Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

International Conference on Communications

Quantum Graph Neural Network for Joint Optimization of Pinching and Fluid Antenna Systems

Okzata Recy, Bhaskara Narottama, Simon L. Cotton, Trung Q. Duong

Published May 24, 2026• 0 citations
Publisher

Abstract

This work optimizes a flexible antenna environment, that incorporates both pinching antenna systems (PASS) and fluid antenna systems (FAS) at the transmitter and the user sides. Optimizing PASS and FAS presents a non-trivial challenge, as the wireless network exhibits a highly dynamic topology and an extensive solution search space. To address this, this work proposes a quantum graph neural network (QGNN)based solution workflow designed to optimize the user’s quality of service by learning the spatial and structural dependencies between the pinching antenna elements and user ports as a quantum state. The QGNN approach encodes the wireless network topology, encompassing both edge and vertex relationships, into the quantum state space, and performs optimization through a tunable single-qubit rotation gate. Overall, the QGNN-based solution shows a faster loss reduction, especially during the early learning episodes, compared to the benchmark model. The QGNN achieves an average training loss approximately 3.4× lower than that of the typical parameter quantum circuit model during model training. In summary, this work points out the feasibility of applying graph-based, data-driven learning, realized through the quantum graph neural network workflow, to the problem of flexible antenna optimization.

DOI
10.1109/icc59461.2026.11587536

Related papers

Topology-Aware Quantum Graph Neural Networks for Sum-Rate Maximization in Fluid Antenna Systems

Okzata Recy, Bhaskara Narottama, Trung Q. Duong

This paper presents a quantum graph-based solution that leverages a quantum circuit to improve learning efficiency, towards maximizing the sum-rate of wireless communication with fluid antennas in dynamic environments. The employed quantum graph neural networks (QGNN) consists of three main blocks, including (i) a quantum encoding layer, (ii) a quantum graph neural network layer, and (iii) an optimizer layer, which collectively comprise the end-to-end learning workflow. The QGNN adjusts parameters through a quantum graph neural network layer, utilizing basic linear gates on a parameterized quantum circuit (PQC) platform. Additionally, the QGNN circuit is designed with shallow depth and optimized gate composition to reduce quantum resource usage and accelerate convergence during training. The results demonstrate that the proposed QGNN offers competitive performance relative to the existing PQC model. Furthermore, this paper highlights the versatility of quantum graph-based solutions for addressing dynamic, topology-aware wireless network problems.

IEEE Communications Letters•Jan 1, 2026•1 citation

Quantum Graph Learning for Fluid Antenna Systems: Enabling Hyper-Reliable Low-Latency Communication

Okzata Recy, Bhaskara Narottama, Simon L. Cotton +2 more

Fluid antenna systems (FAS) unlock enhanced spatial adaptability by dynamically adjusting the position of their radiating ports within a compact footprint, supporting capabilities beyond the reach of conventional fixed antennas. However, this flexibility may impose real-time challenges, particularly when underlying algorithms are unable to track rapid spatial and temporal variations in adjusting continuous port reconfiguration. In response, this article proposes quantum graph learning (QGL) as a key enabler toward the goal of near-zero latency wireless operation in FAS. By representing the wireless network as a unified relational structure, QGL captures signal information and topological connections while leveraging the inherent parallelism of quantum processing. Subsequently, this article demonstrates how a FAS network can be transformed into a quantum graph representation, establishing a robust framework for more advanced QGL models. This article also identifies some pivotal research directions, charting a roadmap for QGL to drive FAS evolution in the near and long term.

