No description available.

Contemporary research in digital signal processing (DSP) is focused on issues of computational complexity, very high data rate and large quantum of data. Thus, the success in newer applications and areas hinge on handling these issues. Conventional ways to address these challenges are to develop newer structures like Multirate signal processing, Multiple Input Multiple Output(MIMO), bandpass sampling, compressed domain sensing etc. In the implementation domain, the approach is to look at floating point over fixed point representation and / or longer wordlength etc., related to number representations and computations. Of these, a simple approach is to look at number representation, perhaps with a simple integer. This automatically guarantees accuracy and zero quantization error as well as longer wordlength. Thus, it is necessary and interesting to explore viable DSP alternatives that can reduce complexity and yet match the required performance. The main aim of this work is to highlight the importance, use and analysis of integer sequences. Firstly, the thesis explores the use of integer sequences as windowing functions. The results of these investigations show that integer sequences and their convolution, indeed, outperform many of the classical real valued window functions in terms of mainlobe width, sidelobe attenuation etc. Secondly, the thesis proposes techniques to approximate discrete Gaussian distribution using integer sequences. The key idea is to convolve symmetrized integer sequences and examine the resulting profiles. These profiles are found to approximate discrete Gaussian distribution with a mean square error of the order of 10−8 or less. While looking at integer sequences to approximate discrete Gaussian, Fibonacci sequence was found to exhibit some interesting properties. The third part of the thesis proves certain fascinating optimal probabilistic limit properties (mean and variance) of Fibonacci sequence. The thesis also provides complete generalization of these properties to probability distributions generated by second order linear recurrence relation with integer coefficients and any kth order linear recurrence relation with unit coefficients. In addition to the above, the thesis also throws light on possible architectural implications of using integer sequences in DSP applications and ideas for further exploration.

A new communication paradigm is foreseen for beyond 2020 society, due to the emergence of new broadband services and the Internet of Things era. The set of requirements imposed by these new applications is large and diverse, aiming to provide a ubiquitous broadband connectivity. Research community has been working in the last decade towards the definition of the 5G mobile wireless networks that will provide the proper mechanisms to reach these challenging requirements. In this framework, three key research directions have been identified for the improvement of capacity in 5G: the increase of the spectral efficiency by means of, for example, the use of massive MIMO technology, the use of larger amounts of spectrum by utilizing the millimeter wave band, and the network densification by deploying more base stations per unit area. This dissertation addresses densification as the main enabler for the broadband and massive connectivity required in future 5G networks. To this aim, this Thesis focuses on the study of the UDN. In particular, a set of technology enablers that can lead UDN to achieve their maximum efficiency and performance are investigated, namely, the use of higher frequency bands for the benefit of larger bandwidths, the use of massive MIMO with distributed antenna systems, and the use of distributed radio resource management techniques for the inter-cell interference coordination. Firstly, this Thesis analyzes whether there exists a fundamental performance limit related with densification in cellular networks. To this end, the UDN performance is evaluated by means of an analytical model consisting of a 1-dimensional network deployment with equally spaced BS. The inter-BS distance is decreased until reaching the limit of densification when this distance approaches 0. The achievable rates in networks with different inter-BS distances are analyzed for several levels of transmission power availability, and for various types of cooperation among cells. Moreover, UDN performance is studied in conjunction with the use of a massive number of antennas and larger amounts of spectrum. In particular, the performance of hybrid beamforming and precoding MIMO schemes are assessed in both indoor and outdoor scenarios with multiple cells and users, working in the mmW frequency band. On the one hand, beamforming schemes using the full-connected hybrid architecture are analyzed in BS with limited number of RF chains, identifying the strengths and weaknesses of these schemes in a dense-urban scenario. On the other hand, the performance of different indoor deployment strategies using HP in the mmW band is evaluated, focusing on the use of DAS. More specifically, a DHP suitable for DAS is proposed, comparing its performance with that of HP in other indoor deployment strategies. Lastly, the presence of