Search

© 2024 Elsevier. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0, In this paper we present SurvLIMEpy, an open-source Python package that implements the SurvLIME algorithm. This method allows to compute local feature importance for machine learning algorithms designed for modelling Survival Analysis data. The presented implementation uses a matrix-wise formulation, which allows to speed up the execution time. Additionally, SurvLIMEpy assists the user with visualisation tools to better understand the result of the algorithm. The package supports a wide variety of survival models, from the Cox Proportional Hazards Model to deep learning models such as DeepHit or DeepSurv. Two types of experiments are presented in this paper. First, by means of simulated data, we study the ability of the algorithm to capture the importance of the features. Second, we use three open source survival datasets together with a set of survival algorithms in order to demonstrate how SurvLIMEpy behaves when applied to different models., Peer Reviewed, Postprint (author's final draft)

This work presents the first results of canopy height mapping with L -Band SAOCOM data by means of Polarimetric SAR Interferometry (PolInSAR). For this study case, a colocated temporal series of SAOCOM fully polarimetric images covering the year 2021 was acquired, with a temporal baseline of 8 to 16 days between each acquisition. The study area corresponds to managed forests in Corrientes, Argentina, one of the main forest production regions in the country. Field measurements provided by forest owners were available in order to validate the obtained results, along with canopy height measurements from the Global Ecosystem Dynamics Investigation mission. A multi-baseline PolInSAR approach was adopted, following the methodology proposed by previous works. The results show that this multi-baseline selection approach is more appropriate than a single-baseline one, given the high variability that SAOCOM presents in the spatial baseline between acquisitions, and the lack of orbital control over this parameter. The height maps obtained over a validation site yielded an average R2 of 0.72 and a mean RMSE of 2.35 m., in agreement with the figures obtained in similar studies. This study presents, for the first time, canopy height maps of 8 and 16-day temporal baselines, L -band, orbital interferograms. The conclusions yielded by this research represent a qualitative and quantitative evaluation of SAOCOM PolInSAR capabilities for forest height mapping. It is also relevant to assess the suitability of the RVoG model to characterize the structure of these forests with L-band data retrieved by the Argentinean constellation., This work was supported by the Spanish MCIN/AEI/10.13039/501100011033 under the Project INTERACTPID2020-114623RB-C32., Peer Reviewed, Postprint (published version)

Vídeos financiats pel programa Seminarios Catedras Telefónica-UPC Ponent, Matteo Stoico

Vídeos financiats pel programa Seminarios Catedras Telefónica-UPC Ponent, Matteo Stoico

Polarization encoded Quantum Key Distribution (QKD) is attracting great attention as it generates unlimitedly and unconditionally secure keys for different use cases. Despite its theoretical excellence based on quantum physics, commercial optical devices supporting QKD systems lack precision, which highly limits the final Key Exchange Rate (KER) of the system. Beside optical component imperfections, eavesdropping and unpredicted environmental events occurred in the quantum channel increase quantum Bit Error Rate (qBER), which leads to further KER reduction. In this paper, we propose DARIUS, a digital twin for polarization encoded QKD systems that bridges the gap between perfect theoretical QKD systems and real implementations to: i) address optical components' non-ideal behavior; ii) discern eavesdropping from high qBER; and iii) dynamically compensate for environmental events. Taking advantage of DARIUS, even moderate eavesdropping rates can be distinguished from qBER. Moreover, significant improvement in proactive environmental event compensation is achieved, as DARIUS can derive proper optical component tuning., This work was partially supported by the European Commission through the HORIZON ALLEGRO project (G.A. 101092766), from the AEI IBON (PID2020-114135RB-I00) project and from the ICREA Institution., Peer Reviewed, Postprint (published version)

Dihedral is a common structure in polarimetric SAR images and can be found on many man-made targets. Many researchers have proposed different dihedral models, but the accuracy of these models is limited. In this case, the feature extraction methods based on these models are also not effective enough, which affects the subsequent applications such as target detection. Therefore, it is necessary to propose a new and accurate scattering model, which can be applied to dihedral with different orientation angles for feature extraction and target detection. In this paper, a general scattering model called modified oriented dihedral scattering model (MODM) is proposed based on physical optics (PO) and geometric optics (GO) of high-frequency approximation techniques. By analyzing the propagation and reflection of electromagnetic wave, MODM can accurately describe the scattering characteristic of dihedral for all observation conditions. In order to apply the model to real PolSAR images, MODM is introduced into a new feature extraction method, which is called five-scattering component polarimetric decomposition method (MODM-5SD). Feature extraction and target detection experiments of buildings with various oriented dihedral structures are performed using different data sets, which show that dihedral scattering components from oriented dihedral structures can be more effectively extracted by MODM-5SD. In addition, more buildings with oriented dihedral structures can be detected with the features from MODM-5SD. The experimental results show that MODM can more accurately describe the scattering characteristic of dihedral, which can be further applied for scattering characterization and feature extraction of targets with typical dihedral structures., This work was supported in part by the National Natural Science Foundation of China under Grant 62271172 and Grant 61871158, and in part by Aeronautical Science Foundation of China under Grant 20182077008., Peer Reviewed, Postprint (published version)

The ability to utilize resources to meet the need of growing diversity of communication services and user behavior marks the future of cognitive wireless communication systems. Cognitive wireless technologies for vehicular communications, in combination with Orbital Angular Momentum (OAM) modes aim at extending Non-Line-Of-Sight (NLOS) short-distance communications for smart mobility. In this regard, OAM antenna frameworks need to be developed to support these technologies. In this work, we describe a magnetoelectric dipole antenna framework supporting OAM modes. The framework is derived from moment tensors of specific vector spherical harmonic functions synthesized from dipoles. The antenna framework is discussed in terms of OAM generation, and it is validated numerically and experimentally for l = 1 OAM mode, achieving more than 500MHz operation bandwidth at the frequency of operation of 3.5GHz. Also, for l = 1 OAM mode, the null aligns precisely with the anticipated dimensions numerically computed., This work was financially supported by PDR-2014-2022/56-30157-2021-2A, 2021 SGR 01415 Q0818003F and Metropolis PLEC2021-007609 grants of the Generalitat de Catalunya and European Regional Development Fund (FEDER), by CICYT PID2019-107885GB-C31, PID2022-136869NB-C31 and Prueba de Concepto PDC2022-133091-I00 grants of the Agencia Estatal de Investigación (Spain) and Next Generation EU., Peer Reviewed, Postprint (author's final draft)

Today, wireless communications play a crucial role in the telecommunications sector due to the emergence of new technologies such as 5G and smart cities. The amount of data to be transmitted is high, leading to an increased need for bandwidth. One of the key components in these wireless communications is Power Amplifiers (PA), as they amplify the signal to ensure it reaches the receiver correctly. However, a significant challenge with these elements is the high energy consumption they entail. To maximize the efficiency of amplifiers, efforts are made to operate them as efficiently as possible. This involves operating the amplifier in a non-linear region, causing signal distortion. Various solutions exist for balancing efficiency and linearity, with the most prevalent solution being the Digital Predistortion (DPD) technique, which is the focus of this study. The Digital Predistortion technique models the amplifier's behavior to linearize it. Different mathematical models, such as Memory Polynomial (MP), Generalized Memory Polynomial (GMP), and Decomposed Vector Rotation (DVR), have been studied for this purpose. Attempts have been made to combine these models to achieve better results. The DOMP reduction technique, simplifying the number of coefficients in each model, has also been explored. An analysis of the results obtained using these models has been conducted, considering various parameters and the number of coefficients. MATLAB has been used as the programming environment. Subsequently, the best models have been tested using a real digital predistortion environment, meeting expectations by complying with 3GPP standards., Objectius de Desenvolupament Sostenible::7 - Energia Assequible i No Contaminant, Objectius de Desenvolupament Sostenible::11 - Ciutats i Comunitats Sostenibles

Synchronizing the local oscillators in multibeam satellites with the objective of coherent communications is still an open challenge. It has to be addressed to implement full-frequency reuse approaches, such as precoding techniques using the already deployed multibeam satellites. This article addresses the required phase synchronization to enable precoding techniques in multibeam satellite systems. It contains the detailed design of a frequency and phase compensation loop based on the proportional-integral controller, which deals with the phase drift introduced by the hardware components. Specifically, the phase noise of the local oscillators used for up and down conversion at each system element (gateway, satellite, and user terminals). The implementation of the two-state phase noise model used to emulate this phase drift is included in the article. Besides, a comparative analysis of several methods to combine the frequency and phase measurements obtained from the user terminals is also included. Finally, the performance of the proposed closed-loop synchronization method is validated through simulations using our in-house developed MIMO end-to-end satellite emulator based on SDR platforms., This research was funded by the Luxembourg National Research Fund (FNR) under the CORE project COHESAT: Cognitive Cohesive Networks of Distributed Units for Active and Passive Space Applications, grant reference [FNR11689919] and the project DISBuS: Dynamic Beam Forming and In-band Signalling for Next Generation Satellite Systems, grant reference [FNR13778945]. For the purpose of open access, the author has applied a Creative Commons Attribution 4.0 International (CCBY4.0) license to any Author Accepted Manuscript version arising from this submission. Additionally, this work has been supported by the European Space Agency under project number 4000122451/18/NL/NR ”Live Satellite Precoding Demonstration - CCN: MEO case (LiveSatPreDem).”, Peer Reviewed, Postprint (published version)