IEEE Network•May 8, 2026•0 citations

Hybrid Quantum-Classical Transformers for Port Selection and Beamforming in Multiuser Fluid Antenna Systems

Niancong Ji, Gan Zheng, Chengkang Pan +2 more

Fluid antenna systems (FAS) enhance multiple-input multiple-output (MIMO) networks by providing additional spatial degrees of freedom for interference suppression and hardware cost reduction. However, the joint design of port selection and beamforming in two-dimensional FAS presents a challenging high-dimensional, mixed-integer optimization problem with strong variable coupling. To address this, we propose a quantum machine learning (QML)-based hybrid quantum-classical Transformer (HyQ-Former) that achieves end-to-end differentiable optimization. In this framework, complex channel state information is decomposed and projected with positional embedding. The backbone employs a quantum bottleneck projection network that integrates instantaneous quantum polynomial embeddings, data re-uploading, strongly entangling, and multi-observable readout to realize quantum multi-head attention and a quantum feed-forward network with higher-order, nonlocal representations. Furthermore, the Gumbel-Softmax technique relaxes discrete port selection into a differentiable operation, facilitating unsupervised training to maximize the sum rate. Numerical results demonstrate that HyQ-Former consistently outperforms classical baselines across various power and spatial correlation scenarios. Notably, due to the efficient low-dimensional quantum representation, the proposed model reduces the number of trainable parameters to only approximately 8.2%∼19.8% of the classical Transformer without compromising generalization. This approach offers a lightweight solution for FAS design, balancing system performance with computational complexity.

IEEE Transactions on Cognitive Communications and Networking•Jan 1, 2026•0 citations

Inductive Graph Representation Learning with Quantum Graph Neural Networks

Arthur Mendonça Faria, Ignacio Fernández Graña, Savvas Varsamopoulos

Graph Neural Networks (QGNNs) offer a promising approach to combining quantum computing with graph-structured data processing. While classical Graph Neural Networks (GNNs) are scalable and robust, existing QGNNs often lack flexibility due to graph-specific quantum circuit designs, limiting their applicability to diverse real-world problems. To address this, we propose a versatile QGNN framework inspired by GraphSAGE, using quantum models as aggregators to generate quantum-native node embeddings. We integrate inductive representation learning techniques with parameterized quantum convolutional and pooling layers, bridging classical and quantum paradigms. The convolutional layer is flexible, allowing tailored designs for specific tasks. Benchmarked on a node regression task with the QM9 dataset, our framework, using a single minimal circuit for all aggregation steps, handles molecules with varying numbers of atoms without changing qubits or circuit architecture. While classical GNNs achieve higher training performance, our quantum approach remains competitive and often shows enhanced generalization as molecular complexity increases. We also observe faster learning in early training epochs. To mitigate trainability limitations of a single-circuit setup, we extend the framework with multiple quantum aggregators on QM9. Assigning distinct circuits to each hop substantially improves training performance across all cases. Additionally, we numerically demonstrate the absence of barren plateaus as qubit numbers increase, suggesting that the proposed model remains trainable to larger, more complex graph-based problems.

arXiv:2503.24111•Mar 31, 2025•3 citations

Quantum Graph Neural Network for Resource Management in Wireless Communication

Le Tung Giang, Nguyen Xuan Tung, Won–Joo Hwang

Effective resource management in wireless communication systems is a critical requirement, given the interference challenges and the need for efficient allocation strategies. Graph Neural Networks (GNNs) have emerged as powerful tools for tackling this problem due to their scalability and ability to generalize over complex graph-structured data, like that found in wireless networks. However, the computational complexity of GNNs on large-scale systems limits their real-time deployment. Recent advances in quantum computing and quantum machine learning (QML) offer a promising solution by leveraging the unique properties of quantum systems to reduce computational overhead. This paper proposes a Quantum Graph Neural Network (QGNN) model specifically designed for a wireless communication scenario, implemented as a variational quantum circuit (VQC) to realize the message-passing mechanism integral to GNNs. Applied to a supervised resource management task in a wireless network, the proposed QGNN demonstrates encouraging results, showcasing its potential as a scalable and efficient approach for real-time resource management in future wireless systems.

