Search

© 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting /republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works, In this paper, a non-linear HRV analysis was performed to assess fragmentation signatures observed in heartbeat time series after intermittent hypoxia (IH). Three markers quantifying short-term fragmentation levels, PIP, IALS and PSS, were evaluated on R-R interval series obtained in a rat model of recurrent apnea. Through airway obstructions, apnea episodes were periodically simulated in six anesthetized Sprague-Dawley rats. The number of apnea events per hour (AHI index) was varied during the first half of the experiment while apnea episodes lasted 15 s. For the second part, apnea episodes lasted 5, 10 or 15 s, but the AHI index was fixed. Recurrent apnea was repeated for 15-min time intervals in all cases, alternating with basal periods of the same duration. The fragmentation markers were evaluated in segments of 5 minutes, selected at the beginning and end of the experiment. The impact of the heart and breathing rates (HR and BR, respectively) on the parameter estimates was also investigated. The results obtained show a significant increase (from 5 to 10%, p <; 0.05) in fragmentation measures of heartbeat time series after IH, indicating a clear deterioration of the initial conditions. Moreover, there was a strong linear relationship (r > 0.9) between these markers and BR, as well as with the ratio given by HR/BR. Although fragmentation may be impacted by IH, we found that it is highly dependent on HR and BR values and thus, they should be considered during its calculation or used to normalize the fragmentation estimates, Peer Reviewed, Postprint (published version)

Obstructive sleep apnea (OSA) is a prevalent disease, but most patients remain undiagnosed and untreated. Here we propose analyzing smartphone audio signals for screening OSA patients at home. Our objectives were to: (1) develop an algorithm for detecting silence events and classifying them into apneas or hypopneas; (2) evaluate the performance of this system; and (3) compare the information provided with a type 3 portable sleep monitor, based mainly on nasal airflow. Overnight signals were acquired simultaneously by both systems in 13 subjects (3 healthy subjects and 10 OSA patients). The sample entropy of audio signals was used to identify apnea/hypopnea events. The apnea-hypopnea indices predicted by the two systems presented a very high degree of concordance and the smartphone correctly detected and stratified all the OSA patients. An event-by-event comparison demonstrated good agreement between silence events and apnea/hypopnea events in the reference system (Sensitivity = 76%, Positive Predictive Value = 82%). Most apneas were detected (89%), but not so many hypopneas (61%). We observed that many hypopneas were accompanied by snoring, so there was no sound reduction. The apnea/hypopnea classification accuracy was 70%, but most discrepancies resulted from the inability of the nasal cannula of the reference device to record oral breathing. We provided a spectral characterization of oral and nasal breathing to correct this effect, and the classification accuracy increased to 82%. This novel knowledge from acoustic signals may be of great interest for clinical practice to develop new non-invasive techniques for screening and monitoring OSA patients at home, Peer Reviewed, Postprint (published version)

© 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting /republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works, Hypoxia induced by obstructive sleep apnea (OSA) leads to the deregulation of the autonomic nervous system (ANS), resulting in an abnormally increased sympathetic activity. Since ANS modulation varies throughout the night, notably for each sleep stage, the hypno-gram and heart rate signals of 81 OSA patients were collected during a polysomnography. They were classified as mild-moderate (n=44) or severe (n=37) based on their apnea-hypopnea index (AHI). Spectral heart rate variability (HRV) series were extracted by a time-frequency approach. These series were then averaged for each sleep stage, in order to compare the sympathetic modulation of mild-moderate and severe patients at the following phases: rapid eye movement (REM), S1, S2 and SWS (slow wave sleep). According to normalized power at the low-frequency band (LF nu ) values, severe OSA seems to be associated with an increased sympathetic modulation at non-REM sleep. Moreover, a decreased autonomic variability throughout the night may be related to a reduced adaptability of the cardiovascular system, characterizing a more advanced stage of the disease. These results provide further evidence for the role of autonomic alterations induced by hypoxia, suggesting the use of HRV analysis, together with AHI, for the study of OSA severity, Peer Reviewed, Postprint (author's final draft)