The Global Navigation Satellite Systems (GNSS), including the US’s GPS, China’s BDS, the European Union’s Galileo, and Russia’s GLONASS, offer real-time, all-weather, any-time, anywhere and high precision observations by transmitting L band signals continuously, which have been widely used for positioning, navigation and timing. With the development of GNSS technology, it has been found that GNSS-reflected signals can be used to detect Earth’s surface characteristics together with other signals of opportunity. In this paper, the current status and latest advances are presented on Global Navigation Satellite System-Reflectometry (GNSS-R) in theory, methods, techniques and observations. New developments and progresses in GNSS-R instruments, theoretical modeling, and signal processing, ground and space-/air-borne experiments, parameters retrieval (e.g. wind speed, sea surface height, soil moisture, ice thickness), sea surface altimetry and applications in the atmosphere, oceans, land, vegetation, and cryosphere are given and reviewed in details. Meanwhile, the challenges in the GNSS-R development of each field are also given. Finally, the future applications and prospects of GNSS-R are discussed, including multi-GNSS reflectometry, new GNSS-R receivers, GNSS-R missions, and emerging applications, such as mesoscale ocean eddies, ocean phytoplankton blooms, microplastics detection, target recognition, river flow, desert studies, natural hazards and landslides monitoring., This work was supported by the Henan International Science and Technology Cooperation Key Project (Grant No. 241111520700) and Strategic Priority Research Program Project of the Chinese Academy of Sciences (Grant No. XDA23040100)., Peer Reviewed, Postprint (published version)

In this paper a statistical study of cirrus geometrical and optical properties based on 4 years of continuous ground-based lidar measurements with the Barcelona (Spain) Micro Pulse Lidar (MPL) is analysed. First, a review of the literature on the two-way transmittance method is presented. This method is a well-known lidar inversion method used to retrieve the optical properties of an aerosol–cloud layer between two molecular (i.e. aerosol and cloud-free) regions below and above, without the need to make any a priori assumptions about their optical and/or microphysical properties. Second, a simple mathematical expression of the two-way transmittance method is proposed for both ground-based and spaceborne lidar systems. This approach of the method allows the retrieval of the cloud optical depth, the cloud column lidar ratio and the vertical profile of the cloud backscatter coefficient. The method is illustrated for a cirrus cloud using measurements from the ground-based MPL and from the spaceborne Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP). Third, the database is then filtered with a cirrus identification criterion based on (and compared to) the literature using only lidar and radiosonde data. During the period from November 2018 to September 2022, 367 high-altitude cirrus clouds were identified at 00:00 and 12:00 UTC, of which 203 were successfully inverted with the two-way transmittance method. The statistical results of these 203 high-altitude cirrus clouds show that the cloud thickness is 1.8 ± 1.1 km, the mid-cloud temperature is −51 ± 8 ∘C and the linear cloud depolarization ratio is 0.32 ± 0.13. The application of the transmittance method yields an average cloud optical depth (COD) of 0.36 ± 0.45 and a mean effective column lidar ratio of 30 ± 19 sr. Statistical results of the errors associated with the two-way transmittance method retrievals are also provided. The highest occurrence of cirrus is observed in spring and the majority of cirrus clou, This research has been partly funded by the Spanish Agencia Estatal de Investigación (grant no. PID2019103886RB-I00) and the European Commission through the Horizon 2020 Programme (project ACTRIS IMP, grant agreement no. 871115; ATMO-ACCESS, grant agreement no. 101008004; GRASP-ACE, grant agreement no. 778349) and through the Horizon Europe Programme (project REALISTIC, grant agreement no. 101086690)., Peer Reviewed, Postprint (published version)

We show that for a wide range of channels and code ensembles with pairwise-independent codewords, with probability tending to 1 with the code length, expurgating an arbitrarily small fraction of codewords from a randomly selected code results in a code attaining the expurgated exponent., This work was supported in part by the Ramon y Cajal fellowship program (grant RYC2021-033908-I) funded by the Spanish Government and the European Union “NextGenerationEU” Recovery Plan, the European Research Council under ERC Agreement 725411 and by the Spanish Ministry of Economy and Competitiveness under Grant PID2020-116683GB-C22., Peer Reviewed, Postprint (published version)

Reducing the large sidelobes of multicarrier signals is crucial to prevent adjacent channel interference. Spectral precoding is an effective approach toward this goal, at the expense of throughput loss due to precoder redundancy; thus, it is of interest to explore alternative precoder designs with improved performance at lower redundancies. We present a novel precoder which minimizes radiated power within a user-selectable frequency region. The structure of the precoding matrix is chosen to allow efficient mitigation of in-band distortion at the receiver by means of iterative and successive interference cancellation, while completely avoiding distortion to protected and pilot subcarriers. By exploiting the low-rank properties of constituent blocks, computational complexity can be significantly reduced with little impact on sidelobe reduction. Simulation results show the benefits of the proposed design, which is particularly effective in redundancy-limited settings targeting high spectral efficiency., Work supported by grants PID2019-105717RB-C21/C22, PID2022-136512OB-C21/C22 and BES-2017-080305 funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe” (EU)., Peer Reviewed, Postprint (published version)

In May–June 2019, smoke plumes from wildfires in Alberta, Canada, were advected all the way to Europe. To analyze the evolution of the plumes and to estimate the amount of smoke aerosols transported to Europe, retrievals from the spaceborne lidar CALIOP (Cloud-Aerosol LIdar with Orthogonal Polarization) were used. The plumes were located with the help of a trajectory analysis, and the masses of smoke aerosols were retrieved from the CALIOP observations. The accuracy of the CALIOP mass retrievals was compared with the accuracy of ground-based lidars/ceilometer near the source in North America and after the long-range transport in Europe. Overall, CALIOP and the ground-based lidars/ceilometer produced comparable results. Over North America the CALIOP layer mean mass was 30¿% smaller than the ground-based estimates, whereas over southern Europe that difference varied between 12¿% and 43¿%. Finally, the CALIOP mass retrievals were compared with simulated aerosol concentrations from two reanalysis models: MERRA-2 (Modern-Era Retrospective analysis for Research and Applications, Version 2) and CAMS (Copernicus Atmospheric Monitoring System). The simulated total column aerosol optical depths (AODs) and the total column mass concentration of smoke agreed quite well with CALIOP observations, but the comparison of the layer mass concentration of smoke showed significant discrepancies. The amount of smoke aerosols in the model simulations was consistently smaller than in the CALIOP retrievals. These results highlight the limitations of such models and more specifically their limitation to reproduce properly the smoke vertical distribution. They indicate that CALIOP is a useful tool monitoring smoke plumes over secluded areas, whereas reanalysis models have difficulties in representing the aerosol mass in these plumes. This study shows the advantages of spaceborne aerosol lidars, e.g., being of paramount importance to monitor smoke plumes, and reveals the urgent need of future, This research has been supported by the Academy of Finland (grant nos. 339885 and 337552), the European Union (grant no. 101079201), the Hellenic Foundation for Research and Innovation (grant nos. 3995 and 07222), Horizon Europe (grant no. 101086690), Agencia Estatal de Investigación (grant no. PID2019-103886RB-I00), H2020 Environment (grant nos. 871115 and 101008004), and H2020 Excellent Science (grant no. 778349)., Peer Reviewed, Postprint (published version)

The detection of tumoural cells from whole slide images is an essential task in medical diagnosis and research. In this article, we propose and analyse a novel approach that combines computer vision-based models with graph neural networks to improve the accuracy of automated tumoural cell detection in lung tissue. Our proposal leverages the inherent structure and relationships between cells in the tissue. Experimental results on our own curated dataset show that modelling the problem with graphs gives the model a clear advantage over just working at pixel level. This change in perspective provides extra information that makes it possible to improve the performance. The reduction of dimensionality that comes from working with the graph also allows us to increase the field of view with low computational requirements. Code is available at https://github.com/Jerry-Master/lung-tumour-study, models are uploaded to https://huggingface.co/Jerry-Master/Hovernet-plus-Graphs, and the dataset is published on Zenodo https://zenodo.org/doi/10.5281/zenodo.8368122., Peer Reviewed, Postprint (published version)

The 2-D frequency-based omega-K method is known to be a suitable algorithm for fully focused SAR (FF-SAR) radar altimeter processors, as its computational efficiency is much higher than equivalent time-based alternatives without much performance degradation. In this article, we provide a closed-form description of a 2-D frequency-domain omega-K algorithm specific for instruments such as Poseidon-4 onboard Sentinel-6. The processor is validated with real data from point targets and over the open ocean. Applications such as ocean swell retrieval and lead detection are demonstrated, showing the potentiality of the processor for future operational global-scale products., This work was supported by the Industrial Doctorates Plan from the Secretary of Universities and Research of the Department of Business and Knowledge of the Government of Catalonia through Project SA.45217 with Expedient Number 094 under Grant Doctorats Industrials 2020., Peer Reviewed, Postprint (published version)

