Resource Allocation in Fronthaul-Constrained Cell-Free Networks Using Edge-Graph Attention Networks
IEEE Transactions on Wireless Communications · 2026 · 25 · 13305-13318
DOI: 10.1109/twc.2026.3669944 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Wireless Communications, 2026. Verfasst von Xiaoyu Liu und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 40 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Jiangsu Provincial Key Research and Development Program, SRIBD Innovation Fund.
Abstract (Originalsprache)
Cell-free networks represent a transformative approach for 6G communications. However, they are subject to stringent capacity requirements on the fronthaul links. In this study, we investigate the downlink transmission of cell-free networks with limited fronthaul capacity, aiming to jointly optimize the beamforming matrices of access points (APs) and the AP-user equipment (UE) association strategies to maximize the weighted sum rate. We first present a model-based approach leveraging alternating optimization and successive convex approximation. To facilitate practical deployment, we then propose a learning-based approach using edge-graph attention networks (EGAT). Unlike conventional graph attention networks that primarily focus on node features while overlooking crucial edge information, the proposed EGAT approach effectively captures both node and edge interactions, thereby improving learning efficiency. We also introduce the multi-head attention mechanism and residual connections to improve the expression ability of the model and to avoid over-smoothing, respectively. Moreover, we solve the constrained optimization problem through primal-dual training and adopt a policy gradient scheme to obtain the AP-UE association. Simulations show that the proposed EGAT approach achieves superior UE rate performance with very low inference time. Notably, the proposed EGAT demonstrates stronger generalization capability across network sizes compared to conventional graph neural networks, thanks to its edge-attention mechanism.
Förderung
- National Natural Science Foundation of China
- Jiangsu Provincial Key Research and Development Program
- SRIBD Innovation Fund
- Jiangsu Major Project on Fundamental Research
- Gusu Innovation Project
- High-Tech District of Suzhou City
Zitieren
APA
Liu, X., Zhao, J., Shen, F., Yang, K., & Sun, S. (2026). Resource Allocation in Fronthaul-Constrained Cell-Free Networks Using Edge-Graph Attention Networks. IEEE Transactions on Wireless Communications, 25, 13305-13318. https://doi.org/10.1109/twc.2026.3669944
Vancouver
Liu X, Zhao J, Shen F, Yang K, Sun S. Resource Allocation in Fronthaul-Constrained Cell-Free Networks Using Edge-Graph Attention Networks. IEEE Transactions on Wireless Communications. 2026;25:13305-13318. doi:10.1109/twc.2026.3669944
BibTeX
@article{Liu2026,
author = {Liu, Xiaoyu and Zhao, Jian and Shen, Furao and Yang, Kun and Sun, Sumei},
title = {Resource Allocation in Fronthaul-Constrained Cell-Free Networks Using Edge-Graph Attention Networks},
journal = {IEEE Transactions on Wireless Communications},
year = {2026},
volume = {25},
pages = {13305--13318},
doi = {10.1109/twc.2026.3669944}
}
EUVIMED