Blind estimation of unknown channel parameters and data symbol detection represent major open problems in non-cooperative communication systems such as automatic modulation classification (AMC). This thesis focuses on estimating the symbol rate and detecting the data symbols. A blind oversampling-based signal detector under the circumstance of unknown symbol period is proposed. The thesis consists of two parts: a symbol rate estimator and a symbol detector. First, the symbol rate is estimated using the EM algorithm. In the EM algorithm, it is difficult to obtain the closed form of the log-likelihood function and the density function. Therefore, both functions are approximated by using the Particle Filter (PF) technique. In addition, the symbol rate estimator based on cyclic correlation is proposed as an initialization estimator since the EM algorithm requires initial estimates. To take advantage of the cyclostationary property of the received signal, there is a requirement that the sampling period should be at least four times less than the symbol period on the receiver side. Second, the blind data symbol detector based on the PF algorithm is designed. Since the signal is oversampled at the receiver side, a delayed multi-sampling PF detector is proposed to manage inter-symbol interference, which is caused by over- sampling, and to improve the demodulation performance of the data symbols. In the PF algorithm, the hybrid importance function is used to generate both data samples and channel model coe±cients, and the Mixture Kalman Filter (MKF) algorithm is used to marginalize out the fading channel coe±cients. At the end, two resampling schemes are adopted.

Echo state network (ESN) is a specific neural network structure composed of high dimensional nonlinear dynamics and learned readout weights. This thesis considers applying ESN for symbol detection in multiple-input, multiple-output orthogonal frequency-division multiplexing (MIMO-OFDM) systems. A new ESN structure, namely, windowed echo state networks (WESN) is introduced to further improve the symbol detection performance. Theoretical analysis justifies WESN has an enhanced short-term memory (STM) compared with the standard ESN such that WESN can offer better computing ability. Additionally, the bandwidth spent as the training set is the same as the demodulation reference signals defined in 3GPP LTE/LTE-Advanced systems for the ESN/WESN based symbol detection. Meanwhile, a unified training framework is developed for both comb and scattered pilot patterns. Complexity analysis demonstrates the advantages of ESN/WESN based symbol detector compared to conventional symbol detectors such as linear minimum mean square error (LMMSE) and sphere decoder when the system is employed with a large number of OFDM sub-carriers. Numerical evaluations show that ESN/WESN has an improvement of symbol detection performance as opposed to conventional methods in both low SNR regime and power amplifier (PA) nonlinear regime. Finally, it demonstrates that WESN can generate a better symbol detection result over ESN. / Artificial neural networks (ANN) are widely used in recognition tasks such as recommendation systems, robotics path planning, self-driving, video tracking, image classifications, etc. To further explore the applications of ANN, this thesis considers using a specific ANN, echo state network (ESN) for a wireless communications task: MIMO-OFDM symbol detection. Furthermore, it proposed an enhanced version of the standard ESN, namely, windowed echo state network (WESN). Theoretical analyses on the short term memory (STM) of ESN and WESN show that the later one has a longer STM. Besides, the training set size of this ESN/WESN based method is chosen the same as the pilot symbols used in conventional communications systems. The algorithm complexity analysis demonstrates the ESN/WESN based method performs with lower complexity compared with conventional methods, such as linear mean square error (LMMSE) and sphere decoding. Comprehensive simulations examine how the symbol detection performance can be improved by using ESN and its variant WESN when the transmission link is non-ideal.

