In this study, a neural network(NN) based algorithm is proposed for real time multiple source tracking problem based on a previously reported work. The proposed algorithm namely modified neural network based multiple source tracking algorithm (MN-MUST) performs direction of arrival(DoA) estimation in three stages which are the detection, filtering and DoA estimation stages. The main contributions of this proposed system are: reducing the input size for the uncorrelated source case (reducing the training time) of NN system without degradation of accuracy and insertion of a nonlinear spatial filter to isolate each one of the sectors where sources are present, from the others. MN-MUST algorithm finds the targets correctly no matter whether the targets are located within the same angular sector or not. In addition as the number of targets exceeds the number of antenna elements the algorithm can still perform sufficiently well. Mutual coupling in array does not influence MN-MUST algorithm performance. iv MN-MUST algorithm is further improved for a cylindrical microstrip patch antenna array by using the advantages of directive antenna pattern properties. The new algorithm is called cylindrical patch array MN-MUST(CMN-MUST). CMN-MUST algorithm consists of three stages as MN-MUST does. Detection stage is exactly the same as in MN-MUST. However spatial filtering and DoA estimation stage are reduced order by using the advantages of directive antenna pattern of cylindirical microstrip patch array. The performance of the algorithm is investigated via computer simulations, for uniform linear arrays, a six element uniform dipole array and a twelve element uniform cylindrical microstrip patch array. The simulation results are compared to the previously reported works and the literature. It is observed that the proposed algorithm improves the previously reported works. The algorithm accuracy does not degrade in the presence of the mutual coupling. A uniform cylindrical patch array is successfully implemented to the MN-MUST algorithm. The implementation does not only cover full azimuth, but also improv the accuracy and speed. It is observed that the MN-MUST algorithm provides an accurate and efficient solution to the targettracking problem in real time.

In this thesis, two-dimensional (2-D) direction-of-arrival (DOA) estimation problem is considered. Usually, DOA estimation is considered in one dimension assuming a fixed elevation angle. While this assumption simplifies the problem, both the azimuth and elevation angles, namely, the 2-D DOA estimates are required in practical scenarios. In this thesis, planar array structures are considered for 2-D DOA estimation. In this context, V-shaped arrays are discussed and some of the important features of these arrays are outlined. A new method for the design of V-shaped arrays is presented for both isotropic and directional beam patterns. The design procedure is simple and can be applied for both uniform and nonuniform V-shaped sensor arrays. Closed form expressions are presented for the V-angle in order to obtain isotropic angle performance. While circular arrays have the isotropic characteristics, V-shaped arrays present certain advantages due to their large aperture for the same number of sensors and inter-sensor distance. The comparison of circular and V-shaped arrays is done by considering the azimuth and elevation Cramer-Rao Bounds (CRB). It is shown that V-shaped and circular arrays have similar characteristics for the sensor position errors while the uniform isotropic (UI) V-array performs better when there is mutual coupling and the sources are correlated. In the literature, there are several techniques for 2-D DOA estimation. Usually, fast algorithms are desired for this purpose since a search in two dimensions is a costly process. These algorithms have a major problem, namely, the pairing of the azimuth-elevation couples for multiple sources. In this thesis, a new fast and effective technique for this purpose is proposed. In this technique, a virtual array output is generated such that when the ESPRIT algorithm is used, the eigenvalues of the rotational transformation matrix have the 2-D angle information in both magnitude and phase. This idea is applied in different scenarios and three methods are presented for these cases. In one case, given an arbitrary array structure, array interpolation is used to generate the appropriate virtual arrays. When the antenna mutual coupling is taken into account, a special type of array structure, such as circular, should be used in order to apply the array interpolation. In general, the array mutual coupling matrix (MCM) should have a symmetric Toeplitz form. It is shown that the 2-D DOA performance of the proposed method approaches to the CRB by using minimum number of antennas in case of mutual coupling. This method does not require the estimation of the mutual coupling coefficients. While this technique is effective, it has problems especially when the number of sources increases. In order to improve the performance, MCM is estimated in the third approach. This new approach performs better, but it cannot be used satisfactorily in case of multipath signals. In this thesis, the proposed idea for fast 2-D DOA estimation is further developed in order to solve the problem when mutual coupling and multipath signals jointly exist. In this case, real arrays with some auxiliary sensors are used to generate a structured mutual coupling matrix. It is shown that the problem can be effectively solved when the array structure has a special form. Specifically, parallel uniform linear arrays (PULA) are employed for this purpose. When auxiliary sensors are used, a symmetric banded Toeplitz MCM is obtained for the PULA. This allows the application of spatial smoothing and ESPRIT algorithm for 2-D DOA estimation. The proposed algorithm uses triplets and presents closed form paired 2-D DOA estimates in case of unknown mutual coupling and multipath signals. Several simulations are done and it is shown that the proposed array structure and the method effectively solve the problem.

