Aquesta tesi doctoral mostra la recerca feta en l'àrea de la diarització de locutor per a sales de reunions. En la present s'estudien els algorismes i la implementació d'un sistema en diferit de segmentació i aglomerat de locutor per a grabacions de reunions a on normalment es té accés a més d'un micròfon per al processat. El bloc més important de recerca s'ha fet durant una estada al International Computer Science Institute (ICSI, Berkeley, Caligornia) per un període de dos anys.La diarització de locutor s'ha estudiat força per al domini de grabacions de ràdio i televisió. La majoria dels sistemes proposats utilitzen algun tipus d'aglomerat jeràrquic de les dades en grups acústics a on de bon principi no se sap el número de locutors òptim ni tampoc la seva identitat. Un mètode molt comunment utilitzat s'anomena "bottom-up clustering" (aglomerat de baix-a-dalt), amb el qual inicialment es defineixen molts grups acústics de dades que es van ajuntant de manera iterativa fins a obtenir el nombre òptim de grups tot i acomplint un criteri de parada. Tots aquests sistemes es basen en l'anàlisi d'un canal d'entrada individual, el qual no permet la seva aplicació directa per a reunions. A més a més, molts d'aquests algorisms necessiten entrenar models o afinar els parameters del sistema usant dades externes, el qual dificulta l'aplicabilitat d'aquests sistemes per a dades diferents de les usades per a l'adaptació.La implementació proposada en aquesta tesi es dirigeix a solventar els problemes mencionats anteriorment. Aquesta pren com a punt de partida el sistema existent al ICSI de diarització de locutor basat en l'aglomerat de "baix-a-dalt". Primer es processen els canals de grabació disponibles per a obtindre un sol canal d'audio de qualitat major, a més dínformació sobre la posició dels locutors existents. Aleshores s'implementa un sistema de detecció de veu/silenci que no requereix de cap entrenament previ, i processa els segments de veu resultant amb una versió millorada del sistema mono-canal de diarització de locutor. Aquest sistema ha estat modificat per a l'ús de l'informació de posició dels locutors (quan es tingui) i s'han adaptat i creat nous algorismes per a que el sistema obtingui tanta informació com sigui possible directament del senyal acustic, fent-lo menys depenent de les dades de desenvolupament. El sistema resultant és flexible i es pot usar en qualsevol tipus de sala de reunions pel que fa al nombre de micròfons o la seva posició. El sistema, a més, no requereix en absolute dades d´entrenament, sent més senzill adaptar-lo a diferents tipus de dades o dominis d'aplicació. Finalment, fa un pas endavant en l'ús de parametres que siguin mes robusts als canvis en les dades acústiques. Dos versions del sistema es van presentar amb resultats excel.lents a les evaluacions de RT05s i RT06s del NIST en transcripció rica per a reunions, a on aquests es van avaluar amb dades de dos subdominis diferents (conferencies i reunions). A més a més, es fan experiments utilitzant totes les dades disponibles de les evaluacions RT per a demostrar la viabilitat dels algorisms proposats en aquesta tasca. / This thesis shows research performed into the topic of speaker diarization for meeting rooms. It looks into the algorithms and the implementation of an offline speaker segmentation and clustering system for a meeting recording where usually more than one microphone is available. The main research and system implementation has been done while visiting the International Computes Science Institute (ICSI, Berkeley, California) for a period of two years. Speaker diarization is a well studied topic on the domain of broadcast news recordings. Most of the proposed systems involve some sort of hierarchical clustering of the data into clusters, where the optimum number of speakers of their identities are unknown a priory. A very commonly used method is called bottom-up clustering, where multiple initial clusters are iteratively merged until the optimum number of clusters is reached, according to some stopping criterion. Such systems are based on a single channel input, not allowing a direct application for the meetings domain. Although some efforts have been done to adapt such systems to multichannel data, at the start of this thesis no effective implementation had been proposed. Furthermore, many of these speaker diarization algorithms involve some sort of models training or parameter tuning using external data, which impedes its usability with data different from what they have been adapted to.The implementation proposed in this thesis works towards solving the aforementioned problems. Taking the existing hierarchical bottom-up mono-channel speaker diarization system from ICSI, it first uses a flexible acoustic beamforming to extract speaker location information and obtain a single enhanced signal from all available microphones. It then implements a train-free speech/non-speech detection on such signal and processes the resulting speech segments with an improved version of the mono-channel speaker diarization system. Such system has been modified to use speaker location information (then available) and several algorithms have been adapted or created new to adapt the system behavior to each particular recording by obtaining information directly from the acoustics, making it less dependent on the development data.The resulting system is flexible to any meetings room layout regarding the number of microphones and their placement. It is train-free making it easy to adapt to different sorts of data and domains of application. Finally, it takes a step forward into the use of parameters that are more robust to changes in the acoustic data. Two versions of the system were submitted with excellent results in RT05s and RT06s NIST Rich Transcription evaluations for meetings, where data from two different subdomains (lectures and conferences) was evaluated. Also, experiments using the RT datasets from all meetings evaluations were used to test the different proposed algorithms proving their suitability to the task.