This document is the partial result of an ongoing satellite research activity carried out by Universitat Politècnica de Catalunya NanoSat Lab and i2CAT Foundation, submitted to the Horizon Europe Program in 2022., Thousands of rockets, spaceships, and satellites have been launched, resulting in a steady increase in the number of space debris notably in Low Earth Orbit (LEO) and Geostationary Orbit (GEO). Among the objects comprising space debris, defunct satellites are commonly found, which may have been abandoned in space due to various anomalies. While these satellite failures may not be catastrophic, certain subsystems could still be repurposed for alternative uses beyond their initial design. This study aims to assess the feasibility of “recycling” partially failed satellites, also known as “Zombie Satellites”, by statistically evaluating the severity of their anomalies and quantifying their remaining capabilities and limitations. By conducting a comprehensive statistical analysis, the anomalies have been categorized based on subsystem, orbit type, occurrence date during the satellite’s lifespan, and severity. Additionally, an analysis is presented to identify failed satellites from some of the current large constellations., This work was sponsored in part by the I+D+i project of UPC GENESIS project PID 2021-1264360B-C21 financed by MCIN/AEI. Additionally, this work was supported by the Spanish Ministry of Economic Affairs and Digital Transformation and the European Union–NextGeneration EU , in the framework of the Recovery Plan, Transformation and Resilience (PRTR) (Call UNICO I+D 5G 2021, ref. number TSI-063000-2021-8-6GSatNet-SeS)., Peer Reviewed, Postprint (published version)

Breast cancer is one of the most prevalent diseases in our society. The survival rate is increasing thanks to medical advances and the early detection of the disease through various diagnostic tests. To achieve this, a tissue sample is extracted from a biopsy, stained with different immunohistochemical techniques and analysed to quantify the tumour cells. For instance, the Estrogen Receptor (ER), Progesterone Receptor (PR), and Ki67 are some of the most commonly used stains to identify tumour grade. This work is developed within the DigiPatICS project, which aims to provide algorithms for the Institut Català de la Salut (ICS). Our objective is to develop a method to segment and classify cells present in breast tissue to obtain meaningful indicators that will support pathologists in the diagnosis of tumour. The proposed method uses a Convolutional Neural Network (specifically, the Hover-Net) followed by a Graph Neural Network (GNN) to analyse these immunohistochemical stains. In the first stage, the cells are segmented and classified, and then the resulting prediction is used to construct a graph. This information is then processed by the GNN to obtain the final classification of the identified instances. The results indicate that the proposed method helps to classify the cells in the three analysed stains: ER, PR, and Ki67. We studied the effect of several factors of the method, including the Hover-Net and the Graph Neural Network, as well as the features used to represent each segmented instance and in the conditions to generate the graph. In addition, we have interpreted the results and identified significant aspects, such as the influence of certain nodes in classifying their neighbours in some situations. This information is provided by the attention weights assigned to each pair of neighbours by certain layers used in the GNN.

Monitoring vegetation moisture conditions is paramount to better understand and assess drought impacts on vegetation, enhance crop yield predictions, and improve ecosystem models. Passive microwave remote sensing allows retrievals of the vegetation optical depth (VOD; [unitless]), which is directly proportional to the vegetation water content (VWC; in units of water mass per unit area [kg/m2]). However, VWC is largely dependent on the dry biomass and structure imprints on the VOD signal. Previously, statistical models have been used to isolate the water component from the biomass and structure components. Physically-based approaches have not yet been proposed for this goal. In this study, we present a multi-sensor semi-physical approach to retrieve the vegetation moisture from the VOD and express it as Live Fuel Moisture Content (LFMC [%]; the percentage of water mass per dry biomass unit). The study is performed in the western United States for the period April 2015 – December 2018. There, in situ LFMC samples are available for assessment. We rely on a VOD model based on vegetation height data from GEDI/Sentinel-2 and radar backscatter from Sentinel-1, which account for the biomass and structure components. Vegetation moisture is retrieved at L-, X- and Ku-bands by minimizing the difference between the modeled VOD and the VOD estimates from SMAP (L-band) and AMSR-2 (X- and Ku-band) satellites. Results show that the LFMC retrievals are independent of canopy height, land cover, and radar backscatter, demonstrating the capability of the proposed algorithm to separate water dynamics from the biomass/structure component in VOD. LFMC estimates at X- and Ku-bands reproduce well the expected spatio-temporal dynamics of in situ LFMC. Results show good agreement with in situ at a regional scale, with Pearson's correlations (r) between in situ LFMC samples and LFMC estimates of 0.64 (Ku-band), 0.60 (X-band) and 0.47 (L-band). Similar results are obtained independently for shr, The work of D. Chaparro was supported by the XXXIII Ramón Areces Postdoctoral Fellowship and by MIT and the “la Caixa” Foundation (ID 100010434) under Grant LCF/ PR/MIT19/51840001 (MIT-Spain Seed Fund; D. Entekhabi, D. Chaparro). M. Piles thanks the support of Conselleria de Innovación, Universidades, Ciencia y Sociedad Digital through the project AI4CS CIPROM/2021/56. M. Vall-llossera acknowledges funding from the Grant PID2020-114623RB-C32, funded by MCIN/AEI/10.13039/ 501100011033, and from the ERDF under Grant RTI2018-096765-A- 100. Also, the authors are grateful to MIT for supporting this research with the MIT-Germany Seed Fund (D. Entekhabi, T. Jagdhuber) and with the MIT-Belgium Seed Fund (D. Entekhabi, F. Jonard). A.F. Feldman was supported by both the ECOSTRESS science team and by a NASA Terrestrial Ecology scoping study for a dryland field campaign., Peer Reviewed, Postprint (author's final draft)

The multi-temporal polarimetric SAR (PolSAR) data contains the scattering change information during the growth of crops. However, the current classification methods usually directly use the addition of features extracted at single-temporal or use the temporal and spatial variations of certain features, not really exploring the complete scattering variation information. The specific data representation models for multi-temporal PolSAR data should combine time with polarimetry to characterize the scattering variations. However, the characterization and utilization of such kind of models are inadequate. In this article, we construct data representation model based on the power form of coherence matrix to comprehensively represent all kinds of scattering mechanism variation, which is full-rank positive semi-definite Hermitian matrix. We extract new time-variant scattering features and design vision transformer classifier accordingly for crop classification. Experiment results on RADARSAT-2 datasets show that the proposed power representation model outperforms other models., Peer Reviewed, Postprint (published version)

Background: Neuroimaging-based brain-age delta has been shown to be a mediator linking cardiovascular risk factors to cognitive function. We aimed to assess the mediating role of brain-age delta in the association between modifiable risk factors of dementia and longitudinal cognitive decline in middle-aged and older individuals who are asymptomatic, stratified by Alzheimer's disease pathology. We also explored whether the mediation effect is specific to cognitive domain. Methods: In this cohort study, we included participants from the ALFA+ cohort aged between 45 years and 65 years who were cognitively unimpaired and who had available structural MRI, cerebrospinal fluid ß-amyloid (Aß)42 and Aß40 measurements obtained within 1 year of each other, modifiable risk factors assessment, and cognitive evaluation over 3 years. Participants were recruited from the Barcelonaßeta Brain Research Center (Barcelona, Spain). Included individuals underwent a first assessment between Oct 25, 2016, and Jan 28, 2020, and a follow-up cognitive assessment 3·28 (SD 0·27) years later. We computed brain-age delta and composites of different cognitive function domains (preclinical Alzheimer's cognitive composite [PACC], attention, executive function, episodic memory, visual processing, and language). We used partial least squares path modelling to explore mediation effects in the associations between modifiable risk factors (including cardiovascular, mental health, mood, metabolic or endocrine history, and alcohol use) and changes in cognitive composites. To assess the role of Alzheimer's disease pathology, we computed separate models for Aß-negative and Aß-positive individuals. Findings: Of the 419 participants enrolled in ALFA+, 302 met our inclusion criteria, of which 108 participants were classified as Aß-positive and 194 as Aß-negative. In Aß-positive individuals, brain-age delta partially mediated (percent mediation proportion 15·73% [95% CI 14·22–16·66]) the association between modif, The ALFA+ study receives funding from La Caixa Foundation (100010434; LCF/PR/GN17/50300004) and the Alzheimer’s Association and an international anonymous charity foundation through the TriBEKa Imaging Platform project (TriBEKa 17 519007). Additional support has been received from the Universities and Research Secretariat, Ministry of Business and Knowledge of the Catalan Government (2021 SGR 009132017-SGR-892). MS-C receives funding from the European Research Council under the EU Horizon 2020 research and innovation programme (948677; project PI19/00155), funded by Instituto de Salud Carlos III and co-funded by the EU, and from a fellowship from La Caixa Foundation (100010434) and from the EU Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant (847648; LCF/BQ/PR21/11840004). DL-M is supported by Instituto de Salud Carlos III (PI19/00117; co-funded by European Regional Development Fund, European Social Fund A Way to Make Europe, and Investing in your Future). EMA-U is supported by the Spanish Ministry of Science and Innovation State Research Agency (RYC2018-026053-I), cofunded by the European Social Fund and the Spanish Ministry of Science and Innovation (PID2019-111514RA-I00). VV has been supported by the Spanish Research Agency (PID2020-116907RB-I00 of the call MCIN/ AEI/10.13039/501100011033). JDG is supported by the Spanish Ministry of Science and Innovation (RYC-2013-13054). JDG has also received research support from the EU and European Federation of Pharmaceutical Industries and Associations Innovative Medicines Initiative Joint Undertaking AMYPAD (115952), European Institute of Innovation and Technology Digital (2021), and from Ministerio de Ciencia y Universidades (RTI2018-102261). GS-B receives funding from the Ministerio de Ciencia e Innovacion, Spanish Research Agency (PID2020119556RA-I00). OG-R receives funding from the Alzheimer’s Association Research Fellowship (2019-AARF-644568), from Instituto de Salud Carlos III (P, Peer Reviewed, Article signat per 19 autors/autores: Irene Cumplido-Mayoral, Anna Brugulat-Serrat, Gonzalo Sánchez-Benavides, Armand González-Escalante, Federica Anastasi, Marta Milà-Alomà, David López-Martos, Muge Akinci, Carles Falcón, Mahnaz Shekari, Raffaele Cacciaglia, Eider M Arenaza-Urquijo, Carolina Minguillón, Karine Fauria, José Luis Molinuevo, Marc Suárez-Calvet, Oriol Grau-Rivera, Verónica Vilaplana, Juan Domingo Gispert, on behalf of the ALFA study*. * Membres del grup "ALFA study": Annabella Beteta, Alba Cañas, Marta del Campo, Carme Deulofeu, Ruth Dominguez, Maria Emilio, Ana Fernández-Arcos, Sherezade Fuentes, Patricia Genius, Laura Hernández, Jordi Huguet, Wiesje Pelkmans, Albina Polo, Sandra Pradas, Blanca Rodríguez-Fernández, Anna Soteras, Laura Stankeviciute, Marc Vilanova, Natalia Vilor-Tejedor., Postprint (published version)

