2 - Détection multiple par les valeurs propres de la matrice spectrale

Show full item record

Files in this item

PDF 002.PDF TEXTE.pdf 899.3Kb

Pour citer ce document :
URI: http://hdl.handle.net/2042/1598
Title: 2 - Détection multiple par les valeurs propres de la matrice spectrale
Author: TAS (I.); LATOMBE (C.)
Abstract: Critères de détection, résultats sur des simulations
Description: In this study we treat the problem of detecting from multidimensional data, the number of uncorrelated signais in passive array treatment as it is the case in underwater acoustics, array processing and seismology . We use four detection criteria. Some of them are known, like AIC and MDL criteria where direct Kullback's divergence is the information measure; we prolong them using the inverse Kullback's divergence. We also adapt a new criterion using the logarithm of the likelihood ratio that has a chi square distribution and we suggest a simplified threshold criterion that uses the eigenvalues of the spectral matrix of the data . We study and compare the performances of these criteria in realistic simulations . The first one is inspired by the problems of array processing and the second one by seismic problems. Finally we study the robustness of these criteria when the classical hypothesis of uncorrelated noises having equal variances is not fulfilled . Thus we outline some application limits of these criteria .
Subject: Détection signal; Matrice mathématique; Valeur propre; Détection multiple; Matrice spectrale
Publisher: GRETSI, Saint Martin d'Hères, France
Date: 1986

This item appears in the following Collection(s)

Show full item record





Advanced Search