Detection and localization of sound sources in an ocean environment can be achieved with a distributed array of passive acoustic sensors. Utilizing an array of autonomous littoral gliders, which offer long-term and quiet operation, and vector sensors, which measure both acoustic pressure and particle velocity, the array's localization performance can be improved. However, vector sensors are susceptible to errors induced by acoustic noise, and autonomous gliders as a sensor platform introduce positional errors. Through both simulations and at-sea data, the localization performance of four processing methods are evaluated under various noisy conditions. In both simulated and at-sea data results, a new cross-coherent method outperforms traditional methods by mitigating the effects of acoustic noise, provided sufficient positional accuracy of the array elements.

Underwater acoustic pressure transducers measure pressure fluctuations, a scalar parameter of the acoustic field. Acoustic vector sensors contain an omnidirectional pressure transducer (omni) and also bi- or tri-axial sensing elements that respond to either the particle velocity or pressure gradient of the acoustic field; which are vector quantities. The amplitude of the signal output of each directional channel of a vector sensor is proportional to the orientation relative to the direction of acoustic pressure propagation. The ratio of the signal amplitudes between two directional channels and the cross-spectra between the vector sensor omni and directional channels enable one to estimate the bearing to the source from a single point measurement. In order to accurately estimate the bearing across the usable frequency band of the vector sensor, the complex sensitivities of the omni and directional channels must be known. Since there is no standard directional reference transducer for a comparative calibration, the calibration must be performed in an acoustic field with a known relationship between the acoustic pressure and the acoustic particle velocity. Free-field calibrations are advantageous because this relationship is known for both planar and spherical wave fronts. However, reflections from waveguide boundaries present a practical limitation for free-field calibrations, especially at low frequencies. An alternative approach is to perform calibration measurements in a standing-wave field, where the relationship between pressure and particle velocity is also known. The calibration facility described in this thesis is composed of a laboratory-based, vertically-oriented, water-filled, elastic-walled waveguide with a piston velocity source at the bottom end and a pressure release boundary condition at the air/water interface at the top end. Some of the challenges of calibrating vector sensors in such an apparatus are discussed, including designing the waveguide to mitigate dispersion, mechanically isolating the apparatus from floor vibrations, understanding the impact of waveguide structural resonances on the acoustic field, and developing the calibration algorithms. Data from waveguide characterization experiments and calibration measurements are presented along with engineering drawings and calibration software. / text

In the field of applied electromagnetics, there is always a need to create new methods for electrical characterization of materials, systems, devices, etc. Many applications need small and/or inexpensive equipment in performing these characterizations. The current method for making measurements of electrical properties at frequencies above 300 MHz, the transmission/reflection method, has severe limitations in these areas due large size and high price of the necessary equipment for making them. Therefore, presented herein is the conceptualization, design and analysis of a complex reflection coefficient detector which is relatively small, lightweight, and inexpensive. A reflection coefficient detector is a device designed to isolate and compare a driving signal against a reflected signal. The reflection of the second signal is caused by a mismatch between the device's output impedance and a load's input impedance. By comparing the driving, or transmitted, signal and the reflected signal, the reflection coefficient at the boundary can be calculated. This coefficient can be used to calculate a load's input impedance, or a material's permittivity when combined with an attached probe's characteristics. The reflection coefficient detector presented is built using microstrip and surface mount components. This makes the device comparably cheap. Its design is based upon five Wilkinson Power Dividers which lends itself to be scaled down for implementation in on-chip, and other micro- and nano- scale systems. The accuracy and functionality of the device will be demonstrated through the use of S-Parameters measurements and CAD simulations. Through this, it will be shown that the device is a practical form of making measurements in applications which are otherwise restricted to certain limitations. In closing, applications, alternative designs and future advancements of the complex reflection coefficient detector will be discussed.