International Conference on Artificial Intelligence in Information and Communication (ICAIIC)•Feb 18, 2025•8 citations

D2D Power Allocation via Quantum Graph Neural Network

Le Tung Giang, Nguyen Xuan Tung, Won–Joo Hwang

Increasing wireless network complexity demands scalable resource management. Classical GNNs excel at graph learning but incur high computational costs in large-scale settings. We present a fully quantum Graph Neural Network (QGNN) that implements message passing via Parameterized Quantum Circuits (PQCs). Our Quantum Graph Convolutional Layers (QGCLs) encode features into quantum states, process graphs with NISQ-compatible unitaries, and retrieve embeddings through measurement. Applied to D2D power control for SINR maximization, our QGNN matches classical performance with fewer parameters and inherent parallelism. This end-to-end PQC-based GNN marks a step toward quantum-accelerated wireless optimization.

arXiv:2511.15246•International Conference on Mobile Computing and Ubiquitous Network (ICMU)•Sep 10, 2025•1 citation
This work optimizes a flexible antenna environment, that incorporates both pinching antenna systems (PASS) and fluid antenna systems (FAS) at the transmitter and the user sides. Optimizing PASS and FAS presents a non-trivial challenge, as the wireless network exhibits a highly dynamic topology and an extensive solution search space. To address this, this work proposes a quantum graph neural network (QGNN)based solution workflow designed to optimize the user’s quality of service by learning the spatial and structural dependencies between the pinching antenna elements and user ports as a quantum state. The QGNN approach encodes the wireless network topology, encompassing both edge and vertex relationships, into the quantum state space, and performs optimization through a tunable single-qubit rotation gate. Overall, the QGNN-based solution shows a faster loss reduction, especially during the early learning episodes, compared to the benchmark model. The QGNN achieves an average training loss approximately 3.4× lower than that of the typical parameter quantum circuit model during model training. In summary, this work points out the feasibility of applying graph-based, data-driven learning, realized through the quantum graph neural network workflow, to the problem of flexible antenna optimization.
This paper presents a quantum graph-based solution that leverages a quantum circuit to improve learning efficiency, towards maximizing the sum-rate of wireless communication with fluid antennas in dynamic environments. The employed quantum graph neural networks (QGNN) consists of three main blocks, including (i) a quantum encoding layer, (ii) a quantum graph neural network layer, and (iii) an optimizer layer, which collectively comprise the end-to-end learning workflow. The QGNN adjusts parameters through a quantum graph neural network layer, utilizing basic linear gates on a parameterized quantum circuit (PQC) platform. Additionally, the QGNN circuit is designed with shallow depth and optimized gate composition to reduce quantum resource usage and accelerate convergence during training. The results demonstrate that the proposed QGNN offers competitive performance relative to the existing PQC model. Furthermore, this paper highlights the versatility of quantum graph-based solutions for addressing dynamic, topology-aware wireless network problems.
Fluid antenna systems (FAS) unlock enhanced spatial adaptability by dynamically adjusting the position of their radiating ports within a compact footprint, supporting capabilities beyond the reach of conventional fixed antennas. However, this flexibility may impose real-time challenges, particularly when underlying algorithms are unable to track rapid spatial and temporal variations in adjusting continuous port reconfiguration. In response, this article proposes quantum graph learning (QGL) as a key enabler toward the goal of near-zero latency wireless operation in FAS. By representing the wireless network as a unified relational structure, QGL captures signal information and topological connections while leveraging the inherent parallelism of quantum processing. Subsequently, this article demonstrates how a FAS network can be transformed into a quantum graph representation, establishing a robust framework for more advanced QGL models. This article also identifies some pivotal research directions, charting a roadmap for QGL to drive FAS evolution in the near and long term.