This master’s thesis presents the first stages of a project which aims to build an accurate and efficient capacitive bee-counter, a device that can count the number of bees entering or exiting a hive. This device has an essential role in protecting this species from hazardous pesticides, but current systems in the market are not convenient due to their lack of accuracy or their unsuitable functionality. First, capacitive sensor development is presented. Taking into account the design constraints of this device and its environment, a capacitive sensor is developed to ensure high precision into the measurements. The study follows testing out the devised capacitor with different Capacitive-to-Digital converters. From this experimentation, the sensor performance is evaluated and a suitable converter is chosen. Employing these elements, a prototype is built to gather data in a real environment. Then, data samplings are performed using the prototype in an apiary. Inspecting the signals collected, conclusions regarding the measurement technique are achieved and series of bees’ behaviours are detected. After this first experimentation, the prototype is upgraded so it can be employed in any hive, improving its design and attaining new information during following testing sessions. Finally, the different signal processing techniques trialled out are described. Their objective was to perform or aid in the bee-counting process. The results of their application are displayed, explaining its dismissal or possible utilization. Later, a suitable methodology for the end product is proposed. Along with its definition, the best results obtained by its implementation employing MATLAB® are presented and pointers for future improvements in its performance are described, Outgoing

© 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., The security in satellite communications is a key issue due to the large footprint of the beams. This is specially critical in IoT devices that transmit data directly to satellite. Take into account that IoT devices are characterized by transmitting packets of short length. Consequently, it means that it is not feasible to augment the security level of the IoT packets via complex cryptographic algorithms. Otherwise, their packet lengths may be increased in a non-negligible way which could augment their collision probabilities, latencies and energy consumptions. For this reason, this paper proposes to take advantage of the time-packing technique. By doing so, it is possible to use the overlapping degree among the pulse-shapes to boost the secrecy-capacity. In particular, the overlapping degree between the pulse-shapes introduces an artificial interference that degrades the eavesdropper's channel. In this regard, it is necessary to highlight that there is a residual co-channel interference in the satellite beams. So, it means that these two sources of impairments make difficult to estimate the legitimate user's transmission parameters by the eavesdropper., Peer Reviewed, Postprint (author's final draft)

La estimación experimental de la receptancia dinámica requiere de mediciones precisas para la determinación de los parámetros dinámicos tales como el factor de amortiguamiento y las frecuencias de resonancia en superestructuras ferroviarias. En este trabajo se propone una metodología de medición basada en un dispositivo excitador con un sistema de masa excéntrica colocado sobre un elastómero ferroviario. Este dispositivo se alimenta con una señal de tipo barrido sinusoidal y como resultado del ensayo se obtienen señales de aceleración. Estas señales se procesan para obtener las densidades de potencia espectral (PSD) y se comparan con el ruido de fondo para escoger la mejor relación señal-ruido (SNR). Además, los diferentes parámetros experimentales como tasa de barrido, tiempo de ensayo y amplitud de fuerza son obtenidos de acuerdo a la menor varianza de las mediciones para la estimación de la receptancia, y considerando la presencia del ruido de fondo durante las mediciones experimentales. Como resultado final de la metodología propuesta, son obtenidas de la muestra ensayada, la receptancia dinámica y la frecuencia de resonancia debido a las mediciones que son procesadas considerando el análisis espectral de la mejor relación señal-ruido y la menor varianza, Postprint (published version)

Over the last years, i-vectors have been the state-of-the-art approach in speaker recognition. Recent improvements in deep learning have increased the discriminative quality of i-vectors. However, deep learning architectures require a large amount of labeled background data which is difficult in practice. The aim of this paper is to propose an alternative scheme in order to reduce the need of labeled data. We propose the use of autoencoder pre-training in a speaker verification task. First, we train an autoencoder in an unsupervised way, using a large amount of unlabeled background data. Then, we train a Deep Neural Network (DNN) initialized with the parameters of the pre-trained autoencoder. The DNN training is carried out in a supervised way using relatively small labeled background data. In the testing phase, we extract speaker embeddings as the output of an intermediate layer of the DNN. The training and evaluation were performed on VoxCeleb-2 and VoxCeleb1 databases, respectively. The experimental results have shown that by initializing DNN with the parameters of the pre-trained autoencoder, we have achieved a relative improvement of 21%, in terms of Equal Error Rate (EER), over the baseline i-vector/PLDA system., Peer Reviewed, Postprint (published version)

© 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Matrix completion and extrapolation (MCEX) are dealt with here over reproducing kernel Hilbert spaces (RKHSs) in order to account for prior information present in the available data. Aiming at a fast and low-complexity solver, the task is formulated as one of kernel ridge regression. The resultant MCEX algorithm can also afford online implementation, while the class of kernel functions also encompasses several existing approaches to MC with prior information. Numerical tests on synthetic and real datasets show that the novel approach is faster than widespread methods such as alternating least-squares (ALS) or stochastic gradient descent (SGD), and that the recovery error is reduced, especially when dealing with noisy data., Peer Reviewed, Postprint (author's final draft)

