Search

Electroencephalography (EEG) is widely used as a non-invasive technique for the diagnosis of several brain disorders, including Alzheimer’s disease and epilepsy. Until recently, diseases have been identified over EEG readings by human experts, which may not only be specific and difficult to find, but are also subject to human error. Despite the recent emergence of machine learning methods for the interpretation of EEGs, most approaches are not capable of capturing the underlying arbitrary non-Euclidean relations between signals in the different regions of the human brain. In this context, Graph Neural Networks (GNNs) have gained attention for their ability to effectively analyze complex relationships within different types of graph-structured data. This includes EEGs, a use case still relatively unexplored. In this paper, we aim to bridge this gap by presenting a study that applies GNNs for the EEG-based detection of Alzheimer’s disease and discrimination of two different types of seizures. To this end, we demonstrate the value of GNNs by showing that a single GNN architecture can achieve state-of-the-art performance in both use cases. Through design space explorations and explainability analysis, we develop a graph-based transformer that achieves cross-validated accuracies over 89% and 96% in the ternary classification variants of Alzheimer’s disease and epilepsy use cases, respectively, matching the intuitions drawn by expert neurologists. We also argue about the computational efficiency, generalizability and potential for real-time operation of GNNs for EEGs, positioning them as a valuable tool for classifying various neurological pathologies and opening up new prospects for research and clinical practice., This work is supported by Fa.Per.M.E project (project code: T3AN-15, CUP C33C22000390006, CUP MASTER H53C22000640006) funded by the Italian Ministry of Health. This work was also granted by the Programma Operativo Nazionale (PON) ‘‘Ricerca e Innovazione’’ 2014-2020: CCI2014IT16M2OP005 (CUP C35F21001220009 code: I05)., Peer Reviewed, Postprint (published version)

This paper investigates the efficient application of half-precision floating-point (FP16) arithmetic on GPUs for boosting LU decompositions in double (FP64) precision. Addressing the motivation to enhance computational efficiency, we introduce two novel algorithms: Pre-Pivoted LU (PRP) and Mixed-precision Panel Factorization (MPF). Deployed in both hybrid CPU-GPU setups and native GPU-only configurations, PRP identifies pivot lists through LU decomposition computed in reduced precision and subsequently reorders matrix rows in FP64 precision before executing LU decomposition without pivoting. Two variants of PRP, namely hPRP and xPRP, are introduced, differing in their computation of pivot lists in full half-precision or mixed half-single precision. The MPF algorithm generates FP64 LU factorization while internally utilizing hPRP for panel factorization, showcasing accuracy on par with standard DGETRF but with superior speed. The study further explores auxiliary functions required for the native mode implementation of PRP variants and MPF., The authors acknowledge the funding received from project PID2019-107255GB of the Spanish Ministerio de Economía Competitividad, and project 2021-SGR-01007 from the Generalitat de Catalunya, Spain. Sandra Catalán was supported by the grant RYC2021-033973-I, funded by MCIN/AEI/10.13039/501100011033 and the European Union “NextGenerationEU”/PRTR, and UJI-2023-04, funded by Universitat Jaume I., Peer Reviewed, Postprint (published version)

The Joint Laboratory on Extreme-Scale Computing (JLESC) was initiated at the same time lossy compression for scientific data became an important topic for the scientific communities. The teams involved in the JLESC played and are still playing an important role in developing the research, techniques, methods, and technologies making lossy compression for scientific data a key tool for scientists and engineers. In this paper, we present the evolution of lossy compression for scientific data from 2015, describing the situation before the JLESC started, the evolution of this discipline in the past 8 years (until 2023) through the prism of the JLESC collaborations on this topic and some of the remaining open research questions., This research was supported by the Exascale Computing Project (ECP), Project Number: 17-SC-20-SC , a collaborative effort of two DOE organizations – the Office of Science and the National Nuclear Security Administration, responsible for the planning and preparation of a capable exascale ecosystem, including software, applications, hardware, advanced system engineering, and early testbed platforms, to support the nation’s exascale computing imperative. The material was supported by the U.S. Department of Energy, Office of Science, Advanced Scientific Computing Research (ASCR), under contract DE-AC02-06CH11357 , and supported by the National Science Foundation under Grant OAC-2003709/2303064 , OAC-2104023/2247080 , OAC-2311875/2311876/2311877 , OAC-2312673 , and OAC-2034169 . We acknowledge the computing resources provided on Bebop (operated by the Laboratory Computing Resource Center at Argonne). Some of the experiments presented in this paper were carried out using the PlaFRIM experimental testbed, supported by Inria, CNRS (LABRI and IMB), Université de Bordeaux, Bordeaux INP and Conseil Régional d’Aquitaine (see https://www.plafrim.fr ). TEZip - This work has been supported by the COE research grant in computational science from Hyogo Prefecture and Kobe City through the Foundation for Computational Science. XIOS-SZ - Mario Acosta and Xavier Yepes-Arbós have received co-funding from the State Research Agency through OEMES ( PID2020-116324RA-I00 )., Peer Reviewed, Article signat per 20 autors/es: Franck Cappello (a), Sheng Di (a), Robert Underwood (a), Dingwen Tao (b), Jon Calhoun (c), Yoshii Kazutomo (a), Kento Sato (d), Amarjit Singh (d), Luc Giraud (e), Emmanuel Agullo (e), Xavier Yepes (f), Mario Acosta (f), Sian Jin (b), Jiannan Tian (b), Frédéric Vivien (e), Boyuan Zhang (b), Kentaro Sano (d), Tomohiro Ueno (d), Thomas Grützmacher (g), Hartwig Anzt (g) / (a) Argonne National Laboratory, United States of America; (b) Indiana University, Bloomington, United States of America; (c) Clemson University, United States of America; (d) RIKEN Center for Computational Science, Japan; (e) National Research Institute for Computing and Automation, France; (f) Barcelona Supercomputing Center, Spain; (g) Karlsruhe Institute of Technology, Germany, Postprint (author's final draft)

The advent of Wi-Fi in the 1990s stimulated the creation of Community networks (CNs), a new networking paradigm deployed and maintained by their users. The non-profit and grassroots self-organization of CNs has shown significant potential in enhancing communication technologies grounded in ethical values, transcending mere Internet expansion to contribute to broader social impacts. Since its conception, hundreds of CNs have been deployed worldwide. However, many have faltered, primarily due to a lack of solid economic models and legal setup. This paper presents Expansió Xarxa Oberta (eXO) association, a commons Internet Service Provider (ISP). In contrast to commercial ISPs, eXO follows the same grassroots principles of CNs. Establishing a commons-based ISP is challenging, yet it is a key component for the sustainability and success of CNs. Therefore, our paper aims to dissect the complexity of a commons ISP by explaining eXO, a pioneering example that has already been in production for more than 14 years. To the best of our knowledge, this is the first time that a commons ISP is described in all its components. To describe eXO, we have proceeded in two parts. In the first one, we delineate the organizational structure, economics, funding mechanisms, activities, legal framework, and infrastructure. To gain more insight on eXO, in the second part, we conduct a technical analysis of traffic data collected from eXO’s core router, elucidating subscriber usage patterns. Our study encompasses various subscriber categories, including regular and financially subsidized users, Wi-Fi and optical fiber access, and traffic characteristics such as volume, TCP/UDP usage, services accessed, and traffic distribution across countries., This work was funded by the Spanish Ministerio de Ciencia Innovación y Universidades through the grant PDC2023-145809-I00/AEI/ 10.13039/501100011033; and PID2023-146066OB-I00 (AEI 2023 Knowledge Generation Projects); the Recovery and Resilience Mechanism of the European Union, and by the European Union NextGenerationEU; the Generalitat de Catalunya through the grants 2021-SGR01059, 2021-SGR-01476, and Comunalitats Urbanes 2022-2024; and the Universitat Politècnica de Catalunya through the grant ALECTORS 0BV05-1., Peer Reviewed, Postprint (published version)

Aquest projecte de Treball de Fi de Grau, titulat Extensión de las aplicaciones para la detección de ataques OSINT en HuntDown, té com a objectiu afegir nous atacs de naturalesa OSINT a l'aplicació de HuntDown. D'aquesta manera, a part de tenir la capacitat de detectar vulnerabilitats amb els atacs ja disponibles, es conseguirà recaptar informació dels objectius. El projecte es desenvolupa a la Universitat Politècnica de Catalunya (UPC), més concretament a la Facultat d'Informàtica de Barcelona (FIB). Es realitza com a part final del Grau d'Enginyeria Informàtica i és una continuació d'una sèrie de projectes d'estudiants de la mateixa facultat. HuntDown va néixer amb l'objectiu d'automatitzar auditories de ciberseguretat basades en proves de penetració (pentesting) per a les empreses. A més, és una eina gratuïta i senzilla que pretén ajudar les organitzacions amb pocs recursos a fer front als problemes cibernètics i la transformació digital., This Final Degree Project, entitled Extensión de las aplicaciones para la detección de ataques OSINT en HuntDown, aims to add new OSINT (Open Source Intelligence) attacks to the HuntDown application. In this way, in addition to being able to detect vulnerabilities with the already available attacks, it will be possible to gather information about the targets. The project is being developed at the Universitat Politècnica de Catalunya(UPC), specifically at the Facultat d'Informàtica de Barcelona(FIB). It is carried out as part of the completion of the Computer Engineering Degree and is a continuation of a series of projects by students from the same faculty. HuntDown was created with the aim of automating cybersecurity audits based on penetration testing for companies. Additionally, it is a free and simple tool designed to assist organizations with limited resources in addressing cyber issues and digital transformation.