MU-MIMO theory predicts manyfold capacity gains by leveraging many antennas (e.g. M >> 10) on wireless base stations to serve many users simultaneously through multi-user beamforming (MUBF). However, realizing such a large-scale design is nontrivial, and has yet to be achieved in the real world. We present the design, realization, and evaluation of Argos, the first reported large-scale base station that is capable of serving many (e.g., 10s of) terminals simultaneously through MUBF. Designed with extreme flexibility and scalability in mind, Argos exploits hierarchical and modular design principles, properly partitions baseband processing, and holistically considers real-time requirements of MUBF. To achieve unprecedented scalability, we devise a novel, completely distributed, beamforming technique, as well as an internal calibration procedure to enable implicit beamforming across large arrays. We implement a prototype with 64 antennas, and demonstrate that it can achieve up to 6.7 fold capacity gains while using a mere 1/64th the transmission power.

Experimental aeroacoustic research was conducted on a wind turbine specific airfoil at low Reynolds numbers. The goal of this thesis was to study trailing edge noise generation from the airfoil and investigate correlations between the noise and the flow field. Before experiments were performed the current wind tunnel had to be modified in order to make it more suitable for aeroacoustic tests. Sound absorbing foam was added to the inside of the tunnel to lower the background noise levels and turbulence reduction screens were added which lowered the turbulence. An S822 airfoil was chosen because it is designed for low Reynolds flows attainable in the wind tunnel which are on the order of 104. Smoke wire flow visualization was used to gain insight into the airfoil wake development and oil film flow visualization was used to qualitatively assess the boundary layer development. Laser Doppler anemometry (LDA) was used to measure two components of velocity at high data rates in the airfoil wake. Wake profiles were measured in addition to single point measurements to determine the velocity spectrum. A microphone was mounted inside the test section in order to measure the trailing edge noise. Initial plans included measuring the trailing edge noise with a microphone array capable of quantifying and locating noise sources. Although an array was built and beamforming code was written it was only used in preliminary monopole source tests. Oil film results showed the behaviour of the boundary layer to be consistent with previous low Reynolds number experiments. LDA results revealed sharp peaks in the velocity spectra at 1100 Hz from U0 = 15–24 m/s, and 3100 and 3800 Hz, from U0 = 25–35 m/s, which were inconsistent with vortex shedding results of previous researchers. Also present were a series of broad peaks in the spectra that increase from 1200–1700 Hz in the U0 = 25–35 m/s range. The shedding frequency from the smoke wire flow visualization was calculated to be 1250 Hz at U0 = 26 m/s. These sharp peaks were also present in the acoustic spectrum. It was reasoned that these peaks are due to wind tunnel resonance which is a common occurrence in hard wall wind tunnels. In particular the tone at 1100 Hz is due to a standing wave with a wavelength equal to half the tunnel width. The shedding frequency from the smoke wire flow visualization was calculated to be 1100 Hz at U0 = 20 m/s. These tones exhibited a “ladder-like” relationship with freestream velocity, another aspect indicative of wind tunnel resonance. It was reasoned that the wind tunnel resonance was forcing the shedding frequency of the airfoil in the U0 = 15–24 m/s range, and in the U0 = 25–35 m/s range, the shedding frequency corresponded to the broad peaks in the LDA spectra.