practical limitations and hardware impairments in the use of hybrid architectures is also investigated. Finally, the investigation of UDN is completed with the study of their main limitation, which is the increasing inter-cell interference in the network. In order to tackle this problem, an eICIC scheduling algorithm based on resource partitioning techniques is proposed. Its performance is evaluated and compared to other scheduling algorithms under several degrees of network densification. After the completion of this study, the potential of UDN to reach the capacity requirements of 5G networks is confirmed. Nevertheless, without the use of larger portions of spectrum, a proper interference management and the use of a massive number of antennas, densification could turn into a serious problem for mobile operators. Performance evaluation results show large system capacity gains with the use of massive MIMO techniques in UDN, and even greater when the antennas are distributed. Furthermore, the application of ICIC techniques reveals that, besides the increase in system capacity, it brings significant energy savings to UDNs. / A partir del año 2020 se prevé que un nuevo paradigma de comunicación surja en la sociedad, debido a la aparición de nuevos servicios y la era del Internet de las cosas. El conjunto de requisitos impuesto por estas nuevas aplicaciones es muy amplio y diverso, y tiene como principal objetivo proporcionar conectividad de banda ancha y universal. En las últimas décadas, la comunidad científica ha estado trabajando en la definición de la 5G de redes móviles que brindará los mecanismos necesarios para garantizar estos requisitos. En este marco, se han identificado tres mecanismos clave para conseguir el necesario incremento de capacidad de la red: el aumento de la eficiencia espectral a través de, por ejemplo, el uso de tecnologías MIMO masivas, la utilización de mayores porciones del espectro en frecuencia y la densificación de la red mediante el despliegue de más estaciones base por área. Esta Tesis doctoral aborda la densificación como el principal mecanismo que permitirá la conectividad de banda ancha y universal requerida en la 5G, centrándose en el estudio de las Redes Ultra Densas o UDNs. En concreto, se analiza el conjunto de tecnologías habilitantes que pueden llevar a las UDNs a obtener su máxima eficiencia y prestaciones, incluyendo el uso de altas frecuencias para el aprovechamiento de mayores anchos de banda, la utilización de MIMO masivo con sistemas de antenas distribuidas y el uso de técnicas de reparto de recursos distribuidas para la coordinación de interferencias. En primer lugar, se analiza si existe un límite fundamental en la mejora de las prestaciones en relación a la densificación. Con este fin, las prestaciones de las UDNs se evalúan utilizando un modelo analítico de red unidimensional con BSs equiespaciadas, en el que la distancia entre BSs se disminuye hasta alcanzar el límite de densificación cuando ésta se aproxima a 0. Las tasas alcanzables en redes con distintas distancias entre BSs son analizadas, considerando distintos niveles de potencia disponible en la red y varios grados de cooperación entre celdas. Además, el comportamiento de las UDNs se estudia junto al uso masivo de antenas y la utilización de anchos de banda mayores. Más concretamente, las prestaciones de ciertas técnicas híbridas MIMO de precodificación y beamforming se examinan en la banda milimétrica. Por una parte, se analizan esquemas de beamforming en BSs con arquitectura híbrida en función de la disponibilidad de cadenas de radiofrecuencia en escenarios exteriores. Por otra parte, se evalúan las prestaciones de ciertos esquemas de precodificación híbrida en escenarios interiores, utilizando distintos despliegues y centrando la atención en los sistemas de antenas distribuidos o DAS. Además, se propone un algoritmo de precodificación híbrida específico para DAS, y se evalúan y comparan sus prestaciones con las de otros algoritmos de precodificación utilizados. Por último, se investiga el impacto en las prestaciones de ciertas limitaciones prácticas y deficiencias introducidas por el uso de dispositivos no ideales. Finalmente, el estudio de las UDNs se completa con el análisis de su principal limitación, el nivel creciente de interferencia en la red. Para ello, se propone un algoritmo de control de interferencias basado en la partición de recursos. Sus prestaciones son evaluadas y comparadas con las de otras técnicas de asignación de recursos. Tras este estudio, se puede afirmar que las UDNs tienen gran potencial para la consecución de los requisitos de la 5G. Sin embargo, sin el uso conjunto de mayores porciones del espectro, adecuadas técnicas de control de la interferencia y el uso masivo de antenas, las UDNs pueden convertirse en serios obstáculos para los operadores móviles. Los resultados de la evaluación de prestaciones de estas tecnologías confirman el gran aumento de la capacidad de las redes mediante el uso masivo de antenas y la introducción de mecanismos de I / A partir de l'any 2020 es preveu un nou paradigma de comunicació en la societat, degut a l'aparició de nous serveis i la era de la Internet de les coses. El conjunt de