DOCTORAL DEGREE IN BIOMEDICAL ENGINEERING

• BARRANCO ALTIRRIBA, MARIA: Computational metabolomics to uncover alterations in diabetes mellitus and its complications
  
  Author: BARRANCO ALTIRRIBA, 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 BIOMEDICAL ENGINEERING
  Department: Department of Automatic Control (ESAII)
  Mode: Article-based thesis
  Deposit date: 12/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: PERERA LLUNA, ALEXANDRE | MAURICIO PUENTE, DIDAC
  Thesis abstract: Metabolomics is the discipline focused on the study of the metabolome, defined as the complete set of small molecules present in a biological system. Metabolites play key roles in cellular processes and interact with other omic layers such as the genome, transcriptome, and proteome. For these reasons, metabolomics provides a powerful approach to uncover underlying biological mechanisms related to specific medical conditions or environmental exposures. Lipidomics, a subfield of metabolomics, focuses on the lipidome, the full lipid content of a system. Metabolomics has shown great potential in understanding metabolic diseases, a growing public health concern that includes both type 1 (T1D) and type 2 diabetes (T2D). T1D is the most prevalent form of diabetes in childhood, while T2D is more frequently acquired during adulthood, and its incidence is increasing at an alarming rate. Both forms can lead to serious microvascular and macrovascular complications.Analytical techniques such as liquid chromatography coupled with mass spectrometry (LC-MS) enable metabolome characterization. Untargeted metabolomics aims to detect the widest possible range of metabolites. However, the resulting data are complex and require careful handling, highlighting the importance of computational metabolomics for preprocessing, filtering, analysis, and interpretation. Interpretation remains the most challenging stage, encompassing LC-MS annotation and enrichment analysis. In lipidomics, the lack of lipid-specific pathway databases further complicates this step.This thesis is broadly divided into two parts: one focusing on computational metabolomics and the other on the more specific field of computational lipidomics. In the first part, diffusion in biological knowledge-based graphs is shown to support LC-MS annotation by prioritizing potential candidates. Building on this approach, the mWISE R package is proposed for the annotation of LC-MS data. mWISE is then applied to annotate a diabetes-related dataset, and the resulting annotations are used to perform pathway enrichment analysis. This analysis reveals the association between T2D incidence and guanine, pregnenolone sulfate, and citrulline, as well as nucleotide metabolism and ABC transporter pathways.In the lipidomics section, a broad panel of disrupted lipids in T1D and T2D is identified. Additionally, sex-specific differences are uncovered, along with lipids that show progressive changes from normoglycemia to prediabetes and T2D. A set of lipids associated with T2D exclusively in women, and others linked to T1D only in men, are proposed to contribute to sex-specific differences in the prevalence of diabetes-related complications. In the same population, a disrupted lipid profile associated with subclinical carotid atherosclerosis is identified exclusively in individuals with T2D. These lipids are hypothesized to underlie the more severe progression of atherosclerosis in T2D through specific molecular mechanisms. Moreover, this atherosclerosis-associated lipid alteration is more pronounced in individuals with T2D who are current or former smokers, as well as in those not receiving lipid-lowering treatment. Finally, a functional enrichment algorithm for lipidomics data is developed and used to interpret both lipidomics studies. The algorithm is based on molecular embeddings of lipids, obtained through a language model that was benchmarked against similar models. The enrichment results strongly align with previous interpretations and additionally highlight an overlooked relationship between T1D and reduced levels of docosahexaenoic acid.
  
• FUENTES LLANOS, JUDITH: Development of 3D Bioengineered Skeletal Muscle-based Bioactuators for Biorobotic and Biomedical Applications
  
  Author: FUENTES LLANOS, JUDITH
  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 BIOMEDICAL ENGINEERING
  Department: Department of Materials Science and Engineering (CEM)
  Mode: Article-based thesis
  Deposit date: 11/02/2026
  Reading date: 10/04/2026
  Reading time: 11:00
  Reading place: Sala d'Actes, Edifici Vèrtex, Campus Diagonal Nord, Vèrtex (VX), Plaça d'Eusebi Güell, 6, 08034 Barcelona
  Thesis director: SÁNCHEZ ORDÓÑEZ, SAMUEL | GUIX NOGUERA, MARIA
  Thesis abstract: Biohybrid systems are engineered constructs that integrate living materials, like cells or tissues, with synthetic components, such as electronics, mechanical structures, or other artificial materials. This integration aims to leverage the unique capabilities of biological materials, including self-assembly, responsiveness to certain stimuli, adaptability, and self-repair, being features challenging to replicate in their synthetic counterparts. By combining these traits with the robustness and compliance of synthetic structures, biohybrid systems emerge as versatile platforms capable to both adapt and actuate in dynamic environments. Among the various biological components explored, skeletal muscle stands out due to its high force-to-weight ratio, controllable contraction, and compatibility with 3D biofabrication strategies, making it ideal for applications in soft robotics, drug testing, and regenerative medicine.This PhD thesis focuses on the development of 3D bioengineered skeletal muscle bioactuators, addressing key aspects from the biofabrication to functionality improvement, actuation control, and regenerative capacity.We developed and optimized a Pluronic-Assisted Coaxial 3D bioprinting (PACA-3D) method capable of biofabricating fascicle-like skeletal muscle bioactuators with improved maturation, sarcomere formation and contractile force when compared to muscles generated using conventional extrusion-based 3D bioprinting. To improve actuation and directional control, we introduced magnetic responsiveness into the bioactuators by developing a biocompatible ferrofluid-based bioink, used to fabricate the Ferromuscle, in collaboration with Waterloo and Aalto Universities. Such bioactuators exhibited improved cell alignment and force output, as well as magnetically guided actuation. In collaboration with the University of Cagliari (UniCa), we also evaluated the integration of flexible Organic Field-Effect Transistor (OFET)-based strain sensors for real-time monitoring of contractile behavior. These devices allowed tunable force sensing, showing no signal cross-talk when using standard electrical stimulation protocols, which is paving the way for closed-loop stimulation strategies and automated drug screening platforms. Moreover, we evaluated the self-healing capacity of the bioactuators after damage conditions, designed to mimic the mechanical stress that they typically undergo during manipulation of biohybrid systems. We observed partial to full force recovery and structural remodeling without external interventions, although the underlying biological mechanisms remain to be elucidated.Finally, additional collaborative work that enriched this thesis will be also discussed. Such studies include the integration of tendon-like structures to develop muscle-tendon unit (MTU) actuators with enhanced biomimicry, performance, and stability (ETH Zurich). Moreover, it will also be covered the design of biohybrid flexure mechanisms powered by skeletal muscle, evaluating how skeleton architecture and voltage- or current-based stimulation influence actuation output (SSA and UniCa).Altogether, the work presented in this thesis contributes to the field of biohybrid robotics by combining advanced fabrication, actuation control, integrated sensing, and regenerative capacity. These developments move us closer to robust, autonomous, and functional biohybrid machines for applications in soft robotics, tissue engineering, and biomedical research.
  
