Compressive link acquisition in multiuser communications

Xiao Li, Andrea Rueetschi, Anna Scaglione, Yonina C. Eldar

Research output: Contribution to journalArticle

13 Citations (Scopus)

Abstract

An important sensing operation is to detect the presence of specific signals with unknown transmission parameters. This task, referred to as 'link acquisition,' is typically a sequential search over the transmitted signal space. Recently, the use of sparsity in similar estimation or detection problems has received considerable attention. These works typically focus on the benefits of compressed sensing, but not generally on the cost brought by sparse recovery. Our goal is to examine the tradeoff in complexity and performance when using sparse recovery with compressed or uncompressed samples. To do so, we propose a compressive sparsity aware (CSA) acquisition scheme, where a compressive multichannel sampling (CMS) front-end is followed by a sparsity regularized likelihood ratio test (SR-LRT) module. The CSA scheme borrows insights from the models studied in sub-Nyquist sampling and finite rate of innovation (FRI) signals. We further optimize the CMS front-end by maximizing the average Kullback-Leibler distance of all the hypotheses in the SR-LRT. We compare the CSA scheme vis-à-vis other popular alternatives in terms of performance and complexity. Simulations suggest that one can use the CSA scheme to scale down the implementation cost with greater flexibility than other alternatives. However, we find that they both have overall complexities that scale linearly with the search space. Furthermore, it is shown that compressive measurements used in the SR-LRT lead to a performance loss when noise prevails, while providing better performance in spite of the compression when noise is mild.

Original languageEnglish (US)
Article number6497667
Pages (from-to)3229-3245
Number of pages17
JournalIEEE Transactions on Signal Processing
Volume61
Issue number12
DOIs
StatePublished - 2013
Externally publishedYes

Fingerprint

Sampling
Communication
Recovery
Compressed sensing
Costs
Innovation

Keywords

  • Compressed sensing
  • detection and estimation
  • Kullback-Leibler distance
  • multiuser communications

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing

Cite this

Compressive link acquisition in multiuser communications. / Li, Xiao; Rueetschi, Andrea; Scaglione, Anna; Eldar, Yonina C.

In: IEEE Transactions on Signal Processing, Vol. 61, No. 12, 6497667, 2013, p. 3229-3245.

Research output: Contribution to journalArticle

Li, Xiao ; Rueetschi, Andrea ; Scaglione, Anna ; Eldar, Yonina C. / Compressive link acquisition in multiuser communications. In: IEEE Transactions on Signal Processing. 2013 ; Vol. 61, No. 12. pp. 3229-3245.
@article{45d9c918988b4a20aff999c00f0461f3,
title = "Compressive link acquisition in multiuser communications",
abstract = "An important sensing operation is to detect the presence of specific signals with unknown transmission parameters. This task, referred to as 'link acquisition,' is typically a sequential search over the transmitted signal space. Recently, the use of sparsity in similar estimation or detection problems has received considerable attention. These works typically focus on the benefits of compressed sensing, but not generally on the cost brought by sparse recovery. Our goal is to examine the tradeoff in complexity and performance when using sparse recovery with compressed or uncompressed samples. To do so, we propose a compressive sparsity aware (CSA) acquisition scheme, where a compressive multichannel sampling (CMS) front-end is followed by a sparsity regularized likelihood ratio test (SR-LRT) module. The CSA scheme borrows insights from the models studied in sub-Nyquist sampling and finite rate of innovation (FRI) signals. We further optimize the CMS front-end by maximizing the average Kullback-Leibler distance of all the hypotheses in the SR-LRT. We compare the CSA scheme vis-{\`a}-vis other popular alternatives in terms of performance and complexity. Simulations suggest that one can use the CSA scheme to scale down the implementation cost with greater flexibility than other alternatives. However, we find that they both have overall complexities that scale linearly with the search space. Furthermore, it is shown that compressive measurements used in the SR-LRT lead to a performance loss when noise prevails, while providing better performance in spite of the compression when noise is mild.",
keywords = "Compressed sensing, detection and estimation, Kullback-Leibler distance, multiuser communications",
author = "Xiao Li and Andrea Rueetschi and Anna Scaglione and Eldar, {Yonina C.}",
year = "2013",
doi = "10.1109/TSP.2013.2258014",
language = "English (US)",
volume = "61",
pages = "3229--3245",
journal = "IEEE Transactions on Signal Processing",
issn = "1053-587X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "12",