Higher Order Statistics (HOS) and Second Order Statistics (SOS) approaches have certain advantages and disadvantages in signal processing applications. HOS approach provides more statistical information for non-Gaussian signals. On the other hand, SOS approach is more robust to the estimation errors than the HOS approach, especially when the number of observations is small. In this thesis, HOS and SOS approaches are jointly used in order to take advantage of both methods. In this respect, the joint use of HOS and SOS approaches are introduced for online calibration of sensor arrays with arbitrary geometries. Three different problems in online array calibration are considered and new algorithms for each of these problems are proposed. In the first problem, the positions of the randomly deployed sensors are completely unknown except the two reference sensors and HOS and SOS approaches are used iteratively for the joint Direction of Arrival (DOA) and sensor position estimation. Iterative HOS-SOS algorithm (IHOSS) solves the ambiguity problem in sensor position estimation by observing the source signals at least in two different frequencies and hence it is applicable for wideband signals. The conditions on these frequencies are presented. IHOSS is the first algorithm in the literature which finds the DOA and sensor position estimations in case of randomly deployed sensors with unknown coordinates. In the second problem, narrowband signals are considered and it is assumed that the nominal sensor positions are known. Modified IHOSS (MIHOSS) algorithm uses the nominal sensor positions to solve the ambiguity problem in sensor position estimation. This algorithm can handle both small and large errors in sensor positions. The upper bound of perturbations for unambiguous sensor position estimation is presented. In the last problem, an online array calibration method is proposed for sensor arrays where the sensors have unknown gain/phase mismatches and mutual coupling coefficients. In this case, sensor positions are assumed to be known. The mutual coupling matrix is unstructured. The two reference sensors are assumed to be perfectly calibrated. IHOSS algorithm is adapted for online calibration and parameter estimation, and hence CIHOSS algorithm is obtained. While CIHOSS originates from IHOSS, it is fundamentally different in many aspects. CIHOSS uses multiple virtual ESPRIT structures and employs an alignment technique to order the elements of rows of the actual array steering matrix. In this thesis, a new cumulant matrix estimation technique is proposed for the HOS approach by converting the multi-source problem into a single source one. The proposed algorithms perform well even in the case of correlated source signals due to the effectiveness of the proposed cumulant matrix estimate. The iterative procedure in all the proposed algorithms is guaranteed to converge. Closed form expressions are derived for the deterministic Cram&acute / er-Rao bound (CRB) for DOA and unknown calibration parameters for non-circular complex Gaussian noise with unknown covariance matrix. Simulation results show that the performances of the proposed methods approach to the CRB for both DOA and unknown calibration parameter estimations for high SNR.

Signal acquisition, signal reconstruction and analysis of spectrum of the signal are the three most important steps in signal processing and they are found in almost all of the modern day hardware. In most of the signal processing hardware, the signal of interest is sampled at uniform intervals satisfying some conditions like Nyquist rate. However, in some cases the privilege of having uniformly sampled data is lost due to some constraints on the hardware resources. In this thesis an important problem of signal reconstruction and spectral analysis from nonuniformly sampled data is addressed and a variety of methods are presented. The proposed methods are tested via numerical experiments on both artificial and real-life data sets. The thesis starts with a brief review of methods available in the literature for signal reconstruction and spectral analysis from non uniformly sampled data. The methods discussed in the thesis are classified into two broad categories - dense and sparse methods, the classification is based on the kind of spectra for which they are applicable. Under dense spectral methods the main contribution of the thesis is a non-parametric approach named LIMES, which recovers the smooth spectrum from non uniformly sampled data. Apart from recovering the spectrum, LIMES also gives an estimate of the covariance matrix. Under sparse methods the two main contributions are methods named SPICE and LIKES - both of them are user parameter free sparse estimation methods applicable for line spectral estimation. The other important contributions are extensions of SPICE and LIKES to multivariate time series and array processing models, and a solution to the grid selection problem in sparse estimation of spectral-line parameters. The third and final part of the thesis contains applications of the methods discussed in the thesis to the problem of radial velocity data analysis for exoplanet detection. Apart from the exoplanet application, an application based on Sudoku, which is related to sparse parameter estimation, is also discussed.

