Passive tracking in underwater acoustic

Abstract

This paper provides a novel passive underwater acoustic method to track a moving object called source or target, with the following constraints: the sensors location is fixed and imposed, and classical array processing techniques cannot be applied. The method proposed has been successfully used to track a surface vessel (2D problem) or an underwater target (3D problem). All the results presented have been obtained with real signals. The localization of the target requires the estimation of propagation delays, that means the duration between the instant of the signal emission and its reception on each receiver. The cross-correlation function is a suitable tool when the target is motionless, but needs to be extended to the ambiguity function when it is moving. The signal to noise ratio, the uniformity of power spectral density and the integration time are determining factors for the accuracy of the localization. We show that a whitening method and a Doppler compensation are necessary, and we propose a way to eliminate the significant problem of the reflected signals. Furthermore, a new configuration of the receivers is proposed, based on the idea of coupling receivers whose distance is chosen from experiment derived results. Furthermore, the algorithm proposed is susceptible to parallel implementation, thereby facilitating real-time uses. Experimental results with real time domain data are presented and compared to trajectories obtained by an active method. RésuméCet article présente une nouvelle méthode d'écoute passive destinée à trajectographier un objet en mouvement appelé la "source" ou la "cible", avec les contraintes suivantes : le réseau de capteurs est fixe et imposé, et les méthodes classiques de traitement d'antenne ne peuvent pas s'appliquer. La méthode proposée ici a été utilisée avec succès pour trajectographier un bâtiment de surface (problème à deux dimensions) ou une cible sous-marine (problème à trois dimensions). Tous les résultats présentés ont été obtenus sur des signaux réels. Le calcul de la position de la cible nécessite l'estimation des temps de propagation, c'est-à-dire du temps écoulé entre l'instant d'émission du signal et sa réception sur chaque capteur. La fonction d'intercorrélation est un outil opportun lorsque la cible est immobile mais elle doit être étendue à la notion d'ambiguïté quand la source est en mouvement. Le rapport signal à bruit, l'uniformité de la densité spectrale de puissance et la durée d'intégration sont des facteurs déterminants pour la précision de la localisation. Nous montrons qu'une méthode de blanchiment et une compensation de l'effet Doppler sont nécessaires et nous proposons un moyen d'éliminer le problème des trajets multiples. Par ailleurs, une nouvelle configuration des capteurs est proposée, fondée sur l'idée d'associer des hydrophones séparés d'une distance déterminée de manière expérimentale. Des résultats de trajectographies sont présentés et comparés aux trajectoires obtenues par une méthode active.

Experimental Results of Passive Imaging of Moving Continuous Broadband Sound Sources Within a Sensor Field

“…In applications of sparse arrays with many hydrophones, [7] is the most similar publication in comparison to the work presented in this paper. In [7], moving broadband targets were localized by 16 hydrophones in a sparse array.…”

“…In [7], Doppler is found using the ambiguity function and is valid when acceleration is negliable. To show how the ambiguity function can be used to find the Doppler, and thereby the velocity and speed, Fig.…”

“…In [7], moving broadband targets were localized by 16 hydrophones in a sparse array. The article focused mainly on localization and used a time of arrival method for single targets.…”

Comparison between whitened generalized cross correlation and adaptive filter for time delay estimation with scattered arrays for passive positioning of moving targets in Baltic Sea shallow waters

“…This is of interest since we, in this work, have not been compensating for the Doppler shifts between the sensors connected to the large baseline and nor for the spreading of sound which gets more complicated for large baselines. In [1] Doppler compensation clearly improved the TDE and this has to be considered in future work on scattered arrays. Methods that consider the spreading of sound, such as Matched Field Processing, should be investigated.…”

“…Little work seems to be conducted about widely and randomly distributed sensors. Xerri, Cavassilas and Burloz [1] have though positioned moving broad band sources with sensors distributed about 400 m in average from each other at dept 250 m.…”