requeriments imposat per aquestes noves aplicacions és ampli i divers, i té com a principal objectiu proporcionar connectivitat universal i de banda ampla. En les últimes dècades, la comunitat científica ha estat treballant en la definició de la 5G, que proveirà els mecanismes necessaris per a garantir aquests exigents requeriments. En aquest marc, s'han identificat tres mecanismes claus per a aconseguir l'increment necessari en la capacitat: l'augment de l'eficiència espectral a través de, per exemple, l'ús de tecnologies MIMO massives, la utilització de majors porcions de l'espectre i la densificació mitjançant el desplegament de més estacions base per àrea. Aquesta Tesi aborda la densificació com a principal mecanisme que permetrà la connectivitat de banda ampla i universal requerida en la 5G, centrant-se en l' estudi de les xarxes ultra denses (UDNs). Concretament, el conjunt de tecnologies que poden dur a les UDNs a la seua màxima eficiència i prestacions és analitzat, incloent l'ús d'altes freqüències per a l'aprofitament de majors amplàries de banda, la utilització de MIMO massiu amb sistemes d'antenes distribuïdes i l'ús de tècniques distribuïdes de repartiment de recursos per a la coordinació de la interferència. En primer lloc, aquesta Tesi analitza si existeix un límit fonamental en les prestacions en relació a la densificació. Per això, les prestacions de les UDNs s'avaluen utilitzant un model analític unidimensional amb estacions base equidistants, en les quals la distància entre estacions base es redueix fins assolir el límit de densificació quan aquesta distància s'aproxima a 0. Les taxes assolibles en xarxes amb diferents distàncies entre estacions base s'analitzen considerant diferents nivells de potència i varis graus de cooperació entre cel·les. A més, el comportament de les UDNs s'estudia conjuntament amb l'ús massiu d'antenes i la utilització de majors amplàries de banda. Més concretament, les prestacions de certes tècniques híbrides MIMO de precodificació i beamforming s'examinen en la banda mil·limètrica. D'una banda, els esquemes de beamforming aplicats a estacions base amb arquitectures híbrides és analitzat amb disponibilitat limitada de cadenes de radiofreqüència a un escenari urbà dens. D'altra banda, s'avaluen les prestacions de certs esquemes de precodificació híbrida en escenaris d'interior, utilitzant diferents estratègies de desplegament i centrant l'atenció en els sistemes d' antenes distribuïdes (DAS). A més, es proposa un algoritme de precodificació híbrida distribuïda per a DAS, i s'avaluen i comparen les seues prestacions amb les de altres algoritmes. Per últim, s'investiga l'impacte de les limitacions pràctiques i altres deficiències introduïdes per l'ús de dispositius no ideals en les prestacions de tots els esquemes anteriors. Finalment, l' estudi de les UDNs es completa amb l'anàlisi de la seua principal limitació, el nivell creixent d'interferència entre cel·les. Per tractar aquest problema, es proposa un algoritme de control d'interferències basat en la partició de recursos. Les prestacions de l'algoritme proposat s'avaluen i comparen amb les d'altres tècniques d'assignació de recursos. Una vegada completat aquest estudi, es pot afirmar que les UDNs tenen un gran potencial per aconseguir els ambiciosos requeriments plantejats per a la 5G. Tanmateix, sense l'ús conjunt de majors amplàries de banda, apropiades tècniques de control de la interferència i l'ús massiu d'antenes, les UDNs poden convertir-se en seriosos obstacles per als operadors mòbils. Els resultats de l'avaluació de prestacions d' aquestes tecnologies confirmen el gran augment de la capacitat de les xarxes obtingut mitjançant l'ús massiu d'antenes i la introducci / Giménez Colás, S. (2017). Ultra Dense Networks Deployment for beyond 2020 Technologies [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/86204 / TESIS

Dans les réseaux mobiles du future, un déploiement plus dense des points d’accés radio est prévu pour satisfaire la demande accrue de débit, mais les terminaux utilisateurs peuvent être affectés par une interférence inter-cellulaire plus forte. Par chance, la centralisation des traitements de signal en bande de base dans l’achitecture Cloud RAN (C-RAN) offre la possibilité de la coordination et du traitement conjoint de plusieurs cellules. Pour réellement permettre de déployer ces techniques, une étude bout-à-bout du CRAN est nécessaire selon plusieurs aspects, notamment l’architecture fonctionnelle, la stratégie de coordination, l’implémentation du traitement de signal multiutilisateur et les optimisations possibles pour un fonctionnement plus efficace.Dans cette thèse, nous proposons en premier une architecture qui définit le placement des fonctions du traitement en bande de base entre les unités distribuées et le serveur central. Le but de ce design est de permettre la réalisation des fonctions multi-utilisateurs en transmettant avec la moins de débit possible sur les liens de fronthaul reliant les différentes entités. Dans un second temps, nous présentons comment il est possible de coordiner les différentes cellules servies par le C-RAN en utilisant le concept de réseaux définis par logiciels adapté pour les réseaux d’accès