Doctoral School

Reading date: 21/05/2024

• BALLESTA GARCIA, MARIA: Propagation of polarized light through turbid media: Application of lidar technology in foggy environments
  
  Author: BALLESTA GARCIA, MARIA
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN OPTICAL ENGINEERING
  Department: Department of Optics and Optometry (OO)
  Mode: Normal
  Deposit date: 22/04/2024
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: ROYO ROYO, SANTIAGO
  Committee:
       PRESIDENT: ARTEAGA BARRIEL, ORIOL
       SECRETARI: VILASECA RICART, MERITXELL
       VOCAL: BIJELIC, MARIO
  Thesis abstract: In recent times, there has been a growing interest in LiDAR imaging systems for outdoor applications involving computer vision, such as automotive systems, surveillance, and robotics. LiDAR sensors have the ability to capture 3D data, that is, the geometry (volume, distances) of the scenes involved, complementing the 2D projections of scenes available in conventional cameras. However, their limited tolerance to adverse weather conditions, particularly fog, stays as one of the obstacles that hinders their complete settlement. This Thesis aims to evaluate the potential of utilizing the polarization properties of light and the digitization of the signal to improve the system¿s imaging capabilities in such challenging conditions. Additionally, our research offers valuable insights in the domain of imaging through fog. Understanding the interaction of polarized light with turbid media and recognizing the importance of the targets¿ polarimetric properties within the imaged scene is essential for optimizing the performance of polarimetric imaging systems.To achieve our goal, a preliminary investigation to examine the characteristics of polarimetric imaging through fog is undertaken. Our findings indicate that polarimetric imaging modes provide higher contrast compared to intensity-based imaging modes, facilitating the identification and segmentation of different targets. Additionally, experimental characterization of the depolarizing behavior of light through fog is conducted for both reflection and transmission imaging modes. The results suggest that, in this scenario, light behavior falls within the scattering regime of the polarization memory effect, with a significantly reduced depolarization in circularly polarized beams when compared to linearly polarized ones. To the best of our knowledge, this Thesis quantifies for the first time the differences between the performance of both polarization modes in fog conditions. Next, a Monte Carlo-based model is developed to meet the requirements of our LiDAR prototype. Considering the resource-intensive nature of experiments conducted in fog conditions and the dynamic nature of fog, the model¿s ability to accurately simulate the physics of the problem, including a realistic fog environment, helps to guide the definition of the future experimental actions. Subsequently, the model is utilized to simulate and analyze various aspects relevant to the design of the system, including polarization configurations, interactions with targets, and irregularities in the media (in practice, generalizing the scattering media beyond fog to e.g. sand or smog), together with the characteristics of the acquired signal. Finally, this Thesis presents a novel polarized LiDAR imager prototype and evaluates its performance in fog conditions. It conclusively shows that using circularly polarized light and a cross-configuration detection setup significantly improves system performance in such scenarios. This system effectively tackles challenges induced by scattered light, reducing saturation effects from backscattering, mitigating scattering noise in point clouds, and enhancing target detection, especially for highly reflective surfaces like metallic targets. This approach offers an innovative, straightforward, and efficient method for signal stabilization and enhancement of the point cloud quality by relying on the inherent physics of the problem.
  

• CEPEDA PACHECO, JUAN CARLOS: Contribution to the enhancement of IoT-based application development and optimization of underwater communications, by artificial intelligence, edge computing, and 5G networks and beyond, in smart cities/seas
  
