Pinching-Antenna Assisted Sensing (PASS)

• PASS is a reconfigurable antenna design that employs mechanically adjustable pinching antennas along dielectric waveguides for precise beamforming and spatial control.
• It enables agile beam steering and enhanced spatial diversity by dynamically controlling radiating element positions, significantly improving SNR and localization accuracy.
• Optimization techniques such as alternating optimization, semidefinite relaxation, and reinforcement learning are used to efficiently manage PA placement and ISAC trade-offs.

Pinching-Antenna Assisted Sensing (PASS) refers to a class of reconfigurable antenna architectures in which mechanical or electromechanical “pinching antennas” (PAs) are deployed along low-loss dielectric waveguides to implement large-scale, position-agile, highly reconfigurable radio-frequency apertures for communication and sensing. By enabling programmable spatial distribution and activation of radiating elements, PASS fundamentally departs from canonical phased arrays and RIS platforms. It provides unprecedented path-loss control, agile beam steering, and spatial diversity with minimal hardware and power overhead—crucial for integrated sensing and communications (ISAC), covert communication, and location services.

1. Physical Principles, Architecture, and Modeling

A typical PASS platform consists of one or more dielectric waveguides fed by RF chains, with multiple small dielectric PAs (“bumps” or “slots”) that can be placed, moved, or activated dynamically anywhere along the waveguide. These PAs function as controllable coupling sites that radiate guided-wave energy into free space. The electromagnetic behavior of each PA–waveguide coupling is rigorously described using coupled-mode theory. The waveguide-confined field couples into each bump with position-dependent amplitude and phase: $$\mathbf{E}(x) = A(x)\mathbf{E}_\mathrm{guide}(x) + B(x)\mathbf{E}_\mathrm{pinch}(x),$$ where the transfer from the guide to the “pinch” is governed by the coupling coefficient $\kappa$ and propagation constant mismatch $\Delta\beta$ (Wang et al., 9 Feb 2025).

The aggregate radiated field from $M$ PAs located at $x_{p,m}$ is: $$y = \sum_{m=1}^M \frac{\eta\,e^{-j\beta_0 r_m}}{r_m} s_{\mathrm{rad},m} + n,$$ with each $s_{\mathrm{rad},m}$ determined by the local coupling, the remaining guided power, and the phase shift from the waveguide. Two design paradigms are common: equal-power (every PA outputs equal field) and proportional-power (each PA taps the same fraction of the residual guided power), both of which yield nearly identical macroscopic performance with appropriate parameterization (Wang et al., 9 Feb 2025).

This architecture scales naturally to support multiple waveguides, each with numerous PAs distributed optimally for the intended application—ISAC, multi-user MIMO, or sensing.

2. Pinching Beamforming: Flexible Control of Propagation Geometry

PASS fundamentally enables “pinching beamforming,” i.e., the spatial reconfiguration of the system's effective aperture by adjusting both large-scale PA positions and fine-scale in-guide phase. Unlike conventional digital or phased arrays, PASS alters the physical location of radiating elements—enabling direct control over both free-space path loss and near-field/spherical propagation (Ouyang et al., 15 May 2025).

The capacity to dynamically bring PAs near users or targets dramatically enhances SNR and spatial resolution, especially in large, distributed scenarios or environments with severe blockage. Pinching beamforming is leveraged for both downlink and uplink, multi-user spatial multiplexing (Gan et al., 3 Jun 2025), and robust null steering for covert or secure communications (Jiang et al., 7 Sep 2025).

3. Sensing: Signal Models, Performance Bounds, and Optimization

PASS can be configured as a highly agile distributed sensing array. The fundamental bounds on estimation accuracy are characterized by both the classical and Bayesian Cramér-Rao Bound (CRB/BCRB). The received signal from a target at $\mathbf{r}$ is: $$y = \sum_{m=1}^M g_m(\mathbf x)\,h_m(\mathbf r)\,s + n,$$ with the in-guide phase $g_m(\mathbf x)$ and free-space channel $h_m(\mathbf r)$. The BCRB for position estimation is

$$\mathrm{BCRB}(\mathbf x) = \mathrm{tr}[\mathbf J_\mathbf r(\mathbf x)^{-1}],$$

where $\mathbf J_\mathbf r$ fuses data and prior information (Jiang et al., 10 Oct 2025).

For multi-target or high-dimensional scenarios, the CRB is often minimized jointly over PA positions and sensing waveform covariances via large-scale nonconvex optimization (Li et al., 27 Aug 2025, Wang et al., 21 May 2025): $$\min_{\{\mathbf{x}, R_s\}} \operatorname{Tr}[\mathrm{CRB}(\theta)],$$ subject to power, aperture, and discreteness/spatial constraints. Alternating optimization leveraging semidefinite relaxations, penalty methods, SCA, and swarm optimization are the preferred approaches.

Numerical results consistently show that optimized PASS—with dynamic, scenario-specific PA placement—achieves an order-of-magnitude improvement in localization accuracy and robustness over uniform or fixed arrays (Li et al., 27 Aug 2025, Wang et al., 21 May 2025, Jiang et al., 10 Oct 2025), with benefits saturating after $M\approx 5$–$10$ PAs.

