Compressed CSI Acquisition in FDD Massive MIMO

How Much Training is Needed?

Juei Chin Shen, Jun Zhang, Emad Alsusa, Khaled Letaief

Research output: Contribution to journalArticle

36 Citations (Scopus)

Abstract

Massive multiple-input-multiple-output (MIMO) is a promising technique for providing unprecedented spectral efficiency. However, it has been well recognized that the excessive training overhead required for obtaining the channel side information is a major handicap in frequency-division duplexing (FDD) massive MIMO. Several attempts have been made to reduce this training overhead by exploiting the sparsity structures of massive MIMO channels. So far, however, there has been little discussion about how to exploit the partial support information of these channels to achieve further overhead reductions. Such information, which is a set of indices of the significant elements of a channel vector, can be acquired in advance and hence is an important option to explore. In this paper, we examine the impact on the required training overhead when this information is applied within a weighted l1 minimization framework, and analytically show that a sharp estimate of the reduced overhead size can be successfully obtained. Furthermore, we examine how the accuracy of the partial support information impacts the achievable overhead reduction. Numerical results for a wide range of sparsity and partial support information reliability levels are presented to quantify our findings and main conclusions.

Original languageEnglish
Article number7420744
Pages (from-to)4145-4156
Number of pages12
JournalIEEE Transactions on Wireless Communications
Volume15
Issue number6
DOIs
Publication statusPublished - 1 Jun 2016
Externally publishedYes

Fingerprint

Multiple-input multiple-output (MIMO)
Division
Sparsity
Partial
Side Information
Spectral Efficiency
Quantify
Training
Acquisition
Numerical Results
Estimate
Range of data

Keywords

  • channel estimation
  • compressed sensing
  • FDD
  • Massive MIMO
  • partial support information
  • phase transition
  • pilot contamination
  • weighted l minimization

ASJC Scopus subject areas

  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Compressed CSI Acquisition in FDD Massive MIMO : How Much Training is Needed? / Shen, Juei Chin; Zhang, Jun; Alsusa, Emad; Letaief, Khaled.

In: IEEE Transactions on Wireless Communications, Vol. 15, No. 6, 7420744, 01.06.2016, p. 4145-4156.

Research output: Contribution to journalArticle

@article{ed7905d344004e18bc83c8f010589e1f,
title = "Compressed CSI Acquisition in FDD Massive MIMO: How Much Training is Needed?",
abstract = "Massive multiple-input-multiple-output (MIMO) is a promising technique for providing unprecedented spectral efficiency. However, it has been well recognized that the excessive training overhead required for obtaining the channel side information is a major handicap in frequency-division duplexing (FDD) massive MIMO. Several attempts have been made to reduce this training overhead by exploiting the sparsity structures of massive MIMO channels. So far, however, there has been little discussion about how to exploit the partial support information of these channels to achieve further overhead reductions. Such information, which is a set of indices of the significant elements of a channel vector, can be acquired in advance and hence is an important option to explore. In this paper, we examine the impact on the required training overhead when this information is applied within a weighted l1 minimization framework, and analytically show that a sharp estimate of the reduced overhead size can be successfully obtained. Furthermore, we examine how the accuracy of the partial support information impacts the achievable overhead reduction. Numerical results for a wide range of sparsity and partial support information reliability levels are presented to quantify our findings and main conclusions.",
keywords = "channel estimation, compressed sensing, FDD, Massive MIMO, partial support information, phase transition, pilot contamination, weighted l minimization",
author = "Shen, {Juei Chin} and Jun Zhang and Emad Alsusa and Khaled Letaief",
year = "2016",
month = "6",
day = "1",
doi = "10.1109/TWC.2016.2535310",
language = "English",
volume = "15",
pages = "4145--4156",
journal = "IEEE Transactions on Wireless Communications",
issn = "1536-1276",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "6",

}

TY - JOUR

T1 - Compressed CSI Acquisition in FDD Massive MIMO

T2 - How Much Training is Needed?

AU - Shen, Juei Chin

AU - Zhang, Jun

AU - Alsusa, Emad

AU - Letaief, Khaled

PY - 2016/6/1

Y1 - 2016/6/1

N2 - Massive multiple-input-multiple-output (MIMO) is a promising technique for providing unprecedented spectral efficiency. However, it has been well recognized that the excessive training overhead required for obtaining the channel side information is a major handicap in frequency-division duplexing (FDD) massive MIMO. Several attempts have been made to reduce this training overhead by exploiting the sparsity structures of massive MIMO channels. So far, however, there has been little discussion about how to exploit the partial support information of these channels to achieve further overhead reductions. Such information, which is a set of indices of the significant elements of a channel vector, can be acquired in advance and hence is an important option to explore. In this paper, we examine the impact on the required training overhead when this information is applied within a weighted l1 minimization framework, and analytically show that a sharp estimate of the reduced overhead size can be successfully obtained. Furthermore, we examine how the accuracy of the partial support information impacts the achievable overhead reduction. Numerical results for a wide range of sparsity and partial support information reliability levels are presented to quantify our findings and main conclusions.

AB - Massive multiple-input-multiple-output (MIMO) is a promising technique for providing unprecedented spectral efficiency. However, it has been well recognized that the excessive training overhead required for obtaining the channel side information is a major handicap in frequency-division duplexing (FDD) massive MIMO. Several attempts have been made to reduce this training overhead by exploiting the sparsity structures of massive MIMO channels. So far, however, there has been little discussion about how to exploit the partial support information of these channels to achieve further overhead reductions. Such information, which is a set of indices of the significant elements of a channel vector, can be acquired in advance and hence is an important option to explore. In this paper, we examine the impact on the required training overhead when this information is applied within a weighted l1 minimization framework, and analytically show that a sharp estimate of the reduced overhead size can be successfully obtained. Furthermore, we examine how the accuracy of the partial support information impacts the achievable overhead reduction. Numerical results for a wide range of sparsity and partial support information reliability levels are presented to quantify our findings and main conclusions.

KW - channel estimation

KW - compressed sensing

KW - FDD

KW - Massive MIMO

KW - partial support information

KW - phase transition

KW - pilot contamination

KW - weighted l minimization

UR - http://www.scopus.com/inward/record.url?scp=84976334291&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84976334291&partnerID=8YFLogxK

U2 - 10.1109/TWC.2016.2535310

DO - 10.1109/TWC.2016.2535310

M3 - Article

VL - 15

SP - 4145

EP - 4156

JO - IEEE Transactions on Wireless Communications

JF - IEEE Transactions on Wireless Communications

SN - 1536-1276

IS - 6

M1 - 7420744

ER -