Machine learning (ML) has shown its success in many areas such as computer vision, natural language processing, robot control, and gaming. ML also draws significant attention in the wireless communication society. However, applying ML schemes to wireless communication networks is not straightforward, there are several challenges need to addressed: 1). Training data in communication networks, especially in physical and MAC layer, are extremely limited; 2). The high-dynamic wireless environment and fast changing transmission schemes in communication networks make offline training impractical; 3). ML tools are treated as black boxes, which lack of explainability. This dissertation tries to address those challenges by selecting training-efficient neural networks, devising online training frameworks for wireless communication scenarios, and incorporating communication domain knowledge into the algorithm design. Training-efficient ML algorithms are customized for three communication applications: 1). Symbol detection, where real-time online learning-based symbol detection algorithms are designed for MIMO-OFDM and massive MIMO-OFDM systems by utilizing reservoir computing, extreme learning machine, multi-mode reservoir computing, and StructNet; 2) Channel estimation, where residual learning-based offline method is introduced for WiFi-OFDM systems, and a StructNet-based online method is devised for MIMO-OFDM systems; 3) Radio resource management, where reinforcement learning-based schemes are designed for dynamic spectrum access, as well as ORAN intelligent network slicing management. All algorithms introduced in this dissertation have demonstrated outstanding performance in their application scenarios, which paves the path for adopting ML-based solutions in practical wireless networks. / Doctor of Philosophy / Machine learning (ML), which is a branch of computer science that trains machine how to learn a solution from data, has shown its success in many areas such as computer vision, natural language processing, robot control, and gaming. ML also draws significant attention in the wireless communication society. However, applying ML schemes to wireless communication networks is not straightforward, there are several challenges need to addressed: 1). Training issue: unlike areas such as computer vision where large amount of training data are available, the training data in communication systems are limited; 2). Uncertainty in generalization: ML usually requires offline training, where the ML models are trained by artificially generated offline data, with the assumption that offline training data have the same statistical property as the online testing one. However, when they are statistically different, the testing performance can not be guaranteed; 3). Lack of explainability, usually ML tools are treated as black boxes, whose behaviors can hardly be explained in an analytical way. When designed for wireless networks, it is desirable for ML to have similar levels of explainability as conventional methods. This dissertation tries to address those challenges by selecting training-efficient neural networks, devising online training frameworks for wireless communication scenarios, and incorporating communication domain knowledge into the algorithm design. Training-efficient ML algorithms are customized for three communication applications: 1). Symbol detection, which is a critical step of wireless communication receiver processing, it aims to recover the transmitted signals from the corruption of undesired wireless channel effects and hardware impairments; 2) Channel estimation, where transmitter transmits a special type of symbol called pilot whose value and position are known for the receiver, receiver estimates the underlying wireless channel by comparing the received symbols with the known pilots information; 3) Radio resource management, which allocates wireless resources such bandwidth and time slots to different users. All algorithms introduced in this dissertation have demonstrated outstanding performance in their application scenarios, which paves the path for adopting ML-based solutions in practical wireless networks.

Bridging machine learning technologies to multiple-input-multiple-output (MIMO) communications systems is a primary driving force for next-generation wireless systems. This dissertation introduces a variety of neural network structures for symbol detection/equalization tasks in MIMO systems configured with two different waveforms, orthogonal frequency-division multiplexing (OFDM) and orthogonal time frequency and space (OTFS). The former one is the major air interface in current cellular systems. The latter one is developed to handle high mobility. For the sake of real-time processing, the introduced neural network structures are incorporated with inductive biases of wireless communications signals and operate in an online training manner. The utilized inductive priors include the shifting invariant property of quadrature amplitude modulation, the time-frequency relation inherent in OFDM signals, the multi-mode feature of massive antennas, and the delay-Doppler representation of doubly selective channel. In addition, the neural network structures are rooted in reservoir computing - an efficient neural network computational framework with decent generalization performance for limited training datasets. Therefore, the resulting neural network structures can learn beyond observation and offer decent transmission reliability in the low signal-to-noise ratio (SNR) regime. This dissertation includes comprehensive simulation results to justify the effectiveness of the introduced NN architectures compared with conventional model-based approaches and alternative neural network structures. / Doctor of Philosophy / An important topic for next-generation wireless systems is the integration of machine learning technologies with conventional communications systems. This dissertation introduces several neural network architectures to solve the transmission problems in wireless communications systems. The discussion focuses on the following major modern communications technologies: multiple-input-multiple-output (MIMO), orthogonal frequency-division multiplexing (OFDM), and orthogonal time frequency space (OTFS). In today's cellular networks, MIMO and OFDM are the major air-interface. OTFS is a novel technique that has been designed to work in a high-mobility setting. The implemented neural network structures are integrated with inductive biases of wireless communications signals and operate in an online training mode with limited training datasets. The neural network architectures, in particular, are based on reservoir computing, which is an efficient neural network computational system. A learning algorithm's inductive bias (also known as learning bias) is a collection of assumptions that the learner makes to infer outputs from unknown inputs. The dissertation introduces four different inductive priors from four different perspectives of MIMO communications systems. As a result, the neural network architectures can learn beyond observation and provide good generalization output in scenarios having model mismatch issues. The dissertation provides extensive simulation results to support the efficacy of the implemented NN architectures compared to alternative neural network models and traditional model-based approaches.