Aquest projecte de final de grau està dedicat a millorar la visualització de dades ràster a l'empresa HYDS. Actualment, la plataforma utilitza capes ràster amb algunes limitacions. La proposta implica l'adopció de 'Vector Tiles' per aconseguir una visualització més eficient i millorar l'experiència de l'usuari. L'objectiu principal és integrar aquesta tecnologia sense afectar negativament els sistemes existents de l'empresa. El projecte es desenvoluparà seguint una línia d'adaptació a 'Vector Tiles', posterior implementació i proves per avaluar-ne el rendiment comparativament amb la solució actual. En resum, aquesta iniciativa té com a finalitat optimitzar la visualització de dades ràster a través de l'ús de 'Vector Tiles', amb un enfocament especial en la integració eficient amb els serveis existents de HYDS., In this senior project, we focus on improving the visualization of raster data for the company HYDS. Currently, the platform uses raster layers with some limitations. The proposal involves adopting 'Vector Tiles' to achieve more efficient visualization and enhance the user experience. The main goal is to integrate this technology without negatively impacting the company's existing systems. The project will be developed following a path of adaptation to 'Vector Tiles,' followed by implementation and testing to assess its performance compared to the current solution. In summary, this initiative aims to optimize the visualization of raster data using 'Vector Tiles,' with a special focus on efficient integration with HYDS's existing services.

El projecte titulat GESTIÓ D'IDENTITATS I ACCÉS DE LA XARXA correspon al Treball Final de Grau (TFG) per a la Facultat d'Informàtica de Barcelona dins del Grau d'Enginyeria Informàtica i l'especialitat de tecnologies de la informació amb la col·laboració de la consultora Ernst & Young. La gestió d'identitats i accessos (IAM) és un conjunt de polítiques, procediments i tecnologies dissenyades per assegurar de manera segura l'accés als recursos digitals dins d'una organització. Això inclou sistemes, aplicacions, dades i altres actius crítics. El nostre objectiu principal és evitar que cap usuari, rol o actiu de l'empresa sigui un punt d'entrada vulnerable per a possibles atacants sense que les restriccions impedeixin el bon funcionament del client., The project titled NETWORK IDENTITY AND ACCESS MANAGEMENT corresponds to the Final Degree Project (TFG) for the Faculty of Informatics of Barcelona within the Bachelor's Degree in Computer Engineering and the specialization in Information Technologies, in collaboration with the consulting firm Ernst & Young. Identity and Access Management (IAM) is a set of policies, procedures, and technologies designed to securely ensure access to digital resources within an organization. This includes systems, applications, data, and other critical assets. Our main objective is to prevent any user, role, or asset of the company from being a vulnerable entry point for potential attackers without the restrictions hindering the client's proper functioning.

(English) The emergence of new applications such as autonomous machines (e.g., robots or self-driving cars) and XR (Extended Reality) promises to revolutionize how society interacts with technology in the rapidly advancing digital era. These technologies, deployed on edge devices, often rely on mobile or embedded SoCs (Systems-on-a-Chip) operating CV (Continuous Vision) pipelines that periodically capture and analyze environmental light. A typical CV SoC comprises two main components: a frontend for image capture and a backend for processing vision algorithms. The frontend usually includes an off-chip camera sensor and an ISP (Image Signal Processor), which processes the pixel stream, converting raw sensor data into high-quality images. The backend—comprising components such as the CPU, GPU, or specialized accelerators—analyzes the images stored in the main memory's framebuffer to extract perception insights and enable advanced decision-making. Existing research identifies visual localization, object detection, and tracking as the primary bottlenecks in these emerging applications. Those algorithms face two principal challenges when deployed in mobile CV systems: latency and energy consumption. For example, an XR headset uses visual localization to track head motion for accurate frame rendering, where latency can cause discomfort. In self-driving cars, localization ensures centimeter-level precision, with delays compromising safety, especially at higher speeds. Additionally, high energy consumption limits the operation of battery-powered mobile systems. This thesis embarks on a strategic journey to elevate mobile CV systems' performance and energy efficiency through several key contributions: We begin by analyzing a state-of-the-art visual localization engine on a CPU. The localization engine processes camera images, extracting and tracking features to estimate camera pose. Our evaluations reveal feature extraction as the primary bottleneck, accounting for 60% to 90, (Català) L’aparició de noves aplicacions com màquines autònomes (per exemple, robots o cotxes autònoms) i RE (Realitat Estesa) promet revolucionar la interacció de la societat amb la tecnologia una era digital. Aquestes aplicacions, desplegades en dispositius edge, sovint depenen de SoCs mòbils que operen amb pipelines de VC (Visió Contínua) per capturar i analitzar la llum ambiental. Un SoC típic de VC consta de dos components principals: un frontend per a la captura d’imatges i un backend per al processament dels algoritmes de visió. El frontend inclou un sensor d’imatges i un ISP (Image Signal Processor), que converteix les dades sense processar del sensor en imatges d’alta qualitat. El backend, que inclou components com la CPU, GPU o acceleradors, analitza les imatges emmagatzemades en un buffer de la memòria principal per permetre la presa de decisions. Les investigacions actuals identifiquen la localització i la detecció i seguiment d’objectes com els principals colls d’ampolla en les aplicacions emergents de visió. Aquests algorismes s’enfronten a dos desafiaments en sistemes mòbils de VC: latència i consum d’energia. La latència pot causar marejos als usuaris d’un casc de RE que utilitzi la localització per estimar la posició del cap i renderitzar els frames. En els cotxes autònoms, la localització assegura una precisió de centímetres, de manera que un error o retard compromet greument la seguretat. A més, el consum d’energia limita l’operació dels sistemes alimentats per bateries. Aquesta tesi emprèn un viatge estratègic per millorar el rendiment i l’eficiència energètica dels sistemes mòbils de VC a través de diverses contribucions clau: Comencem analitzant un motor de localització visual d’avantguarda en una CPU. El motor de localització processa imatges de càmera, extraient i seguint característiques per estimar la posició de la càmera. Les nostres avaluacions revelen que l’extracció de característiques és el principal coll d’ampolla, representant del 60, (Español) La aparición de nuevas aplicaciones como máquinas autónomas (p. ej., robots o coches autónomos) y RE (Realidad Extendida) promete revolucionar la interacción de la sociedad con la tecnología en la era digital. Estas aplicaciones, desplegadas en dispositivos edge, a menudo dependen de SoCs móviles que operan pipelines de VC (Visión Continua) para capturar y analizar la luz ambiental. Un SoC típico de VC consta de dos componentes principales: un frontend para la captura de imágenes y un backend para el procesamiento de los algoritmos de visión. El frontend incluye un sensor de imágenes y un ISP (Image Signal Processor), que convierte los datos sin procesar del sensor en imágenes de alta calidad. El backend, que incluye componentes como la CPU, GPU o aceleradores, analiza las imágenes almacenadas en un buffer de la memoria principal para permitir la toma de decisiones. Las investigaciones actuales identifican la localización y la detección y seguimiento de objetos como los principales cuellos de botella en las aplicaciones emergentes de visión. Estos algoritmos enfrentan dos desafíos en sistemas móviles de VC: latencia y consumo de energía. La latencia puede causar mareos en los usuarios en un casco de RE que utilice la localización para estimar la posición de la cabeza para renderizar los frames. En los coches autónomos, la localización asegura una precisión de centímetros por lo que un error o retraso compromete gravemente la seguridad. Además, el consumo de energía limita la operación de los sistemas alimentados por baterías. Esta tesis emprende un viaje estratégico para mejorar el rendimiento y la eficiencia energética de los sistemas móviles de VC a través de varias contribuciones clave: Comenzamos analizando un motor de localización visual de vanguardia en una CPU. El motor de localización procesa imágenes de cámara, extrayendo y rastreando características para estimar la pose de la cámara. Nuestras evaluaciones revelan que la extracción de característi, Postprint (published version)

L'aprenentatge automàtic s'ha tornat molt rellevant en múltiples indústries i un de les que destaca és la ciberseguretat. El Centre d'Operacions de Seguretat Global (GSOC) de Nestlé ha estat aplicant aquest camp de la Intel·ligència Artificial per millorar el seu entorn de ciberseguretat i protegir l'empresa dels ciberatacs. Tanmateix, un dels reptes als quals s'enfronta actualment el SOC de Nestlé és que els analistes de seguretat no sempre resolen correctament les incidències que revisen. En aquest context, l'objectiu d'aquesta tesi és desenvolupar un nou producte que aprofiti un model d'aprenentatge automàtic per detectar inconsistències en les accions dels analistes de seguretat del SOC de Nestlé. Aquest document exposa el desenvolupament de la primera versió d'aquest producte, explicant detalladament com hem construït, entrenat i analitzat múltiples models mitjançant diverses tècniques de processament de text. Al final, durem a terme una anàlisi exhaustiva dels resultats obtinguts, comparant els models desenvolupats i explicant-ne els diferents resultats., Machine Learning has become highly relevant across multiple industries and one that stands out is cybersecurity. Nestlé's Global Security Operation Center (GSOC) has been applying this field of Artificial Intelligence to enhance their cybersecurity environment and protect the company from cyberattacks. However, one of challenges that Nestlé's SOC is currently facing is that the security analysts do not always correctly resolve the incidents they review. In this context, the objective of this thesis is to develop a new product that leverages a machine learning model to detect inconsistencies in the actions of Nestlé's SOC security analysts. This document follows the development of the first version of this product, explaining in detail how we built, trained and analyzed multiple models using various text processing techniques. In the end, we will conduct a thorough analysis of the results obtained, comparing the developed models and explaining their different outcomes.

Després d'introduir Yocto i SWUpdate, aquest projecte mostra com es construeix un sistema operatiu Linux encastat complet per a un dispositiu amb pantalla. Primerament, es dissenya i implementa l'entorn de desenvolupament des de la perspectiva del desenvolupador, i, en segon lloc, es tria i es prova un procés robust per a les actualitzacions de programari. Finalment, es desenvolupa una aplicació gràfica com a demostració del procés d'actualització des de la perspectiva de l'usuari final., After introducing Yocto and SWUpdate, this project shows how a complete Embedded Linux Operating system is built for a screen-enabled device. Firstly, the development environment and build system is designed and implemented from a developer's perspective, and secondly, a robust process for software updates is chosen and tested. Finally, a graphical application is developed as a demonstration of the update process from an end user's perspective.