Relevant services envisaged for beyond 5G (B5G) systems, such as extended reality and holographic communications, have extremely demanding user experience requirements with significant computational and communication demands. While edge computing aims to address the computation requirements by offloading the computational tasks to edge servers near the user, the communication will take advantage of the technologies developed for 5G New Radio jointly with an never-before-seen degree of network densification. This paper proposes the use of relays with edge computing capabilities. The approach's potential for B5G are identified, and a system model is defined to characterize both computational and communications viewpoints. Based on this, results are provided to highlight the gains and limitations of the proposed approach from a system-level perspective. Finally, the main challenges for enabling relays with computing capabilities in B5G deployments are discussed., This work is part of VERGE project, which has received funding from the Smart Networks and Services Joint Undertaking under the European Union’s Horizon Europe research and innovation programme under Grant Agreement 101096034. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union. Neither the European Union nor the granting authority can be held responsible for them. This paper has also been partly funded by the Spanish Ministry of Science and Innovation MCIN/AEI/ 10.13039/501100011033 under ARTIST project (ref. PID2020- 115104RB-I00). The work of Irene Vila ` has been funded by the European Union-NextGenerationEU, the Spanish Ministry of Universities and the Plan for Recovery, Transformation and Resilience, through the call for Margarita Salas Grants of the Universitat Politècnica de Catalunya (ref. 2022UPC-MSC- 94079)., Peer Reviewed, Postprint (published version)

This study proposes a new method for the detection of facial expressions of pain using a 3D profiler that combines a multiple-input-multiple-output (MIMO) radar system with a machine learning (ML) model (ML-MIMO radar profiler). It offers a solution for pain detection of facial expressions in a non-invasive, non-intrusive, and cost-effective manner. The ML-MIMO radar profiler employs six radars behind a lens to monitor changes in six facial regions and build a 3D facial profile with real-time facial activity information. A dielectric lens was used to ensure an optimal beam size to effectively illuminate each facial region. Signal processing is performed using dynamic time deformation to determine the longitudinal distance and a discrete stationary wavelet transform to filter the signal and improve accuracy. The information from the 3D profiler was compared with the facial action coding system (FACS) to determine actual facial expressions. A machine learning algorithm was trained to learn action units from the FACS and compare them with the information provided by the ML-MIMO radar profiler, thereby performing facial expression classification. In this study, we analyzed four facial expressions: hapiness, sadness, anger, and pain. Identification and classification were performed using a machine-learning model based on multilayer perceptrons. The results revealed 92% accuracy of the system for pain expression, whereas expressions of happiness, sadness, and anger were detected with 88, 86, and 87% accuracy, respectively., This work was supported in part by the Spanish ‘‘Comision Interministerial de Ciencia y Tecnologia’’ (CICYT) under Project PID2019-107885GB-C31, Project PID2022-136869NB-C31, and Project MDM2016-0600; in part by Metropolis under Grant PLEC2021-007609; in part by the Catalan Research Group, Centre Específic de Recerca en Comunicació i Detecció-UPC (CommSensLab-UPC), under Grant 2021 SGR 01415; and in part by the ‘‘Secretaría Nacional de Educación Superior, Ciencia, Tecnología e Innovación’’ (SENESCYT) from the Ecuadorian Government., Peer Reviewed, Postprint (published version)

The expressiveness of neural networks highly depends on the nature of the activation function, although these are usually assumed predefined and fixed during the training stage. Under a signal processing perspective, in this paper we present Expressive Neural Network (ENN), a novel model in which the non-linear activation functions are modeled using the Discrete Cosine Transform (DCT) and adapted using backpropagation during training. This parametrization keeps the number of trainable parameters low, is appropriate for gradient-based schemes, and adapts to different learning tasks. This is the first non-linear model for activation functions that relies on a signal processing perspective, providing high flexibility and expressiveness to the network. We contribute with insights in the explainability of the network at convergence by recovering the concept of bump, this is, the response of each activation function in the output space. Finally, through exhaustive experiments we show that the model can adapt to classification and regression tasks. The performance of ENN outperforms state of the art benchmarks, providing above a 40% gap in accuracy in some scenarios., This work is part of the project IRENE (PID2020-115323RB-C31), funded by MCIN/AEI/10.13039/501100011033., Peer Reviewed, Postprint (published version)

The purpose of this project is to conduct a detailed evaluation of the coverage and quality of service of a 5G network in an urban environment at various heights, using raytracing-based simulations. The realistic deployment of 5G technology has been carried out in the city of Barcelona, particularly in the Eixample district. MATLAB will be used to create a model that simulates the network's performance for both ground-level and elevated users, which will serve as a reference for other studies. The simulations will be highly detailed, incorporating propagation calculations that account for reflections and diffractions, and will require code optimization to speed up the process. The project proposes incorporating beamforming technology to overcome the limitations of current deployments. This advanced technique adjusts and directs the power of signals precisely, providing the three-dimensional coverage necessary to keep both ground and aerial users connected. The study will develop in two distinct phases. The first phase will use a trisectorization strategy, to use this scenario to provide comparative data for the second phase, which will introduce beamforming technology. This second phase will include simulations using three different antenna geometries. Defining this first scenario requires careful network planning. This planning will include determining the positions of the base stations to adequately cover the study area. Planning of the radiation diagrams for the different antenna geometries will also be conducted, taking into account the \acs{ssb} (Synchronization Signal Block) and \acs{pdsch} (Physical Downlink Shared Channel) channels. Key metrics such as received power and \acs{sinr} (Signal to Interference plus Noise Ratio) will be measured. Results will show an improvement in performance with beamforming scenarios compared to trisectorization. On the other hand, for the different geometries, very similar results are observed, making it difficult to highlight

Microwaves have proven their imaging capabilities to visualize the body composition for medical applications, thanks to their penetration inside biological structures. In this context, this article presents a novel methodology that aims to extract not just the internal morphology, but also the brain’s functional activity using the UWB pulse amplitude modulation (PAM) technique to have simultaneously functional monitoring and imaging capability and apply it to monitor the Parkinson’s disease (PD). The radio-frequency system is composed of two orthogonal sets of double UWB probes operating in the frequency range of 0.5–1.5 GHz. An experimental setup has been devised that avoids complex in vivo testing, albeit allows a system proof-of-concept validation. A bio-tag (BT) consisting of an optically modulated photodiode is used to emulate local medium changes associated with cell activity. The proposed system is used to first extract the modeled brain action potential (AP) to validate the performance of the BT and then to monitor the PD based on the beta frequency band character within basal ganglia–thalamocortical (BGTC) which is a key marker for the PD. The results show a good capability of locating and differentiating the signals generated within the phantom by the BT, alternatively emulating the healthy and PD’s state, based on the frequency. The obtained results of the functional monitoring technique on distinguishing the healthy from nonhealthy brain model activity, as well as in the phantom mimicking the average proprieties of a human head, will serve as a basis for detecting functional diseases in the future., This work was supported in part by Comision Interministerial de Ciencia y Tecnologia (CICYT) PID2019-107885GB-C31, PID2022-136869NB-C31, Catalan Research Group 2021 SGR 01415, Metropolis PLEC2021-007609, (FI-SDUR) under Grant 2021 FISDU 00195; in part by the PDR-2014-2022/56-30157-2021-2A grant of the Generalitat de Catalunya; and in part by the Prueba de Concepto PDC2022-133091-I00 grants of the Ministerio de Ciencia, Innovación y Universidades (Spain)., Peer Reviewed, Postprint (published version)

