations were conducted in a $100 \times 100$ meter square environment with varying drone altitudes ($H$) and horizontal positions.
Performance Analysis and Spatial Insights
The study reveals a critical performance trade-off governed by altitude and spatial geometry. At low altitudes (e.g., $H=10$m), STAR-RIS is the clear superior. Its ability to transmit and reflect allows it to serve users on both sides of the surface, effectively eliminating the "blind spots" that plague conventional horizontally-deployed RIS at low elevations.
However, an "altitude crossover" occurs as the drone flies higher. At an altitude of $H=40$m, conventional RIS begins to outperform STAR-RIS in regions close to the base station. This is because the horizontal orientation of the standard RIS aligns more effectively with signals arriving from below at steep angles. Furthermore, while STAR-RIS offers broader coverage, it must split incident signal energy between transmission and reflection. In specific high-altitude geometries, the pure reflection of a standard RIS proves more efficient than the split-energy approach of STAR-RIS.
Another key finding is the extreme sensitivity of STAR-RIS to its orientation angle ($\eta$). In certain scenarios, adjusting the orientation from $0$ to $\pi/4$ can nearly double the achievable sum-rate, highlighting the necessity for precise UAV control in practical deployments.
Conclusion and Future Directions
While STAR-RIS provides superior 360-degree coverage at low altitudes, conventional RIS can be more effective at higher altitudes when positioned near the base station. This suggests that the choice between these technologies should be driven by the specific deployment scenario and required service altitude.
Future research will look toward joint trajectory optimization for mobile users, accounting for hardware impairments like phase-shift errors, and exploring the coordination of multiple aerial surfaces to provide city-scale coverage.
Summary updated