Fluid antenna systems (FAS) enhance multiple-input multiple-output (MIMO) networks by providing additional spatial degrees of freedom for interference suppression and hardware cost reduction. However, the joint design of port selection and beamforming in two-dimensional FAS presents a challenging high-dimensional, mixed-integer optimization problem with strong variable coupling. To address this, we propose a quantum machine learning (QML)-based hybrid quantum-classical Transformer (HyQ-Former) that achieves end-to-end differentiable optimization. In this framework, complex channel state information is decomposed and projected with positional embedding. The backbone employs a quantum bottleneck projection network that integrates instantaneous quantum polynomial embeddings, data re-uploading, strongly entangling, and multi-observable readout to realize quantum multi-head attention and a quantum feed-forward network with higher-order, nonlocal representations. Furthermore, the Gumbel-Softmax technique relaxes discrete port selection into a differentiable operation, facilitating unsupervised training to maximize the sum rate. Numerical results demonstrate that HyQ-Former consistently outperforms classical baselines across various power and spatial correlation scenarios. Notably, due to the efficient low-dimensional quantum representation, the proposed model reduces the number of trainable parameters to only approximately 8.2%∼19.8% of the classical Transformer without compromising generalization. This approach offers a lightweight solution for FAS design, balancing system performance with computational complexity.
Graph Neural Networks (QGNNs) offer a promising approach to combining quantum computing with graph-structured data processing. While classical Graph Neural Networks (GNNs) are scalable and robust, existing QGNNs often lack flexibility due to graph-specific quantum circuit designs, limiting their applicability to diverse real-world problems. To address this, we propose a versatile QGNN framework inspired by GraphSAGE, using quantum models as aggregators to generate quantum-native node embeddings. We integrate inductive representation learning techniques with parameterized quantum convolutional and pooling layers, bridging classical and quantum paradigms. The convolutional layer is flexible, allowing tailored designs for specific tasks. Benchmarked on a node regression task with the QM9 dataset, our framework, using a single minimal circuit for all aggregation steps, handles molecules with varying numbers of atoms without changing qubits or circuit architecture. While classical GNNs achieve higher training performance, our quantum approach remains competitive and often shows enhanced generalization as molecular complexity increases. We also observe faster learning in early training epochs. To mitigate trainability limitations of a single-circuit setup, we extend the framework with multiple quantum aggregators on QM9. Assigning distinct circuits to each hop substantially improves training performance across all cases. Additionally, we numerically demonstrate the absence of barren plateaus as qubit numbers increase, suggesting that the proposed model remains trainable to larger, more complex graph-based problems.
Effective resource management in wireless communication systems is a critical requirement, given the interference challenges and the need for efficient allocation strategies. Graph Neural Networks (GNNs) have emerged as powerful tools for tackling this problem due to their scalability and ability to generalize over complex graph-structured data, like that found in wireless networks. However, the computational complexity of GNNs on large-scale systems limits their real-time deployment. Recent advances in quantum computing and quantum machine learning (QML) offer a promising solution by leveraging the unique properties of quantum systems to reduce computational overhead. This paper proposes a Quantum Graph Neural Network (QGNN) model specifically designed for a wireless communication scenario, implemented as a variational quantum circuit (VQC) to realize the message-passing mechanism integral to GNNs. Applied to a supervised resource management task in a wireless network, the proposed QGNN demonstrates encouraging results, showcasing its potential as a scalable and efficient approach for real-time resource management in future wireless systems.
Increasing wireless network complexity demands scalable resource management. Classical GNNs excel at graph learning but incur high computational costs in large-scale settings. We present a fully quantum Graph Neural Network (QGNN) that implements message passing via Parameterized Quantum Circuits (PQCs). Our Quantum Graph Convolutional Layers (QGCLs) encode features into quantum states, process graphs with NISQ-compatible unitaries, and retrieve embeddings through measurement. Applied to D2D power control for SINR maximization, our QGNN matches classical performance with fewer parameters and inherent parallelism. This end-to-end PQC-based GNN marks a step toward quantum-accelerated wireless optimization.