Submitted by Cláudia Bueno (claudiamoura18@gmail.com) on 2015-11-12T20:02:55Z No. of bitstreams: 2 Dissertação - Jonas Augusto Kunzler - 2015.pdf: 5934373 bytes, checksum: a736817202816bba60673f1a39184580 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2015-11-13T10:40:49Z (GMT) No. of bitstreams: 2 Dissertação - Jonas Augusto Kunzler - 2015.pdf: 5934373 bytes, checksum: a736817202816bba60673f1a39184580 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2015-11-13T10:40:49Z (GMT). No. of bitstreams: 2 Dissertação - Jonas Augusto Kunzler - 2015.pdf: 5934373 bytes, checksum: a736817202816bba60673f1a39184580 (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2015-04-14 / New technologies that have emerged from the transistor advent enabled digital signal processing techniques were aggregated to systems operated or partially in analog improving performance of these systems. Added to this the use of sensor arrangements which made it possible to identify the directions of signals of interest and are used in critical areas of knowledge, such as in tracking radar systems; in astronomy; in sonar systems; mobile communications; the estimation of direction of arrival; in seismology and medical diagnosis and treatment. This work aims to study specific features of an estimation method of direction of arrival based on a linear array of sensors with special attention to mobile communications. Some methods have been proposed in order to get the position of a source of electromagnetic waves and can cite the MUSIC, the MODEX and SEAD, the latter of fundamental importance to this work because it is the basis of conducted research and because it is a new method, lacked further clarification with regard to the differential spectrum, their origin, meaning and importance, as well as obtaining an analytical expression to describe their conduct on the variables that compose the system. Based on eigenvalue decomposition of the correlation matrix has been observed that the differential spectrum is basically a matrix norm calculation, with the largest eigenvalue of the matrix is ​​also standard 2-induced vector. It was proposed to use the Frobenius norm which is simpler to be computed, and consequently requires less computational effort. Moreover, the behavior of the angular spectrum calculated using the Frobenius norm is fully described by the sum of cosines with the formulation described for each part which composes the calculation of the standard. Through this outcome was possible to analyze aspects related to angular resolution, the number of signal sources, the number of sensors, the influence of noise and the correlation between sources. / Novas tecnologias que surgiram a partir do advento transistor permitiram técnicas de processamento digital de sinais fossem agregadas a sistemas que operavam ou parcialmente de forma analógica aprimorando o desempenho desses sistemas. Soma-se a isto a utilização de arranjos de sensores que possibilitaram a identificação das direções dos sinais de interesse e são empregados em importantes áreas do conhecimento, como por exemplo, em sistemas de rastreamento por radar; na Astronomia; em Sistemas sonares; nas comunicações móveis; na estimação de direção de chegada; na sismologia e no diagnóstico e tratamento médico. Este trabalho tem como objetivo estudar características específicas de um método de estimação de direção de chegada baseado em um arranjo linear de sensores com atenção especial em comunicações móveis. Alguns métodos foram propostos com o fim de obter a posição de uma fonte de ondas eletromagnéticas, podendo-se citar o MUSIC, o MODEX e o SEAD, este último de fundamental importância para este trabalho, pois, ele é a base da investigação conduzida e por se tratar de um método novo, carecia de mais esclarecimentos no que diz respeito ao espectro diferencial, sua origem, significado e importância, como também a obtenção de uma expressão analítica que descrevesse seu comportamento em função das variáveis que compões o sistema. Baseado na decomposição em autovalores da matriz de correlação observou-se que o espectro diferencial é basicamente um cálculo de norma matricial, sendo que o maior autovalor da matriz é também a norma 2 induzida por vetor. Propôs-se a utilização da norma de Frobenius que é mais simples de ser calculada e, consequentemente, exige menos esforço computacional. Além disso, o comportamento do espectro angular calculado com a norma de Frobenius é totalmente descrito através da soma de cossenos com a formulação descrita para cada parcela que compõe o cálculo da norma. Através deste resultado foi possível analisar aspectos referentes à resolução angular, ao número de fontes de sinal, ao número de sensores, à influência do ruído e à correlação entre as fontes.