• PRATS BISBE, ALBA: Nous entorns interactius de realitat virtual immersiva aplicats a la neurorehabilitació de funcions cognitives i sensoriomotores
  
  Author: PRATS BISBE, ALBA
  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 BIOMEDICAL ENGINEERING
  Department: Department of Automatic Control (ESAII)
  Mode: Normal
  Deposit date: 29/01/2026
  Reading date: 08/04/2026
  Reading time: 12:00
  Reading place: Sala d'Actes de l'Institut Guttmann (Carretera de Can Ruti, s/n, 08916 Badalona, Barcelona)
  Thesis director: JANE CAMPOS, RAIMON | OPISSO SALLERAS, ELOY
  Thesis abstract: Recovery following an acquired brain injury (ABI) represents a major challenge with significant implications for health, quality of life, and socioeconomic burden. Advances in medicine and neurorehabilitation —particularly personalized, patient-centred clinical practice— have contributed to improved outcomes by promoting autonomy and participation in society. Technology has become a cornerstone in this continuous improvement, and virtual reality (VR) has emerged as a promising therapeutic tool. VR immersive properties, ergonomic interaction capabilities, ecological validity, and intrinsic advantages of digitalisation can enhance motor learning and cognitive improvement. However, there is still no conclusive evidence regarding VR clinically significant effects and consistent adoption in routine hospital practice.The aim of this thesis is to identify the barriers to VR implementation in neurorehabilitation and to elicit the specific requirements for its integration as an effective support tool. To achieve this goal, a systematic review of current applications was first conducted to identify methodologies and features associated with significant therapeutic outcomes. This analysis revealed considerable heterogeneity across clinical protocols, as well as the hardware and software used, highlighting the lack of standards and quality criteria that hinder result generalization. To address this gap, a new conceptual framework for evaluating VR tools in clinical contexts was developed: the Virtual Reality-tools Quality Assessment Framework (VR-tools QAF), which defines technical, functional, and safety requirements for VR equipment and virtual environments. In parallel, a comprehensive methodology for user-centred design, iterative development, and clinical validation was created, termed the Virtual Reality-tools Design and Development Guide (VR-tools DDG).Through the combined model (VR-tools QAF + DDG), eleven multimodal VR environments were developed to support the rehabilitation of cognitive and sensorimotor functions in individuals who have suffered an ABI. These environments integrate principles of repetitive practice, multisensory feedback, adaptive difficulty, and mechanisms for tailoring the experience to the patient’s profile. Validation was carried out in collaboration with a leading hospital in neurorehabilitation and brain health, where usability and feasibility tests were conducted with healthcare professionals and patients with ABI. Proof-of-concept trials demonstrated good acceptance and ease of use among clinicians (n = 26) from different specialties, as well as good tolerance and absence of adverse effects among the 20 patients. Moreover, a longitudinal VR intervention conducted within an efficacy study confirmed the feasibility and safety of delivering a 9-hour VR-based cognitive rehabilitation protocol with 21 patients with ABI, distributed in 20–30-minute sessions, 2–3 times per week.Overall, this research establishes a comprehensive methodological model for effectively integrating immersive VR into neurorehabilitation hospital settings, combining scientific rigour with technical and clinical feasibility. The results lay a solid foundation for future efficacy studies aimed at developing standardized treatment protocols for patients with ABI.
  

DOCTORAL DEGREE IN CIVIL ENGINEERING

• FORERO ORTIZ, EDWAR ANDRÉS: Long-term renewal framework for urban water distribution networks: an application in the Barcelona Metropolitan Area
  