We address the increasing complexity of scientific workflows in the context of high-performance computing (HPC) and their associated need for robust, adaptable, and flexible computational support systems. We explore five key trends as well as future challenges and opportunities for scientific workflows and HPC technologies., Peer Reviewed, Postprint (author's final draft)

La necesidad de realizar testing a las APIs (interfaz de programación de aplicaciones) desarrolladas es crucial para garantizar su correcto funcionamiento y seguridad. Sin embargo, hacerlo de forma manual es un proceso tedioso y propenso a errores. En este proyecto, se analiza y desarrolla un software que permite realizar pruebas a estas APIs a través de peticiones HTTP. Estas pruebas se basan en comprobar los resultados recibidos con los esperados. Combinado con GitHub, se automatizan estas pruebas, de forma que al hacer un push en el repositorio, se envía un aviso al sistema e inicia el proceso de testing. Además, se utilizan tres herramientas distintas para realizar las pruebas, permitiendo así una comparación exhaustiva de estas herramientas. También se ha desarrollado una API específica que servirá como ejemplo para poder realizar pruebas y verificar su correcto funcionamiento., The necessity of performing testing on developed APIs (application programming interfaces) is crucial to ensure their correct functionality and security. However, doing this manually is a tedious and error-prone process. In this project, software is analyzed and developed to enable testing these APIs through HTTP requests. These tests are based on comparing the received results with the expected ones. Combined with GitHub, these tests are automated, so when a push is made to the repository, a notification is sent to the system, and the testing process is initiated. Additionally, three different tools are used to perform the tests, allowing for a comprehensive comparison of these tools. A specific API has also been developed that will serve as an example to conduct tests and verify its correct operation.

Les xarxes neuronals de grafs (GNNs) excel·leixen en tasques com la classificació de nodes, la predicció d'arestes i la classificació de grafs, amb aplicacions en sistemes de recomanació, anàlisi de xarxes i bioinformàtica. La necessitat d'una computació eficient ha motivat l'augment dels acceleradors de hardware per a GNNs. Com es comporten diferents configuracions d'acceleradors en grafs amb diferents propietats és un tema inexplorat, limitant l'adaptabilitat dels acceleradors de GNN a diferents grafs d'entrada. Aquesta tesi introdueix una metodologia basada en dades per predir la latència i millorar les decisions de configuració adaptativa tant per a acceleradors amb un flux de dades flexibles com fixe. Les contribucions inclouen la creació d'un conjunt de dades, el disseny de models de predicció de latència, el desenvolupament d'un marc per determinar la importància de les propietats dels grafs i el desenvolupament de mètodes de planificació per a entorns d'un sol accelerador i multi-acceleradors heterogenis amb tècniques de programació lineal entera (ILP). Les avaluacions experimentals mostren que la precisa predicció de la latència porta a millores en el rendiment tant en entorns d'un sol accelerador com en entorns multi-acceleradors., Graph Neural Networks (GNNs) excel in tasks like node classification, edge predic- tion, and graph classification, with applications in recommendation systems, network ana- lysis, and bioinformatics. The need for efficient computation motivated the rise of GNN hardware accelerators. How different accelerator configurations perform on graphs with different properties remains unexplored, limiting the adaptability of GNN accelerators to different graph inputs. This thesis introduces a data-driven approach to predict latency and improve adaptive configuration decisions for both flexible-dataflow and fixed-dataflow GNN accelerators. Contributions include creating a comprehensive dataset, designing latency prediction models, developing a framework for feature importance and develop- ing scheduling methods for single-accelerator and heterogeneous multi-accelerator settings with Integer Linear Programming (ILP) techniques. Experimental evaluations show that accurate latency prediction leads to performance improvements in both single and multi- accelerator settings.

The global priority for sustainable societies drives the transition to green energy, with geothermal power as a promising alternative. Latin-American countries beneft from the active volcanism along the Pacifc Rim, which fuels their signifcant geo- thermal potential. Geothermal electricity production in the region is steadily growing and currently represents approximately 11% of global output (16 GW). This paper pro- vides details on the installed capacity of electrical generation in the most geothermally signifcant Latin-American countries, as well as the estimated potential production from existing prospects in the region. We also discuss the multiple challenges that limit the widespread development and exploitation of this valuable resource in LatinAmerica. As México stands as the top electricity producer in the region and ranks sixth worldwide, we ofer an overview of its geothermal potential, the use of electromag- netic imaging technologies to enhance Mexican geothermal resource exploration, and the challenges and limitations associated with traditional exploration techniques. Additionally, we present recent case studies on the combined use of these technolo- gies in México, highlighting best practices and lessons learned. The paper identifes open questions and outlines future research directions, particularly in México, to unlock the geothermal potential of the entire region., The work of O.C-R. has received funding from the Ministerio de Educación y Ciencia (Spain) under Project TED2021131882B-C42. This work was supported by Project PAPIIT, IN107223 - Implementación del análisis de Play Fairway para la identifcación de zonas geotérmicas en las provincias de vulcanismo intraplaca. This research is partially funded by the European Union S+T+ARTS programme from call CNECT/2022/3482066 – Art and the digital: Unleashing creativity for European industry, regions, and society under grant agreement LC-01984767., Peer Reviewed, Postprint (published version)

For decades safety-related functionalities in automotive, avionics and space systems were implemented in low-performance microcontrollers. But the increasing software complexity and computational demanding processes are guiding the safety-critical domain towards using GPUs. This work aims at providing simulator tools for the safety-critical domain approach to the usage of GPUs for high computation demanding task. It will be done by modeling a GPU Full System simulator and validating its architecture against a real GPU model.

Les aplicacions cada cop s'estan tornant més complexes, deixant enrere els típics models d'aplicacions monolítiques desplegades en servidors, per passar-se a models basats en microserveis desplegats en contenidors. Això provoca que la manera de monitorar les aplicacions i serveis es torni inevitablement més complexa. Per aquest motiu des de UPCNet s'ha decidit començar un projecte per implementar un sistema d'APM (Application Performance Monitoring) que ajudi als equips de desenvolupadors a saber en tot moment l'estat de les aplicacions. En aquest treball es podrà veure tot el procés que s'ha portat a terme per tal d'implementar un sistema APM. El producte principal d'aquest sistema és OpenTelemetry, una solució de codi obert, i que tot i que no porta gaire temps en el mercat, està evolucionant ràpidament i guanyant molta popularitat entre desenvolupadors i empreses. S'ha decidit desplegar el sistema d'APM en un clúster de Kubernetes, ja que permet fer desplegaments ràpids i escalables., Applications are becoming increasingly complex, moving away from typical monolithic application models deployed on servers to models based on microservices deployed in containers. This means that the way to monitor applications and services inevitably becomes more complex. For this reason, UPCNet has decided to start a project to implement an APM (Application Performance Monitoring) system that helps developers to know the status of applications. In this report, you will be able to see the whole process that has been carried out to implement an APM system. The main product of this system is OpenTelemetry, an open source solution, which despite not being on the market for a long time, is rapidly evolving and gaining popularity among developers and companies. It has been decided to deploy the APM system on a Kubernetes cluster, as it allows fast and scalable deployments.

Aquest treball de fi de grau titulat "Desenvolupament d'un sistema NLP a Cloud i Edge amb distribució de carregues dinàmica", realitzat a la Facultat d'Informàtica de la Universitat Politècnica de Catalunya (UPC), té com a objectiu desenvolupar un sistema de Processament del Llenguatge Natural (NLP) en entorns Cloud i Edge, utilitzant Kubernetes i KubeEdge per a una distribució de càrregues dinàmica. En definitiva resulta en una distribució de càrregues més eficient, temps de resposta reduïts i una millor utilització dels recursos. Aquesta solució innovadora aporta beneficis en rendiment i eficàcia, establint una base per a futures investigacions en el camp., This final degree project, titled "Development of an NLP System in Cloud and Edge with Dynamic Load Distribution," carried out at Facultat d'Informàtica of Universitat Politècnica de Ctalunya (UPC), aims to develop a Natural Language Processing (NLP) system in Cloud and Edge environments, using Kubernetes and KubeEdge for dynamic load distribution. The iterative methodology allowed continuous improvement of the system. The results show more efficient load distribution, reduced response times, and better resource utilization. This innovative solution provides benefits in performance and efficiency, establishing a foundation for future research in the field.

Aquest projecte de final de grau s'emmarca dintre la plataforma CloudStream, desenvolupada per inLab FIB, un laboratori de recerca i innovació de la Facultat d'Informàtica de Barcelona a la UPC, i CloudStream, una proveïdor cloud punter en l'allotjament de serveis que opera en l'àrea metropolitana de Barcelona. L'objectiu principal és desenvolupar una solució d'observabilitat que aporti coneixement, com la salut del sistema o el rendiment de les aplicacions, sobre una plataforma innovadora prèviament desenvolupada i basada en Kubernetes sota un model PaaS (Platform-as-a-Service). Gràcies a aquest sistema, tant clients com administradors podran assegurar fiabilitat així com optimitzar el rendiment de les aplicacions en un entorn cloud dinàmic responent al monitoratge en temps real., This final degree project is part of the CloudStream platform developed by inLab FIB, a research and innovation company of Barcelona School of Informatics at UPC and CloudStream, a leading hosting and cloud provider in the Greater Barcelona Metropolitan Area. The primary goal is to develop an observability solution that provides and integrates comprehensive insights, such as system health or application performance, into the existing cutting-edge Kubernetes-based PaaS (Platform-as-a-Service) previously developed. With this system, both clients and administrators will be able to ensure reliability and optimize the performance of its applications in a dynamic cloud environment responding to real-time monitoring.

