RIS-Assisted Downlink Pinching-Antenna Systems: GNN-Enabled Optimization Approaches

2025-11-25

AI Summary

This work addresses performance optimization for reconfigurable intelligent surface (RIS)-assisted multi-user downlink systems employing patch-antenna substrate-supported (PASS) waveguide-integrated antennas. Method: We propose a graph neural network (GNN)-based joint design framework featuring a three-stage GNN architecture that explicitly models system topology and enables end-to-end co-optimization of PASS antenna placement, RIS phase shifts, and base station beamforming. The framework integrates unsupervised pretraining with convex optimization fine-tuning to achieve low-latency, hardware- and communication-constrained, generalizable resource allocation. Contribution/Results: Experiments demonstrate substantial gains in sum rate and energy efficiency. The approach exhibits strong generalization across diverse scenarios, robustness to channel variations, and real-time inference capability (average latency <5 ms). To the best of our knowledge, this is the first learnable and deployable joint optimization paradigm for RIS-PASS cooperative communications.

Technology Category

• Search and Optimization: Learning to Search
• Intelligent Robots: Learning & Optimization for ROB
• Computer Vision: Learning & Optimization for CV

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

This paper investigates a reconfigurable intelligent surface (RIS)-assisted multi-waveguide pinching-antenna (PA) system (PASS) for multi-user downlink information transmission, motivated by the unknown impact of the integration of emerging PASS and RIS on wireless communications. First, we formulate sum rate (SR) and energy efficiency (EE) maximization problems in a unified framework, subject to constraints on the movable region of PAs, total power budget, and tunable phase of RIS elements. Then, by leveraging a graph-structured topology of the RIS-assisted PASS, a novel three-stage graph neural network (GNN) is proposed, which learns PA positions based on user locations, and RIS phase shifts according to composite channel conditions at the first two stages, respectively, and finally determines beamforming vectors. Specifically, the proposed GNN is achieved through unsupervised training, together with three implementation strategies for its integration with convex optimization, thus offering trade-offs between inference time and solution optimality. Extensive numerical results are provided to validate the effectiveness of the proposed GNN, and to support its unique attributes of viable generalization capability, good performance reliability, and real-time applicability. Moreover, the impact of key parameters on RIS-assisted PASS is illustrated and analyzed.

Problem

• Optimizing sum rate and energy efficiency in RIS-assisted multi-waveguide antenna systems
• Determining optimal antenna positions and RIS phase shifts for multi-user downlink transmission
• Developing GNN-based approaches to handle movable antenna and power constraints

Innovation

• GNN optimizes antenna positions and phase shifts
• Unsupervised training enables real-time beamforming optimization
• Graph neural network integrates with convex optimization strategies