  Author: FORERO ORTIZ, EDWAR ANDRÉS
  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 CIVIL ENGINEERING
  Department: Barcelona School of Civil Engineering (ETSECCPB)
  Mode: Normal
  Deposit date: 17/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: SANCHEZ JUNY, MARTI | MARTÍNEZ GOMARIZ, EDUARDO
  Thesis abstract: Context and BackgroundManaging urban water infrastructure in the 21st century involves balancing systemic asset ageing with rising service expectations and increasing climate stressors. In established metropolitan networks, traditional renewal policies based on reactive maintenance or simple age-based heuristics are becoming less effective at preventing cascading risks of structural failure. Ensuring the sustainability of Water Distribution Networks (WDNs) now requires a shift to proactive stewardship frameworks that can optimise capital allocation over multiple decades while addressing the complex interdependencies of the water-energy nexus and urban mobility.Problem StatementA fundamental disconnect persists between asset-specific stochastic deterioration modelling and system-wide strategic renewal planning. Existing failure prediction models often operate independently of operational decision-support systems, failing to translate probabilistic outputs into actionable long-term policies. Additionally, the integrity of infrastructure datasets is compromised by administrative fragmentation of pipe records and survival bias, which can lead to a significant underestimation of failure risk. There is a critical methodological gap in combining high-fidelity Machine Learning (ML) with Survival Analysis (SA) to project Level-of-Service (LoS) trajectories that incorporate social and environmental externalities as equally important factors in planning.Research ObjectivesThe main aim of this doctoral thesis is to develop and validate a multi-layered computational framework, the Renovation Strategy Simulator (comparator) 'REPIPE', designed to systematically evaluate different pipe renovation strategies for the drinking water distribution system in the Barcelona Metropolitan Area. Specifically, the thesis aims to: (i) improve short-term failure classification through business-oriented hyperparameter optimisation; (ii) create stratified survival models that consider dual temporal censoring and recurrent failure events; and (iii) design an iterative algorithmic framework to convert asset-level prognoses into internally consistent network performance trajectories over a 25-year planning period.MethodologyThis research establishes a unified "data-to-decision" pipeline supported by three key computational pillars. First, a novel Genealogical Tree-Based Inventory (GTBI) was designed to trace pipe segments back to their Original Pipe Section (OPS) identifiers, resolving historical gaps in the asset register. Second, a hybrid predictive framework was developed, combining an eXtreme Gradient Boosting (XGBoost) algorithm for tactical screening with a stratified Cox Proportional Hazards (CPH) model for multi-decadal prognosis. Third, these models were integrated into a simulation engine that employs MIVES-based value functions and conditional survival probabilities to project ten LoS indicators across six annual investment scenarios (€0M–€30M).Main ResultsThe implementation of the GTBI corrected 22% of accumulated failures that databases had previously misattributed, strengthening the evidence base for failure risk modelling. The study applied a 5% annual renewal threshold to keep all scenarios directly comparable and to support a consistent calculation of the failure-avoidance rate under a fixed yearly renovation capacity. Under this constraint, the XGBoost classifier, tuned through a brute-force hyperparameter search using a business-oriented Lorenz metric, achieved a reported 32% relative increase in failure-prevention effectiveness over default settings, capturing 30.2% of expected failures within the 5% renewal limit.
  

DOCTORAL DEGREE IN COMPUTATIONAL AND APPLIED PHYSICS

• PERA I FERRERUELA, JORDI: Itinerant ferromagnetism and polarons in SU(N) Fermi gases
  
  Author: PERA I FERRERUELA, JORDI
  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 COMPUTATIONAL AND APPLIED PHYSICS
  Department: Department of Physics (FIS)
  Mode: Normal
  Deposit date: 13/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: CASULLERAS AMBROS, JOAQUIN | BORONAT MEDICO, JORDI
  Thesis abstract: This thesis presents an in-depth study of the magnetic properties of repulsive Fermi gases in the continuum. Once a purely academic subject, Fermi gases can now be realized experimentally in ultracold atom laboratories, making a theoretical understanding of their behavior both timely and necessary. Our analysis focuses primarily on whether such systems favor a transition to a ferromagnetic state or remain paramagnetic. As technique, we use a perturbative approach that systematically includes the effects of higher-order terms, temperature, and mass imbalance. In the final part of the thesis, we also study the Fermi polaron problem. First, we examine a simple case: a Fermi gas with equal masses at zero temperature up to second order in perturbation theory. As this chapter can be considered an introduction to the magnetic phenomenology of Fermi gases, we compare the outcome of our second order analysis with the well-known scenario of the Stoner model, which is the first-order model. In both cases, a ferromagnetic transition always occurs. Moving from the Stoner model to the second-order one, there are two differences to point out: the continuous phase transition observed at S=1/2 becomes discontinuous, and the ferromagnetic phase arises at lower densities. The next step is to improve the description of the potential interaction. We do so by going up to third order in perturbation theory. However, this improvement comes with a cost; as we now have more information about the potential, the complexity of the problem is higher. We split the analysis into two parts. We first consider a hard-sphere potential without p-wave intra-particle interaction. We observe the same tendencies that we encounter when we move from first order to second order, but, more importantly, we recover the continuous phase transition for S=1/2. Having now at hand the third-order energies, we explore a wide range of different potentials. Surprisingly, for each spin S we observe a richness of scenarios: continuous transitions, discontinuous ones, combinations of the other two, stairs-like cases, and even, no transition at all. Next, we introduce another variable to our problem: the temperature. Thermal effects do not eliminate the ferromagnetic phase for any potential, but they clearly go against it: either they force the system to remain paramagnetic or make the transition to occur at higher densities, which probably hinders the ferromagnetic phase. Nevertheless, larger spins better resist these thermal effects. Additionally, temperature reduces the diversity of magnetic behavior, leading to more universal patterns: continuous transitions for S=1/2 and discontinuous ones for larger spins. Up to this point, we assumed that all the particles have the same mass. The next step is to consider different relations of masses. Adding another variable increases even more the complexity of the problem and, hence, we only inspect it at second order and at zero temperature. As we have commented above, at second order there is always a ferromagnetic transition. Now, this is also the case, but with some nuances. The massive particles are always the promoted ones, even at low densities. For S=1/2, this is the end of the story. But for larger spins, as we can have many mass relations, we can obtain lots of configurations, which may give us more tunability. Finally, we move beyond magnetic properties to study the Fermi polaron problem: an impurity immersed in a Fermi sea. We focus on how impurities affect our system. We examine the main properties of the Fermi polaron at third order and the effects of mass. Specifically, the massive polaron works just fine, but the light polaron is unstable. Altogether, this work offers a detailed and thorough exploration of Fermi gases, providing both qualitative insight and quantitative predictions relevant to experimental efforts.
  