This research received funding of the Mexican National Council of Science and Technology (CONACYT), Grant (no. 490180). Also, this work was supported by the Program for Professional Development Teacher (PRODEP)., In this work, three specific machine learning techniques (neural networks, expectation maximization and k-means) are applied to a multiband spectrum sensing technique for cognitive radios. All of them have been used as a classifier using the approximation coefficients from a Multiresolution Analysis in order to detect presence of one or multiple primary users in a wideband spectrum. Methods were tested on simulated and real signals showing a good performance. The results presented of these three methods are effective options for detecting primary user transmission on the multiband spectrum. These methodologies work for 99% of cases under simulated signals of SNR higher than 0 dB and are feasible in the case of real signals, Peer Reviewed, Postprint (published version)

The study of the ECG signalsis essential to detect several diseases. The present projectaims to analyze the ECG signals in order to detect different types of heart beat sassociated with arrhythmia, using data from the MIT-BIH Arrhythmia database. The proposed pipeline is the following: (a) design and program a data visualizer for the MIT-BIH Arrhythmia database; (b) reduce the dimensionality of the data using a Principal Component Analysis; and (c) compress thebeats using a deep autoencoder.The first step of the analysis consists of an approach to the MIT-BIH Arrhythmia databasefor better understanding on how the data is presented. For this purpose, an ECG data visualizer is designed. This application allows the user to select several graph settings and records of the database and plots the selected signals.Diverse graphic and GUI libraries are considered, such as WFDB, TkInter and Bokeh. But the package that best fits the requirements in this case is Dash, which includes both the graph and the user interface and updates the changed settings in real time.The second step of the analysis consists of reducing the dimensionality of the data by means of a Principal Component Analysis. PCA has been performed for each one of the following types of heartbeats: normal beat (N); left bundle branch block beat (L); right bundle branch block beat (R); premature ventricular contraction (V); and atrial premature beat (A). The importance of every principal component for the shape of each beat is discussed in this section.The final step is to compress the data using a deep auto encoder. The encoder pretends to represent an input heart beat in a reduced number of neurons. The decoder reconstructs the initial beat using the same configuration of layers but arranged in reverse order. The firstproposal is a convolutional autoencoder with 10 neurons. The obtained loss is0.228 (22.8%). The correlation coefficient between the input vector and the output of the autoencoder is 0.99; 0.89; 0.

© 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., This paper deals with the distributed subspace agreement problem for opportunistic communications in time division duplex (TDD) distributed networks. Since scenario-adapted opportunistic transmission schemes rely on locally sampled observations from the wireless environment, degrees-of-freedom (DoF) sensed as available at any node may differ. Transmitting information without agreeing the common active subspace may incur in a performance loss due to noise enhancement, energy loss and inter-system interference. In this context, we propose two subspace concurrence schemes with and without side information about neighboring user's DoF., Peer Reviewed, Postprint (published version)

This paper shows that de Bruijn identity, which relates entropy with Fisher information, can be obtained as a particular case of an immediate generalization of Price’s theorem, which is a tool used in the analysis of nonlinear memoryless systems with Gaussian inputs. It is shown that, while the general Price’s theorem follows since the density of the perturbation satisfies the heat equation, the particular case of de Bruijn identity follows since the score function is zero-mean, which is thewell-knownconditionthatprovidestheinsightfulCrame ´r- Rao bound expression based on the negative second derivative of the log-likelihood function. The unified framework uses the characteristic function as a main tool and becomes a more intuitive alternative to the classical technical proof obtained by integrating by parts. Second-order Tsallis entropy is also briefly explored under this general framework., Peer Reviewed, Postprint (published version)

The paper proposes a robust faults detection and forecasting approach for a centrifugal gas compressor system, the mechanism of this approach used the Kalman filter to estimate and filtering the unmeasured states of the studied system based on signals data of the inputs and the outputs that have been collected experimentally on site. The intelligent faults detection expert system is designed based on the interval type-2 fuzzy logic. The present work is achieved by an important task which is the prediction of the remaining time of the system under study to reach the danger and/or the failure stage based on the Auto-regressive Integrated Moving Average (ARIMA) model, where the objective within the industrial application is to set the maintenance schedules in precisely time. The obtained results prove the performance of the proposed faults diagnosis and detection approach which can be used in several heavy industrial systems, Peer Reviewed, Postprint (published version)