En este Trabajo de Fin de Grado, realizado en colaboración con el Barcelona Supercomputing Center y la Universitat Politècnica de Catalunya, abordé el desafío de instrumentar el entorno de programación HIP de AMD para generar trazas de ejecución en aplicaciones que utilizan GPUs de AMD. Este enfoque es importante debido al aumento en la adopción de HIP como alternativa a CUDA de NVIDIA, especialmente en sistemas que solo incorporan hardware de AMD. El objetivo principal del proyecto ha sido ampliar la funcionalidad de la herramienta Extrae, que se utiliza para la instrumentación de aplicaciones y la recopilación de datos de ejecución, para incluir soporte de instrumentación al runtime HIP. Esta ampliación implicó además tener en cuenta la integración con la herramienta Paraver, permitiendo la visualización y análisis de las trazas generadas. Se hizo especial énfasis en soportar múltiples runtimes y entornos de ejecución, específicamente HIP, OpenMP y OpenMPI, y su interoperabilidad; lo cual es esencial para el análisis de rendimiento en entornos de computación de alto rendimiento a nivel de acelerador, nodo y clúster. Los objetivos secundarios incluyeron la evaluación de herramientas existentes de análisis de rendimiento, el desarrollo de un profiler personalizado para HIP, y la evaluación de diversos métodos de interposición para capturar eventos y operaciones específicas en HIP. Para gestionar eficazmente el proyecto, adopté una metodología Agile organizada en sprints que permitieron una planificación flexible y adaptativa. Utilicé herramientas de gestión de proyectos como Trello y técnicas de planificación como diagramas de Gantt para supervisar el progreso y asegurar una ejecución sistemática. Tambien se incluye en el alcance la relacición de un estudio detallado sobre la sostenibilidad económica del proyecto, lo cual fue importante para entender si se ha realizado un uso eficiente de los recursos y la viabilidad de los costos a nivel empresarial. El proyecto ha, In this Final Degree Project, carried out in collaboration with the Barcelona Supercomputing Center and the Universitat Politècnica de Catalunya, I tackled the challenge of instrumenting AMD's HIP programming environment to generate execution traces in applications using AMD GPUs. This approach is significant due to the increasing adoption of HIP as an alternative to NVIDIA's CUDA, particularly in systems that solely incorporate AMD hardware. The main objective of the project was to enhance the functionality of the Extrae tool, which is used for application instrumentation and execution data collection, to include support for the HIP runtime. This extension also involved integration with the Paraver tool, enabling the visualization and analysis of the generated traces. Special emphasis was placed on supporting multiple runtimes and execution environments, specifically HIP, OpenMP, and OpenMPI, and their interoperability; this is essential for performance analysis in high-performance computing environments at the accelerator, node, and cluster levels. Secondary objectives included the evaluation of existing performance analysis tools, the development of a customized profiler for HIP, and the assessment of various interposition methods to capture specific events and operations in HIP. To manage the project effectively, I adopted an Agile methodology organized into sprints, which allowed for flexible and adaptive planning. I utilized project management tools such as Trello and planning techniques like Gantt charts to monitor progress and ensure systematic execution. The scope also included conducting a detailed study on the economic sustainability of the project, which was important to understand if resources were used efficiently and if the costs were viable at a business level. The project concluded with the validation of the developed tool, providing a comprehensive analysis of its functionality and effectiveness. Tests were conducted on systems with available graph

Recent advancements in communication technology have transformed the way the industrial system works. This digitalization has improved the way of communication between different actors involved in cyber physical production systems (CPPS), such as users, suppliers, and manufacturers, thus making the whole process transparent. The utilization of emerging new technologies in CPPS can cause vulnerable spots that can be exploited by attackers to launch sophisticated distributed denial of service (DDoS) attacks, hence threatening the availability of the production systems. Existing machine learning based intrusion detection systems (IDS) often rely on unrealistic datasets for training and validation, thus missing the crucial testing phase with real-time scenarios. The results generated by the ML models are based on predictions at each flow level and cannot provide summarized information about malicious entities. To address this limitation, this study proposed an efficient IDS system that uses both rule-based detection and ML-based approaches to detect DDoS attacks damaging the infrastructure of CPPS. For training and validation of the system, we use real-time network traffic extracted from a real industrial scenario, referred to as Farm-to-Fork (F2F) supply chain system. Both, attacks and normal traffic were captured, and bidirectional features were extracted through CIC-FLOWMETER. We make use of 8 ML supervised and unsupervised approaches to detect the malicious flows; and then a rule-based detection mechanism is used to calculate the frequency of the malicious flows and to assign different severity levels based on the computed frequency. The overall results show that supervised models outperform unsupervised approaches and achieve an accuracy 99.97% and TPR 99.96%. Overall, the weighted accuracy when tested and deployed in a real-time scenario is around 98.71%. The results prove that the system works better when considering real-time scenarios and provides comprehensive, This work was supported in part by European Union’s Horizon Europe (PHOENi2X) under Grant 101070586, in part by the Spanish Ministry of Science and Innovation funded by MCIN/AEI/10.13039/501100011033 under Grant PID2021-124463OB-I00, in part by ERDF a way of making Europe, and in part by the Catalan Government under Contract 2021 SGR 00326., Peer Reviewed, Postprint (published version)

El documento proporciona una descripción detallada de la creación de una herramienta de autorización utilizando JSON, basada en el estándar XACML (eXtensible Access Control Markup Language) v.3.0 y su perfil JSON para mejorar la eficiencia, flexibilidad y facilidad de integración en aplicaciones modernas. La investigación se centra en el desarrollo de un esquema para definir políticas de control de acceso en formato JSON, incluyendo la creación de una librería específica en TypeScript que implementa este nuevo enfoque. La solución propuesta busca superar las limitaciones de la implementación tradicional de XACML, especialmente en su complejidad y adaptabilidad a entornos modernos, a través de una alternativa más ligera y fácilmente comprensible. Este trabajo constituye un aporte significativo en el campo del control de acceso, ofreciendo una herramienta adaptada a las necesidades de seguridad informática contemporáneas, demostrando su aplicabilidad mediante una aplicación demostrativa para un contexto médico., The document provides a detailed description of creating an authorization tool using JSON, based on the XACML (eXtensible Access Control Markup Language) v.3.0 standard and its JSON profile to improve efficiency, flexibility and ease of integration in modern applications. The research focuses on the development of a scheme to define access control policies in JSON format, including the creation of a specific TypeScript library that implements this new approach. The proposed solution seeks to overcome the limitations of the traditional XACML implementation, especially in its complexity and adaptability to modern environments, through a lighter and easily understandable alternative. This work constitutes a significant contribution to the field of access control, offering a tool adapted to contemporary computer security needs, demonstrating its applicability through a demonstrative application for a medical context.

El monitoratge i anàlisis dels processos que s'executen en un sistema operatiu sempre ha sigut una tasca costosa i complicada. Tot i que hi ha aplicacions que faciliten aquest tasca, sovint tenen limitacions que impedeixen explotar-la al màxim. A més, el propi sistema conté estructures que impossibiliten o dificulten aquest monitoratge. És per aquest motiu que sorgeix aquest projecte, que té com a objectiu oferir una aplicació d'usuari capaç d'instrumentar, en temps real, les crides a sistema executades en el sistema operatiu Windows 10. Per realitzar-ho, serà necessari treballar a baix nivell per tal de recollir tota la informació necessària i evitar les estructures que entorpeixen el recolliment exhaustiu d'informació. Per fer-ho, el treball consta de tres parts: una aplicació d'usuari, un driver i el component principal que permet dur a terme tot això, un Hypervisor de Tipus 2. Aquest és una estructura situada al sistema operatiu, la qual permet gestionar la virtualització del sistema a un nivell d'abstracció més elevat. Per tant, partint d'un Hypervisor de Tipus 2, el qual permet redefinir certes estructures del sistema treballant a nivell de sistema operatiu, es modificarà el sistema per tal de poder establir aquest punt d'entrada, recollint la informació que es necessita i evitant, satisfactòriament, els mecanismes de seguretat., The monitoring and analysis of the processes running in an operating system have always been a costly and complicated task. Although there are applications that facilitate this task, they often have limitations that prevent full exploitation. Moreover, the system itself contains structures that make this monitoring difficult or impossible. This is the reason why this project arises, which aims to provide a user application capable of instrumenting, in real-time, the system calls executed in the Windows 10 operating system. To achieve this, it will be necessary to work at a low level to collect all the necessary information and avoid the structures that hinder the exhaustive collection of information. The work consists of three parts: a user application, a driver, and the main component that makes all this possible, a Type 2 Hypervisor. This is a structure located in the operating system, which allows managing the system virtualization at a higher level of abstraction. Therefore, based on a Type 2 Hypervisor structure, which allows redefining certain system structures by working at the operating system level, the system will be modified to establish this entry point, collecting the necessary information and successfully avoiding security mechanisms.

Aquest projecte és el Treball de Fi de Grau (TFG) corresponent al grau d'Enginyeria Informàtica (GEI) de la Facultat d'Informàtica de Barcelona (FIB), de la Universitat Politècnica de Catalunya (UPC). Es tracta d'un projecte de modalitat B, realitzat a l'empresa Light Eyes SL. L'objectiu és el de dissenyar i implementar un software que automatitzi la generació d'informes de diferents serveis que ofereix Light Eyes SL. Específicament, aquest projecte contempla la implementació de dos mòduls per a la creació automàtica d'informes de seguretat de l'email dels clients i informes OSINT (Open Source Intelligence) sobre un domini i/o una pàgina web., This project is the Bachelor's Thesis (TFG) corresponding to the degree in Computer Engineering (GEI) from the Faculty of Informatics of Barcelona (FIB), at the Polytechnic University of Catalonia (UPC). It is a type B project, carried out at the company Light Eyes SL. The objective is to design and implement software that automates the generation of reports for different services offered by Light Eyes SL. Specifically, this project involves the implementation of two modules for the automatic creation of security reports for clients' email and OSINT (Open Source Intelligence) reports on a domain and/or website.

El treball se centra en la Infraestructura com a Codi (IaC), explorant el seu impacte tant en equips de plataforma com de desenvolupament. Primerament, aborda el perquè de IaC, juntament amb la seva implicació i relació amb la filosofia DevOps. A continuació, per donar suport a aquesta aproximació teòrica, es realitzen dues seccions tècniques. En primer lloc, es duu a terme la creació d'un repositori de codi per desplegar automàticament infraestructura seguint la filosofia GitOps. Seguidament, s'explica com Mango implementa IaC per gestionar la seva infraestructura i apropar-la als desenvolupadors. I finalment, per acompanyar la part de Mango i propulsar aquesta filosofia, s'explotarà una eina del seu flux CI/CD per avaluar els costos d'una merge request abans d'aplicar cap canvi real., This project focuses on Infrastructure as Code (IaC), exploring its impact on both platform and development teams. First, the reason for IaC is addressed, along with its implication and relationship to the DevOps philosophy. In addition, to support this theoretical approach, two technical sections are made. First, the creation of a code repository to automatically deploy infrastructure following the GitOps philosophy is carried out. Next, we explain how Mango implements IaC to manage its infrastructure and bring it closer to developers. And finally, to complement Mango's part and promote this philosophy, a tool from its CI/CD workflow will be explored to evaluate the costs of a merge request before applying any real change.