• PUCH GINER, IGNASI: Optimization of Monte Carlo and Molecular Dynamics Techniques for Receptor–Ligand Binding Studies
  
  Author: PUCH GINER, IGNASI
  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 COMPUTATIONAL AND APPLIED PHYSICS
  Department: Department of Physics (FIS)
  Mode: Normal
  Deposit date: 16/02/2026
  Reading date: 16/04/2026
  Reading time: 10:00
  Reading place: Sala TeleensenyamentEdifici B3 - Ricardo Valle Sala 103 Planta 1Campus Nord
  Thesis director: GUALLAR TASIES, VÍCTOR
  Thesis abstract: Drug discovery is a complex and resource-intensive process that requires the identification of molecular candidates with optimal binding properties to their corresponding targets. Computational techniques play a central role in streamlining this process by predicting biomolecular interactions, optimizing molecular properties, and accelerating early-stage drug development.This thesis focuses on advancing computational strategies for biomolecular modeling, integrating molecular mechanics techniques—such as Monte Carlo and molecular dynamics simulations—with state-of-the-art machine learning tools. The objective is to improve scoring functions, enhance simulation accuracy, and develop a framework that enables the application of these methods to real-world pharmacological challenges.A key contribution of this work is the systematic characterization of the PELE (Protein Energy Landscape Exploration) simulation framework, including the rationalization of binding energy estimators and the assessment of their predictive performance and computational efficiency. These benchmarks address existing knowledge gaps in scoring functions and facilitate more accurate predictions of protein–ligand interactions, thereby supporting the prioritization of promising molecules for experimental evaluation.Beyond methodological advances, this thesis applies the developed computational approaches to pharmacologically relevant case studies. The structural characterization of the amino acid transporter Asc1/CD98hc was carried out using an integrated computational pipeline, providing new insights into its binding and transport mechanisms. Furthermore, the study of the Meis1–Hoxb13 transcription factors explores novel drug–DNA interactions with potential implications for targeted therapies. Another application focuses on Charcot–Marie–Tooth disease, where computational simulations contributed to understanding GDAP1–LAMP1 tethering dysfunction and possible therapeutic interventions. These examples highlight the versatility of the proposed computational strategies across diverse biological contexts.The results of this research propose a practical framework for improving ligand-binding predictions. The explored methodologies strengthen molecular modeling by combining physics-based techniques with varying speed–accuracy trade-offs and data-driven approaches, bridging the gap between traditional simulation methods and AI-driven predictions. The findings have implications for both academic research and industrial drug discovery, offering scalable and generalizable solutions for structure-based and ligand-based design.
  

DOCTORAL DEGREE IN COMPUTER ARCHITECTURE

• DOBLAS FONT, MAX: High-Performance Sequence Alignment: Co-Designing Algorithms and Hardware Architectures for Efficient and Scalable Acceleration
  
  Author: DOBLAS FONT, MAX
  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 COMPUTER ARCHITECTURE
  Department: Department of Computer Architecture (DAC)
  Mode: Normal
  Deposit date: 20/02/2026
  Reading date: 07/04/2026
  Reading time: 16:00
  Reading place: C6-E106
  Thesis director: MORETÓ PLANAS, MIQUEL | MARCO SOLA, SANTIAGO
  Thesis abstract: Over the past decade, the exponential growth of genomic data has driven significant advancements in genomics and healthcare, enabling breakthroughs in large-scale genomic studies, personalized medicine, and epidemiological surveillance. However, this rapid data expansion has also posed substantial computational challenges, particularly in sequence alignment, a cornerstone of genomic analysis. Sequence alignment is critical for applications such as disease diagnostics, population-wide genetic research, and outbreak tracking, yet it demands scalable and efficient solutions to handle the increasing data volumes and diverse use cases.This thesis addresses the efficiency-flexibility gap in sequence alignment through three key contributions. First, it introduces QuickEd, a novel sequence alignment algorithm that reduces computational complexity by combining heuristic bounding with optimal alignment. QuickEd achieves O(nŝ) complexity, where n is the sequence length and ŝ is an estimated upper bound of the alignment score, significantly improving upon the O(n²) complexity of traditional dynamic programming algorithms. By efficiently bounding the maximum alignment score, QuickEd reduces the computational burden while maintaining accuracy, making it well-suited for long-read sequencing and other demanding applications.Second, this thesis proposes GMX, a set of instruction set architecture (ISA) extensions designed to enhance the efficiency of dynamic programming-based alignment algorithms. GMX provides fundamental building-block operations for fast, tile-wise computations of the DP matrix, reducing memory footprint and computational overhead. These extensions enable seamless integration into widely used algorithms and tools, offering a cost-effective alternative to domain-specific accelerators (DSAs) while achieving comparable performance improvements.Third, the thesis presents SMX, a heterogeneous architecture that balances flexibility and performance to address the diverse requirements of real-world sequence alignment tasks. SMX integrates an ISA extension (SMX-1D) for irregular and sequential tasks and a specialized coprocessor (SMX-2D) for regular and parallel tasks, all orchestrated by a general-purpose core. This architecture supports various configurations for different sequence types (DNA, protein, and ASCII text) and alignment models, including weighted gaps and substitution matrices. By combining high performance with adaptability, SMX enables efficient acceleration of a wide range of sequence alignment applications.Together, these contributions advance the state of the art in sequence alignment, providing scalable, flexible, and efficient solutions to meet the demands of modern genomic analysis. The innovations presented in this thesis pave the way for faster and more reliable genomic analyses, facilitating critical applications such as personalized medicine, population-scale sequencing, and epidemiological surveillance in the era of long-read sequencing technologies.
  
• VALDÉS JIMÉNEZ, ALEJANDRO MAURICIO: Design, parallelization and acceleration of algorithms to discover three-dimensional patterns in proteins
  
