Pinching Antennas for ISAC: Waveguide Architectures, Solutions, and Future Directions

Qian Gao, Ruikang ZhongarXiv 2026 · arXiv

TL;DR

This paper introduces Pinching Antenna Systems (PASS) as a transformative architecture for 6G Integrated Sensing and Communication (ISAC). By guiding waves along dielectric waveguides and radiating them through position-adaptive "pinching points," PASS allows the network to physically relocate its radiation origin, overcoming the performance bottlenecks of fixed-geometry antenna arrays.

What Problem Does This Paper Address?

Traditional wireless infrastructure relies on fixed antenna arrays. While these can steer beams electronically, they are physically anchored to a single point. This creates a "one-size-fits-all" geometry that struggles to serve elongated regions (like tunnels or factory aisles) or to simultaneously optimize for two very different tasks: communication (which prioritizes signal strength and interference suppression) and sensing (which requires diverse look-angles and stable illumination).

The authors address this by proposing a geometry-adaptive infrastructure. Instead of asking a fixed array to serve every scenario, the system adjusts the physical origin of radiation to suit the immediate needs of users and sensing targets.

Fig. 1. PASS-enabled ISAC combines guided-wave and free-space propagation. PA positions determine path loss, phase, and sensing look angle.

Methods

The paper categorizes PASS into three main architectures:

  1. Single Waveguide: A low-cost setup where multiple antennas share one feed.
  2. Multiple Waveguides: Uses "Waveguide Division Multiple Access" (WDMA) to provide independent feeds for different users.
  3. Segmented Waveguide (SWAN): Divides long waveguides into short, separately accessible segments to reduce internal signal loss and increase protocol flexibility.
Fig. 2. Waveguide architectures and operating modes for PASS-enabled ISAC. The comparison highlights the progression from shared-feed single waveguides to independent multi-waveguide resources and segmented SWAN operation.

To manage this complex system, the authors propose a two-timescale control framework. A "slow" controller makes geometric decisions (which segments to activate, where to place antennas), while a "fast" controller handles signal-level decisions (beamforming, power allocation).

A key technical contribution is the Segment-Hysteresis Reinforcement Learning (SHRL) algorithm. It uses a "hysteresis gate" to prevent the system from constantly switching physical hardware configurations for marginal gains, which would otherwise lead to instability and high overhead.

Datasets

The study is evaluated using a representative simulation scenario rather than a standard public dataset. The setup includes:

  • A 50m x 60m service area.
  • A carrier frequency of 28 GHz.
  • A SWAN system with 3 segments and 10 pinching antennas per segment.
  • 6 communication users and 1 sensing target.
  • Physics-consistent modeling that accounts for both free-space propagation and internal waveguide attenuation.

Key Insights and Ideas

  • Geometry as a Degree of Freedom: PASS turns the physical location of the antenna into a controllable variable, allowing the network to "follow" users or surround a sensing target for better perspective.
  • The Waveguide Loss Tradeoff: There is a non-monotonic relationship between waveguide length and performance. While longer guides offer better spatial matching, internal attenuation eventually dominates, leading to a "sweet spot" for total length.
  • Aperture Gain vs. Proximity: In segmented systems, increasing the number of segments improves sensing (by reducing the distance signals travel inside the guide) but can hurt communication by reducing the "aperture gain" available to each individual segment.
  • Stability Over Sophistication: The case study showed that for reconfigurable hardware, stabilizing the physical topology (via hysteresis) is more critical for performance than using a more complex reinforcement learning policy.

Key Results

The authors demonstrated the effectiveness of their SHRL approach through a SWAN-ISAC case study:

  • Superior Learning: SHRL achieved an average reward of 1115, significantly outperforming standard algorithms like A2C (880) and PPO (588).
  • Optimal Length: For a 3-segment system, the communication sum rate peaked at 17.07 bps/Hz with an 80m waveguide but plummeted to 6.41 bps/Hz at 100m due to waveguide loss.
  • Granularity Tradeoff: Increasing the number of segments from 3 to 12 boosted the sensing illumination metric by ~4.5x but reduced the communication rate from 13.15 to 10.59 bps/Hz.
Quantitative results for SWAN-ISAC

Limitations and Future Work

The authors acknowledge several areas for further exploration:

  • Model Realism: The current study uses line-of-sight models. Future work needs to incorporate complex multipath environments, clutter rejection, and actual hardware switching latencies.
  • Digital Twins: There is a need for measurement-calibrated "digital twins" to track waveguide health and environmental uncertainty in real-time.
  • Bidirectional Operation: Most current research focuses on the transmit side; joint transmit-receive PASS design is a critical next step for practical radar-like sensing.
  • Security: As the network geometry becomes dynamic, protecting the control plane from "malicious geometry updates" and ensuring location privacy becomes paramount.

Summary updated

Abstract

Integrated sensing and communication (ISAC) is a core design principle for sixth-generation (6G) wireless systems, yet fixed antenna geometry limits its performance. Pinching antenna systems (PASS) guide electromagnetic waves along dielectric waveguides and radiate them through position-adaptive antennas, allowing communication links and sensing observations to benefit from spatial reconfiguration. This article reviews PASS-enabled ISAC through single-waveguide, multi-waveguide, and segmented-waveguide architectures, comparing their feed structures, spatial degrees of freedom, operating protocols, and uplink/downlink support. We organize solution pathways around physics-consistent modeling, joint spatial-signal optimization, dynamic channel acquisition, and stable real-time control. A segmented waveguide-enabled pinching-antenna array (SWAN) case study demonstrates joint segment selection, antenna po?sitioning, and transmit beamforming using segment-hysteresis reinforcement learning. The results illustrate how waveguide loss and per-segment aperture gain create communication-sensing tradeoffs. We conclude with research directions in joint transmit?receive operation, measurement-calibrated digital twins, and trustworthy control.
BibTeX
@misc{gao2026pinching,
  author = {Qian Gao and Ruikang Zhong},
  title = {{Pinching Antennas for ISAC: Waveguide Architectures, Solutions, and Future Directions}},
  year = {2026},
  eprint = {2610.41380},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.41380},
}
Figure