Loading…

Millimeter-Wave Channel Estimation Based on 2-D Beamspace MUSIC Method

Due to the spatial sparsity caused by severe propagation loss, mm-wave channel estimation can be performed by estimating the directions and gains of the paths that have significant power. In this paper, we apply the beamspace 2-D multiple signal classification (MUSIC) method to estimate the path dir...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on wireless communications 2017-08, Vol.16 (8), p.5384-5394
Main Authors: Guo, Ziyu, Wang, Xiaodong, Heng, Wei
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Due to the spatial sparsity caused by severe propagation loss, mm-wave channel estimation can be performed by estimating the directions and gains of the paths that have significant power. In this paper, we apply the beamspace 2-D multiple signal classification (MUSIC) method to estimate the path directions (the angles of departure and arrival) and use the least-squares method to estimate the path gains. Different from its element-space counterpart, the beamspace MUSIC method may exhibit spectrum ambiguity caused by the beamformers. In this paper, we therefore analyze the sufficient conditions on the beamformers under which the MUSIC spectrum has no ambiguity which also leads to the maximum number of resolvable path directions. Moreover, based on the uniform linear array with half-wavelength spacing, we show that the discrete Fourier transform beamformers, which are naturally analog and often employed in the mm-wave communication systems with hybrid precoding structure, can avoid the spectrum ambiguity and maximize the number of resolvable path directions. Simulation results demonstrate that the proposed 2-D beamspace MUSIC mm-wave channel estimator significantly outperforms existing estimators that are based on beam training and sparse recovery; and in the meantime, it requires much less training slots than these existing methods.
ISSN:1536-1276
1558-2248
DOI:10.1109/TWC.2017.2710049