Computer-Aided Drug Design revolutionized pharmaceutical research by leveraging computational tools to expedite the drug discovery process. It employs different techniques to identify, optimize, and predict the interactions between chemical compounds and living beings. It facilitates the exploration of vast chemical space while minimizing costly experimental iterations. By integrating computational and experimental approaches, Computer-Aided Drug Design enhances the efficiency and success rate of drug development, leading to the discovery of safer, more effective therapeutics for various diseases, from cancer to infectious illnesses, thus benefiting patients worldwide. PharmScreen is a proprietary software from the company Pharmacelera S. L. It employs a Computer-Aided Drug Design method called Ligand-Based Drug Design which uses the similarity of molecules with common physicochemical properties to identify alternatives to current drugs. This way, new better and cheaper drugs can be obtained. Its main function is the Virtual Screening tool which is used to obtain the aforementioned scoring of molecules. However, it is a very complete program with more tools that are used in the drug design process. The main objective of this thesis aims to improve the computational performance of PharmScreen. This will be done by optimizing a key task of the program which is executed at the core of the Virtual Screening process. As the program is already parallelized at all levels and there are no more available computational resources, this optimization has to be done by means of algorithmic improvements. Within the program, it has been implemented a novel algorithm called spatial hashing to reduce the number of computations that are done when the program has to obtain some fields required to get the similarity score. The idea of the method is to consider that the atoms of a molecule only affect to the space that is closer to them. Thus, the program can obtain very similar results

The upcoming sixth-generation (6G) networks are expected to be more heterogeneous, scalable, reliable, secure and energy-efficient. 6G services and applications will benefit from the fast data transmission rates, global coverage, precise positioning, and widespread intelligence capabilities provided by 6G technology. In this complex scenario, new attack surfaces and vectors will emerge, due to the distributed, disaggregated, dynamic, programmable and open nature of the entire end-to-end 6G infrastructure along with the fragmentation of data, as well as the need for supporting cross-platforms interoperability, making the application of security and trust fundamentally challenging. To securely manage services in the future 6G mobile networks along with the set of functions ruling its management, conceptual work is needed to defining functional blocks supporting a secure end-to-end 6G systems management. This is especially critical in handling specific functionalities, such as network disaggregation, risk and threats business impact, energyefficiency, self-configuration or dynamic discovery. This paper positions an innovative 6G security services management architecture, which builds on a set of innovative building blocks, leveraging key technologies, such as AI-assisted models and Digital Twin, while enabling a human-centric approach toward an end-to-end security solution. Two canonical workflows along with two illustrative application scenarios are proposed, Light Rail Transit and Extended Reality, to conceptually highlight the proposed architecture impact and its expected benefits when high bandwidth, ultra-low latency, and secure communications are required., This work was supported by the European union’s Horizon Europe, under the project HORSE, Grant 101096342. The work of Eva Rodriguez and Xavi Masip-Bruin was supported in part by the Spanish Ministry of Science and Innovation funded by MCIN/AEI/10.13039/501100011033 and ERDF: A way of making Europe under Grant PID2021-124463OB-I00, and in part by the Catalan Government under Contract 2021 SGR 00326., Peer Reviewed, Postprint (published version)

Wireless Network-on-Chip (WNoC) has been proposed as a low-latency, versatile, and broadcast-capable complement to current interconnects in the quest for satisfying the ever increasing communications needs of modern computing systems. However, to realize the promise of WNoC, multiple wireless links operating at several tens of Gb/s need to be created within a computing package. Unfortunately, the highly integrated and enclosed nature of such computing packages incurs significant Co-Channel Interference (CCI) and Inter-Symbol Interference (ISI), not only preventing the deployment of multiple spatial channels, but also severely limiting the symbol rate of each individual channel. In this work, Time Reversal (TR) is proposed as a means to compensate the channel impairments and enable multiple concurrent high-speed links at the chip scale. We offer evidence, via full-wave simulations at 140 GHz, that TR can increase the symbol rate by an order of magnitude and allow the deployment of multiple concurrent links towards achieving aggregate speeds in excess of 100 Gb/s. Finally, the challenges relative to the realization of TR at the chip scale are analyzed from the implementation, protocol support, and architectural perspectives., The authors acknowledge support from the European Research Council (ERC) under grant agreement 101042080 (WINC) as well as the European Innovation Council (EIC) under grant agreement No 101099697 (QUADRATURE)., Peer Reviewed, Postprint (author's final draft)

Aquest Treball de Fi de Grau (TFG) explora l'àmbit dels Sistemes de Conducció Autònoms, on la seguretat és essencial, i les fallades del sistema poden comportar greus conseqüències. Garantir la previsibilitat d'aquests sistemes complexos suposa un repte substancial. Les investigacions prèvies han identificat vulnerabilitats en sistemes autònoms, especialment en relació amb els atacs basats en interferències dins de sistemes de control. Si bé s'han introduït noves mètriques relacionades amb el temps per a la correcció funcional i no funcional en aplicacions d'automoció, hi ha una manca de solucions a nivell de codi i proves exhaustives d'aquests sistemes amb proteccions de temps implementades. La nostra recerca aborda aquests buits centrant-nos en la identificació i mitigació de patrons de violació temporal en un marc d'automoció representatiu desplegat sobre Cyber-RT, un middleware distribuït de qualitat industrial. Aquests frameworks de middleware permeten el disseny d'aplicacions complexes i modulars que, d'una banda, faciliten el desplegament de sistemes de control funcionalment rics basats en IA, però d'altra banda introdueixen una capa d'abstracció. Aquest últim complica la definició de mètriques de temps i la verificació de la correcció temporal. Ens centrem en el problema de la verificació de les propietats temporals en els sistemes crítics de seguretat sobre middleware distribuït considerant solucions que es poden aplicar directament a nivell de middleware i per tant poden adaptar-se a la semàntica d'aplicacions diverses. En aquest projecte, considerem tres problemes temporals específics: "Temporal Displacement", fallada persistent en el lliurament de missatges i degradació de freqüència semànticament rellevant. Per afrontar aquests reptes, hem desenvolupat i implementat mecanismes de protecció personalitzats. Aquests inclouen mecanismes de verificació de "timestamps" i de reintent per mitigar el desplaçament temporal, models predictius integrats amb salvagu, This Final Degree Project (TFG) explores the realm of Autonomous Driving Systems, where safety is essential, and system failures can lead to severe consequences. Ensuring the predictability of these complex systems poses a substantial challenge. Previous research has identified vulnerabilities in autonomous systems, particularly concerning interference-based attacks within control pipelines. While new timing-related metrics for both functional and non-functional correctness in automotive applications have been introduced, there is a lack of solutions at the code level and comprehensive testing of these systems with implemented timing protections. Our research addresses these gaps by focussing on the identification and mitigation of timing violation patterns in a representative automotive framework deployed on top of Cyber-RT, an industrial-quality distributed middleware. These middleware frameworks enable the design of complex and modular applications that on one hand makes it easier to deploy functionally-rich control systems based on AI, but on the other hand introduce an abstraction layer. The latter complicates the definition of timing metrics and the verification of timing correctness. We focus on the problem of verification of timing properties in safety critical systems on top of distributed middleware by considering solutions that can be applied directly at the middleware level and therefore can adapt to diverse application semantics. In this project, we consider three specific timing issues: temporal displacement, persistent failure in message delivery, and semantically relevant frequency degradation. To address these challenges, we developed and implemented customised protection mechanisms. These include timestamp verification and retry mechanisms to mitigate temporal displacement, predictive models integrated with safeguards to compensate for message delivery failures, and a data purging mechanism to mitigate the effects of frequency degradation. Through

This master's thesis aims to improve the efficiency of Vlasiator, the world's first simulation model that uses the Vlasov equation to model Earth's plasma environment in its global context. The research focuses on porting the application to Rhea, the high-performance / low-consumption microprocessor that SiPearl is developing, and that is being designed to be used for European extreme-scale systems. This work aims to compile Vlasiator to ARM, the base architecture behind Rhea, and identify the possible bottlenecks present in the application. The application's hotpots will be analyzed to identify potential performance improvements. The final goal is to compare the performance gains achieved using Scalable Vector Extension (SVE) and Vector Length Agnostic (VLA) implementations against traditional approaches. Future research directions may explore alternative approaches to overcome the identified limitations and enhance performance by exploiting other techniques.

Transcripciones de las 25 entrevistas realizadas para la tesis doctoral titulada “Between rigor and imagination: Creativity in structures education”., Este estudio presenta los resultados de 25 entrevistas realizadas para analizar las perspectivas de profesionales sobre la creatividad en el ámbito de las estructuras. La investigación empleó una metodología cualitativa iterativa basada en un enfoque fenomenológico hermenéutico. Se recopilaron datos a través de entrevistas semiestructuradas con expertos en creatividad y estructuras, con el objetivo de comprender sus experiencias y encontrar puntos en común para promover la creatividad en la educación de la ingeniería estructural. Las entrevistas, que duraron entre 30 y 60 minutos, se llevaron a cabo entre febrero y octubre de 2022, con 13 realizadas en línea y 12 de manera presencial. Se grabaron y transcribieron manualmente en Microsoft Word, con 10 entrevistas en inglés y 15 en español. Los participantes, codificados según su orden y especialización (arquitectos, ingenieros y profesionales creativos), firmaron formularios de consentimiento para la publicación de sus datos. La investigación concluye con recomendaciones para fomentar la creatividad en el campo de la ingeniería estructural., Preprint

Aquest treball ha explorat la possibilitat d'incloure en un cicle de desenvolupament d'un producte informàtic, un equip de seguretat encarregat de guiar i comprovar que el producte final té uns estàndards dignes de l'era i context actual, on tot producte informàtic pot esperar ser atacat per usuaris tant curiosos com malintencionats. Per a fer aquesta comprovació, s'ha simulat el procés de desenvolupament de principi a fi d'un pàgina web, amb un equip de seguretat incorporat des de l'inici, que ha fet les aportacions que ha considerat necessàries, seguint la metodología empreada. Gràcies a això, s'ha pogut obtenir contribucions que ha fet aquest equip, i valorar si els canvis i millores de seguretat al producte final paguen la pena de mantenir aquest equip., This project has explored the possibility of including in a development cycle of a it product, a security team in charge of guiding and verifying that the final product has standards worthy of the current era and context, where any computer product can expect to be attacked by users both curious and malicious. To carry out this check, the development process from beginning to end of a website has been simulated, with a security team incorporated from the beginning, which has made the contributions it has considered necessary, following the methodology used. Thanks to this, it has been possible to obtain contributions made by this team, and assess whether changes and safety improvements to the final product are worth maintaining this team.