No description available.

Un radar MIMO (Multiple Input Multiple Output) est un système radar qui utilise plusieurs antennes émettrices et réceptrices, dans lequel les formes d'ondes émises peuvent être indépendantes. Par rapport aux radars utilisant des antennes en réseaux phasés, les radars MIMO offrent davantage de degrés de liberté, ce qui permet d'améliorer les performances du système en termes de détection et localisation. La technique MIMO offre également la possibilité de synthétiser un diagramme de rayonnement désiré par une définition judicieuse des formes d'ondes émises. Dans la mesure où les paramètres des cibles (positions, vitesses, directions d'arrivée (DOA), ...) sont estimés à partir des échos des signaux émis, on comprend aisément que les formes d'ondes employées jouent un rôle clé dans les performances du système. Cette thèse porte sur l'estimation de DOA et sur la conception des formes d'ondes pour un radar MIMO. Le cadre d'étude est restreint au cas où les antennes sont colocalisées et les cibles sont immobiles et supposées ponctuelles. La plupart des travaux antérieurs (au commencement de la thèse) portaient sur le radar MIMO bande étroite et faisaient l'hypothèse d'émetteurs-récepteurs idéaux et indépendants. Cette thèse contribue à élargir le cadre d'étude en s'intéressant d'une part au passage en large bande et d'autre part à la modélisation et à la prise en compte de la non-indépendance des émetteurs-récepteurs et autres imperfections. Dans la mesure où le recours à des signaux large bande est nécessaire lorsqu'une résolution importante est souhaitée, nous nous sommes attachés dans cette thèse à adapter le modèle d'un système de radar MIMO au cas large bande et à proposer de nouvelles techniques visant à améliorer les performances d'estimation de DOA dans le cas de signaux large bande. Cette thèse analyse également l'influence de conditions non idéales comme l'impact des phénomènes de couplage électromagnétique sur les diagrammes de rayonnement dans un réseau d'antennes. Cette étude est menée dans le cas bande étroite. En particulier, nous étudions l'influence du couplage direct entre les réseaux d'antennes d'émission et de réception (appelé « crosstalk ») sur les performances des techniques proposées. Nous établissons un modèle du signal permettant de prendre en compte ce phénomène et proposons une technique de réduction du « crosstalk » qui permet une estimation efficace des DOA des cibles. Nous montrons par ailleurs comment améliorer les performances d'estimation de DOA en présence de diagrammes de rayonnement incluant le couplage entre antennes. Le dernier apport principal de cette thèse est la conception et réalisation d'une plateforme expérimentale comportant une seule architecture d'émetteur-récepteur, qui permet de simuler un système MIMO utilisant des antennes colocalisées en appliquant le principe de superposition. Cette plateforme nous a permis d'évaluer les performances des techniques proposées dans des conditions plus réalistes / A Multiple-Input Multiple-Output (MIMO) radar is a system employing multiple transmitters and receivers in which the waveforms to be transmitted can be totally independent. Compared to standard phased-array radar systems, MIMO radars offer more degrees of freedom which leads to improved angular resolution and parameter identifiability, and provides more flexibility for transmit beampattern design. The main issues of interest in the context of MIMO radar are the estimation of several target parameters (which include range, Doppler, and Direction-of-Arrival (DOA), among others). Since the information on the targets is obtained from the echoes of the transmitted signals, it is straightforward that the design of the waveforms plays an important role in the system accuracy. This document addresses the investigation of DOA estimation of non-moving targets and waveform design techniques for MIMO radar with colocated antennas. Although narrowband MIMO radars have been deeply studied in the literature, the existing DOA estimation techniques have been usually proposed and analyzed from a theoretical point of view, often assuming ideal conditions. This thesis analyzes existing signal processing algorithms and proposes new ones in order to improve the DOA estimation performance in the case of narrowband and wideband signals. The proposed techniques are studied under ideal and non-ideal conditions considering punctual targets. Additionally, we study the influence of mutual coupling on the performance of the proposed techniques and we establish a more realistic signal model which takes this phenomenon into account. We then show how to improve the DOA estimation performance in the presence of distorted radiation patterns and we propose a crosstalk reduction technique, which makes possible an efficient estimation of the target DOAs. Finally, we present an experimental platform for MIMO radar with colocated antennas which has been developed in order to evaluate the performance of the proposed techniques under more realistic conditions. The proposed platform, which employs only one transmitter and one receiver architectures, relies on the superposition principle to simulate a real MIMO system

