Variational Learning-Based Channel Estimators for Uplink Massive MIMO–OTFS Systems
IEEE Transactions on Wireless Communications · 2026 · 25 · 19384-19401
DOI: 10.1109/twc.2026.3709795 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Wireless Communications, 2026. Verfasst von Mudasir Ahmad Sheikh und 2 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Gefördert von: Ministry of Electronics and Information Technology (MeitY), Government of India, under the “6G Research and Standardization Project.".
Abstract (Originalsprache)
We consider a massive multiple-input multiple-output orthogonal time-frequency space system with fractional delay/Doppler, which induces block and burst sparsity respectively, along the delay-Doppler (DD) and angular dimension of the delay-Doppler-angular (DDA) domain channel. This leads to structured two-dimensional (2D) burst-block sparsity in the DDA channel. We capture this 2D sparsity structure by constructing a novel user-coupled Gaussian mixture prior. We then use this 2D prior to develop a coupled prior variational Bayesian learning (CPVBL) estimator for DDA channel, which dynamically learns the coupling probabilities of neighboring channel entries in the burst. This flexibility enables our prior to capture variable burst sizes, and effectively adapt to varying burst sparsity patterns. The CPVBL has a high complexity due to high-dimensional covariance matrix inversion. We then design a low-complexity fast CPVBL (F-CPVBL) algorithm, which updates parameters of the significant channel entries alone. We numerically show that the CPVBL and F-CPVBL algorithms vastly outperform their existing state-of-the-art counterparts. Crucially, F-CPVBL algorithm yields this performance improvement by maintaining a similar runtime as that of the existing algorithms.
Förderung
- Ministry of Electronics and Information Technology (MeitY), Government of India, under the “6G Research and Standardization Project."
Zitieren
APA
Sheikh, M. A., Choudhary, B., & Budhiraja, R. (2026). Variational Learning-Based Channel Estimators for Uplink Massive MIMO–OTFS Systems. IEEE Transactions on Wireless Communications, 25, 19384-19401. https://doi.org/10.1109/twc.2026.3709795
Vancouver
Sheikh MA, Choudhary B, Budhiraja R. Variational Learning-Based Channel Estimators for Uplink Massive MIMO–OTFS Systems. IEEE Transactions on Wireless Communications. 2026;25:19384-19401. doi:10.1109/twc.2026.3709795
BibTeX
@article{Sheikh2026,
author = {Sheikh, Mudasir Ahmad and Choudhary, Bhuvanesh and Budhiraja, Rohit},
title = {Variational Learning-Based Channel Estimators for Uplink Massive MIMO–OTFS Systems},
journal = {IEEE Transactions on Wireless Communications},
year = {2026},
volume = {25},
pages = {19384--19401},
doi = {10.1109/twc.2026.3709795}
}
EUVIMED