Bayesian End-to-End Learning for FDD-Massive MIMO Physical-Layer Design
IEEE Transactions on Signal Processing · 2026 · 74 · 1124-1140
DOI: 10.1109/tsp.2026.3669707 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Signal Processing, 2026. Verfasst von Yiyi Gao und 2 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Gefördert von: Research Grants Council (RGC), Hong Kong, via the Areas of Excellence Scheme, National Foreign Expert Project under Project, Guangdong Provincial Project.
Abstract (Originalsprache)
This paper proposes a variational Bayesian learning framework for the end-to-end design of frequency-division duplex (FDD) massive Multiple Input Multiple Output (MIMO) communication systems. By factorizing the posterior distribution of involved variables, we effectively model various components of the system, including MIMO precoding and channel state information (CSI) at the transmitter (CSIT) recovery at the transmitter, along with channel estimation (CE), CSIT compression, and MIMO detection at the receiver. This factorization enhances adaptability and incorporates estimation errors during training, as each block outputs a probability distribution rather than deterministic values. Furthermore, we exploit the inherent sparsity of the MIMO channel using a two-layer hierarchical prior model to achieve efficient estimation. To address the intractability of the original problem, we propose a low-complexity unsupervised variational Bayesian training algorithm that does not require true channel realizations. Extensive simulations demonstrate that the proposed framework significantly outperforms existing methods while minimizing signaling overhead.
Förderung
- Research Grants Council (RGC), Hong Kong, via the Areas of Excellence Scheme
- National Foreign Expert Project under Project
- Guangdong Provincial Project
- Guangzhou Municipal Science and Technology Project
- Guangdong Provincial Key Laboratory of Integrated Communications, Sensing and Computation for Ubiquitous Internet of Things
- National Foreign Expert Project
Zitieren
APA
Gao, Y., Tsang, D. H. K., & Lau, V. K. N. (2026). Bayesian End-to-End Learning for FDD-Massive MIMO Physical-Layer Design. IEEE Transactions on Signal Processing, 74, 1124-1140. https://doi.org/10.1109/tsp.2026.3669707
Vancouver
Gao Y, Tsang DHK, Lau VKN. Bayesian End-to-End Learning for FDD-Massive MIMO Physical-Layer Design. IEEE Transactions on Signal Processing. 2026;74:1124-1140. doi:10.1109/tsp.2026.3669707
BibTeX
@article{Gao2026,
author = {Gao, Yiyi and Tsang, Danny H. K. and Lau, Vincent K. N.},
title = {Bayesian End-to-End Learning for FDD-Massive MIMO Physical-Layer Design},
journal = {IEEE Transactions on Signal Processing},
year = {2026},
volume = {74},
pages = {1124--1140},
doi = {10.1109/tsp.2026.3669707}
}
EUVIMED