One-snapshot on-grid and off-grid channel estimation in uplink large-scale millimeter-wave MIMO using an exponential penalty function

Large-scale multiple-input multiple-output (MIMO) technology is well-recognized for enhancing throughput in modern communication systems. In these systems, the accurate estimation of the direction of arrival (DoA) and uplink channel estimation is essential to leverage spatial multiplexing before the...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Signal processing Ročník 239; s. 110261
Hlavní autoři: Nargesi, Mohammadreza, Olfat, Ali
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier B.V 01.02.2026
Témata:
Iterative shrinkage/thresholding algorithm
Lambert-W function
Massive MIMO
Millimeter-wave
Proximal gradient algorithm
Single measurement vector
Sparsity inducing priors
Sparsity inducing priors
Millimeter-wave
Iterative shrinkage/thresholding algorithm
Lambert-W function
Massive MIMO
Single measurement vector
Proximal gradient algorithm
ISSN:0165-1684
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Large-scale multiple-input multiple-output (MIMO) technology is well-recognized for enhancing throughput in modern communication systems. In these systems, the accurate estimation of the direction of arrival (DoA) and uplink channel estimation is essential to leverage spatial multiplexing before the precoding phase. This study focuses on DoA-based channel estimation in a time division duplex (TDD) massive MIMO communication system particularly in scenarios with a limited number of observations. The DoA estimation is conducted during the uplink phase using a single measurement vector (SMV). Both on-grid and off-grid formulations of the problem are considered. The on-grid scenario is formulated as a sparse estimation problem regularized by a non-smooth exponential function. To solve this, a proximal gradient algorithm (PGA) is utilized, incorporating a Lambert-W function in the pruning process. The algorithm designed for the on-grid formulation can be effectively applied to solve other sparse signal recovery problems. For the off-grid problem, an alternating algorithm is proposed, leveraging initial estimates from the on-grid solution to enhance accuracy and mitigate grid mismatch issues. Simulation results show that the proposed approach consistently outperforms representative algorithms from the compressive sensing literature, including convex relaxation techniques such as ℓ1-minimization via LASSO and ADMM, and greedy algorithms such as OMP and CoSaMP.
ISSN:0165-1684
DOI:10.1016/j.sigpro.2025.110261