The concept of cooperation in communications has drawn a lot of research attention in recent years due to its potential to improve the efficiency of wireless networks. This new form of communications allows some users to act as relays and assist the transmission of other users' information signals. The aim of this thesis is to apply optimization techniques in the design of multi-relay wireless networks employing cooperative communications. In general, the thesis is organized into two parts: ``Distributed space-time coding' (DSTC) and ``Distributed beamforming', which cover two main approaches in cooperative communications over multi-relay networks. <br><br> In Part I of the thesis, various aspects of distributed implementation of space-time coding in a wireless relay network are treated. First, the thesis proposes a new fully-diverse distributed code which allows noncoherent reception at the destination. Second, the problem of coordinating the power allocation (PA) between source and relays to achieve the optimal performance of DSTC is studied and a novel PA scheme is developed. It is shown that the proposed PA scheme can obtain the maximum diversity order of DSTC and significantly outperform other suboptimal PA schemes. Third, the thesis presents the optimal PA scheme to minimize the mean-square error (MSE) in channel estimation during training phase of DSTC. The effect of imperfect channel estimation to the performance of DSTC is also thoroughly studied. <br><br> In Part II of the thesis, optimal distributed beamforming designs are developed for a wireless multiuser multi-relay network. Two design criteria for the optimal distributed beamforming at the relays are considered: (i) minimizing the total relay power subject to a guaranteed Quality of Service (QoS) measured in terms of signal-to-noise-ratio (SNR) at the destinations, and (ii) jointly maximizing the SNR margin at the destinations subject to power constraints at the relays. Based on convex optimization techniques, it is shown that these problems can be formulated and solved via second-order conic programming (SOCP). In addition, this part also proposes simple and fast iterative algorithms to directly solve these optimization problems.

Adaptive antenna arrays have a great importance in reduction of the effect of interference and increase the capacity for the mobile satellite communication. Interference and multipath fading remain a main problem for reception of signals. These two problems obviously affect the overall capacity.  Adaptive antenna arrays in the handheld mobile apparatus will be the solution for the above two problems.   Satellite mobile communication is one of the growing fields in the communication area where terrestrial infrastructures are unable or ineffective to supply. Maritime, aeronautical and land mobile are some of the applications. During natural disasters where ground services are stopped, mobile satellite communications has great importance. Following the hurricane season, the Asian Tsunami and the devastating Haiti earthquake, mobile satellite communications had played a great role to fill the communication gaps.  The satellites can be tracked automatically by adaptive antenna array when it moves in its orbital plane.   In this thesis the methods that how the adaptive antenna array combats interferers is presented and simulated using MATLAB software. The performance of the adaptive antenna array is evaluated by simulating the directivity pattern of the antenna and Mean Square Error (MSE) graph for different scenario like Signal to Interference Noise ratio (SINR), number of iterations, antenna array elements and convergence factor (μ), assuming the signals are coming from different Direction of Arrival (DOA).

Recent information theoretical results indicate that dirty-paper coding (DPC) achieves the entire capacity region of the Gaussian multiple-input multiple-output (MIMO) broadcast channel (BC). This thesis presents practical code designs for Gaussian BCs based on DPC. To simplify our designs, we assume constraints on the individual rates for each user instead of the customary constraint on transmitter power. The objective therefore is to minimize the transmitter power such that the practical decoders of all users are able to operate at the given rate constraints. The enabling element of our code designs is a practical DPC scheme based on nested turbo codes. We start with Cover's simplest two-user Gaussian BC as a toy example and present a code design that operates 1.44 dB away from the capacity region boundary at the transmission rate of 1 bit per sample per dimension for each user. Then we consider the case of the multiple-input multiple-output BC and develop a practical limit-approaching code design under the assumption that the channel state information is available perfectly at the receivers as well as at the transmitter. The optimal precoding strategy in this case can be derived by invoking duality between the MIMO BC and MIMO multiple access channel (MAC). However, this approach requires transformation of the optimal MAC covariances to their corresponding counterparts in the BC domain. To avoid these computationally complex transformations, we derive a closed-form expression for the optimal precoding matrix for the two-user case and use it to determine the optimal precoding strategy. For more than two users we propose a low-complexity suboptimal strategy, which, for three transmit antennas at the base station and three users (each with a single receive antenna), performs only 0.2 dB worse than the optimal scheme. Our obtained results are only 1.5 dB away from the capacity limit. Moreover simulations indicate that our practical DPC based scheme significantly outperforms the prevalent suboptimal strategies such as time division multiplexing and zero forcing beamforming. The drawback of DPC based designs is the requirement of channel state information at the transmitter. However, if the channel state information can be communicated back to the transmitter effectively, DPC does indeed have a promising future in code designs for MIMO BCs.