  Author: CEPEDA PACHECO, JUAN CARLOS
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN NETWORK ENGINEERING
  Department: Department of Network Engineering (ENTEL)
  Mode: Normal
  Deposit date: 22/04/2024
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: DOMINGO ALADREN, MARIA DEL CARMEN
  Committee:
       PRESIDENT: LLORET MAURI, JAIME
       SECRETARI: REMONDO BUENO, DAVID
       VOCAL: HUERTA, MÓNICA KAREL
  Thesis abstract: 6G networks have emerged as a revolutionary breakthrough, promising ultra-fast and reliable connectivity that redefines the way we interact with the digital world. This new generation of networks not only drives communication between devices but is also the backbone of the Internet of Things. In addition, the learning and adaptive capabilities of Artificial Intelligence systems are driving process automation and efficiency. Similarly, Edge Computing complements this landscape by decentralizing data processing, bringing computing capacity closer to the sources of information. This allows for reducing latency and improving efficiency byprocessing data in real-time, driving critical applications that require instantaneous responses.This thesis focuses on two important points: 1) Improving the efficiency of applications in smart cities, and 2) Enhancing the efficiency of underwater communications in smart coastal cities by applying artificial intelligence, edge computing, and 5G and beyond. To achieve these objectives, an exhaustive study of the existing literature on 5G and beyond networks, smart cities, and artificial intelligence has been carried out. In addition, technical documentation to obtain an updated view of the different technologies that enable the development of applications based on 5G and beyond has been analyzed. Aiming to generate newand innovative alternatives in the field of tourism, security, improved underwater communications, and marine discovery that drive promote development to meet the needs of citizens in smart cities and ocean/sea. As a result of this study, the first contribution has emerged. It involves the analysis, design, and implementation of a tourist attraction recommendation system employing a deep learning algorithm tailored for smart cities. The primary objective is to improve how tourist attraction recommendations are made so that they are tailored to the requirements of each visitor in a given city and thereby reduce the time it may take a visitor to search for possible places to visit.The second contribution arises in surveillance and security, which consists of a distraction detection system for the prevention of drowning in aquatic places, developed in a 5G and beyond network environment. For this goal, an approach of surveillance cameras capturing images of people in charge of minors in swimming pools or beaches was proposed; and employing an ML algorithm (convolutional neural networks) to classify the type of distraction that a person in charge of a minor may have.Finally, the third contribution is presented, called reinforcement learning and mobile edge computing for 6G-based underwater wireless networks. In this approach, a submerged edge mobile computing architecture is presented in which an AUV is used as a mobile platform (MEC), in addition, several local AUVs equipped with computational resources that collect tasks from sensor nodes and can make the decision to process them locally or partially or fully offload them to the mobile edge computing AUV device. To this end, an algorithm based on deep reinforcement learning (DDPG) is proposed for trajectory control, task offloadingstrategy, and computational resource allocation, combined with mobile edge computing and AUVs to improve underwater communication; aiming to minimize the sum of maximum processing delays and energy consumption during the whole process of executing a task.The contributions presented in this doctoral thesis are of singular importance, since to date they continue to be innovative. The contributions presented not only represent significant advances in their respective areas but also lay the groundwork for future research and developments in smart city construction and underwater communications optimization, thereby reinforcing the transformative potential of artificial intelligence, edge computing, and advanced wireless networks in these domains.
  

• HAASTRUP, ADEBANJO: Enhanced Dynamic Bandwidth Algorithms for Passive Optical Networks
  
