Pinching-Antenna Systems

• Pinching-antenna systems are reconfigurable wireless architectures using dielectric radiators along waveguides to dynamically shape electromagnetic fields.
• They enable strong line-of-sight links and optimal beamforming through joint adjustment of antenna positions and digital weighting without extra phase shifters.
• Practical implementations leverage low-cost, scalable designs with advanced algorithms to enhance energy efficiency and spectral performance for next-generation networks.

Pinching-antenna systems are a class of reconfigurable, flexible wireless architectures in which small dielectric radiators—termed pinching antennas (PAs)—are positionally activated along a dielectric waveguide to dynamically tailor the radiated electromagnetic field. By leveraging physical reconfiguration over distances far exceeding those of traditional phased arrays or fluid/movable antennas, pinching-antenna systems facilitate the formation of strong, low-loss line-of-sight (LoS) paths, significantly mitigate large-scale path loss, and offer unique adaptability for advanced wireless, sensing, and integrated communication applications. This paradigm enables new physical-layer design strategies, combining mechanically or electronically reconfigurable aperture placement with low-cost, scalable hardware and novel algorithmic frameworks for optimizing energy efficiency, multicast/unicast rates, and physical-layer security.

1. Physical Architecture and Signal Modeling

The architecture uses dielectric waveguides as almost lossless transmission conduits; PAs—dielectric particles or adjunct structures—can be attached or pinched onto the waveguide at arbitrary positions, creating a spatially reconfigurable array of radiating elements (Ding et al., 3 Dec 2024, Liu et al., 30 Jan 2025). The waveguide carries the input RF signal, which is coupled out into free space through each PA. The coupling mechanism is rigorously modeled as an open-ended directional coupler or, equivalently, via coupled-mode theory:

$$\begin{align*} E(x, y, z) &= A(x) E_{\text{wg}}(x, y, z) + B(x) E_{\text{PA}}(x, y, z), \ \frac{dA(x)}{dx} &= -j\kappa B(x) e^{-j\Delta\beta x}, \quad \frac{dB(x)}{dx} = -j\kappa A(x) e^{j\Delta\beta x} \end{align*}$$

where $A(x)$ and $B(x)$ are the modal amplitudes in the waveguide and pinching antenna, $\kappa$ is the coupling coefficient, and $\Delta\beta$ is the propagation constant difference. For phase-matched coupling ($\Delta\beta = 0$), full radiation is achieved at the coupling length $L = \pi/(2\kappa)$.

The received field at a user location $\psi$ from a PA positioned at $\psi_n$ is modeled as:

$$h_n = \frac{\eta^{1/2} e^{-\alpha \|\psi_0 - \psi_n\|}}{\|\psi - \psi_n\|} \exp\left\{ -j \left[ \frac{2\pi}{\lambda} \|\psi - \psi_n\| + \frac{2\pi}{\lambda_g} \|\psi_0 - \psi_n\| \right] \right\},$$

where $\alpha$ is the in-waveguide attenuation, $\lambda_g$ is the guided wavelength, and $\psi_0$ is the feed point of the waveguide (Xu et al., 30 Jun 2025).

Key system advantages:

• Near-zero additional cost for activating multiple PAs on a single waveguide (sharing a single RF feed).
• Spatial reconfiguration over meter-scale apertures.
• Mechanically or electronically controlled activation for dynamic channel shaping, with no need for phase shifters or digital-to-analog converters at all elements.

2. Beamforming, Array Gain, and Optimization

Pinching-antenna systems support a novel "pinching beamforming" regime, in which both the PA positions and traditional digital weights are optimized jointly, yielding spatial degrees of freedom not accessible in conventional phased arrays (Liu et al., 30 Jan 2025). For a set of $N$ PAs, the total array gain (for equally spaced PAs near the user) can be maximized by balancing near-field focusing and mutual coupling effects. The array gain $a$ (SNR-based) is characterized by:

$$a = \frac{\eta}{N d^2} \left| \sum_{n=1}^N \exp\left( -j \frac{2\pi}{\lambda} \sqrt{d^2 + \Delta_n^2} - j \frac{2\pi \Delta_n}{\lambda_g} \right) \right|^2,$$

with $\Delta_n$ the transverse PA-user offset (Ouyang et al., 10 Jan 2025).

