Zero attracting recursive least square algorithms
Zero attracting recursive least square algorithms
The $l_1$-norm sparsity constraint is a widely used technique for constructing sparse models. In this contribution, two zero-attracting recursive least squares algorithms, referred to as ZA-RLS-I and ZA-RLS-II, are derived by employing the $l_1$-norm of parameter vector constraint to facilitate the model sparsity. In order to achieve a closed-form solution, the $l_1$-norm of the parameter vector is approximated by an adaptively weighted $l_2$-norm, in which the weighting factors are set as the inversion of the associated $l_1$-norm of parameter estimates that are readily available in the adaptive learning environment. ZA-RLS-II is computationally more efficient than ZA-RLS-I by exploiting the known results from linear algebra as well as the sparsity of the system. The proposed algorithms are proven to converge, and adaptive sparse channel estimation is used to demonstrate the effectiveness of the proposed approach.
213-221
Hong, Xia
e6551bb3-fbc0-4990-935e-43b706d8c679
Gao, Junbin
a3dcab84-9675-402c-a19e-d41ea9973f3a
Chen, Sheng
9310a111-f79a-48b8-98c7-383ca93cbb80
January 2017
Hong, Xia
e6551bb3-fbc0-4990-935e-43b706d8c679
Gao, Junbin
a3dcab84-9675-402c-a19e-d41ea9973f3a
Chen, Sheng
9310a111-f79a-48b8-98c7-383ca93cbb80
Hong, Xia, Gao, Junbin and Chen, Sheng
(2017)
Zero attracting recursive least square algorithms.
IEEE Transactions on Vehicular Technology, 66 (1), .
(doi:10.1109/TVT.2016.2533664).
Abstract
The $l_1$-norm sparsity constraint is a widely used technique for constructing sparse models. In this contribution, two zero-attracting recursive least squares algorithms, referred to as ZA-RLS-I and ZA-RLS-II, are derived by employing the $l_1$-norm of parameter vector constraint to facilitate the model sparsity. In order to achieve a closed-form solution, the $l_1$-norm of the parameter vector is approximated by an adaptively weighted $l_2$-norm, in which the weighting factors are set as the inversion of the associated $l_1$-norm of parameter estimates that are readily available in the adaptive learning environment. ZA-RLS-II is computationally more efficient than ZA-RLS-I by exploiting the known results from linear algebra as well as the sparsity of the system. The proposed algorithms are proven to converge, and adaptive sparse channel estimation is used to demonstrate the effectiveness of the proposed approach.
Text
TVT2017-Jan.pdf
- Version of Record
Restricted to Repository staff only
Request a copy
Text
TVT-2015-01004.pdf
- Accepted Manuscript
More information
Accepted/In Press date: 20 February 2016
e-pub ahead of print date: 24 February 2016
Published date: January 2017
Organisations:
Southampton Wireless Group
Identifiers
Local EPrints ID: 404853
URI: http://eprints.soton.ac.uk/id/eprint/404853
ISSN: 0018-9545
PURE UUID: 3fb882aa-ebe7-4625-a094-2c10679bf210
Catalogue record
Date deposited: 24 Jan 2017 13:11
Last modified: 15 Mar 2024 04:17
Export record
Altmetrics
Contributors
Author:
Xia Hong
Author:
Junbin Gao
Author:
Sheng Chen
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics