Hybrid Spatial and Contextual Learning for Port Selection in Fluid Antenna Systems
IEEE Communications Letters · 2026 · 30 · 2455-2459
DOI: 10.1109/lcomm.2026.3709008 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Communications Letters, 2026. Verfasst von Wenkai Yan und 6 weiteren Autor:innen. Die Arbeit nennt 15 Literaturangaben. Gefördert von: Open Fund of Anhui Province Key Laboratory of Cyberspace Security Situation Awareness and Evaluation.
Abstract (Originalsprache)
This letter investigates fast antenna port selection in multiuser transmit-side multiple-input single-output (MISO) fluid antenna systems (Tx-MISO-FAS), where the combinatorial nature of the search space creates a fundamental tradeoff between optimality and latency. To address this challenge, we propose a hybrid spatial and contextual learning framework. Specifically, a one-dimensional convolutional neural network (1D-CNN) is employed to extract spatial features from instantaneous channel observations, while a frozen large language model (LLM) provides contextual guidance based on statistical descriptors. The two components are integrated through a residual connection to derive the final port selection decision. Numerical results demonstrate that the proposed method achieves sum-rate performance close to that of the brute-force search and is highly competitive with the greedy method. Moreover, the execution latency can be reduced to approximately 1/150 of that of the greedy method.
Förderung
- Open Fund of Anhui Province Key Laboratory of Cyberspace Security Situation Awareness and Evaluation
Zitieren
APA
Yan, W., Liang, K., Wang, X., Liu, J., Zheng, G., Wong, K., & Chae, C. (2026). Hybrid Spatial and Contextual Learning for Port Selection in Fluid Antenna Systems. IEEE Communications Letters, 30, 2455-2459. https://doi.org/10.1109/lcomm.2026.3709008
Vancouver
Yan W, Liang K, Wang X, Liu J, Zheng G, Wong K, et al. Hybrid Spatial and Contextual Learning for Port Selection in Fluid Antenna Systems. IEEE Communications Letters. 2026;30:2455-2459. doi:10.1109/lcomm.2026.3709008
BibTeX
@article{Yan2026,
author = {Yan, Wenkai and Liang, Kai and Wang, Xinyu and Liu, Jiaxin and Zheng, Gan and Wong, Kai-Kit and Chae, Chan-Byoung},
title = {Hybrid Spatial and Contextual Learning for Port Selection in Fluid Antenna Systems},
journal = {IEEE Communications Letters},
year = {2026},
volume = {30},
pages = {2455--2459},
doi = {10.1109/lcomm.2026.3709008}
}
EUVIMED