Most pedestrian detection methods focus on bounding boxes based on fusing RGB with lidar. These methods do not relate to how the human eye perceives objects in the real world. Furthermore, lidar and vision can have difficulty detecting pedestrians in scattered environments, and radar can be used to overcome this problem. Therefore, the motivation of this work is to explore, as a preliminary step, the feasibility of fusing lidar, radar, and RGB for pedestrian detection that potentially can be used for autonomous driving that uses a fully connected convolutional neural network architecture for multimodal sensors. The core of the network is based on SegNet, a pixel-wise semantic segmentation network. In this context, lidar and radar were incorporated by transforming them from 3D pointclouds into 2D gray images with 16-bit depths, and RGB images were incorporated with three channels. The proposed architecture uses a single SegNet for each sensor reading, and the outputs are then applied to a fully connected neural network to fuse the three modalities of sensors. Afterwards, an up-sampling network is applied to recover the fused data. Additionally, a custom dataset of 60 images was proposed for training the architecture, with an additional 10 for evaluation and 10 for testing, giving a total of 80 images. The experiment results show a training mean pixel accuracy of 99.7% and a training mean intersection over union of 99.5%. Also, the testing mean of the IoU was 94.4%, and the testing pixel accuracy was 96.2%. These metric results have successfully demonstrated the effectiveness of using semantic segmentation for pedestrian detection under the modalities of three sensors. Despite some overfitting in the model during experimentation, it performed well in detecting people in test mode. Therefore, it is worth emphasizing that the focus of this work is to show that this method is feasible to be used, as it works regardless of the size of the dataset. Also, a bigger dataset w, "The project on which these results are based has received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No. 801342 (Tecniospring INDUSTRY). Also, the project was supported by funding from projects PID2020-119484RB-I00 and PDC2021-121038-I00 and by MCIN/AEI/10.13039/501100011033, as well as the European Union’s NextGeneration EU/PRTR and the Government of Catalonia’s Agency for Business Competitiveness (ACCIÓ).", Peer Reviewed, Postprint (published version)

Reconfigurable intelligent surfaces (RISs) have become a promising candidate for the development of future mobile systems. In the context of massive machine-type communications (mMTC), a RIS can be used to support the transmission from a group of sensors to a collector node. Due to the short data packets, we focus on the design of the RIS for maximizing the weighted sum and minimum rates in the finite blocklength regime. Under the assumption of non-orthogonal multiple access, successive interference cancelation is considered as a decoding scheme to mitigate interference. Accordingly, we formulate the optimizations as non-convex problems and propose two suboptimal solutions based on gradient ascent (GA) and sequential optimization (SO) with semi-definite relaxation (SDR). In the GA, we distinguish between Euclidean and Riemannian gradients. For the SO, we derive a concave lower bound for the throughput and maximize it sequentially applying SDR. Numerical results show that the SO can outperform the GA and that strategies relying on the optimization of the classical Shannon capacity might be inadequate for mMTC networks., This work was supported by the project ROUTE56 (PID2019-104945GB-I00 / MCIN / AEI / 10.13039/501100011033) funded by the Agencia Estatal de Investigación (Spanish Ministry of Science and Innovation); and the FPI grant BES-2017-079994, funded by the Spanish Ministry of Science, Innovation, and Universities., Peer Reviewed, Postprint (author's final draft)

Solidly Mounted Resonators (SMRs) for high frequency RF filters and sensing applications often display spurious resonances that distort their frequency response. In this work, we try to identify the origin of spurious resonances accompanying the main series resonances in AlN-based SMRs with the help of modified Butterworth Van Dyke (BVD) and Mason’s models. By manufacturing SMRs of different sizes and shapes and studying the influence of the position of the electrical probing spot, we have demonstrated both theoretically and experimentally that devices with larger areas are more likely to display these additional peaks. Our updated models accurately simulate the frequency response of the SMRs, revealing that spurious peaks are mostly related to the resistance of the electrodes. Our study clarifies the origin of the spurious resonances and offers solutions for both, the optimal design and measurement method of SMRs., This work was supported in part by the UPC University. Department of Signal and communications [Grant BS123456], CONACYT [grant 291234] Conacyt-Secretaría de Energía-Hidrocarburos 2017 octavo período, R&D National Plan of Spanish Government [PID-2020- 118410RB-C22] and Comunidad de Madrid through the call Research Grants for Young Investigators from Universidad Politécnica de Madrid, ACUEX, APOYO-JOVENES-21-TUGJDU-108-WNZK0H., Peer Reviewed, Postprint (published version)

Because of the capitalist system and the increasing demand for products, a complex industrialization process took place in the late 17th and early 19th century, starting with the use of simple machines and culminating today with the process of industrial automation and digital transformation. In other words, with the fourth industrial revolution, also known as Industry 4.0. Throughout this project, we define what Industry 4.0 is, its component parts (automation and digitalization), along with a detailed analysis of its advantages and disadvantages. Essentially, the goal of Industry 4.0 is to increase efficiency and responsiveness to customers. Thanks to digitization, errors can be detected, and line stoppages avoided, preventive and predictive maintenance can be adopted, and processes can be optimized. This report details the main systems that make digitization possible in companies. To demonstrate the importance of Industry 4.0 today, a case study has been implemented for a fictitious company in the food sector, which prepares boxes of cherries. The demonstration applies the MES (Manufacturing Execution System) Legato Sapient, a CMMS (Computer Aided Maintenance Management), a web page to access the different systems and, finally, develops a 3D simulation with Unreal Engine on the factory that has been previously monitored and digitized., Como consecuencia del sistema capitalista y el incremento de la demanda de productos, a finales del siglo XVII y principio del XIX se produjo un proceso de industrialización complejo, empezando con el uso de máquinas sencillas y culminando en la actualidad con el proceso de automatización industrial y la transformación digital. Es decir, con la cuarta revolución industrial, también denominada Industria 4.0. A lo largo de este proyecto se define qué es la industria 4.0, las partes que la componen (automatización y digitalización), junto con un análisis detallado de las ventajas e inconvenientes que reporta. Esencialmente, el objetivo de la Industria 4.0 es aumentar la eficiencia y la capacidad de respuesta a los clientes. Gracias a la digitalización, se puede detectar errores y evitar paros en la línea, adoptar mantenimientos preventivos y predictivos y optimizar los procesos. En esta memoria se detallan los principales sistemas que hacen posible la digitalización en las empresas. Para demostrar la importancia de la Industria 4.0 en la actualidad se ha implementado un caso práctico para una empresa ficticia del sector alimenticio, que prepara cajas de cerezas. La demostración aplica el sistema MES (Manufacturing Execution System) Legato Sapient, un GMAO (Gestión de Mantenimiento Asistida por Ordenador), una página web para acceder a los diferentes sistemas y, por último, desarrolla una simulación 3D con Unreal Engine sobre la fábrica que se ha monitorizado y digitalizado anteriormente., A conseqüència del sistema capitalista i l'increment de la demanda de productes, a la fi del segle XVII i principi del XIX es va produir un procés d'industrialització complex, començant amb l'ús de màquines senzilles i culminant en l'actualitat amb el procés d'automatització industrial i la transformació digital. És a dir, amb la quarta revolució industrial, també denominada Indústria 4.0. Al llarg d'aquest projecte es defineix què és la indústria 4.0, les parts que la componen (automatització i digitalització), juntament amb una anàlisi detallada dels avantatges i inconvenients que reporta. Essencialment, l'objectiu de la Indústria 4.0 és augmentar l'eficiència i la capacitat de resposta als clients. Gràcies a la digitalització, es pot detectar errors i evitar aturs en la línia, adoptar manteniments preventius i predictius i optimitzar els processos. En aquesta memòria es detallen els principals sistemes que fan possible la digitalització en les empreses. Per a demostrar la importància de la Indústria 4.0 en l'actualitat s'ha implementat un cas pràctic per a una empresa fictícia del sector alimentós, que prepara caixes de cireres. La demostració aplica el sistema MES (Manufacturing Execution System) Legato Sapient, un GMAO (Gestió de Manteniment Assistida per Ordinador), una pàgina web per a accedir als diferents sistemes i, finalment, desenvolupa una simulació 3D amb Unreal Engine sobre la fàbrica que s'ha monitorat i digitalitzat anteriorment.

Graph topology inference is a research field that studies how to learn the structure of a graph from data related to it. It is a very popular research topic in which algorithms are being developed faster and more efficient year after year. Graphs are a mathematical tool that allows us to model relationships between data obtained from complex structures, such as the human brain. In this thesis, we have developed a graph topology inference algorithm based on the estimation of the precision matrix with Laplacian constraints, where we have used real data obtained by f-MRI to evaluate the algorithm. Then, we have applied network science parameters, such as the small-world parameter, in order to compare different graph network topologies and provide valuable information about the structure of these graphs., La inferencia de topología de grafos es un campo de investigación en el que se estudia cómo aprender la estructura de un grafo a partir de datos relacionados con él. Es un tema de investigación muy popular en el que se desarrollan algoritmos cada vez más rápidos y eficientes año tras año. Los grafos son una herramienta matemática que nos permite modelar las relaciones entre datos obtenidos en estructuras complejas, como por ejemplo el cerebro humano. En esta tesis, hemos desarrollado un algoritmo de inferencia de topología de grafos basado en la estimación de la matriz de precisión con restricciones de Laplaciana, donde hemos utilizado datos reales obtenidos por f-MRI para evaluar el algoritmo. Después, hemos aplicado parámetros de ciencia de redes, como por ejemplo el parámetro "small-world", para poder comparar diferentes topologías de grafos y proporcionar una información valiosa sobre la estructura de estos grafos., La inferència de topologia de grafs és un camp de recerca en el qual s'estudia com aprendre l'estructura d'un graf a partir de dades relacionades amb ell. És un tema de recerca molt popular en el qual es desenvolupen algorismes cada vegada més ràpids i eficients any rere any. Els grafs són una eina matemàtica que ens permet modelar les relacions entre dades obtingudes en estructures complexes, com per exemple el cervell humà. En aquesta tesi, hem dut a terme un algorisme d'inferència de topologia de grafs basat en l'estimació de la matriu de precisió amb restriccions de Laplaciana, on hem utilitzat dades reals aconseguides per f-MRI per avaluar l'algorisme. Després, hem aplicat paràmetres de ciència de xarxes, com per exemple el paràmetre "small-world", per poder comparar diferents topologies de grafs i proporcionar una informació valuosa sobre l'estructura d'aquests grafs.