Two key findings:

• There is an optimal number $N^*$ of PAs and optimal spacing $\Delta^*$ that jointly maximize array gain; excessive numbers or proximity induce diminishing per-element power and pronounced mutual coupling, leading to oscillatory performance and eventual gain reduction as $N \to \infty$.
• Power models—"equal power" (tuned coupling for uniform output) and "proportional power" (identical elements)—yield comparable performance, especially at moderate-to-large $N$ (Wang et al., 9 Feb 2025).

Optimization tasks include:

• Joint antenna position and power allocation (Xie et al., 8 Feb 2025), often pursued via alternating minimization, graph neural networks (BGAT), or distributed algorithms suited for highly coupled, non-convex objective landscapes.
• Two-stage approaches: first minimizing large-scale (geometric) path loss (uniform PA spacing about target users), then refining positions for coherent constructive combination (phase alignment to multiples of $2\pi$) (Xu et al., 18 Feb 2025).
• Extensions to multiple waveguides (multi-user MIMO, multicast) utilize alternating optimization (AO) and majorization-minimization (MM) frameworks combined with (second-order cone) convex programming for transmit and pinching beamformer co-design (Shan et al., 31 May 2025).

3. Application Domains and Protocols

Pinching-antenna systems have been applied to a diverse range of wireless scenarios:

• Multi-user communications: By activating close-to-user PAs (whether in OMA, NOMA, or hybrid MIMO architectures), strongly improved LoS links and sum-rates are achievable, with analysis showing rate gains scale quadratically (or at least superlinearly) with deployment aperture and number of users (Ding et al., 3 Dec 2024, Liu et al., 30 Jan 2025). Multiple access in PASS can be realized via waveguide switching, division, or multiplexing (Liu et al., 11 Aug 2025).
• Wideband/OFDMA and ISI mitigation: For mmWave/THz wideband regimes, the multi-tap nature of the channel due to distributed PAs (modeled as FIR filters) introduces strong frequency selectivity and ISI. OFDMA frameworks, coupled with subcarrier assignment and power allocation algorithms, exploit frequency diversity to achieve robust, fair user rates even under severe LoS blockage (Oikonomou et al., 26 May 2025).
• Integrated sensing and wireless power transfer: By optimizing the spatial configuration of PAs/joint transmit waveforms (often via CRB minimization), systems can achieve high-resolution target localization, leveraging large, reconfigurable apertures for both communication and sensing (Wang et al., 21 May 2025).
• Secure and covert communications: Pinching beamforming is tuned to create constructive superposition at legitimate receivers and destructive superposition at eavesdroppers (enforcing secrecy or covertness constraints), with artificial noise (AN) and joint baseband-PA design for resistance to interception (Zhu et al., 18 Apr 2025, Wang et al., 14 Jul 2025, Jiang et al., 14 Apr 2025).
• Operation under blockage and interference: Selective activation and assignment algorithms allow dynamic bypassing of LoS obstructions and even deliberate use of NLoS links for inter-user interference suppression, significantly boosting throughput in cluttered environments (Wang et al., 14 Jul 2025).

4. Practical Implementation, Hardware Considerations, and Algorithmic Strategies

PASS requires only single-RF-chain-driven dielectric waveguides and passive PA elements. Activation may be realized mechanically (moveable clamps, discrete tracks) or via controllable switches for pre-installed arrays, depending on timescale and feedback constraints (Liu et al., 30 Jan 2025, Yang et al., 18 Jan 2025).