Despite the progress in understanding of neural codes, the studies of the cortico-muscular coupling still largely rely on interferential electromyographic (EMG) signal or its rectification for the assessment of motor neuron pool behavior. This assessment is non-trivial and should be used with precaution. Direct analysis of neural codes by decomposing the EMG, also known as neural decoding, is an alternative to EMG amplitude estimation. In this study, we propose a fully-deterministic hybrid surface EMG (sEMG) decomposition approach that combines the advantages of both template-based and Blind Source Separation (BSS) decomposition approaches, a.k.a. guided source separation (GSS), to identify motor unit (MU) firing patterns. We use the single-pass density-based clustering algorithm to identify possible cluster representatives in different sEMG channels. These cluster representatives are then used as initial points of modified gradient Convolution Kernel Compensation (gCKC) algorithm. Afterwards, we use the Kalman filter to reduce the noise impact and increase convergence rate of MU filter identification by gCKC. Moreover, we designed an adaptive soft-thresholding method to identify MU firing times out of estimated MU spike trains. We tested the proposed algorithm on a set of synthetic sEMG signals with known MU firing patterns. A grid of 9 × 10 monopolar surface electrodes with 5-mm inter-electrode distances in both directions was simulated. Muscle excitation was set to 10, 30, and 50%. Colored Gaussian zero-mean noise with the signal-to-noise ratio (SNR) of 10, 20, and 30 dB, respectively, was added to 16 s long sEMG signals that were sampled at 4,096 Hz. Overall, 45 simulated signals were analyzed. Our decomposition approach was compared with gCKC algorithm. Overall, in our algorithm, the average numbers of identified MUs and Rate-of-Agreement (RoA) were 16.41 ± 4.18 MUs and 84.00 ± 0.06%, respectively, whereas the gCKC identified 12.10 ± 2.32 MUs with the average R, Peer Reviewed, Postprint (published version)

Automatic Dependent Surveillance-Broadcast (ADS_B) system is one of the most important innovation of radar system which is used for direct state information exchange between aircraft. The project consists of the study of ADS-B signal received from Barcelona airport. The low-cost device RTL-SDR is used to capture the signal and then be processed by Matlab. Through the Matlab demo execution such information can be extracted, per example, altitude, position, velocity, aircraft identification, etc. These information during the flight usually are displayed to cockpit, to enhance the situational awareness of pilot and its safety. Through the Simulink the spectrum of the ADS-B signal can be obtained and be studied. The capture of signal is taken in different places to see the influence the distance at the receiving signal process. The present paper is organized as following structure: firstly, there is an overview of ADS-B system in the world. Second chapter talks about the background of the ADS-B system and deeper study of it. After it comes the study of the captured signal by RTL-SDR. The study of aerial traffic is followed the study of the signal. At the final part, the conclusion is reported., El sistema automático de transmisión de vigilancia dependiente (ADS_B) es una de las innovaciones más importantes del sistema de radar que se utiliza para el intercambio directo de información de estado entre aeronaves. El proyecto consiste en el estudio de la señal ADS-B recibida desde el aeropuerto de Barcelona. El dispositivo de bajo costo RTL-SDR se utiliza para capturar la señal y luego ser procesado por Matlab. A través de la ejecución de la demo de Matlab, se puede extraer dicha información, por ejemplo, altitud, posición, velocidad, identificación de la aeronave, etc. Esta información durante el vuelo generalmente se muestra a la cabina, para mejorar el conocimiento de la situación del piloto y su seguridad. A través de Simulink, se puede obtener y estudiar el espectro de la señal ADS-B. La captura de la señal se toma en diferentes lugares para ver la influencia de la distancia en el proceso de la recepción de señal. El presente documento está organizado según la siguiente estructura: en primer lugar, se presenta una descripción general del sistema ADS-B en el mundo. El segundo capítulo habla sobre los antecedentes del sistema ADS-B y su estudio más profundo. Luego viene el estudio de la señal capturada por RTL-SDR. El estudio de tráfico aéreo con los datos obtenidos viene después del estudio de señal. Por lo último, la conclusión del trabajo.

Laser-generated ultrasound is a modern non-destructive testing technique. It has been investigated over recent years as an alternative to classical ultrasonic methods, mainly in industrial maintenance and quality control procedures. In this study, the detection and reconstruction of internal defects in a metallic sample is performed by means of a time-frequency analysis of ultrasonic waves generated by a laser-induced thermal mechanism. In the proposed methodology, we used wavelet transform due to its multi-resolution time frequency characteristics. In order to isolate and estimate the corresponding time of flight of eventual ultrasonic echoes related to internal defects, a density-based spatial clustering was applied to the resulting time frequency maps. Using the laser scan beam’s position, the ultrasonic transducer’s location and the echoes’ arrival times were determined, the estimation of the defect’s position was carried out afterwards. Finally, clustering algorithms were applied to the resulting geometric solutions from the set of the laser scan points which was proposed to obtain a two-dimensional projection of the defect outline over the scan plane. The study demonstrates that the proposed method of wavelet transform ultrasonic imaging can be effectively applied to detect and size internal defects without any reference information, which represents a valuable outcome for various applications in the industry. View Full-Text, Peer Reviewed, Postprint (published version)