Aquest treball proposa la creació d'una plataforma automatitzada per a pentesting amb l'objectiu de simplificar i millorar els processos de ciberseguretat per a empreses de qualsevol mida, especialment pimes sense la capacitat de contactar a un equip especialitzat. Tot i que existeixen eines semiautomàtiques com Burpsuite, Metasploit o Nessus aquestes són sovint especialitzades en aspectes concrets o costoses, dificultant el seu ús per a petites i mitjanes empreses, que constitueixen el 99,8% del teixit empresarial català. L'objectiu principal és crear una plataforma unificada que integri les principals eines de pentesting, automatitzant les tasques per facilitar-ne l'ús i la interpretació dels resultats per part dels usuaris. Aquesta plataforma ha de ser accessible, fàcil d'utilitzar per a tècnics no tan especialitzats en el sector i econòmica per a les pimes. En definitiva, tot i que el resultat final no aporta funcionalitats o eines noves al camp de la ciberseguretat, la creació d'una plataforma automatitzada i unificada facilita enormement el procés de pentesting, fent-lo més accessible i eficient. Això representa una millora significativa en la seguretat digital per a moltes empreses, que sovint no poden permetre's les eines actuals disponibles al mercat., This paper proposes the creation of an automated platform for pentesting with the aim of simplifying and improving cybersecurity processes for companies of any size, especially SMEs without the ability to contact a specialized team. Although there are semi-automatic tools such as Burpsuite, Metasploit or Nessus, these are often specialized in specific aspects or have hight costs, making their use difficult for small and medium-sized companies, which make up 99.8% of the Catalan business fabric. The main goal is to create a unified platform that integrates the main pentesting tools, automating tasks to make it easier for users to use and interpret the results. This platform must be accessible, easy to use for non-specialized technicians in the sector and economical for SMEs. Ultimately, while the end result does not bring any new functionalities or tools to the cybersecurity field, creating an automated and unified platform greatly facilitates the pentesting process, making it more accessible and efficient. This represents a significant improvement in digital security for many businesses, which often cannot afford the current tools available on the market.

Aquest treball analitza les possibilitats que la Intel·ligència Artificial generativa ofereix en l’actualitat per a elaborar una programació didàctica d’un tema concret de l’assignatura de Tecnologia i Digitalització de 2n d’ESO. Des de l’aparició de la Intel·ligència Artificial generativa l’any 2022 i l’obertura d’aquesta al gran públic, hem observat una popularització massiva d’aquests tipus d’eines, que ofereixen un gran ventall de possibilitats pels usuaris en quant a creació i redacció de text, respostes a preguntes, generació d’imatges, audio o vídeo, entre d’altres. Concretament, però, aquest treball se centra en l’elaboració d’una Situació d’Aprenentatge mitjançant la Intel·ligència Artificial generativa d’OpenAI, coneguda com a ChatGPT, en la seva versió gratuïta 3.5, comparant el currículum actual amb l’antic, que són molt diferents, per tal d’observar les possibilitats que cadascun d’ells ofereix per a que una Intel·ligència Artificial generativa pugui elaborar una programació en forma de Situació d’Aprenentatge.

En l'era digital actual, l'accés a Internet s'ha convertit en una necessitat essencial, impulsant el desenvolupament social, econòmic i cultural de les societats modernes. La infraestructura de xarxa és clau per garantir una connectivitat constant i eficient, i la seva monitorització és fonamental per assegurar la disponibilitat i seguretat dels serveis digitals. Aquest projecte, desenvolupat com a part del Grau d'Enginyeria Informàtica a la Universitat Politècnica de Catalunya (UPC), té com a objectiu millorar el sistema de monitorització de la infraestructura de xarxa de la UPC, gestionada per UPCnet. El projecte sorgeix de la necessitat de substituir l'actual sistema de monitorització monolític basat en Observium, que no pot gestionar eficientment el volum de dades generat pels equips de xarxa. Es proposa una solució distribuïda per repartir la càrrega de treball entre diverses màquines, millorant així la velocitat, flexibilitat, escalabilitat i tolerància a fallades del sistema. L'objectiu principal és desplegar una plataforma distribuïda, és a dir, que diverses màquines virtuals col·laborin per a realitzar conjuntament les tasques de monitorització. La plataforma, basada en Observium, haurà de ser capaç de monitoritzar la xarxa de la UPC de manera eficient. Addicionalment, es preveu actualitzar el protocol SNMP a la versió 3 per augmentar la seguretat, utilitzar Ansible per automatitzar tasques repetitives de configuració d'equips de xarxa, i implementar una solució escalable que permeti una futura ampliació del sistema., In the current digital era, access to the Internet has become an essential necessity, driving the social, economic, and cultural development of modern societies. Network infrastructure is key to ensuring constant and efficient connectivity, and its monitoring is fundamental to ensuring the availability and security of digital services. This project, developed as part of the Computer Engineering Degree at the Polytechnic University of Catalonia (UPC), aims to improve the monitoring system of the UPC's network infrastructure, managed by UPCnet. The project arises from the need to replace the current monolithic monitoring system based on Observium, which cannot efficiently manage the volume of data generated by network devices. A distributed solution is proposed to distribute the workload among several machines, thus improving the system's speed, flexibility, scalability, and fault tolerance. The main objective is to deploy a distributed platform where multiple virtual machines collaborate to perform monitoring tasks together. The platform, based on Observium, will efficiently monitor the UPC network. Additionally, it is planned to upgrade the SNMP protocol to version 3 to increase security, use Ansible to automate repetitive network equipment configuration tasks, and implement a scalable solution that allows for future system expansion.

In recent years there has been a surging requirement for precise and effective indoor localization measurements due to the emergence of innovative technologies such as the Internet of Things (IoT) [1] and smart home devices. Traditional localization technologies like GPS have been proven not to be precise in indoor environments. A new approach based on WiFi RTT (Round Trip Time) called FTM protocol (Fine Time Measurement) has gained popularity over the last years due to its low-cost implementation and wide distribution especially indoors. Accuracy issues within FTM have been pointed out by studies. In an environment where the signal bounces into walls and buildings, the signal is received through different paths. As the sampling interval of the radios is below the delay between the different paths, the receiver often sees a mixture of the signals over different paths. Therefore, it is not possible to distinguish the direct path (if it exists) from the reflected signals. The receiver measures a mixture of different signals with different amplitudes, phases, and probably also different frequencies. Recent studies analyzed the increase in accuracy by choosing different burst sizes and multiple positions. As expected, increasing the number of bursts in an FTM frame leads to a reduction in the error. However, this comes at the cost of consuming more channel time. The increase in accuracy is not directly proportional to the amount of channel time required to perform it. This thesis' objective will be to develop an algorithm capable of dynamically adapting its parameters to ensure, simultaneously, channel time optimization and precision in real-case scenarios.

Este proyecto se ha llevado a cabo durante una estancia de prácticas en HP, Hewlett Packard. El problema inicial planteado fue el tiempo excesivo que tomaba adaptar scripts cruciales al desarrollo de nuevas impresoras, dado que cada una posee una estructura única. Esto resultaba en horas de desarrollo dedicadas a modificar el código fuente de estos scripts o a crear nuevas versiones para su uso. Para solucionar este problema, se propuso implementar un método que permitiera al programa detectar el tipo de impresora y adaptarse dinámicamente a sus condiciones. Se eligió Python como lenguaje de programación, ya que las impresoras de HP utilizan este lenguaje. La solución se centró en la reimplementación de los scripts, permitiéndoles recibir información de los archivos sobre la estructura de las impresoras. Además, se desarrollaron librerías que facilitarían esta tarea, y al distribuirlo en librerías, se garantizaría que los futuros scripts pudieran ser programados con la capacidad adaptativa desde el inicio. Los resultados obtenidos fueron satisfactorios, demostrando que el tiempo necesario para realizar una adaptación se ha reducido a no más de diez minutos., This project was carried out during an internship at HP, Hewlett Packard. The initial problem posed was the excessive time it took to adapt crucial scripts for the development of new printers, given that each has a unique structure. This resulted in hours of programming dedicated to modifying the source code of these scripts or creating new versions for their use. To solve this problem, I proposed to implement a method that would allow the program to detect the type of printer and dynamically adapt to its conditions. Python was chosen as the programming language since HP printers use this language. The solution focused on reimplementing the scripts, allowing them to receive information from files about the structure of the printers. Additionally, libraries were developed to facilitate this task, and by distributing it in libraries, it would be ensured that future scripts could be programmed with adaptive capability from the start. The results obtained were satisfactory, demonstrating that the time needed for an adaptation has been reduced to no more than ten minutes.

Aquest treball analitza el problema de la integració de l'anglès als cicles formatius d'FP, en concret al CFGS en Automoció perquè actualment el nivell d'aquesta llengua entre els alumnes que finalitzen els estudis no arriba al nivell mínim necessari per mantenir una conversa tècnica en anglès, segons el feedback que donen els empresaris que contracten els recent titulats els últims anys. L'objectiu final d'aquest projecte és proposar un sistema d'aprenentatge integrat als mòduls principals del cicle formatiu que aconsegueixi motivar els alumnes a assolir uns coneixements d'anglès tècnic suficients per desenvolupar la seva feina sense problemes un cop obtinguin el títol.

