Waveguide to Meaning: Semantic-Aware NOMA for Pinching-Antenna Systems

1. Problem Statement

The paper addresses the challenge of enhancing spectral efficiency (SE) and Quality of Service (QoS) in next-generation (6G) heterogeneous networks. Specifically, it investigates the integration of Pinching-Antenna Systems (PASS) with Non-Orthogonal Multiple Access (NOMA) for Semantic Communication (SC).

• Context: Traditional fixed-antenna systems (CAS) rely on bit-based communication and lack the flexibility to dynamically shape the wireless channel. While PASS offers flexible antenna placement to create strong Line-of-Sight (LoS) links, its potential for semantic communication (transmitting meaning rather than raw bits) in a NOMA setting remains unexplored.
• The Challenge: The system must serve two types of users simultaneously:
  1. Semantic Users (S): Require high semantic fidelity, often operating in low-to-moderate SNR regimes where traditional mutual information fails.
  2. Bit Users (B): Require strict QoS (minimum data rate) based on traditional bit communication.
• The Goal: Maximize the Semantic Spectral Efficiency (SE) for the semantic user while guaranteeing the minimum rate requirement for the bit user, utilizing the dynamic channel control capabilities of PASS in both single-waveguide and multi-waveguide scenarios.

2. Methodology

The authors propose a unified framework for heterogeneous NOMA users using PASS, divided into two architectural scenarios:

A. System Models

• Single-Waveguide PASS: Multiple pinching antennas are deployed on a single dielectric waveguide. They share a single Radio Frequency (RF) chain. The antennas are spaced at a minimum distance ($\Delta \ge \lambda/2$) to prevent mutual coupling.
• Multi-Waveguide PASS: Multiple waveguides are used, with each waveguide carrying a single pinching antenna. This allows for spatial multiplexing across different waveguides with sufficient lateral spacing to avoid coupling.
• Channel Model: A spherical wave channel model is used to account for near-field effects, which is critical given the large aperture of the waveguide relative to the wavelength.
• Semantic Rate Model: The semantic rate is modeled using a DeepSC-based sigmoid function ($\epsilon_K(\gamma)$), which relates the received SNR to semantic similarity. This captures the diminishing returns of semantic transmission at high SNR.

B. Optimization Framework

The core objective is to maximize the semantic rate of User S subject to the QoS constraints of User B and the decoding order constraints (Bit-to-Semantic Successive Interference Cancellation).

• Decoding Order: The system adopts a Bit-to-Semantic decoding order. The bit user's signal is decoded first (treating the semantic signal as interference), and the semantic user decodes its signal after canceling the bit user's signal (SIC). This is chosen to guarantee the strict QoS of the bit user.
• Single-Waveguide Solution (Alternating Optimization - AO):
  • The problem is non-convex due to the coupling of antenna positions and power allocation.
  • Step 1 (Power Allocation): For fixed antenna positions, a closed-form solution is derived for the power allocation coefficient ($\alpha_S$) based on active constraints (QoS and SIC feasibility).
  • Step 2 (Antenna Positioning): For fixed power, the antenna positions are optimized using a bisection search. This involves two stages:
    1. Large-scale placement: Relocating antennas to maximize the LoS channel gain for the semantic user.
    2. Fine-scale phase alignment: Adjusting positions to ensure constructive interference (phase alignment) for the semantic user while maintaining bit-user constraints.
• Multi-Waveguide Solution (Monotonic Optimization & Minorization-Maximization - MM):
  • The problem is decomposed into three subproblems: Power allocation, antenna positions, and waveguide power allocation ($\beta_k$).
  • Since the objective function is non-concave and variables are strongly coupled, the authors use an MM approach. They construct a tight global lower-bound surrogate function for the waveguide power allocation subproblem.
  • This surrogate is iteratively maximized to ensure a non-decreasing sequence of the objective function, solving for the optimal power split across waveguides.

3. Key Contributions

1. Novel Integration: First study to combine PASS with Semantic Communication in a heterogeneous NOMA framework, leveraging the mobility of pinching antennas to create channel disparities beneficial for SIC.
2. Dual-Scenario Analysis: Provides a comprehensive comparison between Single-Waveguide (dense antenna deployment on one guide) and Multi-Waveguide (distributed antennas across multiple guides) architectures.
3. Algorithmic Innovation:
   • Developed an AO-based algorithm for single-waveguide systems that jointly optimizes power and antenna positions with phase fine-tuning.
   • Proposed an MM-based approach for multi-waveguide systems to solve the complex waveguide power allocation subproblem, ensuring convergence to a local optimum.
4. Bit-to-Semantic Decoding Strategy: Justified and implemented a specific decoding order (Bit first, then Semantic) to strictly satisfy bit-user QoS while exploiting the robustness of semantic transmission at lower SNRs.
5. Performance Validation: Extensive numerical simulations demonstrating that PASS significantly outperforms conventional fixed-antenna systems (CAS).

4. Simulation Results

The authors validated their framework using parameters typical for 6G (e.g., 28 GHz carrier frequency, near-field conditions).

• Semantic SE Gain: PASS consistently outperforms CAS.
  • In single-waveguide scenarios, PASS achieves higher semantic SE, with gains increasing as the number of pinching antennas increases.
  • Phase Alignment: Fine-tuning antenna positions to align phases (tight $\delta_S$) yields significant gains (up to 15% improvement in SE at 10 dBm) compared to coarse alignment or fixed spacing.
• Multi-Waveguide Superiority:
  • The multi-waveguide PASS outperforms the single-waveguide PASS, particularly in stringent conditions (high bit-user QoS requirements and wide coverage areas).
  • Under strict conditions ($R_{min}^B = 1.5$ bps/Hz, wide area), the multi-waveguide system reduces the required transmit power by 8.5 dB compared to the single-waveguide system to meet the same outage probability.
• Outage Probability: PASS demonstrates significantly lower outage probabilities for bit-user QoS and SIC feasibility compared to CAS, attributed to the ability to dynamically optimize channel gains.
• Coverage and Power:
  • Smaller coverage areas yield better SE due to reduced path loss.
  • Multi-waveguide systems show the most advantage in low-power, wide-coverage scenarios.

5. Significance and Conclusion

This work establishes Pinching-Antenna Systems as a critical enabler for 6G Semantic Communication. By dynamically controlling the physical layer (antenna positions) to match the requirements of semantic transmission, PASS can overcome the limitations of fixed antennas and traditional bit-centric communication.

• Practical Implication: The study suggests that for dense urban deployments or scenarios requiring high reliability for diverse user types (IoT bits vs. AI-driven semantic data), a multi-waveguide PASS architecture offers a robust, low-complexity (single RF chain) solution.
• Future Direction: It highlights the synergy between flexible antenna hardware and semantic-aware software (DeepSC), paving the way for "Waveguide to Meaning" networks where the physical channel is optimized not just for data rate, but for information meaning.