Classical signal processing techniques are developed under prior statistical knowledge of the kind of data we are processing. Unfortunately, one does not know too much about reality in practice, therefore inferring information about the statistical behaviour of our data is needed in addition of processing. To properly infer information about the statistics, one may need more than one realization of the experiment, although this is not possible in some real-life environments, as one may have just a single realization of the random process or low amount of available samples. If one would like to tackle both issues at once using classical signal processing approaches, it may lead to an almost impossible problem. This is where Bootstrap statistical inference shines, which suits perfectly this kind of problems. Moreover, we want to fuse all available data, which comes from different sources, to improve our knowledge of the working environment and eventual accuracy in further operations, such as estimating some parameter. However, there is a risk of fusing too much corrupted data without taking into account how contaminated a data set is, so the integrity of the final estimation gets compromised. We will still consider small amount of data available to tackle this issue. In reponse to the mentioned problems, the purpose of this project is to explore and analyze the potentials of the Bootstrap techniques. In particular, we will focus on the issues of data integrity and getting benefits from data redundancy in Precise Point Positioning receivers, whose context suits perfectly this kind of framework., Las técnicas clásicas de procesamiento de señales se desarrollan a partir de conocimiento a priori de los datos que procesamos. Desgraciadamente, en la práctica uno puede desconocer este conocimiento a priori, entonces, tenemos la necesidad de inferir información del comportamiento estadístico de los datos, además del propio procesamiento. Para inferir de manera precisa, es necesaria más de una realización del experimento, pero es posible que el entorno de estudio no nos lo permita o haya datos insuficientes. Si se intentase resolver este problema usando técnicas clásicas, uno es podría encontrar un problema casi imposible. Aquí es donde las técnicas Bootstrap brillan, puesto que se adaptan muy bien a este tipo de problemas. Además del problema mencionado, también queremos fusionar todas los posibles fuentes de información, para mejorar el conocimiento que disponemos del medio de estudio y la eventual precisión en próximas operaciones, como estimar un parámetro dado. Por el contrario, hay un riesgo de fusionar demasiados datos corruptos, entonces la integridad de la estimación final se vería comprometida. Como respuesta de los problemas mencionados, el objetivo de este proyecto es el estudio de las técnicas Bootstrap en entornos que se adapten bien a los problemas mencionados., Les tècniques clàssiques de processament de senyal es desenvolupen a partir de coneixement a priori de les dades que processem. Malauradament, a la pràctica un pot desconèixer aquest coneixement, aleshores, hi ha la necessitat d'inferir informació del comportament estadístic de les dades, a més del propi processament. Per inferir de manera acurada, és necessària més d'una realització de l'experiment, però és possible que l'entorn d'estudi no ens ho permeti o hi hagi dades insuficients. Si s'intentés resoldre aquest problema fent servir tècniques clàssiques, un es podria trobar un problema gairebé impossible. Aquí és on les tècniques Bootstrap brillen, ja que s'adapten molt be a aquest tipus de problemes. A més del problema esmentat, també volem fusionar totes les possibles fonts d'informació, per millorar el coneixement que disposem del medi d'estudi i l'eventual precisió en pròximes operacions, com estimar un paràmetre donat. Per contra, hi ha un risc de fusionar masses dades corruptes, llavors la integritat de l'estimació final es veuria compromesa. Com a resposta dels problemes esmentats, l'objectiu d'aquest projecte és l'estudi de les tècniques Bootstrap en entorns que s'adaptin bé als problemes esmentats.

© 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., This work addresses the optimization of the network spectral efficiency (SE) under successive interference cancellation (SIC) at a given blocklength n. We adopt a proof-of-concept satellite scenario where network users can vary their transmission power and select their transmission rate from a set of encoders, for which decoding is characterized by a known packet error rate (PER) function. In the large-system limit, we apply variational calculus (VC) to obtain the user-energy distribution, the assigned per-user rate and the SIC decoding order maximizing the network SE under a sum-power constraint at the SIC input. We analyze two encoder sets: (i) an infinite set of encoders achieving information-theoretic finite blocklength PER results over a continuum of code rates, where the large-n second order expansion of the maximal channel coding rate is used; (ii) a feasible finite set of encoders. Simulations quantify the performance gap between the two schemes., Peer Reviewed, Postprint (author's final draft)