  Author: VALDÉS JIMÉNEZ, ALEJANDRO MAURICIO
  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 COMPUTER ARCHITECTURE
  Department: Department of Computer Architecture (DAC)
  Mode: Normal
  Deposit date: 03/03/2026
  Reading date: 09/04/2026
  Reading time: 15:00
  Reading place: C6-E101
  Thesis director: JIMENEZ GONZALEZ, DANIEL | NUÑEZ VIVANCO, GABRIEL
  Thesis abstract: The rapid growth of protein structure databases, such as the Protein Data Bank (over 230,000 structures) and AlphaFold (over 200 million structures), requires efficient and scalable algorithms capable of exploiting high-performance computing (HPC) architectures to enable large-scale structural analysis in reasonable times. This thesis focuses on the design and implementation of optimized and parallel algorithms for discovering, analyzing, and clustering conserved three-dimensional amino acid patterns in proteins. The work focuses on the Geomfinder algorithm (A multi-feature identifier of similar three-dimensional protein patterns: a ligand-independent approach), which compares three-dimensional patterns between pairs of proteins, and the novel 3D-PP algorithm (A tool for discovering conserved three-dimensional protein patterns), proposed in this thesis, which discovers and clusters common three-dimensional patterns within protein sets. Both algorithms are ligand and sequence independent and do not require predefined patterns, enabling the identification of previously unknown functional sites. However, their original sequential implementations limit their applicability to large datasets. For Geomfinder, several sequential optimizations were introduced to reduce algorithmic complexity and long-latency operations. The incorporation of a Merge Join–based strategy reduced partial scoring complexity from O(N×M) to O(N+M), ensuring each descriptor element is evaluated only once. Lazy evaluation and reordering of partial scoring function calls further reduced execution time. These optimizations achieved speedups ranging from 6.2x to 19.7x, depending on the search range. Multiple parallelization strategies were then explored, including OpenMP, MPI, hybrid MPI+OpenMP, and CUDA. OpenMP with fine-grained data decomposition and optimized scheduling achieved near-ideal speedups, reaching 32.6x with 64 threads. MPI-based distributed parallelization achieved up to 19.4x speedup with 64 processes, while hybrid MPI+OpenMP further improved performance, reaching 67.4x using 1,024 threads. GPU acceleration using CUDA provided speedups of up to 8.6x, with performance increasing for larger workloads. After applying the algorithmic optimizations to the original sequential version, profiling revealed a change in the computational bottleneck, and an additional OpenMP parallelization stage was applied, achieving up to 494x acceleration over the original sequential version. In one case study, the runtime was reduced from over one hour to approximately 3.4 seconds. For the 3D-PP algorithm, profiling revealed that over 96% of execution time was spent processing protein chains. All major components were parallelized. Three OpenMP approaches were evaluated, with the best solution based on explicit and nested tasks, achieving a 22.3x speedup and reducing execution time from 1.25 hours to 201.5 seconds. Distributed-memory strategies using MPI focused on minimizing communication through early pattern reduction, achieving speedups of approximately 32x with 64 processes. Hybrid MPI+OpenMP implementations further improved performance, with the best approach achieving a 162.5x speedup and reducing runtime to 27 seconds. This hybrid approach mitigated synchronization overhead inherent to pure OpenMP implementations and demonstrated weak-scaling efficiency (90-100%) up to 8 processes, although efficiency dropped to around 72% when using 16 processes due to increased load imbalance and synchronization costs. The results show that explicit task parallelism and early data reduction substantially improve the performance and scalability of 3D-PP. All these improvements will help us address processing and pattern discovery in larger protein databases.
  

DOCTORAL DEGREE IN CONSTRUCTION ENGINEERING

• CLARES GARCIA, ANTONI: A Framework for the Digital Twinning of Doubly Curved Masonry Structures
  
  Author: CLARES GARCIA, ANTONI
  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 CONSTRUCTION ENGINEERING
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Normal
  Deposit date: 19/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: CHACÓN FLORES, ROLANDO ANTONIO | LLORENS SULIVERA, MIQUEL
  Thesis abstract: This thesis proposes an open and interoperable Digital Twin framework for doubly curved masonry vaults, addressing the growing need for accurate assessment and long-term management of complex heritage structures. The framework integrates BIM, structural simulations, and data from the physical asset within an IFC-based environment, enabling continuity across design, construction, and monitoring phases.The research is validated through two case studies—a Catalan vault tested to failure in laboratory conditions and a Gaussian vault analysed from design through construction and structural assessment—developed within an industrial doctoral program involving academia and industry. The methodology combines parametric geometric modelling, material characterisation using destructive and non-destructive testing, point cloud acquisition via photogrammetry and laser scanning, non-linear finite element analysis, and the integration of monitoring sensors within the BIM workflow.By linking computational design, numerical analysis, and digital heritage conservation, this work establishes a data-driven workflow for the digitalisation and preservation of complex masonry vaults and contributes to the advancement of Digital Twin methodologies in structural engineering.
  
• DELGADO ZHAGUI, ERIKA BEATRIZ: Validation of MEMS Accelerometers for Static and Dynamic Bridge Assessment Integrating Automated Synchronization and Modal Parameter Extraction
  
  Author: DELGADO ZHAGUI, ERIKA BEATRIZ
  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 CONSTRUCTION ENGINEERING
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Normal
  Deposit date: 19/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: TURMO CODERQUE, JOSE | KOMARIZADEHASL, SEYEDMILAD
  Thesis abstract: Aging bridge infrastructure worldwide faces growing challenges from increased traffic demand, environmental degradation, and sustained dynamic loads, compromising structural integrity and public safety. In this context, structural health monitoring (SHM) has emerged as a key tool for proactive infrastructure management. However, its widespread implementation remains limited by the high cost and operational complexity of commercial instrumentation systems. As a result, many bridges, especially those with limited resources, remain unmonitored, while the technical feasibility of low-cost alternatives for reliable structural assessment remains scientifically unvalidated.This research addresses this gap through the first multi-scale experimental validation of LARA (Low-Cost Adaptive Reliable Accelerometer) sensors applied to bridge monitoring, contributing to both the theoretical advancement and practical implementation of low-cost SHM systems. This work presents four original contributions.First, a methodology for indirect deformation estimation based on rotation measurement is developed and validated, using the dual-axis inclinometer capabilities of LARA. The methodology is validated through static load tests carried out on a railway bridge (access to Barcelona–El Prat Airport), allowing vertical displacements to be estimated with correlation coefficients greater than 0.95 compared to the reference instrumentation. These results demonstrate that rotations can accurately replace traditional displacement measurements.Second, controlled dynamic tests are carried out comparing the performance of LARA with commercial-grade accelerometers (IOLITEI-3XMEMS-ACC-S). The results show that LARA identifies fundamental frequencies with errors of less than 2% and estimates damping ratios with deviations of less than 4%. These results constitute the first quantitative benchmark on the use of low-cost sensors in the dynamic analysis of bridges, establishing validated performance thresholds previously unavailable in literature.Third, a complete automated procedure for modal identification is developed and validated, integrating noise filtering, peak picking algorithms, frequency domain decomposition, and quality assurance. Applied to a reinforced concrete isostatic bridge in the Basque Country, the procedure made it possible to extract modal parameters under environmental excitation with sufficient accuracy to update a finite element model (MAC>0.90), representing one of the first demonstrations of structural model calibration driven by low-cost sensors under operating conditions.Fourth, the thesis evaluates the sustainability of the LARA system from an environmental, economic, and social perspective. The results indicate a reduced carbon footprint and the possibility of expanding monitoring coverage at a fraction of the cost of commercial systems, while also contributing to several United Nations Sustainable Development Goals.Based on two experimental campaigns on bridges and multiple load scenarios, it has been confirmed that LARA achieves operational accuracy comparable to that of commercial sensors (frequency errors below 3% and modal correlation above 0.85) at approximately 8–10% of the cost. The main contribution of this thesis is the empirical demonstration that low-cost sensors can reliably perform structural identification tasks traditionally reserved for expensive instrumentation, providing a solid scientific foundation for the democratization of structural monitoring.
  