In this work, we report the set-up and results of the Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI) 2017 and the International Conferences on Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2017 and 2018. The image dataset is diverse and contains primary and secondary tumors with varied sizes and appearances with various lesion-to-background levels (hyper-/hypo-dense), created in collaboration with seven hospitals and research institutions. Seventy-five submitted liver and liver tumor segmentation algorithms were trained on a set of 131 computed tomography (CT) volumes and were tested on 70 unseen test images acquired from different patients. We found that not a single algorithm performed best for both liver and liver tumors in the three events. The best liver segmentation algorithm achieved a Dice score of 0.963, whereas, for tumor segmentation, the best algorithms achieved Dices scores of 0.674 (ISBI 2017), 0.702 (MICCAI 2017), and 0.739 (MICCAI 2018). Retrospectively, we performed additional analysis on liver tumor detection and revealed that not all top-performing segmentation algorithms worked well for tumor detection. The best liver tumor detection method achieved a lesion-wise recall of 0.458 (ISBI 2017), 0.515 (MICCAI 2017), and 0.554 (MICCAI 2018), indicating the need for further research. LiTS remains an active benchmark and resource for research, e.g., contributing the liver-related segmentation tasks in http://medicaldecathlon.com/. In addition, both data and online evaluation are accessible via https://competitions.codalab.org/competitions/17094., Bjoern Menze is supported through the DFG funding (SFB 824, subproject B12) and a Helmut-Horten-Professorship for Biomedical Informatics by the Helmut-Horten-Foundation. Florian Kofler is Supported by Deutsche Forschungsgemeinschaft (DFG) through TUM International Graduate School of Science and Engineering (IGSSE), GSC 81. An Tang was supported by the Fonds de recherche du Québec en Santé and Fondation de l’association des radiologistes du Québec (FRQS- ARQ 34939 Clinical Research Scholarship – Junior 2 Salary Award). Hongwei Bran Li is supported by Forschungskredit (Grant NO. FK-21- 125) from University of Zurich., Peer Reviewed, Article signat per 109 autors/es: Patrick Bilic 1,a,b, Patrick Christ 1,a,b, Hongwei Bran Li 1,2,∗,b, Eugene Vorontsov 3,a,b, Avi Ben-Cohen 5,a, Georgios Kaissis 10,12,15,a, Adi Szeskin 18,a, Colin Jacobs 4,a, Gabriel Efrain Humpire Mamani 4,a, Gabriel Chartrand 26,a, Fabian Lohöfer 12,a, Julian Walter Holch 29,30,69,a, Wieland Sommer 32,a, Felix Hofmann 31,32,a, Alexandre Hostettler 36,a, Naama Lev-Cohain 38,a, Michal Drozdzal 34,a, Michal Marianne Amitai 35,a, Refael Vivanti 37,a, Jacob Sosna 38,a, Ivan Ezhov 1, Anjany Sekuboyina 1,2, Fernando Navarro 1,76,78, Florian Kofler 1,13,57,78, Johannes C. Paetzold 15,16, Suprosanna Shit 1, Xiaobin Hu 1, Jana Lipková 17, Markus Rempfler 1, Marie Piraud 57,1, Jan Kirschke 13, Benedikt Wiestler 13, Zhiheng Zhang 14, Christian Hülsemeyer 1, Marcel Beetz 1, Florian Ettlinger 1, Michela Antonelli 9, Woong Bae 73, Míriam Bellver 43, Lei Bi 61, Hao Chen 39, Grzegorz Chlebus 62,64, Erik B. Dam 72, Qi Dou 41, Chi-Wing Fu 41, Bogdan Georgescu 60, Xavier Giró-i-Nieto 45, Felix Gruen 28, Xu Han 77, Pheng-Ann Heng 41, Jürgen Hesser 48,49,50, Jan Hendrik Moltz 62, Christian Igel 72, Fabian Isensee 69,70, Paul Jäger 69,70, Fucang Jia 75, Krishna Chaitanya Kaluva 21, Mahendra Khened 21, Ildoo Kim 73, Jae-Hun Kim 53, Sungwoong Kim 73, Simon Kohl 69, Tomasz Konopczynski 49, Avinash Kori 21, Ganapathy Krishnamurthi 21, Fan Li 22, Hongchao Li 11, Junbo Li 8, Xiaomeng Li 40, John Lowengrub 66,67,68, Jun Ma 54, Klaus Maier-Hein 69,70,7, Kevis-Kokitsi Maninis 44, Hans Meine 62,65, Dorit Merhof 74, Akshay Pai 72, Mathias Perslev 72, Jens Petersen 69, Jordi Pont-Tuset 44, Jin Qi 56, Xiaojuan Qi 40, Oliver Rippel 74, Karsten Roth 47, Ignacio Sarasua 51,12, Andrea Schenk 62,63, Zengming Shen 59,60, Jordi Torres 46,43, Christian Wachinger 51,12,1, Chunliang Wang 42, Leon Weninger 74, Jianrong Wu 25, Daguang Xu 71, Xiaoping Yang 55, Simon Chun-Ho Yu 58, Yading Yuan 52, Miao Yue 20, Liping Zhang 58, Jorge Cardoso 9, Spyridon Bakas 19,23,24, Rickmer Br, Postprint (published version)

This work presents an analytical formulation to assess the six-degrees-of-freedom-motion-induced error in floating Doppler wind LiDARs (FDWLs). The error products derive from the horizontal wind speed bias and apparent turbulence intensity. Departing from a geometrical formulation of the FDWL attitude and of the LiDAR retrieval algorithm, the contributions of the rotational and translational motion to the FDWL-measured total error are computed. Central to this process is the interpretation of the velocity–azimuth display retrieval algorithm in terms of a first-order Fourier series. The obtained 6 DoF formulation is validated numerically by means of a floating LiDAR motion simulator and experimentally in nearshore and open-sea scenarios in the framework of the Pont del Petroli and IJmuiden campaigns, respectively. Both measurement campaigns involved a fixed and a floating ZephIRTM 300 LiDAR. The proposed formulation proved capable of estimating the motion-induced FDWL horizontal wind speed bias and returned similar percentiles when comparing the FDWL with the fixed LiDAR. The estimations of the turbulence intensity increment statistically matched the FDWL measurements under all motional and wind scenarios when clustering the data as a function of the buoy’s mean tilt amplitude, mean translational-velocity amplitude, and mean horizontal wind speed., This research project was part of the project PID2021-126436OB-C21 funded by the Ministerio de Ciencia e Investigación (MCIN)/Agencia Estatal de Investigación (AEI)/10.13039/501100011033 y FEDER “Una manera de hacer Europa”. The work of A. Salcedo-Bosch was supported by grant 2020 FISDU 00455 funded by Generalitat de Catalunya—AGAUR. The work of M.P Araújo da Silva was supported under Grant PRE2018-086054 funded by MCIN/AEI/10.13039/501100011033 and FSE “El FSE invierte en tu futuro”. The European Commission collaborated under projects H2020 ATMO-ACCESS (GA-101008004) and H2020 ACTRIS-IMP (GA-871115)., Peer Reviewed, Postprint (published version)

The Project consists of the analysis, programming and simulation of an on-line scheduler for time-sensitive applications, with different variables to respond to the QoS that each user may have. Thanks to the ability of the algorithm to consider the channel quality of each user, the time delay due to queueing and the different delay and rate requirements, the scheduler is able to minimize the users delay while satisfiying the different per-user QoS constraints. It has been possible to conclude, thanks to the designed simulator, that for a given latency constraint (QoS) bounded to an outage probability the real-time scheduler will be able to give optimal response to the requirements, El proyecto consiste en el análisis, programación y simulación de un planificador en línea para aplicaciones sensibles al tiempo, con diferentes variables para responder a la calidad de servicio (QoS) que cada usuario pueda tener. Gracias a la capacidad del algoritmo para considerar la calidad del canal de cada usuario, el retraso debido a la cola y los diferentes requisitos de retraso y tasa de tráfico, el scheduler es capaz de dar una respuesta óptima y justa, minimizando el retraso de los usuarios mientras satisface las diferentes restricciones de QoS por usuario. Se ha podido concluir, gracias al simulador diseñado, que para una determinada restricción de latencia (QoS) vinculada a una probabilidad de outage, el planificador en tiempo real será capaz de dar una respuesta óptima a los requisitos., El projecte consisteix en l'anàlisi, programació i simulació d'un planificador (scheduler) en temps real per aplicacions sensibles al temps, amb diferents variables per a donar resposta a la qualitat de servei (QoS) que cada usuari pugui tenir. Gràcies a la capacitat de l'algorisme de tenir en compte la qualitat del canal de cada usuari, el retard que porten els paquets en cua i els diferents requisits de retard i tràfic. El scheduler és capaç de donar una resposta òptima i justa, minimitzant el retad pels usuaris mentres satisfà les diferents restriccions de QoS per usuari. S'ha pogut concloure, gràcies al simulador disenyat, que per a una restricció temporal (QoS) lligada a una probabilitat d?outage, el scheduler en temps real, donarà una resposta als requeriments i aquesta serà propera als límits de velocitat establerts per la teoria de la informació.