Direction of Arrival estimation using unsteered antenna arrays, unlike mechanically scanned or phased arrays, requires complex algorithms which perform poorly with small aperture arrays or without a large number of observations, or snapshots. In general, these algorithms compute a sample covriance matrix to obtain the direction of arrival and some require a prior estimate of the number of signal sources. Herein, artificial neural network architectures are proposed which demonstrate improved estimation of the number of signal sources, the true signal covariance matrix, and the direction of arrival. The proposed number of source estimation network demonstrates robust performance in the case of coherent signals where conventional methods fail. For covariance matrix estimation, four different network architectures are assessed and the best performing architecture achieves a 20 times improvement in performance over the sample covariance matrix. Additionally, this network can achieve comparable performance to the sample covariance matrix with 1/8-th the amount of snapshots. For direction of arrival estimation, preliminary results are provided comparing six architectures which all demonstrate high levels of accuracy and demonstrate the benefits of progressively training artificial neural networks by training on a sequence of sub- problems and extending to the network to encapsulate the entire process.

Poaching is a persistent issue that threatens many of earth’s species including therhino. The methods used by poachers are varied, but many use guns to carry outtheir illegal activities. Gunfire is extremely loud and can be heard for kilometres.This thesis investigates whether it is possible to aid anti-poaching efforts in Kenyawith a gunshot detection and estimation device using an array of microphones. Ifsuccessful, the device could be placed around the savannah or any exposed areaand warn if poaching is taking place in the nearby. If a shot is fired within theaudible range of the device’s microphones, a trained machine learning algorithmdetects the shot on the edge using a microprocessor. The detection runs in realtime and achieved an accuracy of 93% on an unbalanced data set, where themajority class was the one without gunshots. Once a detection has been made, thereceived signal power to each microphone is used to produce a direction of arrivalestimate. The estimate can produce an angle estimate with a standard deviationof 66.78° for a gunshot, and with a standard deviation of 7.65° when testing themodel with white noise. Future implementations could use several devices thatdetected the same event, and fuse their estimates to locate the shooter’s position.All of this information, as well as the sound file, can be used to alert and assistlocal wildlife services. The challenges of this project have been centred aroundmaking a system run in real time with only a microprocessor on the edge, whilealso prioritizing low cost components for future deployment. / Project Ngulia

The angle of arrival estimation of multiple sources plays a vital role in the field of array signal processing as MIMO systems can be employed at both the transmitter and the receiver end and the system capacity, reliability and throughput can be significantly increased by using array signal processing. Almost all applications require accurate direction of arrival (DOA) estimation to localize the sources of the signals. Another important parameter of localization systems is the array geometry and sensor design which can be application specific and is used to estimate the DOA. In this work, various array geometries and arrival estimation algorithms are studied and then a new scheme for multiple source estimation is proposed and evaluated based on the performance of subspace and non-subspace decomposition methods. The proposed scheme has shown to outperform the conventional Multiple Signal Classification (MUSIC) estimation and Bartlett estimation techniques. The new scheme has a better performance advantage at low and high signal to noise ratio values (SNRs). The research work also studies different array geometries for both single and multiple incident sources and proposes a geometry which is cost effective and efficient for 3, 4, and 5 antenna array elements. This research also considers the shape of the ground plane and its effects on the angle of arrival estimation and in addition it shows how the mutual couplings between the elements effect the overall estimation and how this error can be minimised by using a decoupling matrix. At the end, a novel miniaturised multi element reconfigurable antenna to represent the receiver base station is designed and tested. The antenna radiation patterns in the azimuth angle are almost omni-directional with linear polarisation. The antenna geometry is uniplanar printed logspiral with striplines feeding network and biased components to improve the impedance bandwidth. The antenna provides the benefit of small size, and re-configurability and is very well suited for the asset tracking applications.