DOCTORAL DEGREE IN ELECTRICAL ENGINEERING

• FORERO QUINTERO, JOSE FERNANDO: Flexibility Management System with Distributed Energy Resources
  
  Author: FORERO QUINTERO, JOSE FERNANDO
  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 ELECTRICAL ENGINEERING
  Department: Department of Electrical Engineering (DEE)
  Mode: Article-based thesis
  Deposit date: 19/02/2026
  Reading date: 20/04/2026
  Reading time: 12:00
  Reading place: Aula 28.8 Sala de Conferències. ETSEIB-UPC
  Thesis director: VILLAFÁFILA ROBLES, ROBERTO | MONTESINOS MIRACLE, DANIEL
  Thesis abstract: The challenge of climate change is ever demanding more technological and political efforts to respond adequately to its adverse effects, which could put life on the planet at risk. Diverse plans have been proposed to incorporate more renewable generation and digitalize the power system in a decarbonizing energy transition framework. However, such plans are either not being accomplished or are insufficient according to the Paris Agreement. At the same time, emerging technologies such as artificial intelligence, high-performance sensing and automation, and real-time information and communication systems, among others, are transforming electrical networks into smart grids, aggregating more complexity.This thesis, based on a papers compendium, first surveys the existing energy management systems with distributed energy resources, delving into their configuration, energy and flexibility sources, information and communication systems, profitability analysis and indicators, and flexibility services. After reviewing the state of the art, this thesis focuses on boosting management systems towards the real-time horizon, proposing a flexibility management system. This system is a hierarchical three-level management system based on an adaptive autoregression algorithm, cost-benefits analysis, and redispatching and unit recommitment submodules to control and manage the power, and short-term flexibility requests.To validate the proposed system, not only several simulations and sensitivity analyses are carried out, but real-world energy assets are addressed in PVZEN microgrid lab in Torino, Italy. The results demonstrate that proposed system can reduce the cost overruns of real-time power deviations of the distributed resources, leverage the excess of renewable energy, and generate profits providing flexibility services to the grid. Chapters 2, 3, and 4 show the papers related to this thesis, starting with reviewing the art status, passing through the simulations and analysis, and finishing with the essays on real-world energy assets. Main conclusions and future work are depicted at the end of the document.
  
• JENÉ VINUESA, MARC: Data-Driven and Generative Methodologies for Enhanced Grid-Edge Visibility in Distribution Grids
  
  Author: JENÉ VINUESA, MARC
  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 ELECTRICAL ENGINEERING
  Department: Department of Electrical Engineering (DEE)
  Mode: Normal
  Deposit date: 12/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: ARAGÜÉS PEÑALBA, MÒNICA | SUMPER, ANDREAS
  Thesis abstract: The rapid deployment of distributed energy resources (DERs), such as residential photovoltaic (PV) systems, heat pumps, and electric vehicles, is accelerating the energy transition while reshaping electrical distribution grids. Although these technologies enable decarbonization and flexibility, their widespread installation behind the meter (BTM), together with limited measurement granularity, restricted data access, and the absence of dedicated metering, creates blind spots that hinder reliable grid operation, planning, and monitoring under high DERs penetration. This doctoral thesis addresses limited grid-edge visibility by developing data-driven methodologies to infer unobserved distribution-level phenomena from low-resolution smart meter data under realistic operational constraints. Grid-edge visibility is framed as the ability to reconstruct both legitimate and illegitimate power exchanges, encompassing BTM DERs behavior and non-technical losses (NTLs). The proposed contributions aim to support actionable and risk-aware decision-making in increasingly decentralized distribution systems. The first part of the thesis focuses on NTL detection and characterization. A comprehensive framework is proposed to identify abnormal losses through energy balances and machine learning using transformer- and customer-level active power measurements. The methodology enables fraud detection and classification, while an unsupervised customer flagging module supports targeted inspection strategies. Validation using real-world data from a Spanish distribution system operator demonstrates robust performance under realistic data availability and class imbalance. The second part addresses customer-level BTM PV disaggregation. A deterministic and adaptive hybrid methodology is introduced to detect PV installations, estimate their capacity, and disaggregate generation from net consumption measurements. By combining data-driven and physics-based models with contextually supervised source separation techniques, the approach captures system-specific and seasonal effects while remaining suitable for practical deployment. Extensive case studies demonstrate robustness to data length, seasonal variability, aggregated metering, and cross-domain generalization. Finally, the thesis extends disaggregation to a probabilistic setting using generative artificial intelligence. A conditional diffusion-based framework is proposed to model the distribution of BTM PV generation conditioned on net consumption and exogenous variables. The methodology provides calibrated and sharp probabilistic estimates and is further extended to jointly disaggregate multiple DERs, such as heat pumps. Validation on real-world datasets demonstrates reliable uncertainty quantification and strong generalization under limited training data.
  