Electromagnetic (EM) imaging aims to produce large-scale, high-resolution soil conductivity maps that provide essential information for Earth subsurface exploration. To rigorously generate EM subsurface models, one must address both the forward problem and the inverse problem. From these subsurface resistivity maps, also referred to as volumes of resistivity distribution, it is possible to extract useful information (lithology, temperature, porosity, permeability, among others) to improve our knowledge about geo-resources on which modern society depends (e.g., energy, groundwater, and raw materials, among others). However, this ability to detect electrical resistivity contrasts also makes EM imaging techniques sensitive to metallic structures whose EM footprint often exceeds their diminutive stature compared to surrounding materials. Depending on target applications, this behavior can be advantageous or disadvantageous. In this work, we review EM modeling and inverse solutions in the presence of metallic structures, emphasizing how these structures affect EM data acquisition and interpretation. By addressing the challenges posed by metallic structures, our aim is to enhance the accuracy and reliability of subsurface EM characterization, ultimately leading to improved management of geo-resources and environmental monitoring. Here, we consider the latter through the lens of a triple helix approach: physics behind metallic structures in EM modeling and imaging, development of computational tools (conventional strategies and artificial intelligence schemes), and configurations and applications. The literature review shows that, despite recent scientific advancements, EM imaging techniques are still being developed, as are software-based data processing and interpretation tools. Such progress must address geological complexities and metallic casing measurements integrity in increasing detail setups. We hope this review will provide inspiration for researchers to study th, The work of O.C-R., P.Q., P.P., and J. L. have received funding from the MCIN/AEI /10.13039/501100011033 (Spain) and from the European Union NextGenerationEU/PRTR under the project GEOTHERPAL-EM_TED2021-131882B-C42 and -C41., Peer Reviewed, Postprint (published version)

En un mon dirigit a la digitalització, cada cop hi ha mes dades personals a la xarxa. Ja sigui a traves de xarxes socials, contractes, mútues, educació, etc., constantment cedim les nostres dades per a que siguin digitalitzades i utilitzades a la xarxa. Per sort, la ciberseguretat ha evolucionat al llarg del temps. Abans, el control sobre les dades a la xarxa era escàs. La regulació era més laxa i la ciberseguretat no rebia tanta atenció com ara. Afortunadament, en els últims anys, hem vist canvis significatius. Recentment es va aprovar la llei GDPR, i moltes grans empreses van haver de pagar grans sancions per no complir amb els mínims establerts i no garantir seguretat a les dades personals emmagatzemades. Es per això, que des de Generali, s'ha decidit dedicar una part del departament de seguretat a intentar rebaixar la vulnerabilitat més gran de la ciberseguretat: l'humà. En concret, en la navegació web i la fuga de dades personals., In a world driven by digitization, more and more personal data is being stored online. Whether through social networks, contracts, insurance, education, etc., we are constantly giving our data to be stored and used online. Fortunately, cybersecurity has evolved over time. In the past, there was little control over data on the network. Regulation was more lax and cybersecurity did not receive as much attention as it does now. In recent years we have seen significant changes. Recently the GDPR law was passed, and many large companies had to pay large penalties for not complying with the established minimums and not guaranteeing the security of stored personal data. That is why Generali has decided to dedicate part of its security department to trying to reduce the greatest vulnerability of cybersecurity: the human factor. More specifically, in web browsing and data leakage.

In today's rapidly evolving landscape of digital systems, where complexity is ever-increasing, the verification of these systems stands as a cornerstone of paramount importance. Addressing this critical need, this master thesis introduces a Database-driven FPGA Workflow for digital system verification, that seamlessly integrates database management with FPGA programming, transcending vendor-specific constraints to enhance the precision and efficiency of the verification process. The thesis outlines the development of a workflow, tailored to interface seamlessly with a vendor-independent database repository. This repository houses test verification results, forming a platform for verification. This strategy is architected to streamline verification procedures, mitigate potential errors, and establish a methodical framework for managing verification data across diverse FPGA platforms. The methodology, comprises a sequence of steps executed by a Rust program. Beginning with the extraction of RTL code from a Git repository, the Rust program automates synthesis, implementation, and generation of bitstream using tools such as AMD Xilinx Vivado or equivalents from other FPGA vendors. It then verifies the output against a reference file and stores the verification data in a database.

In today's hyper-connected world, almost everything we do involves the internet, and dating is no exception. The online dating industry, valued at billions of dollars, is pivotal in shaping how people form connections in the 21st century. With 381 million adults engaging online dating platforms worldwide in 2023 and a growing acceptance of this modern way of meeting partners, the quality of matchmaking algorithms can profoundly impact individuals' ability to find compatible partners. This project, hosted at IThinkUPC's Analytics & AI Department, aims to advance the matchmaking algorithm used by Cites a Cegues, an online platform that merges the traditional matrimonial agency approach with modern dating app features. The project focuses on optimizing the existing algorithm, initially created by IThinkUPC, to improve the accuracy of match suggestions. Key improvements include incorporating user preferences, especially physical preferences, from various data sources such as text, videos, and images, and leveraging Generative AI technologies, resulting in a more personalized and effective matchmaking process. This report provides a comprehensive overview of the project, including the background, the methodology employed, the technical improvements implemented, and the evaluation of the updated algorithm's performance.

A l'hora de desplegar una xarxa IoT hi ha dos factors que s'han de tenir en compte si es vol optimitzar la xarxa. El primer és la distribució dels dispositius. Si aquests estan massa separats o hi ha massa obstacles entre aquests, els missatges no arribaran correctament al dest¿¿. El segon són els paràmetres utilitzats en la comunicació. Si per exemple, la potència que s'ha escollit per transmetre les dades no és prou gran, per molt ben col·locats que estiguin els dispositius, no hi haurà un bon senyal entre els dispositius. El que procura aquest estudi és analitzar el comportament de dues tecnologies IoT: LoRa i una freqüència sense fils de 443MHz. D'aquesta manera, podem indicar quins factors s'han de tenir en compte si es vol obtenir un bon rendiment en la xarxa., If you want to lay out an IoT network, two factors that must be taken into account. The first one is how the devices are allocated. If these devices are too far away or there is a lot of obstruction, the messages will not arrive correctly at the destination. The second factor is the parameters used in the transmissions. If, for example, the used power is not strong enough, the signal won't be good enough to be used. This project focuses on studying two IoT technologies: LoRa and a wireless radio frequency of 443MHz. This way, we can select the best positions and parameters for obtaining the best efficiency.

This thesis examines the integration of 5G technology with edge computing to address the challenge of reducing latency in telecommunications, towards enhancing performance for real-time applications such as autonomous driving, telemedicine, online gaming, and smart city infrastructures. Central to this research is the concept that hosting applications, processing and routing data at the network's edge, closer to its source, can significantly decrease transmission delays, offering critical improvements in safety, decision-making speed, and interactive experiences. By deploying a practical 5G testbed with edge computing capabilities, the study not only demonstrates considerable latency reductions but also validates the feasibility of hosting and managing application data directly at the edge, thereby presenting a more efficient, cost-effective, and secure alternative to conventional cloud processing. This work underscores the transformative potential of combining 5G and edge computing in creating responsive, high-performance telecommunications systems.

Genome sequence analysis is fundamental to medical breakthroughs such as developing vaccines, enabling genome editing, and facilitating personalized medicine. The exponentially expanding sequencing datasets and complexity of sequencing algorithms necessitate performance enhancements. While the performance of software solutions is constrained by their underlying hardware platforms, the utility of fixed-function accelerators is restricted to only certain sequencing algorithms.This paper presents QUETZAL, the first general-purpose vector acceleration framework designed for high efficiency and broad applicability across a diverse set of genomics algorithms. While a commercial CPU’s vector datapath is a promising candidate to exploit the data-level parallelism in genomics algorithms, our analysis finds that its performance is often limited due to long-latency scatter/gather memory instructions. QUETZAL introduces a hardware-software co-design comprising an accelerator microarchitecture closely integrated with the CPU’s vector datapath, alongside novel vector instructions to fully capitalize on the proposed hardware. QUETZAL integrates a set of scratchpad-style buffers meticulously designed to minimize latency associated with scatter/gather instructions during the retrieval of input genome sequences data. QUETZAL supports both short and long reads, and different types of sequencing data formats. A combination of hardware and software techniques enables QUETZAL to reduce the latency of memory instructions, perform complex computation using a single instruction, and transform data representations at runtime, resulting in overall efficiency gain. QUETZAL significantly accelerates a vectorized CPU baseline on modern genome sequence analysis algorithms by 5.7×, while incurring a small area overhead of 1.4% post place-and-route at the 7nm technology node compared to an HPC ARM CPU., Nishil Talati received partial support from the Excelencia Severo Ochoa mobility program CEX2021- 001148-S / funded by MCIN/AEI /10.13039/501100011033. This work has been partially supported by the Spanish Ministry of Science and Innovation PID2019-107255GB-C21 funded by MICIU/AEI /10.13039/501100011033. We acknowledge the generous gifts provided by our industrial partners, including IBM, Google, Huawei, Intel, Microsoft, and VMware. This research was partially supported by the EU Horizon project BioPIM (grant agreement 101047160), the AI Chip Center for Emerging Smart Systems Limited (ACCESS), the Swiss National Science Foundation (SNSF), Semiconductor Research Corporation (SRC), and the ETH Future Computing Laboratory (EFCL)., Peer Reviewed, Postprint (author's final draft)