The study is based on the acoustic data collected by horizontal line array (HLA) of Asian Seas International Acoustic Experiment (ASIAEX) in South China Sea of 2001. Beamforming was used to localize the sound sources during the experiment, such as explosive, to understand its correlation with position deviation from the array. According to previous studies, the horizontal line array has ambiguity in bearing identification, which was often resolved by maneuvering array¡¦s position, therefore it can also be concluded that the well understanding of array¡¦s position is the key to the accurate source localization. Due to the limitation of element spacing of array used in the experiment, 50 Hz is the highest frequency can be analyzed in the study, so the recorded explosive sounds were chosen for analysis. The numerical test of array has shown the width of main lobe in beamforming is reduced by the increasing total length of the array, and the energy in the side lobes would affect the accuracy of source bearing. During the experiment, the horizontal array was found, and proved indirectly by current measurement of acoustic Dopper current profiler, to be moved by tidal currents, so the deviation from the designed deploy position can be computed. The deviation was used to quantify the resolution effect on localization by the movement of array, and when there is a 10 m deviation, the width of main lobe would increase from 8¢X to 16¢X in the studied case. The experimental results match the simulations well, so it can provide a guideline in predicting the accuracy of underwater source localization when the movement of horizontal line array is possible. Key word: underwater source localization, horizontal line array, beamforming, position deviation, array resolution

In this thesis the direction of arrival estimation of wideband signals using frequency invariant beamforming method is examined. The difficulty with the direction of arrival estimation of wideband signals is that it is not possible to obtain a single covariance matrix valid for the whole frequency spectrum of the signal. There are various methods proposed in the literature to overcome this difficulty. The common aim of all the methods is to obtain a composite covariance matrix for the overall band of the signal. In this thesis, we concentrate on a method in [12]. This method is based on a beamforming technique that provides frequency invariant beams in the band of interest. Therefore there is no need for frequency decomposition as it is done with the other wideband methods. A comparison of the frequency invariant beamforming method with coherent signal subspace methods and narrow band methods is also given.

In multicell coordinated networks where multiple base stations cooperate to jointly combat interference from adjacent cells and fading to receivers, one of the outstanding questions is what is the role of receive antenna and receiver processing. Multiple receive antennas not only enable additional degrees of freedom at each receiver to combat the other-cell interference but also can change the transmitter design because transmitter and receiver beamforming design is often closely coordinated. In this dissertation, we investigate the role of the multiple receive antennas in multicell cooperative systems under different interference conditions. We then present novel non-iterative and iterative coordinated beamforming and precoding algorithms with different receiver processing. We present comprehensive performance comparison of various multicell cooperative systems and explore the feasibility of achieving much higher throughput via hyper-densification of heterogeneous and small cell networks with mandatory multicell cooperation. / text

The Generalized Sidelobe Canceller is an adaptive algorithm for optimally estimating the parameters for beamforming, the signal processing technique of combining data from an array of sensors to improve SNR at a point in space. This work focuses on the algorithm’s application to widely-separated microphone arrays with irregular distributions used for human voice capture. Methods are presented for improving the performance of the algorithm’s blocking matrix, a stage that creates a noise reference for elimination, by proposing a stochastic model for amplitude correction and enhanced use of cross correlation for phase correction and time-difference of arrival estimation via a correlation coefficient threshold. This correlation technique is also applied to a multilateration algorithm for an efficient method of explicit target tracking. In addition, the underlying microphone array geometry is studied with parameters and guidelines for evaluation proposed. Finally, an analysis of the stability of the system is performed with respect to its adaptation parameters.