De nos jours, il existe de nombreuses antennes de radiogoniométrie UHF large bande ou multibandes, néanmoins très peu d’entre elles permettent une couverture angulaire 3D. A notre connaissance, la première antenne de radiogoniométrie 3D fût étudiée dans les années 1960, par une équipe de l’université du Michigan. Composée de 17 capteurs positionnés sur une surface hémisphérique, sa taille et son nombre d’éléments en font un dispositif encombrant et complexe à utiliser. De récentes études ont proposé une autre approche basée sur la mesure multicomposante du champ électromagnétique, permettant de réduire la taille des antennes et le nombre d’éléments tout en conservant une couverture angulaire 3D. Cependant, à ce jours, seul des systèmes HF (3MHz-30MHz) ou bande étroite ont été abordés. Cette thèse porte donc sur l’étude et le développement d’une antenne vectorielle UHF multibande appliquée à la radiogoniométrie 3D pour des ondes transverses magnétiques. Tout d’abord, deux techniques de goniométrie adaptées à cette approche sont confrontées : une nouvelle technique basée sur la décomposition en harmonique sphérique du rayonnement de l’antenne qui permet de recomposer le champ électromagnétique reçu à partir d’échantillons mesurés et un algorithme bien connu, MUSIC. Une méthodologie de conception est proposée, en identifiant les critères physiques des antennes vectorielles qui influent sur leurs performances à savoir la précision d’estimation, la sensibilité, le nombre d’éléments et l’encombrement. Cette méthode est utilisée pour développer et réaliser une première antenne vectorielle monobande. La caractérisation de cette antenne réaliste permet d’écarter la première technique de traitement dont les performances sont trop sensibles aux perturbations de rayonnement. Une antenne vectorielle bibande compacte, d’un rayon de λ/4 et d’une hauteur de λ/5.5 à la fréquence la plus basse, composée de seulement six éléments rayonnants couvrant chacun les bandes de fréquences GSM [890MHz-960MHz] et [1710MHZ-1880MHz] est ensuite développée en se basant sur cette méthode de conception. Les capteurs électriques et magnétiques constituant l’antenne sont étudiés séparément puis assemblés selon une répartition spatiale planaire pour restreindre l’encombrement. Les structures rayonnantes sont communes pour les deux bandes de fréquences ce qui permet réduire le nombre d’éléments ainsi que les éventuelles perturbations de rayonnement. Après la caractérisation de l’antenne bibande au travers de simulations numériques, un prototype est réalisé et ses performances d’estimation sont mesurées en chambre anéchoïque afin de valider l’approche par simulation. La sensibilité obtenue est de -110dBW/m² (85μV/m) pour une précision de 5° RMS. Enfin l’étude est élargie au cas général d’antennes multibandes en illustrant le processus d’extension de la couverture fréquentielle par l’ajout d’une troisième bande, [400MHz-430MHz]. Six nouveaux éléments sont donc développés et intégrés aux capteurs GSM existants afin d’obtenir une antenne tribande d’un rayon de λ/3.2 et d’une hauteur de λ/12.5 à 400MHz. Malgré une légère augmentation de l’erreur d’estimation, causée par la présence de ces nouveaux éléments, la caractérisation de cette nouvelle antenne tribande montre de bonnes performances d’estimation avec une sensibilité de -105dBW/m² (155μV/m) pour une précision de 5° RMS. / Nowadays, a lot of wideband or multiband direction finding antennas operating in the UHF band exist. Nevertheless, only few of them allow to estimate the direction of arrival in the full 3D space. At the author’s knowledge, the first 3D direction finding antenna was studied in the 1960s, at the University of Michigan. Composed of 17 sensors, located on a large hemispherical surface, this antenna is bulky and complex to use. Recently, some studies have proposed another approach based on the multicomponent measurement of the electromagnetic field that allows to decrease the antennas size and the number of radiating elements without reducing the 3D angular coverage. However, only HF (3-30MHz) or narrowband systems have been reported. The objective of this PhD is to study and to develop an UHF multiband vector sensor applied to the estimation of the direction of arrival of transverse magnetic waves in the full 3D space. Firstly, two signal processing techniques adapted to this approach are compared : a new technique based on the spherical harmonic decomposition of the antenna radiation which allows to recompose the received electromagnetic field from the measured samples and a well-known high resolution algorithm called MUSIC. A design methodology allowing to identify the physical criteria of vector sensors related to the antenna performances such as the estimation accuracy, the sensitivity, the number of elements and the antenna size is proposed. This method is used for developing and designing a first single-band vector sensor. The results obtained from numerical simulations allow to rule out the first signal processing technique which is too sensitive to the radiation perturbations. Then, a compact dual-band vector sensor operating in the GSM frequency band, [890MHz-960MHz] and [1710MHZ-1880MHz], is developed by using the same design methodology. The antenna size is λ/4 in radius and λ/5.5 in height at the lowest frequency. The electric and magnetic elements which compose the vector sensors are designed separately and then combined according to a planar spatial distribution to retain a compact antenna size. The same radiating structures are used for operating in the two frequency bands in order to reduce the number of elements and the eventual radiation perturbations. After the performances assessment through numerical simulations in each band, a prototype is manufactured and its estimation performances are measured for a validation purpose. The sensitivity is -110dBW.m−2 (85μV.m−1) for a 5◦ RMS angular accuracy. Finally, the study is extended to the general case of multiband antennas by adding a third band, [400MHz-430MHz]. New elements are developed and incorporated into the dual-band GSM sensors to obtain a tri-band vector sensor. The size of this new antenna is λ/3.2 in radius and λ/12.5 in height at 400MHz. Despite a slight increase of the angular errors in the estimation of the direction of arrival caused by the presence of the new antenna elements, the characterization of the tri-band sensor performances by simulation show a good accuracy with a sensitivity valued at -105dBW.m−2 (155μV.m−1) for a 5◦ RMS angular accuracy.

The combination of a video camera with an acoustic vector sensor (AVS) opens new possibilities in environment awareness applications. The goal of this thesis is the design of an algorithm for detection and tracking of low-flying aircraft using a multi-coordinate VAUDEO system. A commercial webcam placed in line with an AVS in a ground array are used to record real low-flying aircraft data at Teuge international airport. Each frame, the algorithm analyzes a matrix of three orthogonal acoustic particle velocity signals and one acoustic pressure signal using the Singular Value Decomposition to estimate the Direction of Arrival, DoA of propeller aircraft sound. The DoA data is then applied to a Kalman filter and its output is used later on to narrow the region of video frame processed. Background subtraction is applied followed by a Gaussian-weighted intensity mask to assign high priority to moving objects which are closer to the sound source estimated position. The output is applied to another Kalman filter to improve the accuracy of the aircraft location estimation. The performance evaluation of the algorithm proved that it is comparable to the performances of state-of-the-art video alone based algorithms. In conclusion, the combination of video and directional audio increases the accuracy of propeller aircraft detection and tracking comparing to reported previous work using audio alone. / +593 980826278