Primary aldosteronism (PA) is one possible cause of hypertension. Primary aldosteronism can be treated, hence the interest for its timely diagnosis. Previous work suggests that the appearance of the carotid artery wall in longitudinal ultrasound images differs between patients with PA and with the so-called essential hypertension. However, the achieved accuracy using classical texture analysis methods is only around 70%. Moreover, this previous work relies on an expert to localize the region of interest (ROI) to be analyzed., El aldosteronismo primario (AP) es una posible causa de hipertensión. El aldosteronismo primario puede ser tratado, de ahí el interés por su diagnóstico oportuno. Trabajos previos sugieren que la apariencia de la pared de la arteria carótida en imágenes longitudinales de ultrasonido difiere entre pacientes con AP y con la llamada hipertensión esencial. Sin embargo, la precisión alcanzada utilizando los métodos clásicos de análisis de texturas es solo de alrededor del 70 %. Además, este trabajo previo cuenta con un experto para localizar la región de interés (ROI) a analizar., L'aldosteronisme primari (AP) és una possible causa d'hipertensió. L'aldosteronisme primari es pot tractar, d'aquí l'interès pel seu diagnòstic oportun. Treballs anteriors suggereixen que l'aparició de la paret de l'artèria caròtida en imatges d'ecografia longitudinal difereix entre pacients amb PA i amb l'anomenada hipertensió essencial. No obstant això, la precisió aconseguida mitjançant els mètodes clàssics d'anàlisi de textures és només al voltant del 70%. A més, aquest treball previ compta amb un expert per localitzar la regió d'interès (ROI) a analitzar.

Utilitzar filtre en grafs per a predir la topologia d'un graf., This project will focus on the development of methods to predict relations (links) between elements (nodes) of a network structured in a graph. In order to study these systems, the graphs will be understood as a signal domain. Therefore, the aim of the project becomes to design a signal processing method to predict missing or future links of a graph, Este proyecto se centra en el desarrollo de métodos para predecir las relaciones (enlaces) entre elementos (nodos) de una red estructurada en un grafo. Para estudiar estos sistemas, los grafos se entenderán como un dominio de señales. Por lo tanto, el objetivo del proyecto se convierte a diseñar un método de procesamiento de señales para predeci enlaces futuros o desconocidos de un grafo, Aquest projecte se centra en el desenvolupament de mètodes per predir les relacions (enllaços) entre elements (nodes) d'una xarxa estructurada en un graf. Per tal d'estudiar aquests sistemes, els grafs s'entendran com un domini de senyals. Per tant, l'objectiu del projecte es converteix en dissenyar un mètode de processament de senyals per predir enllaços futurs o desconeguts d'un graf

These last years have been an increasing need regarding the automatic sport analysis. Being soccer one of the most watched sports around the world, action spotting for soccer videos has become one of the main field studies in computer vision. In this project, we have done a deep research and analysis of the State of the Art regarding human recognition and action recognition in soccer videos. Apart from this more theoretical part, we have also executed and compared two of the most pioneering models in sports action spotting today: CALF[3] and NetVLAD++[4] models. In addition, we have been able to make a modification in one hyperparameter, the window size, to evaluate its effect on performance. Finally, we have concluded that we get the highest mAP performance with the NetVLAD++[4] model and that modifying the window size worsens the overall performance in both models, but the individual class performance can benefit from it as the performance is improved in some classes for different window sizes.

The need for non-ionizing techniques for medical imaging applications has led to the use of microwave signals. Several systems have been introduced in recent years based on increasing the number of antennas and frequency bandwidth to obtain high resolution and good accuracy in locating objects. A novel microwave imaging system that reduces the number of required antennas for precise target location appropriate for medical applications is presented. The proposed system consists of four UWB extended gap ridge horn (EGRH) antennas covering the frequency band from 0.5 GHz to 1.5 GHz mounted on a cylindrical phantom that mimics the brain in an orthogonal set of two EGRH probes. This configuration has the ability to control both the longitudinal and transversal dimensions of the reconstructed target’s image, rather than controlling the spatial resolution, by increasing the frequency band that can be easily affected by medium losses. The system is tested numerically and experimentally by the detection of a cylindrical target within a human brain model., This work was financially supported by CICYT PID2019-107885GB-C31, and support grants for university departments and research units (FI-SDUR)(2021 FISDU 00195). This work was also supported by the National Center for Scientific and Technological Research (CNRST) under grant number PPR2/2015/36., Peer Reviewed, Postprint (published version)

Advances in digital health and phenotyping technologies are crucial to ensureincreased access to high-quality mental health support services and treatment. Speech is uniquely placed in this regard, as no other mobile health signal contains its singular combination of cognitive, neuromuscular and physiological information. It is this complexity which makes speech a suitable marker for different mentalhealth conditions.However, much research exploring links between speech and depression is limited, and co-morbidities with conditions such as anxiety have not been exploited to help improve machine learning models.The purpose of this project is to jointly learn depression and anxietydirectly from speech signals.For this project, speech signals were split into segments that were converted into Mel-spectrograms. Automatic feature extraction was performed using a CNN-LSTM model that can classify into5 severities ofdepression. With transfer learning, this model was then usedas a pre-trained model for other tasks, such as classifying speech signals into different 4 severities of anxiety or improving modelsfor both co-morbiditiesin different languages. Finally, a Multi-Task learning model is used to jointly detect depression and anxiety. Models that use transfer learning to detectanxiety achieve an improvement from 67% to 72% of accuracy, while multi-Task learning models achieve an accuracy of 71% for both co-morbidities, anxiety and depression. The experiments show promising results, discussing the viability of jointly detecting mental health conditions such as depression and anxiety as well as exploiting the viability of using models pre-trained for just one condition, language or task to fine-tune a model for another condition, language or task, demonstrating that co-morbidities can help to improve models for joint learning severities directly from speech signals.

In recent years, the number of infant deaths being left inside a vehicle has considerably increased, most of them due to careless parents unknowingly leaving the infant locked in the car. Therefore, this thesis aims to put forward a solution preventing this kind of infant deaths by detecting human presence inside the vehicle, especially infants and babies. In other words, a solution based on the installation of a Child Presence Detection (CPD) system inside the vehicle. This preventing system is based on different radar technologies together with the use of Deep Leaning techniques, which use multiple layers to extract relevant features of the data gathered by the radar and produce an alarm if necessary. This consists of an alternative low-cost non-contact implementation by using Doppler and FMCW radars, which can easily be integrated in the car structure while delivering high accuracy measurements and presenting lower power consumption. From the multiple Deep Learning structures that exist, this thesis explores the most interesting ones dealing with this kind of radar data, such as CNN, RNN, R-CNN among many others. The implementation using the 24GHz Doppler radar has been able to perform CPD with a considerably high accuracy considering it is a low-cost solution. On the other hand, the higher bandwidth of the 60GHz FMCW radar, not only has allowed to perform CPD, but it has also been able to localize the different target positions in the car., En los últimos años, el número de niños fallecidos encerrados en vehículos ha incrementado considerablemente; en muchos casos debido al descuido de los padres que han dejado encerrado al bebé accidentalmente. Así pues, esta tesis tiene la intención de ofrecer una solución que prevenga este tipo de muertes mediante la detección de presencia humana dentro del vehículo, especialmente babes i niños. En otras palabras, una solución basada en la instalación de lo que se conoce como un sistema Child Presence Detection (CPD). Este sistema de prevención está basado en diferentes tecnologías radar en conjunto con el uso de técnicas de Deep Learning, que usan múltiples capas para extraer características importantes de los datos obtenidos por el radar, y así poder alarmar en caso de que sea conveniente. Más concretamente, consiste en una alternativa de bajo coste y a distancia mediante el uso de radares Doppler y FMCW, que pueden ser fácilmente integrados en la estructura de vehículo y seguir dando medidas de alta fiabilidad. De las múltiples técnicas de Deep Learning que existen, este trabajo explora las más interesantes y adecuadas para tratar este tipo de datos de radar, como pueden ser las CNN, RNN, R-CNN entre otras. La implementación usando el radar Doppler de 24GHz ha sido capaz de realizar una detección de bebés con una precisión considerablemente elevada, teniendo en cuenta que se trata de una solución de bajo coste. Por otro lado, en lo que se refiere al radar de 60GHz FMCW, no solo ha sido capaz de realizar la detección de niños, sino que también ha logrado localizar su posición dentro del coche., En els darrers anys, el nombre d'infants morts tancats a vehicles ha incrementat considerablement, moltes de les quals per culpa del descuit dels pares que han deixat tancat al nadó sense tenir-ne coneixement. Així doncs, aquesta tesi té la intenció d'oferir una solució que previngui aquest tipus de morts mitjançant la detecció de presencia humana dins el vehicle, especialment nadons i nens. En altres paraules, una solució basada en la instal·lació del que es coneix com a un sistema Child Presence Detection (CPD). Aquest sistema de prevenció està basat en diferents tecnologies radar juntament amb l'ús de tècniques de Deep Learning, que fan servir múltiples capes per a extreure característiques importants de les dades obtingudes amb el radar, i així poder alarmar en cas que sigui convenient. Més concretament, consisteix en una alternativa de baix cost i a distància mitjançant l'ús de radars Doppler i FMCW, que poden ser fàcilment integrats en l'estructura del vehicle i continuar donant mesures d'alta fiabilitat. De les múltiples tècniques de Deep Learning que existeixen, aquest treball explora les més interessants per a tractar aquest tipus de dades, com poden ser les CNN, RNN, R-CNN entre d'altres. La implementació fent servir el radar Doppler de 24GHz ha estat capaç de realitzar una detecció de nadons amb una precisió considerablement elevada tenint en compte que es tracta d'una solució de baix cost. D'altra banda, pel que fa referència al radar de 60GHz FMCW, no tan sols ha estat capaç de dur a terme la detecció d'infants, sinó que també ha aconseguit localitzar la seva posició dins el cotxe.