<p>Piping is an essential component in buildings, and its as-built information is critical to facility management tasks. Manually extracting piping information from legacy drawings that are in paper, PDF, or image format is mentally exerting, time-consuming, and error-prone. Symbol recognition and detection are core problems in the computer-based interpretation of piping drawings, and the main technical challenge is to determine robust features that are invariant to scaling, rotation, and translation. This thesis aims to use convolutional neural networks (CNNs) to automatically extract features from raw images, and consequently, to locate and recognize symbols in piping drawings.</p> <p>In this thesis, the Spatial Transformer Network (STN) is applied to improve the performance of a standard CNN model for recognizing piping symbols, and the Faster Region-based Convolutional Neural Network (Faster RCNN) is adopted to exploit its capacity in symbol detection. For experimentation, the synthetic data are generated as follows. Two datasets are generated for symbol recognition and detection, respectively. For recognition, eight types of symbols are synthesized based on the geometric constraints between the primitives. The drawing samples for detection are manually sketched using AutoCAD MEP software and its piping component library, and seven types of symbols are selected from the piping component library. Both sets of samples are augmented with various scales, rotations, and random noises.</p> <p>The experiment for symbol recognition is conducted and the accuracies of the recognition accuracy of the CNN + STN model and the standard CNN model are compared. It is observed that the spatial transformer layer improves the accuracy in classifying piping symbols from 95.39% to 98.26%. For the symbol detection task, the experiment is conducted using a public implementation of Faster RCNN. The mean Average Precision (mAP) is 82.8% when Intersection over Union (IoU) threshold equals to 0.5. Imbalanced data (i.e., imbalanced samples in each class) led to a decrease in the Average Precision in the minority class. Also, the symbol library, the small dataset, and the complex backbone network limit the generality of the model. Future work will focus on the collection of larger set of drawings and the improvement of the network’s geometric invariance.</p>

Echo State Network (ESN) is a recently developed machine-learning paradigm whose processing capabilities rely on the dynamical behavior of recurrent neural networks (RNNs). Its performance metrics outperform traditional RNNs in nonlinear system identification and temporal information processing. In this thesis, we design and implement ESNs through Field-programmable gate array (FPGA) and explore their full capacity of digital signal processors (DSPs) to target low-cost and low-power applications. We propose a cost-optimized and scalable ESN architecture on FPGA, which exploits Xilinx DSP48E1 units to cut down the need of configurable logic blocks (CLBs). The proposed work includes a linear combination processor with negligible deployment of CLBs, as well as a high-accuracy non-linear function approximator, both with the help of only 9 DSP units in each neuron. The architecture is verified with the classical NARMA dataset, and a symbol detection task for an orthogonal frequency division multiplexing (OFDM) system on a wireless communication testbed. In the worst-case scenario, our proposed architecture delivers a matching bit error rate (BER) compares to its corresponding software ESN implementation. The performance difference between the hardware and software approach is less than 6.5%. The testbed system is built on a software-defined radio (SDR) platform, showing that our work is capable of processing the real-world data. / Master of Science / Machine learning is a study of computer algorithms that evolves itself by learning through experiences. Currently, machine learning thrives as it opens up promising opportunities of solving the problems that is difficult to deal with conventional methods. Echo state network (ESN), a recently developed machine-learning paradigm, has shown extraordinary effectiveness on a wide variety of applications, especially in nonlinear system identification and temporal information processing. Despite the fact, ESN is still computationally expensive on battery-driven and cost-sensitive devices. A fast and power-saving computer for ESN is desperately needed. In this thesis, we design and implement an ESN computational architecture through the field-programmablegate array (FPGA). FPGA allows designers to build highly flexible customized hardware with rapid development time. Our design further explores the full capacity of digital signal processors (DSP) on Xilinx FPGA to target low-cost and low-power applications. The proposed cost-optimized and scalable ESN architecture exploits Xilinx DSP48E1 units to cut down the need of configurable logic blocks (CLBs). The work includes a linear combination processor with negligible deployment of CLBs, and a high-accuracy non-linear function approximator, both with the help of only 9 DSP units in each neuron. The architecture is verified with the classical NARMA dataset, and a symbol detection task for an orthogonal frequency division multiplexing (OFDM) system in a wireless communication testbed. In the worst-case scenario, our proposed architecture delivers a matching bit error rate (BER) compares to its corresponding software ESN implementation. The performance difference between the hardware and software approach is less than 6.5%. The testbed system is built on a software-defined radio (SDR) platform, showing that our work is capable of processing the real-world data.

