Autoregressive model for high-resolution wavenumber estimation in a shallow water environment using a broadband source

Courtois, Florent Le; Bonnel, Julien

doi:10.1121/1.4869821

Cited by 26 publications

(5 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…They operate in the transform domain and require impulsive sources (less convenient than continuous wave and triangular pulses [12]) that has to traverse a significant range interval [20], or to be activated very far from the sensors [22]. These techniques are based on modal separability that i) is more pronounced at large ranges where, unfortunately, Signal-to-Noise Ratio (SNR) is rather low, and ii) require the use of non-linear processing techniques (masking [20] or warping [22]) that operate under restrictive conditions [22, (7), (8)] [20, (23), (24), (25)]. Also, they are not fully automatic as they involve some user-defined parameters.…”

Section: This Is Made Possible Thanks To the Use Of Eigen Vector Decomentioning

confidence: 99%

“…Also, they are not fully automatic as they involve some user-defined parameters. Computation requirements of [22], [25], [20] are also high as Fourier transforms or SVDs are repeatedly computed. These drawbacks affect the practicality of these mode extraction methods as well as their resolution, resulting in a limited ability to extract weakly excited modes [20].…”

Section: This Is Made Possible Thanks To the Use Of Eigen Vector Decomentioning

confidence: 99%

“…Accurate estimation of the model parameters is very critical to the performance of geo-acoustic inversion [19], [25], [22], and to source localization algorithms [5], [2], especially high resolution ones. Current techniques identify sampled modal functions (evaluated at the sensor depths) using the output of an array of hydrophones arranged in VLA (Vertical Linear Array) [18], [17], [8], [19], [20] or HLA (Horizontal Linear Array) [25], [2], [6] configurations. On one hand, HLAbased techniques are very restrictive about the source location.…”

mentioning

confidence: 99%

“…Transform-domain techniques [22], [25], [20] are search-based and their resolution is limited by the array size. Subspace techniques operate on (averaged) data arranged in rank deficient matrices.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Closed-Form Estimation of Normal Modes From a Partially Sampled Water Column

Gazzah¹,

Jesus

2020

IEEE J. Oceanic Eng.

View full text Add to dashboard Cite

The output of a vertical linear array is used to infer about the parameters of the normal mode model that describes acoustic propagation in a shallow water. Often, singular vector decomposition is performed on the data and estimates are obtained of the model parameters. Such subspace algorithms deliver the exact modal functions only if the array is covering the total water column. We prove that this very restrictive requirement can be relaxed if two hydrophone arrays are used to sense an array of monochromatic sources, with arrays covering sparsely and partially the water column. Estimates of both the modal functions and the wavenumbers are obtained in a fully-automatic and search-free manner, under no restrictive condition. This methods compares advantageously to both existing subspace techniques that require dense sampling of the full water column, and to transform-domain techniques that require impulsive sources. With two (eigen and singular) vector decompositions, the proposed technique has the complexity of a regular subspace algorithm. Index Terms Shallow water, normal modes, subspace algorithms, vertical linear arrays. I. INTRODUCTION Shallow water is a challenging propagation environment that requires the design of dedicated signal processing techniques in order to perform efficiently the different tasks of an underwater observation system: identification, localization, communication, or also inversion. Such techniques will not be efficient until they are based on a suitable propagation model and that the

show abstract

Section: This Is Made Possible Thanks To the Use Of Eigen Vector Decomentioning

confidence: 99%

Section: This Is Made Possible Thanks To the Use Of Eigen Vector Decomentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Closed-Form Estimation of Normal Modes From a Partially Sampled Water Column

Gazzah¹,

Jesus

2020

IEEE J. Oceanic Eng.

View full text Add to dashboard Cite

show abstract

“…Des méthodes à hautes résolutions (HR) sont souvent préférées pour surmonter ces limites (Marcos, 1998). Les modèles autorégressifs (Shang et al, 1988 ;Becker, Frisk, 2006 ;Philippe et al, 2008 ;Le Courtois, Bonnel, 2014a) et les méthodes en sousespace comme MUSIC, ESPRIT (Rajan, Bhatta, 1993) et les Matrix Pencil (Lu et al, 1998, ont montré leur intérêt pour l'estimation des k rm . Cependant, si ces méthodes améliorent la résolution, elles restent sujettes à une plus grande sensibilité au bruit.…”

Section: Introductionunclassified

Estimation modale large bande avec une petite antenne horizontale

Courtois¹,

Bonnel²

2016

Traitement du signal

Self Cite

View full text Add to dashboard Cite

L'estimation des nombres d'onde acoustiques présente un grand intérêt pour l'inversion environnementale en milieu petit fond. Avec une antenne horizontale, les nombres d'onde sont calculés par estimation spectrale dans la dimension spatiale. La représentation fréquencenombre d'onde (f −k) est obtenue en concaténant les spectres des nombres d'onde sur plusieurs fréquences. Le plan f − k est particulièrement utile pour la caractérisation des milieux dispersifs. Une méthode d'Acquisition Comprimée (CS pour Compressed Sensing) est proposée pour l'amélioration des performances de l'estimation spectrale spatiale. Le CS est adapté en contexte de propagation modale car le champ acoustique est décrit par un petit nombre de modes propagatifs. Cependant, aux plus hautes fréquences le nombre de modes augmente, la séparation des nombres d'onde devient plus difficile, en particulier lorsque l'antenne est petite. Cet article présente l'utilisation d'une méthode de filtrage particulaire (FP) pour séparer les nombres d'onde sur une large plage de fréquences. La méthode repose sur la physique dispersive, vraie pour tous les guides d'onde, pour suivre les courbes de dispersion avec les fréquences. L'utilisation successive du CS et du FP permet alors la représentation f − k pour des données ne respectant pas les dimensions usuelles pour l'analyse spectrale spatiale. La méthode est appliquée avec succès sur les mesures de l'antenne SHARK (32 capteurs) de la campagne SW06.ABSTRACT. Estimation of the acoustic wavenumbers presents great interests in shallow environments. They provide relevant material to infer the sediment parameters. Using a horizontal array, the wavenumbers are obtained with usual spectral estimation methods in the spatial dimension. The computation of the wavenumber spectra at several frequencies leads to a frequencywavenumber (f −k) diagram, which is appropriate for the characterization of dispersive propagations. In this paper, a Compressed Sensing (CS) method is proposed to improve the separation of the wavenumbers. The CS approach is particularly relevant since only few modes are propagating. However, at higher frequencies the number of propagating mode increases and the wavenumber separation becomes more difficult, especially when using short arrays. This paper introduces a wideband particle filtering (PF) algorithm for the wavenumber tracking. It takes advantage of the dispersion relation which is true in every waveguides. The consecutive use of CS and PF allows computing the f-k representation for array measurements that does not respect the usual requirements of array length. An application on the 32 sensor SHARK array of the SW06 campaign illustrates the whole methodology.

show abstract

Difference Factor of Vertical Beam Pattern for Shallow-Water Source Depth Discrimination

Zheng

Zhu

2021

Acoust Aust

View full text Add to dashboard Cite

Autoregressive model for high-resolution wavenumber estimation in a shallow water environment using a broadband source

Cited by 26 publications

References 9 publications

Closed-Form Estimation of Normal Modes From a Partially Sampled Water Column

Closed-Form Estimation of Normal Modes From a Partially Sampled Water Column

Estimation modale large bande avec une petite antenne horizontale

Difference Factor of Vertical Beam Pattern for Shallow-Water Source Depth Discrimination

Contact Info

Product

Resources

About