Introduction
Unmanned Aerial Vehicles (UAVs) are transforming logistics, but they require high-speed, reliable communication links for real-time control and safety. As we move toward higher frequencies (like mmWave) to meet data demands, free-space pathloss becomes a massive hurdle. Traditional fixed-antenna systems struggle because the distance to a moving UAV is constantly changing. While technologies like Reconfigurable Intelligent Surfaces (RIS) or Movable Antennas (MA) exist, they often suffer from "double-fading" effects or very limited ranges of motion.
This paper introduces a novel framework for UAV delivery powered by Pinching-Antenna Systems (PASS). This technology uses a dielectric waveguide—a low-loss signal carrier—and contactless "pinching antennas" (PAs) that can be activated at specific points along a meter-scale track to maintain a strong Line-of-Sight (LoS) link with the UAV, effectively turning "last-mile" wireless challenges into a "last-meter" advantage.
System Model and PASS Mechanism
The PASS framework consists of a ground station connected to a waveguide with multiple PAs distributed along its length. As the UAV follows its delivery trajectory, the system dynamically selects which PAs to activate to minimize the distance to the receiver.
The core of PASS lies in the coupling between the waveguide and the PAs. The signal power is transferred from the waveguide to the $k$-th PA based on a coupling coefficient $\kappa_k$, which is a function of the physical separation $S_{wg,k}$. This proportional power model implies that activating one antenna affects the power available to others further down the waveguide, necessitating sophisticated optimization.
Optimization Framework
The authors developed a Double-Layer Optimization (DLO) algorithm to minimize the total communication energy used during a delivery cycle:
- Outer Layer (Trajectory Planning): This treats the delivery sequence as a Traveling Salesman Problem (TSP). They propose a Hierarchical Alternating Optimization (HAO) scheme where a Genetic Algorithm (GA) performs a broad global search, while Dynamic Programming (DP) performs fine-grained local refinement on subpaths.
- Inner Layer (Antenna Activation): This addresses the Mixed-Integer Nonlinear Programming (MINLP) problem of deciding which PAs to activate at each time slot. The authors offer a Branch-and-Bound (BnB) approach for global optimality and an Incremental Search and Local Refinement (ISLR) heuristic for lower-complexity, real-time applications.
Key Results and Insights
Numerical simulations at a 15 GHz carrier frequency with a 100m waveguide demonstrate several key advantages of the PASS framework:
- Superior Energy Efficiency: PASS significantly outperforms conventional MIMO systems, especially as the required communication rate ($R_{th}$) increases. By carrying the signal via waveguide to a point physically close to the UAV, the final wireless jump is reduced to just a few meters.
- The Importance of Selective Activation: Interestingly, activating all antennas performed the worst. This "Full Activation" strategy wastes energy on antennas far from the UAV and can cause destructive interference, highlighting the necessity of the proposed selective activation algorithms.
- Path Optimization: The HAO-based delivery sequence effectively reduced total flight distance and time compared to standard GA benchmarks, preventing the inefficient "zig-zag" paths often generated by simple stochastic searches.
Limitations and Future Work
While promising, the current model assumes a constant flight speed and perfect Doppler compensation. Future research will explore 2D waveguide grids or non-linear shapes to cover more complex environments and investigate multi-UAV coordination to manage interference and resource scheduling within a shared PASS infrastructure.
Summary updated