Skip to Main content Skip to Navigation

Méthodes à haute résolution pour antennes non uniformes

Abstract : In this thesis, we are interested in the signal processing methods for source localization in the case of nonuniform arrays. The work concerns mainly the high resolution methods which are particularly adapted to the linear nonuniform arrays because they reduce the effect of sidelobes. We show that Root-MUSIC can be directly applied in the case of the nonuniform arrays, particularly the arrays with missing sensors. Moreover, we show that the elimination of a certain number of sensors of a uniform array degrades only very slightly the performance thanks to the use of the high-resolution methods like Root-MUSIC. Furthermore, we propose a new method EM-ESPRIT that has the ability to deal with any type of nonuniform arrays. This iterative method is based on the two algorithms : Expectation-Maximization (EM) and ESPRIT. The study of performance of EM-ESPRIT shows that it presents better performance than most of the existing methods, in particular for high noise levels. In addition, two alternatives : Stochastic EM-ESPRIT and EM-Root-MUSIC are proposed in order to improve the convergence and consistance aspects of EM-ESPRIT. Finally, we consider an application related to the localization of aeroacoustic sources on the body of a car. The implementation of a method of the type Zero-Forcing MUSIC allowed to obtain good performance on simulated and real data, in particular for close spaced sources having different power.
Complete list of metadatas
Contributor : Karine El Rassi <>
Submitted on : Friday, January 15, 2010 - 10:46:17 AM
Last modification on : Monday, December 14, 2020 - 12:38:04 PM
Long-term archiving on: : Friday, June 18, 2010 - 12:58:01 AM


  • HAL Id : tel-00447488, version 1



Carine El Kassis. Méthodes à haute résolution pour antennes non uniformes. Traitement du signal et de l'image [eess.SP]. Université Paris Sud - Paris XI, 2009. Français. ⟨tel-00447488⟩



Record views


Files downloads