Examinando por Autor "Barquero-Pérez, Óscar"

Mostrando 1 - 3 de 3

Deep Neural Network: An Alternative to Traditional Channel Estimators in Massive MIMO Systems
(IEEE, 2022-04-05) Melgar, Antonio; de la Fuente, Alejandro; Carro-Calvo, Leopoldo; Barquero-Pérez, Óscar; Morgado, Eduardo
Fifth-generation (5G) requires a highly accurate estimate of the channel state information (CSI) to exploit the benefits of massive multiple-input-multiple-output (MaMIMO) systems. 5G systems use pilot sequences to estimate channel behaviour using traditional methods like least squares (LS), or minimum mean square error (MMSE) estimation. However, traditional methods do not always obtain reliable estimations: LS exhibits a poor estimation when inadequate channel conditions (i.e., low- signal-to-noise ratio (SNR) region) and MMSE requires prior statistical knowledge of the channel and noise (complex to implement in practice). We present a deep learning framework based on deep neural networks (DNNs) for fifth-generation (5G) MaMIMO channel estimation. After a first preliminary scheme with which we verify the good estimation capacity of our DNN-based approach, we propose two different models, which differ in the information processed by the DNN and benefit from lower computational complexity or greater flexibility for any reference signal pattern, respectively. The results show that, compared to the LS-based channel estimation, the DNN approach decreases the mean square error (MSE) and the system’s spectral efficiency (SE) increases, especially in the low- SNR region. Our approach provides results close to optimal MMSE estimation but benefits from not requiring any prior channel statistics information.
High diagnostic quality ECG compression and CS signal reconstruction in body sensor networks
(IEEE, 2016) Chidean, Mihaela I.; Barquero-Pérez, Óscar; Zhang, Qi; Jacobsen, Rune Hylsberg; Caamaño, Antonio J
Compression of electrocardiograms (ECG) in wireless environments, with diagnostic quality, has shown limited potential. This lack of quality preservation, using Wavelet Transform (WT), is due to the fact that the multiple levels of detail that can be achieved in the time domain are not exploited. In the present work, we propose to fully exploit the wavelet capability to operate at different levels of signal detail at different time scales. WT with an appropriate Compressed Sensing (CS) matrix is used in the electrode nodes of body sensor networks to encode and compress the ECG. Then, the signal is reconstructed using a basis pursuit denoise algorithm. Preservation of the diagnostic quality by means of standardized metrics is then tested for multiple wavelet bases and levels. High quality ECGs from 50 healthy patients are used to statistically show that diagnostic quality preservation is possible even at high compression rates. In these cases suitable ECG wavelets are required.
Non-invasive estimation of atrial fibrillation driver position using long-short term memory neural networks and body surface potentials
(Elsevier, 2024-04) Gutiérrez-Fernández-Calvillo, Miriam; Cámara-Vázquez, Miguel Ángel; Hernández-Romero, Ismael; Gillem, María S.; Climent, Andreu M.; Fambuena-Santos, Carlos; Barquero-Pérez, Óscar
Background and Objective Atrial Fibrillation (AF) is a supraventricular tachyarrhythmia that can lead to thromboembolism, hearlt failure, ischemic stroke, and a decreased quality of life. Characterizing the locations where the mechanisms of AF are initialized and maintained is key to accomplishing an effective ablation of the targets, hence restoring sinus rhythm. Many methods have been investigated to locate such targets in a non-invasive way, such as Electrocardiographic Imaging, which enables an on-invasive and panoramic characterization of cardiac electrical activity using recording Body Surface Potentials (BSP) and a torso model of the patient. Nonetheless, this technique entails some major issues stemming from solving the inverse problem, which is known to be severely ill-posed. In this context, many machine learning and deep learning approaches aim to tackle the characterization and classification of AF targets to improve AF diagnosis and treatment. Methods In this work, we propose a method to locate AF drivers as a supervised classification problem. We employed a hybrid form of the convolutional-recurrent network which enables feature extraction and sequential data modeling utilizing labeled realistic computerized AF models. Thus, we used 16 AF electrograms, 1 atrium, and 10 torso geometries to compute the forward problem. Previously, the AF models were labeled by assigning each sample of the signals a region from the atria from 0 (no driver) to 7, according to the spatial location of the AF driver. The resulting 160 BSP signals, which resemble a 64-lead vest recording, are preprocessed and then introduced into the network following a 4-fold cross-validation in batches of 50 samples. Results The results show a mean accuracy of 74.75% among the 4 folds, with a better performance in detecting sinus rhythm, and drivers near the left superior pulmonary vein (R1), and right superior pulmonary vein (R3) whose mean sensitivity bounds around 84%-87%. Significantly good results are obtained in mean sensitivity (87%) and specificity (83%) in R1. Conclusions Good results in R1 are highly convenient since AF drivers are commonly found in this area: the left atrial appendage, as suggested in some previous studies. These promising results indicate that using CNN-LSTM networks could lead to new strategies exploiting temporal correlations to address this challenge effectively