radio. Nous avons mis en place un prototype démontrant la faisabilité de la méthode de contrôle proposée. Finalement, nous étudions l’allocation adaptative du débit sur les liens de fronthaul transportant les symboles numériques quantifiés des utilisateurs en besoin de traitement multi-cellulaire sur la voie montante pour exploiter l’interférence entre eux. Nous proposons un modèle d’optimisation qui inclut le coût des transmissions fronthaul pour maximiser ainsi le gain obtenu par l’opérateur du réseau où la communication multiutilisateur a lieu. Nous réalisons l’optimisation pour différents modèles de coût et en utilisants deux types de données: d’abord les estimations de canal supposées parfaites et disponibles en temps réel, puis seulement les statistiques du canal. Nous montrons que la méthode d’optimisation proposée permet d’exploiter plus efficacement les liens de fronthaul dans l’architecture précedemment définie. / In future mobile networks denser deployment of radio access points is planned to satisfy demand of higher throughput, but an increased number of mobile users can suffer from inter-cell interference. Fortunately, the centralization of base-band processing offered by Cloud Radio Access Network (C-RAN) architecture enables coordination and joint physical layer processing between cells. To make practical deployment of these techniques possible, we have to study C-RAN in an end-to-end view regarding several aspects: the functional architecture of a deployment, the multi-cell coordination strategy, the implementation of multi-user signal processing and possibilities for optimization to increase operational efficiency.In this thesis, first, we propose an architecture defining the placement of base-band processing functions between the distributed remote units and the central processing unit. The aim of this design is to enable multi-cell processing both on the uplink and the downlink while requiring low data rate between the involved entities. Secondly, we study how low latency coordination can be realized inside the central unit using software defined networking adapted to radio access networks. Our demonstration through a real-time prototype deployment shows the feasibility of the proposed control framework. Finally, we investigate adaptive allocation of fronthaul rate that is used for transferring quantized base-band symbols for users participating in uplink multi-cell reception in order to exploit interference between them. We propose an optimization model that includes the cost of fronthaul tranmissions and aims to maximize the gain of network operators from multi-user transmissions in C-RAN. We solve the optimization problem for different fronthaul pricing models, in a scenario where real-time and accurate channel estimates are available and in another where only channel statistics are exploited. Using our method - fitting in the architecture that we have defined - cost efficiency of fronthaul usage can be significantly improved.

The objectives of this work were to investigate, design and implement Multiple-Input Multiple-Output (MIMO) antenna arrays for mobile phones. Several MIMO antennas were developed and tested over various wireless-communication frequency bands. The radiation performance and channel capacity of these antennas were computed and measured: the results are discussed in the context of the frequency bands of interest. A comprehensive study of MIMO antenna configurations such as 2 × 1, 3 × 1, 2 × 2 and 3 × 3, using polarisation diversity as proposed for future mobile handsets, is presented. The channel capacity is investigated and discussed, as applying to Rayleigh fading channels with different power spectrum distributions with respect to azimuth and zenith angles. The channel capacity of 2 × 2 and 3 × 3 MIMO systems using spatial polarisation diversity is presented for different antenna designs. The presented results show that the maximum channel capacity for an antenna contained within a small volume can be reached with careful selection of the orthogonal spatial fields. The results are also compared against planar array MIMO antenna systems, in which the antenna size considered was much larger. A 50% antenna size reduction method is explored by applying magnetic wall concept on the symmetry reference of the antenna structure. Using this method, a triple dual-band inverted-F antenna system is presented and considered for MIMO application. Means of achieving minimum coupling between the three antennas are investigated over the 2.45 GHz and 5.2 GHz bands. A new 2 2 MIMO dual-band balanced antenna handset, intended to minimise the coupling with the handset and human body was proposed, developed and tested. The antenna coupling with the handset and human hand is reported in terms the radiation performance and the available channel capacity. In addition, a dual-polarisation dipole antenna is proposed, intended for use as one of three collocated orthogonal antennas in a polarisation-diversity MIMO communication system. The antenna actually consists of two overlaid electric and magnetic dipoles, such that their radiation patterns are nominally identical but they are cross-polarised and hence only interact minimally.