DOCTORAL DEGREE IN ENVIRONMENTAL ENGINEERING

• CANO LARROTTA, ANA MARIA JOSE CANDELARIA: VALORISATION OF BIOSOLIDS FROM SLUDGE TREATMENT WETLANDS FOR AGRICULTURAL USE
  
  Author: CANO LARROTTA, ANA MARIA JOSE CANDELARIA
  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 ENVIRONMENTAL ENGINEERING
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Normal
  Deposit date: 13/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: UGGETTI, ENRICA
  Thesis abstract: Sludge treatment wetlands (STWs) are nature-based systems increasingly used to treat and dewater sewage sludge, particularly in small and medium-sized wastewater treatment plants. Through alternating feeding and resting periods, these systems promote natural dewatering, organic matter stabilisation, and partial attenuation of contaminants, producing a stabilised by-product commonly referred to as biosolids. Despite the growing use of STWs, the agricultural use of the biosolids they generate remains poorly understood, largely due to the limited number of agronomic studies and persistent concerns about contaminants.In this context, this PhD thesis examined the current state of knowledge on the agricultural use of biosolids derived from STWs, identifying a clear knowledge gap compared with biosolids produced by conventional sludge treatment technologies. This gap is largely attributed to uncertainties about biosolid quality, potential contaminant-related risks, and the lack of experimental and economic evidence supporting their use in agricultural systems.The main objective of this PhD thesis was to assess the feasibility of valorising biosolids from STWs for agricultural applications by analysing their stabilisation process, resulting quality, agronomic functionality, and economic implications. A key aspect addressed was the evaluation of the final resting period required before biosolids can be safely removed from STWs. Biosolids characterised after different resting periods showed that an adequate stabilisation phase is essential, with a resting period of approximately six months after the final feeding cycle under Mediterranean conditions. After this period, biosolids reached an average dewatering degree of around 51% total solids and an organic matter stabilisation of approximately 41% (TS/VS), indicating suitability for agricultural use.The agronomic valorisation of the biosolids was subsequently investigated. As soil amendments, STW-derived biosolids significantly increased total soil organic carbon and total nitrogen reserves, reaching up to 18.1 Mg·ha⁻¹ and 1.8 Mg·ha⁻¹, respectively. They also increased the stable and recalcitrant organic carbon fractions by up to 46% and 57%, respectively, and improved soil biological fertility, with microbial biomass carbon rising by 76% relative to the unamended control. In crop trials with lettuce, radish, and ryegrass, biosolids exhibited fertilising effects comparable to those of commonly used biofertilisers. In radish, biosolid application increased fresh and dry biomass by 93% and 32%, respectively, while protein content exceeded the control by 6%, 9%, and 17% in lettuce, ryegrass, and radish, respectively. Nitrogen losses through leaching were also reduced compared with urea-based fertilisation.Finally, a quantitative techno-economic analysis compared STWs with centrifugation over a 20-year assessment period. Although STWs required higher initial capital investment, their operating costs were substantially lower, with operating expenditures approximately 4.6–6.6 times lower than those of centrifuge systems. An additional economic assessment of fertilisation strategies showed that combining biosolids with inorganic fertilisers reduced fertilisation costs by approximately 29%–116% compared with conventional mineral fertilisation, even under conservative phosphorus-limited scenarios.Overall, this thesis demonstrates that biosolids derived from sludge treatment wetlands can be effectively and sustainably valorised in agriculture when appropriate stabilisation periods and management practices are applied, supporting their integration into circular and sustainable wastewater and agricultural management strategies
  
• CORREA GONZÁLEZ, SANDRA: Anaerobic digestion-based biorefineries to advance circularity in the olive oil sector
  
  Author: CORREA GONZÁLEZ, SANDRA
  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 ENVIRONMENTAL ENGINEERING
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Normal
  Deposit date: 27/02/2026
  Reading date: 21/04/2026
  Reading time: 10:30
  Reading place: Place: ETSECCPBUPC, Campus NordBuilding C1. Classroom: 002C/Jordi Girona, 1-308034 Barcelona
  Thesis director: FERRER MARTI, IVET | PASSOS LOPES, FABIANA
  Thesis abstract: Every year, the olive oil industry generates large quantities of olive pomace, a by-product that currently represents an environmental problem but has great potential for the recovery of bio-based products and bioenergy within a biorefinery framework.In this context, the objective of this PhD thesis was to develop and evaluate anaerobic digestion–based biorefinery strategies for the valorisation of olive pomace. Specifically, the limitations of olive pomace mono-digestion were assessed, and co-digestion was investigated as a strategy to improve process stability and methane production. Biorefinery models for the recovery of bio-based products and biogas were examined, the agronomic potential of the resulting digestates was analysed, and last anaerobic fermentation was explored as an alternative valorisation pathway.The results demonstrated that olive pomace mono-digestion is unstable due to substrate-related properties, including acidic pH, lack of alkalinity and nutrients, high C/N ratio, and the presence of phenolic compounds. These limitations led to the accumulation of volatile fatty acids, inhibition of methanogenesis, and collapse of the microbial community, resulting in very low methane yields.Co-digestion with nitrogen-rich co-substrates, specifically from the swine farming sector, proved to be an effective strategy to overcome these limitations. In particular, pig slurry provided alkalinity, nutrients, and water, enabling stable operation, increased microbial diversity, and enhanced methane yields (145 mL CH₄·g⁻¹ VS). In contrast, co-digestion with pig manure achieved higher methane yields (289 mL CH₄·g⁻¹ VS) but exhibited lower resilience under stress conditions.Two anaerobic digestion–based biorefinery models were developed in this thesis. In the first, an ionic liquid pre-treatment ([Et₃NH][HSO₄], 120 °C, 1 h) was applied to recover lignin nanoparticles from olive pomace, followed by anaerobic digestion of the residual fraction. Analysis of the residual olive pomace revealed a less compact surface structure and greater bioaccessibility for microorganisms, resulting in higher methane yields during co-digestion with pig slurry compared to untreated olive pomace (173 mL CH₄·g⁻¹ VS). In the second biorefinery model, a thermal process (water, 100 °C, 45 min) was used to recover natural dyes for textile applications. The residual olive pomace fraction retained 88% of the initial methane potential. During semi-continuous reactor operation, co-digestion of this fraction with pig slurry achieved a methane yield of 157 mL CH₄·g⁻¹ VS.Digestates derived from the co-digestion of untreated and residual olive pomace with pig slurry exhibited favourable agronomic properties, such as near-neutral pH, balanced nutrient content, and partially stabilised organic matter. However, phytotoxicity at high application rates and elevated Hg concentrations highlight the limitations of these digestates and the need to blend them with other organic fertilisers to ensure safe soil application.Anaerobic fermentation was identified as an alternative or complementary pathway for olive pomace valorisation. Temperature and hydraulic retention time strongly influenced volatile fatty acid yields and profiles, as well as microbial community composition.Overall, this thesis demonstrates the potential of anaerobic digestion–based biorefineries for the valorisation of olive pomace and the production of value-added bio-based products, renewable energy, and organic fertilisers, contributing to the development of circular and resilient agro-industrial systems.
  