Key implementation insights:

• In-waveguide attenuation must be accounted for, especially for large waveguides or high-frequency operation. Closed-form rate-loss expressions enable system designers to identify conditions (region size, loss coefficient $\alpha$) where neglecting attenuation is justified (Xu et al., 30 Jun 2025).
• Scalability is inherent: as user regions grow, additional PAs can be rapidly deployed at negligible hardware cost, without requiring a corresponding increase in RF front ends, unlike massive MIMO.
• Channel state information (CSI) acquisition is non-trivial, due to the coupling between PA position and effective channel, and demands new strategies: sequential PA activation, compressed sensing, or parameter-based learning approaches (Liu et al., 11 Aug 2025).
• To address computational complexity and non-convexity, machine-learning-based methods—transformers, graph (attention) neural networks, and KKT-guided predictors—have been developed for fast, near-optimal PA placement and beamforming (Xie et al., 8 Feb 2025, Zhang et al., 13 Apr 2025, Liu et al., 11 Aug 2025).

5. Theoretical Performance Guarantees and Limitations

Rigorous analytical results reveal the fundamental spectral efficiency and energy efficiency limits of PASS:

• Maximum rates in single-user, single-waveguide scenarios are always strictly higher than those attainable by fixed-antenna systems, with gains increasing as spatial coverage grows or as users are more heterogeneously distributed (Ding et al., 3 Dec 2024, Xu et al., 18 Feb 2025).
• In MISO/MIMO extensions, the ability to reconfigure effective channels (via micro-scale PA position adjustment) enables the system to meet the theoretical upper bounds of classical interference channel performance under constructive phase-alignment and orthogonality conditions (Ding et al., 3 Dec 2024).
• Energy efficiency can be substantially enhanced via BGAT-optimized designs, with robust scalability to untrained PA/user counts and millisecond inference times (Xie et al., 8 Feb 2025).
• In wideband/multicast settings, performance is sensitive to PA number, aperture, and optimization of spatial configuration, with PSO/AO-type methods yielding rapidly convergent, near-optimal solutions (Mu et al., 23 Feb 2025, Shan et al., 31 May 2025).

Notable limitations include:

• Mutual coupling imposes strict constraints on minimum PA spacing, with non-monotonic gain scaling as PAs are packed more closely.
• The diminishing returns from adding excessively many PAs (with fixed total power) due to decreasing per-element power and array “aperture dilution.”
• Channel estimation, hardware precision in PA activation, and large-scale practical integration (e.g., for fast-moving users) remain open challenges.

6. Emerging Directions and Future Research

Key research frontiers for pinching-antenna systems include:

• Advancing CSI acquisition algorithms that account for dynamic, high-dimensional, and non-orthogonal channel structures (Liu et al., 11 Aug 2025).
• Machine-learning-empowered PA assignment and beamforming (Xie et al., 8 Feb 2025, Zhang et al., 13 Apr 2025), including reinforcement learning for real-time adaptation under non-stationary channel and user conditions.
• Further integration with next-generation multiple access (NGMA), ISAC, covert/secure communications, and wireless power transfer.
• Advanced stochastic geometry and spatial statistics for deployment modeling (PA/user density, blockage dynamics, aperture scaling) (Liu et al., 30 Jan 2025).
• Hardware innovation: real-time, low-loss PA motion/activation, analog-digital hybridization, and control circuit scaling (Yang et al., 18 Jan 2025).
• Synergies with other flexible/reconfigurable antennas (fluid antennas, RIS) as well as practical demonstration in urban/indoor/outdoor environments (including vehicular and UAV applications).

Overall, pinching-antenna systems define a new design space for flexible, low-cost, and scalable wireless infrastructure, characterized by strong theoretical performance, rich physical and algorithmic degrees of freedom, and significant practical promise for 6G and beyond (Liu et al., 11 Aug 2025, Liu et al., 30 Jan 2025, Ding et al., 3 Dec 2024).