Els sistemes moderns de seguretat crítica exigeixen un alt rendiment per a suportar funcionalitats avançades, superant les capacitats de les arquitectures tradicionals. Les unitats de processament gràfic de propòsit general (GPGPU) integrades ofereixen el rendiment necessari, però plantegen reptes a causa de la seva naturalesa massivament paral·lela. Aquestes dificultats augmenten els costos de verificació i fan necessari realitzar nombroses proves humanes per a cobrir tot el codi. Aquest projecte pretén resoldre aquests problemes substituint l'actual backend de Brook Auto per un nou backend capaç d'executar codi GPGPU en Vulkan Safety Critical. L'objectiu és millorar la programabilitat de Vulkan permetent als desenvolupadors programar en Brook i executar en Vulkan, eliminant així la necessitat d'aprendre Vulkan en si. Aquest enfocament aprofita les capacitats d'alt rendiment de les GPGPU al mateix temps que redueix la corba d'aprenentatge i els costos de verificació associats a les arquitectures paral·leles complexes. En integrar Brook Auto amb Vulkan SC, preveiem un procés de desenvolupament més àgil per als sistemes de seguretat crítica, oferint una solució més accessible i rendible per a les aplicacions d'alt rendiment. El nostre treball podria aplanar el camí per a una adopció més àmplia de la tecnologia GPGPU en entorns de seguretat crítica, contribuint a l'avanç dels sistemes moderns., Modern safety-critical systems demand high performance to support advanced functionalities, exceeding the capabilities of traditional architectures. Embedded General Purpose Graphics Processing Units (GPGPUs) offer the performance required but present challenges due to their massively parallel nature. These challenges increase verification costs and necessitate extensive human-driven testing for code coverage. This project aims to address these issues by replacing the current backend of Brook Auto with a new backend capable of running GPGPU code on Vulkan Safety Critical. The objective is to improve the programmability of Vulkan by enabling developers to program in Brook and execute on Vulkan, thus eliminating the need to learn Vulkan itself. This approach leverages the high-performance capabilities of GPGPUs while reducing the learning curve and verification costs associated with complex parallel architectures. By integrating Brook Auto with Vulkan SC, we anticipate a streamlined development process for safety-critical systems, offering a more accessible and cost-effective solution for high-performance applications. Our work could pave the way for broader adoption of GPGPU technology in safety-critical environments, contributing to the advancement of modern systems.

Los avances tecnológicos actuales han provocado un aumento significativo en la recopilación e intercambio de datos, sobre todo en el ámbito sanitario. Sin embargo, debido a la sensibilidad y trascendencia de estos datos, tratarlos con la máxima privacidad y seguridad se ha convertido en una necesidad crucial, pues la gestión de datos médicos implica un manejo de información sensible que requiere de protección integral. Así pues, este proyecto se centra en el desarrollo de un módulo para la gestión de consentimientos dinámicos de datos médicos, utilizando una combinación de tecnologías para garantizar la seguridad, flexibilidad y cumplimiento normativo en el manejo de información médica sensible. El sistema utiliza una base de datos para almacenar y gestionar todos los datos de profesionales de la salud, pacientes, datos médicos y consentimientos. A través de una API RESTful se facilita la interacción entre el servidor y los clientes que se desarrollen para este servicio. En el caso de este proyecto, se ha desarrollado un cliente funcional para comprobar que todas las funcionalidades prestadas por la API funcionan de la manera esperada. Por otro lado, se han utilizado diferentes estándares y librerías para desarrollar el módulo. Entre ellos se encuentran HL7 FHIR, que se ha utilizado como formato para definir los consentimientos. FHIR facilita el intercambio de datos de salud entre sistemas, y su modelo basado en recursos se adapta bien a los requisitos específicos de los consentimientos médicos. También se ha utilizado la librería HAPI FHIR, de código abierto, la cual implementa este estándar en Java para su uso en cualquier proyecto. Por último, cabe destacar que este sistema permite a los pacientes controlar quién puede acceder a su información médica y bajo que condiciones, ya que registra cada acceso con el fin de proporcionar la máxima transparencia posible. Esto proporciona a los pacientes un control directo sobre sus datos de salud y como se manejan. En defin, The current technological advances have led to a significant increase in the gathering and exchange of data, particularly in the healthcare field. However, due to the sensitivity and transcendence of this data, dealing with it with the strictest privacy and security has become a crucial necessity, since the management of medical data involves the management of sensitive information that requires end-to-end protection. Therefore, this project focuses on the development of a module for the management of dynamic medical data consents by using a combination of technologies to ensure security, flexibility and legal compliance in the management of sensitive medical information. The system uses a database to store and manage all healthcare professional, patient, medical and consent data. The interaction between the server and the clients developed for this service is facilitated through a RESTful API. In the case of this project, a functional client has been developed to check that all the functionalities provided by the API work as expected. In addition, different standards and libraries have been used to develop the module. Among them, HL7 FHIR must be highlighted, as it has been used as a format for defining consents. FHIR facilitates the exchange of health data between systems, as its resource- based model is well adapted to the specific requirements of medical consents. Moreover, the open source HAPI FHIR library, which implements this standard in Java for its use in any project, has also been used. Finally, it should be noted that this system allows patients to control who has access to their medical information and under what conditions, by registering each access with the aim of providing the maximum possible transparency. This provides patients with direct control over their health data and how it is handled. Ultimately, this project offers a solution for dynamic consent management in healthcare by integrating modern technologies to improve trust and efficiency in

La importància de la seguretat en les aplicacions incrementa constantment a mesura que anualment es troben noves vulnerabilitats. Es poden fer servir diferents mètodes per a mantenir una aplicació segura, entre els quals destaca fer tests i proposar millores al respecte. La tècnica de testeig usada serà Dynamic Application Security Testing (també coneguda com a DAST), un tipus de test basat en escanejar l'aplicació en busca de vulnerabilitats simulant que aquesta està en execució. L'objectiu d'aquesta tesi és determinar fins a quin punt es poden automatitzar els procediments de DAST per una l'aplicació web de l'empresa mitjançant l'execució d'un test DAST que desplegui l'aplicació, executi els tests i que proporcioni un informe de l'escaneig que se n'ha fet. Entre els temes investigats, els quals són necessaris per a l'atomatització, s'inclou una investigació de com millorar els tests que s'executaran. Aquesta ajudarà a identificar quins procediments de DAST es poden automatitzar i quines eines és possible integrar en el projecte., The importance of security in applications is constantly increasing as new types of vulnerabilities are found each year. In order to maintain a secure application different methods may be used, in which active testing and establishing action points are highlighted. The testing technique that will be used for this thesis is Dynamic Application Security Testing (also known as DAST), a type of testing procedure based on scanning for vulnerabilities simulating the running application. The aim of this thesis is to determine how much DAST procedures can be automated for one of our corporation's web application. It will be achieved by running a DAST test that deploys the application, runs the tests and finally provides a report of the scanning. Another topic of investigation, which is a previous step to the automatization, will involve a research aimed at improving the testing done. This previous investigation helps in the identification of which procedures of DAST are likely to be automated and what tools are feasible to integrate into the project.

Este trabajo explora la implementación y evaluación de RINA (Recursive InterNetwork Architecture) en un escenario experimental de red, para superar las limitaciones de la arquitectura TCP/IP tradicional. Se centra en la necesidad de adaptar las infraestructuras de red a las exigencias actuales de eficiencia, seguridad y gestión de la calidad de servicio (QoS). En este contexto, el proyecto aplicó la teoría demostrada en ERASER (Experimenting with Real Application-specific QoS Guarantees in a Large-scale RINA Demonstrator) en un entorno controlado para evaluar su funcionamiento en condiciones de tráfico realistas. A través de la implementación de políticas de QoS avanzadas, como el Quantitative Timeliness Agreement Multiplexor (QTA-Mux). El trabajo resalta la viabilidad técnica de RINA para redefinir las prácticas de comunicación en redes. Sin embargo, se identificaron desafíos y dificultades técnicas que requieren futuras investigaciones para optimizar la implementación de RINA en entornos de red más grandes y complejos. Este estudio sugiere que futuras investigaciones deberían centrarse en mejorar las herramientas de configuración y gestión de RINA, actualiza la documentación soporte y desarrollar métodos más robustos para la depuración de errores., This paper explores the implementation and evaluation of RINA (Recursive InterNetwork Architecture) in an experimental network scenario, to overcome the inherent limitations of the traditional TCP/IP architecture. It focuses on the need to adapt network infrastructures to today's demands for efficiency, security and Quality of Service (QoS) management. In this context, the project applied the theory demonstrated in ERASER (Experimenting with Real Application-specific QoS Guarantees in a Large-scale RINA Demonstrator) in a controlled environment to evaluate its performance under realistic traffic conditions. Through the implementation of advanced QoS policies, such as the Quantitative Timeliness Agreement Multiplexor (QTA-Mux). The work highlights the technical feasibility of RINA to redefine communication practices in networks. However, technical challenges and difficulties were identified that require future research to optimise the implementation of RINA in larger and more complex network environments. This study suggests that future research should focus on improving RINA configuration and management tools, updating supporting documentation, and developing more robust methods for debugging errors.

En el mundo digital contemporáneo, donde el contenido visual inunda las plataformas digitales y juega un papel esencial en la interacción social, la seguridad de las imágenes se convierte en un desafío y una necesidad primordial. Ante esta realidad, se emprende una exploración de sistemas capaces de manejar adecuadamente la seguridad y privacidad de las imágenes y sus metadatos. El proyecto plantea un enfoque inicial de estudio de arquitectura de software, comparando características clave según el caso de estudio para identificar la solución más adecuada. Esta fase de comparación se centró en factores como la escalabilidad, eficiencia, costos de desarrollo, seguridad y rendimiento, fundamentales para la elección de una arquitectura óptima que garantice la seguridad y agilidad en una plataforma digital. Posteriormente, se realizó una comparación de tecnologías, proveedores y formatos, identificando estrategias que mejoren no solo la gestión de las imágenes y sus metadatos, sino también su protección. En esta etapa el proyecto se enfocó en el estándar JPEG Universal Metadata Box Format (JUMBF) y el desarrollo de estrategias de autorización para asegurar el acceso y la transferencia segura de imágenes Joint Photographic Experts Group (JPEG) y sus metadatos. Este análisis condujo a la adopción de una arquitectura de microservicios complementada con principios serverless (sin servidor) como la solución más prometedora para cumplir con los criterios de seguridad, escalabilidad y eficiencia operativa. Finalmente, se desarrolló y desplegó una solución funcional en Amazon Web Services (AWS) que ilustra la aplicación práctica de los hallazgos de la investigación. A través de un enfoque práctico, este proyecto no solo ofrece una visión completa sobre cómo implementar de manera efectiva sistemas serverless de microservicios para el manejo de contenido visual en plataformas digitales, sino que también establece un marco de referencia valioso para futuras investigaciones y desarr