Real-time detection of cycle-slips on undifferenced carrier-phase measurements is an important task to properly exclude wrong phase trackers from precise positioning algorithms. The detection is especially challenging in high-dynamic mobile scenarios, where traditional approaches (as those based on single-channel polynomial fitting) may easily lead to false positives. Using a multi-channel formulation of the problem, the proposed technique takes benefit of the available data redundancy (high number of tracked satellites) in order to ameliorate the false positives. This robustness is accomplished by adaptively estimating the orthogonal subspace spanned by the polynomial time-varying residuals obtained from all available channels (treated as a vector process), and using that subspace to form efficient channel combinations with cancelled satellite-receiver dynamics. The main advantage of the multi-channel approach is that wrong measurements can be discarded without needing any positioning estimate nor phase-ambiguity solver, thus improving the accuracy, reliability and integrity of positioning. The performance improvement is shown by means of theoretical analysis and computer simulations., Peer Reviewed, Postprint (published version)

The procedures for condition monitoring of electromechanical systems are undergoing a reformulation, mainly, due to the current thermographic affordability of infrared cameras to be incorporated in industrial applications. However, high-performing multifault data-driven methodologies must be investigated in order to infer reliable condition information from the thermal distribution of not only electrical motors but also of shafts and couplings. To address this issue, a novel thermography-based methodology is proposed. First, the infrared capture is processed to obtain a thermographic residual image of the kinematic chain. Second, the thermal distribution of the image's regions of interest is characterized by means of statistical features. Finally, a distributed self-organizing map structure is used to model the nominal thermal distribution to subsequently perform a fault detection and identification. The method provides a reliability quantification of the resulting condition assessment in order to avoid misclassifications and identify the actual fault root-causes. The performance and the effectiveness of the proposed methodology is validated experimentally and compared with the classical maximum temperature gradient procedure., Peer Reviewed, Postprint (published version)

Whilst research efforts have traditionally focused on Electrocardiographic (ECG) signals and handcrafted features as potential biometric traits, few works have explored systems based on the raw photoplethysmogram (PPG) signal. This work proposes an end-to-end architecture to offer biometric authentication using PPG biosensors through Convolutional Neural Networks. We provide an evaluation of the performance of our approach in two different databases: Troika and PulseID, the latter a publicly available database specifically collected by the authors for such a purpose. Our verification approach through convolutional network based models and using raw PPG signals appears to be viable in current monitoring procedures within e-health and fitness environments, and shows a remarkable potential as a biometric identifier. When tested on a verification task with one second trials, the approach achieved an AUC of $78.2\%$ and $83.2\%$, averaged among target subjects, on PulseID and Troika datasets respectively. Our experimental results on other small datasets support the usefulness of PPG-extracted biomarkers as viable traits for multi-biometric or standalone biometrics. Furthermore, the approach results in a low input throughput and complexity that allows for continuous authentication in real-world scenarios and implementation in little wearable devices. Nevertheless, the reported experiments also suggest that further research is necessary to develop a definitive system., Si bien los esfuerzos en la investigación se han focalizado tradicionalmente en las señales electrocardiográficas (ECG) y características extraídas manualmente como rasgos biométricos potenciales, pocas operaciones han explorado sistemas basados en la señal fotopletográfica (PPG). Este trabajo propone una arquitectura de extremo a extremo para ofrecer autenticación biométrica mediante biosensores PPG a través de redes convolucionales. Ésta aproximación se ha evaluado en dos bases de datos diferentes: Troika y PulseID, ésta última disponible públicamente y que ha sido recogida por los autores para este propósito. La verificación a través de modelos basados en redes convolucionales y el uso de señales PPG en crudo parecen ser viables en los procedimientos de seguimiento actuales, dentro del entorno de la salud y del deporte, mostrando así un gran potencial para la biometría. El trabajo testeado en la tarea de verifiación, en ensayos de un segundo, consiguen una AUC de $78,2\%$ y $83,2\%$ en media, entre todos los sujetos objetivo, en los conjuntos de datos PulseID y Troika respectivamente. Los resultados experimentales en otros conjuntos de datos pequeños refuerzan la potencial utilidad de estos biomarcadores extraídos de señales PPG como rasgos viables para la caracterización biométrica. Además, este enfoque permite una autenticación contínua debido a su baja complejidad y número de operaciones, haciéndola sostenible para escenarios del mundo real así como para poder ser implementado en dispositivos de reducido tamaño y capacidad computacional. Sin embargo, los experimentos aquí reportados sugieren que son necesarias más investigaciones para poder desarrollar un sistema definitivo., Si bé els esforços en la investigació s'han centrat tradicionalment en senyals electrocardiogràfics (ECG) i característiques artesanals com a trets biomètrics potencials, pocs treballs han explorat sistemes basats en el senyal fotopletogràfic (PPG). Aquest treball proposa una arquitectura d'extrem a extrem per oferir autenticació biomètrica mitjançant biosensors PPG a través de xarxes convolucionals. L'acompliment d'aquest enfocament s'ha avaluat en dues bases de dades diferents: Troika i PulseID, aquesta última disponible públicament i que ha estat recollida pels autors per a aquest propòsit. Aquest enfocament de verificació a través de models basats en xarxes convolucionals i l'ús de senyals de PPG en cru sembla ser viable en els procediments de monitorització actuals, dins d'entorns de salut i esport, mostrant així un gran potencial i atractiu per a la biometria. L'enfocament provat en la tasca de verificació, en assaigs que duren un segon, aconsegueix una AUC de $78,2\%$ i $83,2\%$ en mitjana, entre els subjectes objectiu, en els conjunts de dades de PulseID i Troika, respectivament. Els nostres resultats experimentals en altres conjunts petits de dades recolzen la utilitat dels biomarcadors extrets de PPG com a trets viables per a la biometria multi-biomètrica o autònoma. A més, l'enfocament permet una autenticació contínua degut a la baixa complexitat i nombre d'operacions, que la fan sostenible pels escenaris del món real així com per a ésser implementat en dispositius de reduit tamany i capacitat computacional. No obstant això, els experiments reportats també suggereixen que més investigacions són necessàries per a poder desenvolupar un sistema definitiu.

