arXiv subjects

Le Tung Giang

Publications and source records attributed to Le Tung Giang.

4 recordsLinked to original sources

arXiv2026

Scalable Quantum Message Passing Graph Neural Networks for Next-Generation Wireless Communications: Architectures, Use Cases, and Future Directions

Graph Neural Networks (GNNs) are eminently suitable for wireless resource management, thanks to their scalability, but they still face computational challenges in large-scale, dense networks in classical computers. The integration of quantum computing with GNNs offers a promising pathway for enhancing computational efficiency because they reduce the model complexity. This is achieved by leveraging the quantum advantages of parameterized quantum circuits (PQCs), while retaining the expressive power of GNNs. However, existing pure quantum message passing models remain constrained by the limited number of qubits, hence limiting the scalability of their application to the wireless systems. As a remedy, we conceive a Scalable Quantum Message Passing Graph Neural Network (SQM-GNN) relying on a quantum message passing architecture. To address the aforementioned scalability issue, we decompose the graph into subgraphs and apply a shared PQC to each local subgraph. Importantly, the model incorporates both node and edge features, facilitating the full representation of the underlying wireless graph structure. We demonstrate the efficiency of SQM GNN on a device-to-device (D2D) power control task, where it outperforms both classical GNNs and heuristic baselines. These results highlight SQM-GNN as a promising direction for future wireless network optimization.

cs.IT↗

arXiv2025

Vehicle Routing Problems via Quantum Graph Attention Network Deep Reinforcement Learning

The vehicle routing problem (VRP) is a fundamental NP-hard task in intelligent transportation systems with broad applications in logistics and distribution. Deep reinforcement learning (DRL) with Graph Neural Networks (GNNs) has shown promise, yet classical models rely on large multi-layer perceptrons (MLPs) that are parameter-heavy and memory-bound. We propose a Quantum Graph Attention Network (Q-GAT) within a DRL framework, where parameterized quantum circuits (PQCs) replace conventional MLPs at critical readout stages. The hybrid model maintains the expressive capacity of graph attention encoders while reducing trainable parameters by more than 50%. Using proximal policy optimization (PPO) with greedy and stochastic decoding, experiments on VRP benchmarks show that Q-GAT achieves faster convergence and reduces routing cost by about 5% compared with classical GAT baselines. These results demonstrate the potential of PQC-enhanced GNNs as compact and effective solvers for large-scale routing and logistics optimization.

cs.LG↗

arXiv2025

Graph Neural Networks for Next-Generation-IoT: Recent Advances and Open Challenges

Graph Neural Networks (GNNs) have emerged as a powerful framework for modeling complex interconnected systems, hence making them particularly well-suited to address the growing challenges of next-generation Internet of Things (NG-IoT) networks. Existing studies remain fragmented, and there is a lack of comprehensive guidance on how GNNs can be systematically applied to NG-IoT systems. As NG-IoT systems evolve toward 6G, they incorporate diverse technologies. These advances promise unprecedented connectivity, sensing, and automation but also introduce significant complexity, requiring new approaches for scalable learning, dynamic optimization, and secure, decentralized decision-making. This survey provides a comprehensive and forward-looking exploration of how GNNs can empower NG-IoT environments. We commence by exploring the fundamental paradigms of GNNs and articulating the motivation for their use in NG-IoT networks. Besides, we intrinsically connect GNNs with the family of low-density parity-check codes, modeling the NG-IoT as dynamic constrained graphs. We highlight the distinct roles of node-, edge-, and graph-level tasks in tackling key challenges and demonstrate the GNNs' ability to overcome the limitations of traditional optimization. We examine the application of GNNs across core NG-enabling technologies and their integration with distributed frameworks to support privacy-preservation and distributed intelligence. We then delve into the challenges posed by adversarial attacks, offering insights into defense mechanisms. Lastly, we examine how GNNs can be integrated with emerging technologies. Our findings highlight the transformative potential of GNNs in improving efficiency, scalability, and security. Finally, we summarize the key lessons learned and outline promising future research directions, along with a set of design guidelines tailored for NG-IoT applications.

cs.IT↗

arXiv2025

Joint Access Point Activation and Power Allocation for Cell-Free Massive MIMO Aided ISAC Systems

Cell-free massive multiple-input multiple-output (MIMO)-aided integrated sensing and communication (ISAC) systems are investigated where distributed access points jointly serve users and sensing targets. We demonstrate that only a subset of access points (APs) has to be activated for both tasks, while deactivating redundant APs is essential for power savings. This motivates joint active AP selection and power control for optimizing energy efficiency. The resultant problem is a mixed-integer nonlinear program (MINLP). To address this, we propose a model-based Branch-and-Bound approach as a strong baseline to guide a semi-supervised heterogeneous graph neural network (HetGNN) for selecting the best active APs and the power allocation. Comprehensive numerical results demonstrate that the proposed HetGNN reduces power consumption by 20-25\% and runs nearly 10,000 times faster than model-based benchmarks.

cs.IT↗