Multi-User Multi-Input Multi-Output (MU-MIMO) systems are considered to be the sustainable technologies of the current and future of the upcoming wireless and mobile networks generations. The perspectives of these technologies under several scenarios is the focus of the present thesis. The initial system model covers the MU-MIMO, especially in the massive form that is considered to be the promising ideas and pillars of the 5G network. It is observed that the optimal number of users should be served in the time-frequency resource even though the maximum limitation of the MU-MIMO is governed by the total receiving antennas (K) is less than or equal to the base station antennas (M). The system capacity of the massive MIMO (mMIMO) under perfect channel state information (CSI) of uncorrelated channel is investigated and studied. Two types of precoders were applied, one is directly based on channel inversion, and the other uses the Eigen decomposition that is derived subject to the signal to a leakage maximization problem. The two precoders show a degree of equivalency under certain assumptions for the number of antennas at the user end. The convex optimization of multi-antenna networks to achieve the design model of optimum beamformer (BF) based on the uniform linear array (ULA) is studied. The ULA is selected for its simplicity to analyse many scenarios and its importance to match the future network applied millimetre wave (mmWave) spectrum. The maximum beams generated by the ULA are explored in terms of several physical system parameters. The duality between the MU-MIMO and ULA and how they are related based on beamformer operation are detailed and discussed. Finally, two approaches for overloaded systems are presented when the availability of massive array that is not guaranteed due to physical restrictions since the existence of a large number of devices will result in breaking the dimension rule (i.e., K ≤ M). As a solution, a low complexity users selection algorithm is proposed. The channel considered is uncorrelated with full and perfect knowledge at the BS. In particular, these two channel conditions may not be available in all scenarios. The CSI may be imperfect, and even the instantaneous form does not exist. A hybrid precoder between the mixed CSI (includes imperfect and statistical) and rate splitting approach is proposed to deal with an overloaded system under a low number of BS antennas. / Ministry of Higher Education and Scientific Research of Iraq

<p>The Multiple-input and Multiple-output (MIMO) channel model is very useful for the presentation of a wide range of wireless communication systems. This thesis addresses the joint design of a precoder and a receiver for a MIMO channel model, in a scenario in which perfect channel state information (CSI) is available at both ends. We develop a novel framework for the transmitting-receiving procedure. Under the proposed framework, the receiver decomposes the channel matrix by using a block QR decomposition, where Q is a unitary matrix and R is a block upper triangular matrix. The optimal maximum likelihood (ML) detection process is employed within each diagonal block of R. Then, the detected block of symbols is substituted and subtracted sequentially according to the block QR decomposition based successive cancellation. On the transmitting end, the expression of probability of error based on ML detection is chosen as the design criterion to formulate the precoder design problem. This thesis presents a design of MIMO transceivers in the particular case of having 4 transmitting and 4 receiving antennas with full CSI knowledge on both sides. In addition, a closed-form expression for the optimal precoder matrix is obtained for channels satisfying certain conditions. For other channels not satisfying the specific condition, a numerical method is applied to obtain the optimal precoder matrix.</p> / Master of Applied Science (MASc)

No / An investigation into the wavelet packet transform (WPT) modulation scheme for Multiple Input Multiple Output (MIMO) band-limited systems is presented. The implementation involves using the WPT as the base multiplexing technology at baseband, instead of the traditional Fast Fourier Transform (FFT) common in Orthogonal Frequency Division Multiplexing (OFDM) systems. An investigation for a WPT-MIMO multicarrier system, using the Alamouti diversity technique, is presented. Results are consistent with those in the original Alamouti work. The scheme is then implemented for WPT-MIMO and FFTMIMO cases with extended receiver diversity, namely 2 ×Nr MIMO systems, where Nr is the number of receiver elements. It is found that the diversity gain decreases with increasing receiver diversity and that WPT-MIMO systems can be more advantageous than FFT-based MIMO-OFDM systems.

