par Determe, Jean-François 
;Bourdoux, André ;Horlin, François
Référence (30-31 May, 2013: Leuven, Belgium), 34rd WIC Symposium on Information Theory in the Benelux, 3rd Joint WIC/IEEE Symposium on Information Theory and Signal Processing in the Benelux, WIC
Publication Publié, 2013-05
;Bourdoux, André ;Horlin, François Référence (30-31 May, 2013: Leuven, Belgium), 34rd WIC Symposium on Information Theory in the Benelux, 3rd Joint WIC/IEEE Symposium on Information Theory and Signal Processing in the Benelux, WIC
Publication Publié, 2013-05
Abstract de conférence
| Résumé : | The ever-improving Orthogonal Matching Pursuit (OMP) algorithms have enabled one to deal with a widevariety of l_0-norm problems more and more efficiently. Among all of the applications of OMP, the identification ofactive frequency bandwidths within sparse spectrums has remained a challenge due to the reliability to be achieved.This particular application is generally associated with the so-called cognitive radios which aim at detecting availablebands within sparse multi-gigahertz wireless spectrums to transmit their own data streams. This paper presentsa novel OMP algorithm called EDA-OMPMMV exhibiting memory capabilities which prevent bad detectionsfrom occurring by averaging the results of successive iterations. This algorithm has been especially designed forthe Modulated Wideband Converter (MWC) and exhibits optimal performance for spectrums undergoing limitedmodifications over the time. |