4. Integrated Sensing and Communication (ISAC): Trade-Offs, Dynamic Scheduling, and Rate Regions

PASS provides the hardware flexibility required to realize ISAC systems with fully programmable trade-offs between communication rate and sensing quality. In ISAC applications, the simultaneous optimization of PA activation schedules, information/cognitive beamforming, and waveform allocation is essential (Khalili et al., 3 May 2025, Zhang et al., 10 Apr 2025, Ouyang et al., 15 May 2025).

The typical ISAC-PASS model employs:

• Time-slot-based dynamic PA scheduling: To provide angular/target-diversity, with each time slot activating a distinct PA for distinct look-angles (Khalili et al., 3 May 2025).
• Probabilistic RCS modeling: Outage-based reliability metrics that treat the radar cross-section as a random variable varying with PA angle, yielding realistic, non-deterministic sensing performance predictions.
• Joint communication constraints: Enforcing long-term QoS/lower bounds on accumulated user rates.

The achievable rate region for PASS-ISAC is analytically tractable in both single- and multi-pinch settings. Pareto-optimal beamforming and AO strategies yield rate regions that strictly encompass those attainable with static antenna arrays (Ouyang et al., 15 May 2025).

Dynamic PA activation, compared to fixed or conventional antenna switching, yields significant diversity/sensing reliability improvements—up to an order of magnitude reduction in radar outage (Khalili et al., 3 May 2025), with performance scaling favorably in both the number of PAs and time slots.

5. Algorithmic Frameworks and Prototypical Solutions

PASS deployments require solving multimodal, nonconvex, and often mixed-integer programs for array geometry, waveform design, and scheduling. The prevailing methodology includes:

• Penalty-based alternating optimization (AO): Decoupling digital beamforming, PA placement, and power allocation, leverage convex relaxations or SCA for each subproblem (Li et al., 27 Aug 2025, Wang et al., 9 Feb 2025).
• Semidefinite relaxation (SDR): For joint covariance design under CRB or rate constraints.
• Swarm-based methods (PSO): For high-dimensional, combinatorial PA positioning problems (Wang et al., 21 May 2025, Gan et al., 3 Jun 2025).
• Majorization-Minimization (MM) and dual approaches: For tight, fast beamformer updates (Gan et al., 3 Jun 2025).
• Reinforcement learning (SAC): For real-time PA positioning and adaptation in time-varying channel/interference environments (Jiang et al., 7 Sep 2025).
• Closed-form quadratic (KKT-based) or iterative 1D search algorithms: For BCRB or min-max optimization under geometric and power constraints (Jiang et al., 10 Oct 2025).

Equal- and proportional-power strategies for the in-guide coupling configuration are both practical, with minimal performance penalties for the latter (Wang et al., 9 Feb 2025). Discrete versus continuous activation trade-offs are governed by hardware cost and desired spatial/phase resolution; high position quantization density is required for ultra-narrow beams (Wang et al., 9 Feb 2025).

6. Special Use Cases: Indoor Positioning, IoT Coverage, Covert Communications

Indoor Positioning: PASS's geometric channel is exploited for RSSI-based ranging and weighted least-squares localization algorithms. The architecture attains meter-level or submeter positioning accuracy, especially between and near PAs, with diminishing gains beyond 7–10 PAs (Zhang et al., 11 Aug 2025).

IoT and Coverage Optimization: In circular or complex geometries, PASS allows devices to be dynamically served by proximal PAs, optimizing for SNR/rate under both full and partial coverage, with performance subject to trade-offs between waveguide length, placement, and material attenuation (Zhang et al., 5 Sep 2025). Optimal coverage often balances increased device capture versus waveguide loss.

Covert Communications: PASS enables dynamic null steering and channel obfuscation against adversaries (“wardens”), with EKF-based sensing/CSI estimation and joint deep RL for PA placement and AN design. Covertness and low probability of detection are substantially improved compared to static arrays (Jiang et al., 7 Sep 2025).

7. Energy Efficiency, Dual-Scale Deployment, and Practical Guidelines

PASS achieves up to two-fold energy efficiency gain over cell-free and MIMO architectures by exhaustively optimizing transmit precoding, PA radiation, and physical deployment (Gan et al., 31 Oct 2025). The dual-scale deployment (DSD) protocol—decoupling meter-scale “slide” with micrometer “tune” stages—realizes high precision, high agility, and minimal actuation power. Protocol choice (e.g., SAT: select-base plus tune-PA) is critical for balancing power, complexity, and response (Gan et al., 31 Oct 2025).

The optimality of grid and actuation resolution, minimum inter-PA spacing (for mutual coupling avoidance), and angular/geometric coverage are dominant hardware-design trade-offs.

---

In summary, Pinching-Antenna Assisted Sensing (PASS) represents a paradigm shift in array design for wireless communications and sensing, underpinned by physics-based coupled-mode models, flexible pinching beamforming, and robust algorithmic frameworks for ISAC and spatial inference. It enables new modalities in angular diversity, dynamic coverage, covert operation, and energy-efficient deployment—outperforming conventional platforms in both theoretical bounds and empirical benchmarks (Wang et al., 9 Feb 2025, Khalili et al., 3 May 2025, Wang et al., 21 May 2025, Li et al., 27 Aug 2025, Zhang et al., 5 Sep 2025, Ouyang et al., 15 May 2025, Gan et al., 31 Oct 2025, Jiang et al., 10 Oct 2025, Gan et al., 3 Jun 2025, Zhang et al., 11 Aug 2025, Jiang et al., 7 Sep 2025, Zhang et al., 10 Apr 2025).