This dissertation addresses multi-layer optimization aspects of multiple input multiple output (MIMO) and deep learning-based communication systems. The initial focus is on the rate optimization for multi-user MIMO (MU-MIMO) configurations; specifically, multiple access channel (MAC) and interference channel (IC). First, the ergodic sum rates of MIMO MAC and IC configurations are determined by jointly integrating the error and overhead effects due to channel estimation (training) and feedback into the rate optimization. Then, we investigated methods that will increase the achievable rate for parallel Gaussian IC (PGIC) which is a special case of MIMO IC where there is no interference between multiple antenna elements. We derive a generalized iterative waterfilling algorithm for power allocation that maximizes the ergodic achievable rate. We verified the sum rate improvement with our proposed scheme through extensive simulation tests. Next, we introduce a novel physical layer scheme for single user MIMO spatial multiplexing systems based on unsupervised deep learning using an autoencoder. Both transmitter and receiver are designed as feedforward neural networks (FNN) and constellation diagrams are optimized to minimize the symbol error rate (SER) based on the channel characteristics. We first evaluate the SER in the presence of a constant Rayleigh-fading channel as a performance upper bound. Then, we quantize the Gaussian distribution and train the autoencoder with multiple quantized channel matrices. The channel is provided as an input to both the transmitter and the receiver. The performance exceeds that of conventional communication systems both when the autoencoder is trained and tested with single and multiple channels and the performance gain is sustained after accounting for the channel estimation error. Moreover, we evaluate the performance with increasing number of quantization points and when there is a difference between training and test channels. We show that the performance loss is minimal when training is performed with sufficiently large number of quantization points and number of channels. Finally, we develop a distributed and decentralized MU-MIMO link selection and activation protocol that enables MU-MIMO operation in wireless networks. We verified the performance gains with the proposed protocol in terms of average network throughput. / Doctor of Philosophy / Multiple Input Multiple Output (MIMO) wireless systems include multiple antennas both at the transmitter and receiver and they are widely used today in cellular and wireless local area network systems to increase robustness, reliability and data rate. Multi-user MIMO (MU-MIMO) configurations include multiple access channel (MAC) where multiple transmitters communicate simultaneously with a single receiver; interference channel (IC) where multiple transmitters communicate simultaneously with their intended receivers; and broadcast channel (BC) where a single transmitter communicates simultaneously with multiple receivers. Channel state information (CSI) is required at the transmitter to precode the signal and mitigate interference effects. This requires CSI to be estimated at the receiver and transmitted back to the transmitter in a feedback loop. Errors occur during both channel estimation and feedback processes. We initially analyze the achievable rate of MAC and IC configurations when both channel estimation and feedback errors are taken into account in the capacity formulations. We treat the errors associated with channel estimation and feedback as additional noise. Next, we develop methods to maximize the achievable rate for IC by using interference cancellation techniques at the receivers when the interference is very strong. We consider parallel Gaussian IC (PGIC) which is a special case of MIMO IC where there is no interference between multiple antenna elements. We develop a power allocation scheme which maximizes the ergodic achievable rate of the communication systems. We verify the performance improvement with our proposed scheme through simulation tests. Standard optimization techniques are used to determine the fundamental limits of MIMO communications systems. However, there is still a gap between current operational systems and these limits due to complexity of these solutions and limitations in their assumptions. Next, we introduce a novel physical layer scheme for MIMO systems based on machine learning; specifically, unsupervised deep learning using an autoencoder. An autoencoder consists of an encoder and a decoder that compresses and decompresses data, respectively. We designed both the encoder and the decoder as feedforward neural networks (FNNs). In our case, encoder performs transmitter functionalities such as modulation and error correction coding and decoder performs receiver functionalities such as demodulation and decoding as part of the communication system. Channel is included as an additional layer between the encoder and decoder. By incorporating the channel effects in the design process of the autoencoder and jointly optimizing the transmitter and receiver, we demonstrate the performance gains over conventional MIMO communication schemes. Finally, we develop a distributed and decentralized MU-MIMO link selection and activation protocol that enables MU-MIMO operation in wireless networks. We verified the performance gains with the proposed protocol in terms of average network throughput.