The increasing demand for high data rate transmission over broadband radio channels has imposed significant challenges in wireless communications. Accurate channel estimation has a major impact on the whole system performance. Specifically, reliable estimate of the channel state information (CSI) is more challenging for orthogonal frequency division multiplexing (OFDM) systems in doubly selective fading channels than for the slower fading channels over which OFDM has been deployed traditionally. With the help of a basis expansion model (BEM), a novel multivariate autoregressive (AR) process is developed to model the time evolution of the fast fading channel. Relying on pilot symbol aided modulation (PSAM), a novel Kalman smoothing algorithm based on a second-order dynamic model is exploited, where the mean square error (MSE) of the channel estimator is near to that of the optimal Wiener filter. To further improve the performance of channel estimation, a novel low-complexity iterative joint channel estimation and symbol detection procedure is developed for fast fading channels with a small number of pilots and low pilot power to achieve the bit error rate (BER) performance close to when the CSI is known perfectly. The new channel estimation symbol detection technique is robust to variations of the radio channel from the design values and applicable to multiple modulation and coding types. By use of the extrinsic information transfer (EXIT) chart, we investigate the convergence behavior of the new algorithm and analyze the modulation, pilot density, and error correction code selection for good system performance for a given power level. The algorithms developed in this thesis improve the performance of the whole system requiring only low ratios of pilot to data for excellent performance in fast fading channels. / Graduate

Submitted by isabela.moljf@hotmail.com (isabela.moljf@hotmail.com) on 2017-05-30T12:08:23Z No. of bitstreams: 1 gléversonfabnercondelemos.pdf: 2102819 bytes, checksum: e934ec8e8bf0daaaa39a52749b708828 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-05-30T13:52:51Z (GMT) No. of bitstreams: 1 gléversonfabnercondelemos.pdf: 2102819 bytes, checksum: e934ec8e8bf0daaaa39a52749b708828 (MD5) / Made available in DSpace on 2017-05-30T13:52:51Z (GMT). No. of bitstreams: 1 gléversonfabnercondelemos.pdf: 2102819 bytes, checksum: e934ec8e8bf0daaaa39a52749b708828 (MD5) Previous issue date: 2011-09-12 / A presente dissertação versa sobre técnicas de baixo custo para detecção, modulação e demodulação de constelações M-QAM (quadrature amplitude modulation) de ordem elevada, ou seja, M = 2n, n = {2,3, · · · ,16}. Al´em disso, s˜ao propostas constelações alternativas para M-QAM, M = 22n, n = {1,2, · · · ,8}, que buscam minimizar a PAPR (peak to average power ratio) quando um sistema OFDM (orthogonal frequency division multiplexing) ´e utilizado para a transmissão de dados. Uma implementação, de baixo custo e em dispositivo FPGA (field programmable gate array), de um esquema de modulação constante e adaptativa para sistemas OFDM, quando a modulação é MQAM, M = 22n, n = {1,2, · · · ,8}, é descrita e analisada. O desempenho das técnicas de detecção propostas é avaliado através de simulações computacionais quando o ruído é AWGN (additive white Gaussian noise) e AIGN (additive impulsive Gaussian noise). Os resultados em termos de BER × Eb/N0 indicam que as perdas de desempenho geradas com as técnicas propostas não são significativas e, portanto, tais técnicas são candidatas adequadas para a implementação de um sistema OFDM com elevada eficiência espectral. Os resultados computacionais revelam ainda que as propostas alternativas para constelações M-QAM reduzem a PAPR, mas, em contrapartida, degradam consideravelmente a BER. Finalmente, a análise da complexidade computacional das técnicas de detecção e demodulação, as quais foram implementadas em dispositivo FPGA, indica que há uma redução do custo computacional, ou seja, redução do uso de recursos de hardware do dispositivo FPGA quando tais técnicas são implementadas para a demodulação e detecção de símbolos M-QAM de ordem elevada. / This dissertation deals with low-cost techniques for detection, modulation and demodulation of high order M-QAM (quadrature amplitude modulation) constellations, i.e., M = 2n, n = {2,3, · · · ,16}. In addition, alternative constellations are proposed to M-QAM, M = 22n, n = {1,2, · · · ,8}, which seek to minimize the PAPR (peak to average power ratio) when an OFDM (orthogonal frequency division multiplexing) system is used for data transmission. A low-cost implementation using a FPGA (field programmable gate array) device of a modulation scheme for constant and adaptive OFDM systems when the modulation is M-QAM, M = 22n, n = {1,2, · · · ,8}, is described and analyzed. The performance of the proposed detection techniques is evaluated through computer simulations when the noise is AWGN (additive white Gaussian noise) and AIGN (additive impulsive Gaussian noise). The results in terms of BER × Eb/N0 indicate that the performance losses generated by the proposed techniques are not significant and, therefore, such techniques are appropriate candidates for the implementation of an OFDM system with high spectral efficiency. The computational results reveal that the alternative proposals for M-QAM constellations reduce the PAPR, but, considerably degrade the BER. Finally, the analysis of computational complexity of detection and demodulation techniques, which were implemented in a FPGA device, indicates that there is a computational cost reduction, i.e., a reduction of resource usage of hardware device such as FPGA when these techniques are implemented for the demodulation and detection of high-order M-QAM symbols.