  Author: HAASTRUP, ADEBANJO
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN NETWORK ENGINEERING
  Department: Department of Network Engineering (ENTEL)
  Mode: Normal
  Deposit date: 22/04/2024
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: RINCON RIVERA, DAVID | PINEY DA SILVA, JOSE RAMON
  Committee:
       PRESIDENT: KHALILI, HAMZEH
       SECRETARI: SPADARO, SALVATORE
       VOCAL: PAPAGIANNI, CHRYSA
  Thesis abstract: The telecommunications industry faces rapid changes due to the deployment of ultra-high speed access networks (5G and beyond, fiber-to-the-home), promising unparalleled experiences with high bandwidth and low latency. However, this transition brings challenges. With the surge in smart device numbers and bandwidth demand, optimizing network architecture, management, and resource usage is crucial for cost-efficiency. Passive Optical Networks (PONs) offer efficient broadband access for residential and commercial sectors, with advantages like energy efficiency and robust security and high performance. Leading organizations, such as IEEE and ITU-T, are actively developing standards to increase the capabilities of next-generation PONs. The goal is to meet the demands by implementing innovative mechanisms for efficient management, resource allocation, QoS, energy savings, and low latency.Next-generation PONs have introduced the use of multiple wavelengths based on TWDM techniques. However, managing multiple wavelengths presents challenges, as DBA algorithms need to consider both the time and wavelength dimensions of the network. This follows a Joint Time and Wavelength Scheduling (JTWS) scheme, which requires complex implementation. TWDM-PON also utilizes tunable transceivers in ONUs to switch between wavelengths, but this introduces a delay called Laser Tuning Time (LTT) which is often ignored, but it is an important consideration when designing our DBA algorithms. Additionally, there is a demand to integrate metro and access networks for streamlined telecom infrastructure. Long Reach PON (LRPON) offers a solution by expanding coverage from 20 km to 100 km, enabling high-speed, long-distance data transmission over optical fibers. This reduces the need for central offices, resulting in cost savings. However, the extended reach of LRPONs introduces new challenges, particularly in the area of DBA algorithms. Traditional DBA algorithms like IPACT may not be as efficient for LRPONs due to increased propagation delays and round-trip times (RTT) between the OLT and ONUs. To address these challenges, a novel DBA algorithm called the Distance Weighted DBA (DWDBA) algorithm is proposed.This thesis delves into the limitations of traditional DBA algorithms and proposes novel Enhanced DBA solutions for PONs. Leveraging techniques such as the Longest Processing Time (LPT) scheduling method to minimize queue delays, our DBAs also consider the concept of laser tuning time to bring a practical, real-world approach to our system. The main contributions of this thesis are: - Incorporating the often-overlooked laser tuning time (LTT) concept in our analysis of DBA for TWDM PONs, therefore obtaining more realistic results. - Introducing an innovative algorithm for PONs employing LPT to minimize queue delay and enhance throughput, resulting in a notable reduction (up to 73%) with respect to the queue delay when compared to IPACT. - Developing a Distance Weighted DBA (DWDBA), specifically tailored for LRPONs, aimed at preventing the penalization of ONUs located farther from the OLT. This results in improving up to 30% and 10% the queue delay and throughput, respectively, over IPACT.The effectiveness of these proposed algorithms is rigorously evaluated through comprehensive simulations, demonstrating their potential to meet the demands of future networks.
  

• KARAMANEA, PANAGIOTA: Diachronic terrains: Three landscape narrations on the west Attic coast
  
  Author: KARAMANEA, PANAGIOTA
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN URBANISM
  Department: (DUTP)
  Mode: Normal
  Deposit date: 23/04/2024
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: SABATE BEL, JOAQUIN | GOULA, MARIA
  Committee:
       PRESIDENT: SARDÀ FERRAN, JORDI
       SECRETARI: SPANOU, IOANNA
       VOCAL: MORAITIS, KONSSTANTINOS
  Thesis abstract: The thesis researched for a hybrid interdisciplinary system of tools that could help decode coastal landscapes. The west Attic shoreline, a space of contrasts and surprises, stands as a forgotten landscape awaiting rediscovery. Amidst the interplay of urbanity and landscape Attica, unfolds dynamic perspectives as a cultural ecotone. Until now, a comprehensive cartographic study of the western Attic coast, has been absent. The research delves into the cultural formation of perspectives on the coast, examining how both foreign and local intellectuals have shaped these views over time. It investigates the Grand Tour and its contemporary interpretations, the 1930¿s, seeking to understand their influence on the collective imagination regarding Attica. It provides a synthesized overview of the historical aspect, spanning from antiquity to the touristic evolution of today. The research navigated through toponymy and landscape, historical archives and old maps focusing especially on the west Attic coastal terrain. The research also endeavors to adopt the lens of the seventh art, cinema. The research explores the sensorial gaze Attica and the local cinema 1950¿s selection of coastal shots. From a cultural and sensory standpoint, cinema offers a unique lens through which the coast is perceived. It allows to explore the emotional and experiential aspects of coastal life, capturing the essence of the seaside environment. Cinema, with its storytelling capabilities can help us connect with the coast on a personal and emotional level. By dissecting films where the landscape takes on a protagonist role, the research shifts its focus to cinema as a tool for articulating the intricacies of the coastal environment ¿ a novel and imaginative approach to apprehending its essence. In this context, cinema is not merely regarded as a narrative artistic form; rather, it emerges as a potent medium for both representation and comprehension. Finally, the thesis delves into an exploration of the landscape through original mapping and cartography, aiming to uncover dynamics and qualitative attributes. The overarching goal is to dissect the coastal expanse, unravelling its configuration, fundamental traits, and prominent landscape elements. Through cartography, an effort is made to discern the intricate relationships between various entities, navigating scales and offering an interpretive lens grounded in landscape perspective. On the other hand, the morphological and cartographical approach involves an examination of the physical characteristics and spatial layout of the coast. This includes the topography, land use patterns, and geographical features that shape the coastal region. By delving into these aspects, a deeper understanding of how the coast functions is gained and how it can be harnessed for various purposes, from urban landscape planning to environmental conservation. The spatial configuration of the ground, the surface as infrastructure in the sense of landscape armature and ecological aspects seen as a systemic network, could help integrate the natural landscape in the discussion of rejuvenating the city in a resilient way. Landscape architecture as an interface between city and nature is proposing interdisciplinary processes to apply.
  