With the increasing number of wireless devices and the phenomenal amount of data that is being generated by them, there is a growing interest in the wireless communications community to complement the traditional model-driven design approaches with data-driven machine learning (ML)-based solutions. However, managing the large-scale multi-dimensional data to maintain the efficiency and scalability of the ML algorithms has obviously been a challenge. Tensors provide a useful framework to represent multi-dimensional data in an integrated manner by preserving relationships in data across different dimensions. This thesis studies two new applications of tensors to ML for wireless communications where the tensor structure of the concerned data is exploited in novel ways. The first contribution of this thesis is a tensor learning-based low-complexity precoder codebook design technique for a full-dimension multiple-input multiple-output (FD-MIMO) system with a uniform planar antenna (UPA) array at the transmitter (Tx) whose channel distribution is available through a dataset. Represented as a tensor, the FD-MIMO channel is further decomposed using a tensor decomposition technique to obtain an optimal precoder which is a function of Kronecker-Product (KP) of two low-dimensional precoders, each corresponding to the horizontal and vertical dimensions of the FD-MIMO channel. From the design perspective, we have made contributions in deriving a criterion for optimal product precoder codebooks using the obtained low-dimensional precoders. We show that this product codebook design problem is an unsupervised clustering problem on a Cartesian Product Grassmann Manifold (CPM), where the optimal cluster centroids form the desired codebook. We further simplify this clustering problem to a $K$-means algorithm on the low-dimensional factor Grassmann manifolds (GMs) of the CPM which correspond to the horizontal and vertical dimensions of the UPA, thus significantly reducing the complexity of precoder codebook construction when compared to the existing codebook learning techniques. The second contribution of this thesis is a tensor-based bandwidth-efficient gradient communication technique for federated learning (FL) with convolutional neural networks (CNNs). Concisely, FL is a decentralized ML approach that allows to jointly train an ML model at the server using the data generated by the distributed users coordinated by a server, by sharing only the local gradients with the server and not the raw data. Here, we focus on efficient compression and reconstruction of convolutional gradients at the users and the server, respectively. To reduce the gradient communication overhead, we compress the sparse gradients at the users to obtain their low-dimensional estimates using compressive sensing (CS)-based technique and transmit to the server for joint training of the CNN. We exploit a natural tensor structure offered by the convolutional gradients to demonstrate the correlation of a gradient element with its neighbors. We propose a novel prior for the convolutional gradients that captures the described spatial consistency along with its sparse nature in an appropriate way. We further propose a novel Bayesian reconstruction algorithm based on the Generalized Approximate Message Passing (GAMP) framework that exploits this prior information about the gradients. Through the numerical simulations, we demonstrate that the developed gradient reconstruction method improves the convergence of the CNN model. / Master of Science / The increase in the number of wireless and mobile devices have led to the generation of massive amounts of multi-modal data at the users in various real-world applications including wireless communications. This has led to an increasing interest in machine learning (ML)-based data-driven techniques for communication system design. The native setting of ML is {em centralized} where all the data is available on a single device. However, the distributed nature of the users and their data has also motivated the development of distributed ML techniques. Since the success of ML techniques is grounded in their data-based nature, there is a need to maintain the efficiency and scalability of the algorithms to manage the large-scale data. Tensors are multi-dimensional arrays that provide an integrated way of representing multi-modal data. Tensor algebra and tensor decompositions have enabled the extension of several classical ML techniques to tensors-based ML techniques in various application domains such as computer vision, data-mining, image processing, and wireless communications. Tensors-based ML techniques have shown to improve the performance of the ML models because of their ability to leverage the underlying structural information in the data. In this thesis, we present two new applications of tensors to ML for wireless applications and show how the tensor structure of the concerned data can be exploited and incorporated in different ways. The first contribution is a tensor learning-based precoder codebook design technique for full-dimension multiple-input multiple-output (FD-MIMO) systems where we develop a scheme for designing low-complexity product precoder codebooks by identifying and leveraging a tensor representation of the FD-MIMO channel. The second contribution is a tensor-based gradient communication scheme for a decentralized ML technique known as federated learning (FL) with convolutional neural networks (CNNs), where we design a novel bandwidth-efficient gradient compression-reconstruction algorithm that leverages a tensor structure of the convolutional gradients. The numerical simulations in both applications demonstrate that exploiting the underlying tensor structure in the data provides significant gains in their respective performance criteria.