}

TY - JOUR

T1 - Compressive link acquisition in multiuser communications

AU - Li, Xiao

AU - Rueetschi, Andrea

AU - Scaglione, Anna

AU - Eldar, Yonina C.

PY - 2013

Y1 - 2013

N2 - An important sensing operation is to detect the presence of specific signals with unknown transmission parameters. This task, referred to as 'link acquisition,' is typically a sequential search over the transmitted signal space. Recently, the use of sparsity in similar estimation or detection problems has received considerable attention. These works typically focus on the benefits of compressed sensing, but not generally on the cost brought by sparse recovery. Our goal is to examine the tradeoff in complexity and performance when using sparse recovery with compressed or uncompressed samples. To do so, we propose a compressive sparsity aware (CSA) acquisition scheme, where a compressive multichannel sampling (CMS) front-end is followed by a sparsity regularized likelihood ratio test (SR-LRT) module. The CSA scheme borrows insights from the models studied in sub-Nyquist sampling and finite rate of innovation (FRI) signals. We further optimize the CMS front-end by maximizing the average Kullback-Leibler distance of all the hypotheses in the SR-LRT. We compare the CSA scheme vis-à-vis other popular alternatives in terms of performance and complexity. Simulations suggest that one can use the CSA scheme to scale down the implementation cost with greater flexibility than other alternatives. However, we find that they both have overall complexities that scale linearly with the search space. Furthermore, it is shown that compressive measurements used in the SR-LRT lead to a performance loss when noise prevails, while providing better performance in spite of the compression when noise is mild.

AB - An important sensing operation is to detect the presence of specific signals with unknown transmission parameters. This task, referred to as 'link acquisition,' is typically a sequential search over the transmitted signal space. Recently, the use of sparsity in similar estimation or detection problems has received considerable attention. These works typically focus on the benefits of compressed sensing, but not generally on the cost brought by sparse recovery. Our goal is to examine the tradeoff in complexity and performance when using sparse recovery with compressed or uncompressed samples. To do so, we propose a compressive sparsity aware (CSA) acquisition scheme, where a compressive multichannel sampling (CMS) front-end is followed by a sparsity regularized likelihood ratio test (SR-LRT) module. The CSA scheme borrows insights from the models studied in sub-Nyquist sampling and finite rate of innovation (FRI) signals. We further optimize the CMS front-end by maximizing the average Kullback-Leibler distance of all the hypotheses in the SR-LRT. We compare the CSA scheme vis-à-vis other popular alternatives in terms of performance and complexity. Simulations suggest that one can use the CSA scheme to scale down the implementation cost with greater flexibility than other alternatives. However, we find that they both have overall complexities that scale linearly with the search space. Furthermore, it is shown that compressive measurements used in the SR-LRT lead to a performance loss when noise prevails, while providing better performance in spite of the compression when noise is mild.

KW - Compressed sensing

KW - detection and estimation

KW - Kullback-Leibler distance

KW - multiuser communications

UR - http://www.scopus.com/inward/record.url?scp=84878480901&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84878480901&partnerID=8YFLogxK

U2 - 10.1109/TSP.2013.2258014

DO - 10.1109/TSP.2013.2258014

M3 - Article

AN - SCOPUS:84878480901

VL - 61

SP - 3229

EP - 3245

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 12

M1 - 6497667

ER -