The detection and sizing of fruits with computer vision methods is of interest because it provides relevant information to improve the management of orchard farming. However, the presence of partially occluded fruits limits the performance of existing methods, making reliable fruit sizing a challenging task. While previous fruit segmentation works limit segmentation to the visible region of fruits (known as modal segmentation), in this work we propose an amodal segmentation algorithm to predict the complete shape, which includes its visible and occluded regions. To do so, an end-to-end convolutional neural network (CNN) for simultaneous modal and amodal instance segmentation was implemented. The predicted amodal masks were used to estimate the fruit diameters in pixels. Modal masks were used to identify the visible region and measure the distance between the apples and the camera using the depth image. Finally, the fruit diameters in millimetres (mm) were computed by applying the pinhole camera model. The method was developed with a Fuji apple dataset consisting of 3925 RGB-D images acquired at different growth stages with a total of 15,335 annotated apples, and was subsequently tested in a case study to measure the diameter of Elstar apples at different growth stages. Fruit detection results showed an F1-score of 0.86 and the fruit diameter results reported a mean absolute error (MAE) of 4.5 mm and R2 = 0.80 irrespective of fruit visibility. Besides the diameter estimation, modal and amodal masks were used to automatically determine the percentage of visibility of measured apples. This feature was used as a confidence value, improving the diameter estimation to MAE = 2.93 mm and R2 = 0.91 when limiting the size estimation to fruits detected with a visibility higher than 60%. The main advantages of the present methodology are its robustness for measuring partially occluded fruits and the capability to determine the visibility percentage. The main limitation is that, This work was partly funded by the Departament de Recerca i Universitats de la Generalitat de Catalunya (grant 2021 LLAV 00088), the Spanish Ministry of Science, Innovation and Universities (grants RTI2018-094222-B-I00 [PAgFRUIT project], PID2021-126648OB-I00 [PAgPROTECT project] and PID2020-117142GB-I00 [DeeLight project] by MCIN/AEI/10.13039/501100011033 and by “ERDF, a way of making Europe”, by the European Union). The work of Jordi Gené Mola was supported by the Spanish Ministry of Universities through a Margarita Salas postdoctoral grant funded by the European Union - NextGenerationEU., Peer Reviewed, Postprint (published version)

One of the main topics in artificial intelligence is the speech characterization. Moreover, it is a field of study with the minimal scope when the Catalan language is involved in. In this project, we try to perform an age classification by decades firstly in the Catalan CommonVoice Dataset and then add the Spanish Dataset and English Dataset to have more data. To reach our purpose Deep Learning techniques are used to implement the classifier. The most common backbones are used such as Resnet and VGG. Furthermore, we use an attention encoder to encode the Mel-Spectrogram features. In contrast to statistical pooling methods like average pooling, Attention Pooling layers and various Attention Mechanisms are used in all backbones to perform pooling and reduce the dimensionality of the feature vector derived from the Front-End architecture. In this study, we will compare two different models, the first with an AM-Softmax in the final layer and the other with an AM-Softmax combined with Ordinal Regression.

Breakthroughs in fully autonomous vehicle development are reaching and all-time high. The question is not if, but when this technology is will make it on the market. And the likely answer is soon. Although AVs are just a few years from being launched, many questions remain unanswered. One thing is clear though. Distinct adoption patterns will shape the trajectory of the technology. Most research in this field emphasises the point that an AV future will be shared, shifting to an on-demand mobility future. Nonetheless, polling data shows that many people don?t want to give up their private vehicle for a shared one. This paper will illuminate the often-overlooked side of private AV adoption. Thereby, different private autonomous vehicle deployment scenarios will be simulated. The data will then be analysed to determine the impacts of a private AV adoption on emissions and traffic flow. Simultaneously, research gaps in that area will be filled for a European region.

Nowadays, we use a large number of elements in our daily lives that require connectivity, from communications, to entertainment consumption, to remote home automation. For this, the best way to have a reliable, stable and high-speed connection is fibre optics. This project consists of the design and characterisation of a fibre-optic network to the home (Fiber-to-the-Home or FTTH), seeing and understanding how the light is carried from the optical headends to the terminal boxes near the end consumers. It also describes the elements that make it up, the necessary infrastructure and the GPON technology used. In addition, we will learn about the different possible topologies for our network and the existing versions of GPON. The design is for the area of L'Ardiaca, in the municipality of Cambrils, province of Tarragona. This design is totally realistic, following the existing regulations and the operator's guidelines.

Cloud Computing Course

The advent of Fifth generation (5G) and B5G future mobile networks has generated significant growth in the demand for reliable and low-latency data networks to support a wide range of applications and services with demanding requirements, such as monitoring, remote surgery, and vehicles connectivity. To address these challenges, small cells have emerged as a solution, offering high data rates in dense device environments. However, the increased density of deployed cells leads to frequent handovers and signaling overheads, creating challenges for managing radio resources. This work explores the use of prediction algorithms to address these challenges by accurately predicting handovers to prepare in advance the required resources in the next cells visited by the User Equipment (UE). Various methodologies, including machine learning (ML) algorithms, have been proposed for predicting UE mobility and target base stations. Prediction models based on Markov models, Bayesian Networks, Neural Networks, Support Vector Machines (SVM), and clustering techniques have been developed to utilize mobility patterns and improve prediction accuracy. The main objective of the thesis is to develop a prediction model that utilizes historical user trajectory data to predict the next target base station during a handover process. The proposed methodology utilizes pre-processed subtrajectories as features to identify trajectory patterns and make predictions. Simulated mobility trajectories of vehicles in the city of Cologne are used to test the proposed approach. The model's performance is evaluated based on prediction certainty, anticipation time, and processing time. The performance of different classifiers and subtrajectory sizes is evaluated to determine the most effective model parameters. The Logistic Regression algorithm is found to provide the highest prediction certainty, while subtrajectories of size 2 yield faster processing times. The developed ML classification model is compared

Incoming

A graph-based approach uses a triangular adaptive mesh for simulating the propagation of light beams through the atmosphere. In this approach, the atmospheric turbulence and the beam wavefront are signals in a graph, with vertices representing an irregular distribution of signal points and edges between vertices showing their relationships. The adaptive mesh provides a better representation of the spatial variations in the beam wavefront, resulting in increased accuracy and resolution compared to regular meshing schemes. The adaptability of this approach to the propagated beam characteristics makes it a versatile tool for simulating beam propagation in various turbulence conditions., FUNDING: Funding. Agencia Estatal de Investigación (PID2020-118410RB-C21)., Peer Reviewed, Postprint (published version)

Every day between 7,000 and 10,000 million tons of waste is generated around the world. This work deals with the collection of environmental waste, its identification and subsequent classification in order to contribute to recycling in a world that is increasingly aware of caring for the environment. A public database has been created from scratch with the aim that it can be used by anyone who needs it. For the training of the model, YOLO architecture has been used, specifically the latest version. Thanks to this the detection of different objects can be performed. The purpose of generating this model is to implement it in a real application and make a model using Arduino that simulates the waste classification process. This project has helped me to understand how convolutional neural networks work, learn how to create databases for projects focused on object detection, and generate models to be used in everyday applications., Objectius de Desenvolupament Sostenible::13 - Acció per al Clima

This paper explores the problem of estimating a temporal series measured from multiple independent sensors with unequal and stationary measurement errors with unknown variances. By formulating the data fusion problem as a joint Maximum Likelihood estimation of sensor covariances and a fusion rule, a batch data driven method is derived involving a residual covariance determinant minimization of a diagonal matrix. It is shown that yielding useful learning from data with good generalization properties in the joint regression and fusion approach requires the assumption of some structure on the sensor noises and/or on the temporal series to be estimated. An efficient data driven algorithm is proposed to obtain the best linear sensor combiner, whose performance is numerically analyzed and compared with the Cramer-Rao Lower Bound of the estimated parameters., This work has been supported by the Spanish Ministry of Science and Innovation through project RODIN (PID2019-105717RB-C22 / AEI / 10.13039/501100011033) and by the Catalan Government (AGAUR) under grant 2017 SGR 578., Peer Reviewed, Postprint (author's final draft)