Hal will be stopped for maintenance from friday on june 10 at 4pm until monday june 13 at 9am. More information
Skip to Main content Skip to Navigation
Journal articles

Blind identification of multiuser nonlinear channels using tensor decomposition and precoding

Carlos Alexandre Rolim Fernandes 1, 2 Gérard Favier 1 João Mota 2
1 Laboratoire d'Informatique, Signaux, et Systèmes de Sophia-Antipolis (I3S) / Equipe SIGNAL
Laboratoire I3S - SIS - Signal, Images et Systèmes
2 UFC - GTEL - Grupo de Pesquisa em Telecomunicações Sem Fio
Abstract : This paper presents two blind identification methods for nonlinear memoryless channels in multiuser communication systems. These methods are based on the parallel factor (PARAFAC) decomposition of a tensor composed of channel output covariances. Such a decomposition is possible owing to a new precoding scheme developed for phase-shift keying (PSK) signals modeled as Markov chains. Some conditions on the transition probability matrices (TPM) of the Markov chains are established to introduce temporal correlation and satisfy statistical correlation constraints inducing the PARAFAC decomposition of the considered tensor. The proposed blind channel estimation algorithms are evaluated by means of computer simulations
keyword : Blind nonlinear channel identification Markov chain Multiuser channel PARAFAC decomposition Volterra model
Document type :
Journal articles
Complete list of metadata
https://hal.archives-ouvertes.fr/hal-00417569
Contributor : Thomas Bouilloc Connect in order to contact the contributor
Submitted on : Wednesday, September 16, 2009 - 11:15:35 AM
Last modification on : Tuesday, December 7, 2021 - 4:10:10 PM

Identifiers

Collections

CNRS | I3S | UNIV-COTEDAZUR

Citation

Carlos Alexandre Rolim Fernandes, Gérard Favier, João Mota. Blind identification of multiuser nonlinear channels using tensor decomposition and precoding. Signal Processing, Elsevier, 2009, 89 (12), pp.2644-2656. ⟨10.1016/j.sigpro.2009.05.012⟩. ⟨hal-00417569⟩

Export

BibTeX TEI DC DCterms EndNote Datacite

Share

Metrics

Record views

61