The advancement of RADAR technology has allowed it to expand its use and applications beyond its traditional role. Among them, it highlights its potential as a medical diagnostic tool for biological signals that can allow the development of portable diagnostic systems to improve the speed of diseases detection in its most primitive stages. This thesis is dedicated to the study of various signal processing techniques for respiratory and cardiac signals, as well as to include an initial study on the early detection of epileptic movements. An initial theoretical study of each of the techniques has been carried out, in addition to checking their performance by implementing them in the MATLAB code and carrying out a comparative study to evaluate their possible implementation. It has been used pattern signals extracted from various subjects by a continuous wave radar prototype at 120 GHz developed by the TERAHEALTH team of the Department of Signal Theory and Communications., El avance de la tecnología RADAR ha permitido expandir su uso y aplicaciones más allá de su rol tradicional. Entre ellas, destaca su potencial como herramienta de diagnóstico médico de señales biológicas que puede permitir el desarrollo de sistemas portátiles de diagnóstico para mejorar la velocidad de detección de enfermedades en su estadio más primitivo. Esta tesis está dedicada al estudio de diversas técnicas de procesado de señal para señales respiratorias y cardíacas, además de incluir un estudio inicial sobre la detección temprana de movimientos epilépticos. Se ha realizado un estudio inicial teórico de cada una de las técnicas, además de comprobar sus prestaciones implementándolas en código MATLAB y realizar un estudio comparativo para evaluar su posible implementación. Se ha utilizado señales patrón extraídas de diversos sujetos mediante un prototipo de radar de onda continua a 120 GHz desarrollado por el equipo TERAHEALTH del departamento de Teoría de Señal y Comunicaciones., L'avenç de la tecnologia RADAR ha permès expandir el seu ús i aplicacions més enllà del seu rol tradicional. Entre elles, destaca el seu potencial com a eina de diagnòstic mèdic de senyals biològiques que pot permetre el desenvolupament de sistemes portàtils de diagnòstic per millorar la velocitat de detecció de malalties en el seu estadi més primitiu. Aquesta tesi està dedicada a l'estudi de diverses tècniques de processat de senyal per a senyals respiratòries i cardíaques, a més d'incloure un estudi inicial sobre la detecció primerenca de moviments epilèptics. S'ha realitzat un estudi inicial teòric de cadascuna de les tècniques, a més de comprovar les seves prestacions implementant en codi MATLAB i realitzar un estudi comparatiu per avaluar la seva possible implementació. S'han utilitzat senyals patró estretes de diversos subjectes mitjançant un prototip de radar d'ona contínua a 120 GHz desenvolupat per l'equip TERAHEALTH del departament de Teoria de Senyal i Comunicacions.