Reading date: 24/05/2024

• AGUIRRE RUZ, ALEJANDRO: Numerical approximation of thin structures using stabilized mixed formulations for Infinitesimal and Finite Strain theories, including Fluid-Structure Interaction problem applications.
  
  Author: AGUIRRE RUZ, ALEJANDRO
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN STRUCTURAL ANALYSIS
  Department: (DECA)
  Mode: Normal
  Deposit date: 22/04/2024
  Reading date: 24/05/2024
  Reading time: 12:00
  Reading place: Aula C1002, Edifici C1, Campus Nord ETSECCPB (Escola Tècnica Superior d'Engineria de Camins, Canals i Ports de Barcelona)
  Thesis director: CODINA ROVIRA, RAMON | BAIGES AZNAR, JOAN
  Committee:
       PRESIDENT: ROMERO OLLEROS, IGNACIO
       SECRETARI: ROSSI BERNECOLI, RICCARDO
       VOCAL NO PRESENCIAL: COLOMÉS GENÉ, JOSEP ORIOL
  Thesis abstract: The theories of thin structures can be classified into two main branches depending upon whether shear deformation in the transverse direction is taken into consideration or not. In this context, theories accounting for shear deformations prove suitable for modeling structures with both thin and thick profiles. In the Finite Element context, they are referred to as C0 theories due to the minimum continuity order of shape functions required to pose a discretized approximation. However, there are space incompatibilities in the standard discrete approximation that exhibits spurious solutions, particularly evident in thin structures. These instabilities, known as numerical locking, result in an artificial stiffening of the structure, whose effect becomes more pronounced for thinner structures. Various forms of numerical locking can be triggered, influenced not only by the slenderness of the structure but also by its shape and the nature of the applied loads. In this context, flat structures are prone to shear locking when exposed to transverse loads. Conversely, curved structures may confront different mechanisms leading to various forms of numerical locking, namely membrane, thickness, and trapezoidal locking. The initial part of the study aims to develop a specialized framework to address instabilities in the context of flat structures in the context of Reissner-Mindlin theory. Subsequently, the second part of the study aims to expand the framework to effectively address instabilities arising in of curved structures in the context of Solid-Shell elements. The locking problem is approached by means of a mixed formulation that considers displacements and stress as unknowns in a curvilinear coordinate framework. This approach allows to isolate the components of the stress tensor in order to study the mechanisms in which every type of numerical locking are triggered. The third part of the thesis is dedicated to integrating the previous advancements into Finite Strain analysis by the inclusion of standard hyperelastic constitutive behavior. With this approach, the problem becomes even more difficult to solve because of the non-linearity and the large deformations the shell is subject to. Lastly, the fourth and final part is dedicated to addressing the Fluid-Structure Interaction problem using an embedded mesh approach, which has consistently been a topic of great research interest in the literature, because of its complexity and wide variety of applications. This problem introduces a variety of challenges that have to be properly addressed: the discontinuous pressure field arising for the structure separating the fluid domain, the computation and imposition of transmission conditions between domains, the coupling strategy, and the algorithmic work needed to join all of these ingredients together. This thesis mainly focuses on overcoming challenges associated with thin structures when employing the conventional Galerkin Finite Element approach. It seeks solutions through stabilized methods, specifically within the Variational Multiscale framework. As result, the formulations developed through the investigations have proven to be robust, allowing to model locking-free thin structures efficiently, and to accurately describe the physics of thin shells immersed in fluid flows and being subject to large deformations.