On the one hand, knowledge of symbol timing and frequency offset is required for demodulation in digital communication systems. On the other hand, capabilities of machine learning in the communications domain have not yet been developed so to be able to have an impact in the actual systems and technologies. The goal is to realize a computationally efficient synchronization algorithm based on Neural Networks (NN). The main attractive of NN is the fact that given a structure which is known to be able to converge, it will give the capability to be easily implemented and trained on the concrete type of conditions given in each environment so that it will lead to the best estimator for each case. Also, despite the possibility of designing your own activation functions has already been exploded, it is important that the NN structure to be found is made of Dense feedforward hidden layers with the usual activation functions found on Keras library (those being for instance Linear, Relu, Sigmoid, etc)., Por un lado, el conocimiento del tiempo del símbolo y el desplazamiento de frecuencia es necesario para la demodulación en los sistemas de comunicación digital. Por otro lado, las capacidades de Machine Learning en el dominio de las comunicaciones aún no se han desarrollado para poder tener un impacto en los sistemas y tecnologías actuales. El objetivo es realizar un algoritmo de sincronización computacionalmente eficiente basado en redes neuronales (NN). El principal atractivo de las NN es el hecho que, dada una estructura la cual es sabido que puede converger, se da la capacidad de implementarla fácilmente y entrenarla sobre el tipo concreto de condiciones que se dan en cada entorno para que de lugar al mejor estimador en cada caso Además, a pesar de que ya se ha explotado la posibilidad de diseñar funciones de activación propias, es importante que la estructura NN que se encuentre esté compuesta de feedforward hidden layers con las funciones de activación habituales que se encuentran en la biblioteca Keras (por ejemplo Linear, Relu) , Sigmoid, etc.)., D'una banda, el coneixement del temps del símbol i el desplaçament de freqüència és necessari per a la desmodulació en els sistemes de comunicació digital. D'altra banda, les capacitats de Machine Learning en el domini de les comunicacions encara no s'han desenvolupat per poder tenir un impacte en els sistemes i tecnologies actuals. L'objectiu és realitzar un algoritme de sincronització computacionalment eficient basat en xarxes neuronals (NN). El principal atractiu de les NN és el fet que, donada una estructura la qual és sabut que pot convergir, es dóna la capacitat d'implementar-se fàcilment i entrenar-la sobre el tipus concret de condicions que es donen en cada entorn per a que doni lloc al millor estimador en cada cas A més, tot i que ja s'ha explotat la possibilitat de dissenyar funcions d'activació pròpies, és important que l'estructura NN que es trobi estigui composta de feedforward hidden layers amb les funcions d'activació habituals que es troben a la biblioteca Keras (per exemple linear, Relu), Sigmoid, etc.).

This paper deals with the problem of sparse beamforming for interference mitigation in milimiter wave (mm Wave) frequency bands. Multiantenna solutions in mmWave are generally implemented with phase shifters which are known to have good interference rejection capabilities. However, phase shifters are power-demanding components and, depending on their resolution, they require bulky hardware solutions able to accommodate the control lines. On the other hand, the use of switches leads to a cost-efficient alternative able to provide a sufficiently large interference rejection while substantially reducing the hardware cost and power consumption. This work proposes a beamforming scheme able to maximize the signal-to-interference-plus-noise ratio (SINR) assuming that the beamforming weights can only take 0 and 1 values. The resulting optimization problem is a binary quadratic fractional problem which is a difficult non-convex problem. Two optimization approaches are proposed; namely, the semidefinite relaxation and the penalized convex-concave procedure. We show that both techniques behave well in the considered scenarios and their performance is close to the optimization problem upper bound value., Peer Reviewed, Postprint (published version)

Condition-based maintenance plays an important role to ensure the working condition and to increase the availability of the machinery. The feature calculation and feature extraction are critical signal processing that allow to obtain a high-performance characterization of the available physical magnitudes related to specific working conditions of machines. Aiming to overcome this issue, this research proposes a novel condition monitoring strategy based on the spectral energy estimation and Linear Discriminant Analysis for diagnose and identify different operating conditions in an induction motor-based electromechanical system. The proposed method involves the acquisition of vibration signals from which the frequency spectrum is computed through the Fast Fourier Transform. Subsequently, such frequency spectrum is segmented to estimate a feature matrix in terms of its spectral energy. Finally, the feature matrix is subjected to a transformation into a 2-dimentional base by means of the Linear Discriminant Analysis and the final diagnosis outcome is performed by a NN-based classifier. The proposed strategy is validated under a complete experimentally dataset acquired from a laboratory electromechanical system., Peer Reviewed, Postprint (published version)

We propose a new approach of optical field management based on a local Hilbert transform, where the non-Hermitian potentials generating arbitrary vector fields of directionality p(\vec r), with desired shapes and topologies are designed. We derive a local Hilbert transform to build systematically such potentials, by modifying background potentials (being either regular or random, extended or localized). In particular, we explore particular directionality fields, for instance in the form of a focus to create sinks for probe fields, to generate vortices in the probe fields, and others. Physically, the proposed directionality fields provide a flexible new mechanism for dynamically shaping and precise control over probe fields leading to novel effects in wave dynamics. © 2018 IEEE., Peer Reviewed, Postprint (published version)