• SOUSSE VILLA, RUBEN: Dust in the atmosphere: Integrated modeling of heterogeneous chemistry, mineralogy, and optical properties
  
  Author: SOUSSE VILLA, RUBEN
  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 ENVIRONMENTAL ENGINEERING
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Normal
  Deposit date: 16/02/2026
  Reading date: 29/04/2026
  Reading time: 12:00
  Reading place: Place: ETSECCPBUPC, Campus NordBuilding C1. Classroom: 002C/Jordi Girona, 1-308034 Barcelona
  Thesis director: PEREZ GARCIA-PANDO, CARLOS | JORBA CASELLAS, ORIOL
  Thesis abstract: Airborne mineral dust is one of the dominant aerosol components in the atmosphere and plays a central role in the Earth system through its interactions with atmospheric composition and radiation. Despite its importance, large uncertainties remain in its representation in atmospheric and climate models, particularly concerning dust heterogeneous chemistry. These reactions between dust particles and gas species play a dominant role over dust affected regions in the formation of secondary inorganic aerosols such as particulate nitrate, ammonium, and sulfate. Yet, they are still subject to strong simplifications and parameterizations in atmospheric models, which still show significant discrepancies in their representation.This Thesis aims to advance the understanding of dust atmospheric chemical processes, their dependence on mineralogical composition, and their implications for aerosol optical properties. To this end, multiple dust heterogeneous chemistry mechanisms and mineralogical representations are implemented and systematically evaluated within the MONARCH atmospheric chemistry model. Their impacts on atmospheric composition, aerosol acidity, and optical properties are assessed and evaluated against observations. Notably, this work represents the first integration into an atmospheric chemistry model of observationally constrained maps of surface mineral composition of arid surfaces derived from the recently launched NASA’s Earth Surface Mineral Dust Source Investigation (EMIT) spectrometer.The evaluation of different dust heterogeneous chemistry mechanisms highlights the importance of representing reversible secondary inorganic aerosol formation in both fine and coarse particle modes, together with an accurate treatment of dust and sea-salt alkalinity. This combined approach is shown to be necessary to achieve consistent nitrogen budgets and aerosol acidity, and to better reproduce observed atmospheric nitrate concentrations.The comparison of different mineral atlases focuses on calcite as the primary contributor to dust alkalinity, demonstrating that EMIT provides the most consistent representation of the atmospheric calcite cycle. The use of source-resolved mineralogy compared to the common assumption of globally averaged dust mineral content, reveals strong geographical contrasts in calcite abundance, particularly between the Middle East and other major dust regions such as the Sahara and East Asia. These differences in calcite lead to marked regional and size-dependent implications for aerosol acidity and secondary inorganic aerosol formation. Impacts on particulate nitrate are found to be most pronounced during short-lived, intense formation events in regions affected by both dust and anthropogenic pollution, while global burdens and long-term trends are less sensitive.The implications of dust chemistry and mineralogical representation for aerosol optical depth and single scattering albedo are generally modest, with fine-mode aerosols exerting a larger influence than coarse particles, though both remain secondary compared to strongly absorbing minerals such as iron oxides. Limitations related to the representation of absorbing carbonaceous aerosols and the absence of cloud–aerosol interactions constrain the assessment of radiative effects.Overall, this Thesis demonstrates that combining reversible dust heterogeneous chemistry with source-resolved mineralogical information from EMIT substantially improves the representation of dust–gas interactions and their impacts on atmospheric composition and aerosol optics. Remaining uncertainties highlight the need for continued development of mineralogical datasets and chemical parameterizations, as well as the inclusion of anthropogenic dust sources and additional reactive gas species. Further progress would also benefit from high-resolution modeling of small-scale dust events and from expanded observational coverage over dust-affected urban regions.
  

DOCTORAL DEGREE IN GEOTECHNICAL ENGINEERING

• PINEDO VILCAHUAMÁN, PAUL JOSÉ: Simulation of blast-induced liquefaction with an unconventional elastoplasticity model
  
  Author: PINEDO VILCAHUAMÁN, PAUL JOSÉ
  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 GEOTECHNICAL ENGINEERING
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Normal
  Deposit date: 19/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: ARROYO ALVAREZ DE TOLEDO, MARCOS | GENS SOLE, ANTONIO
  Thesis abstract: Soil liquefaction is a phenomenon that typically occurs following earthquakes, although it may also be triggered by non-seismic loading. During liquefaction, a substantial and rapid reduction in the strength and stiffness of soils occurs due to pore pressure generation. Soil liquefaction is systematically investigated in the laboratory using element and model tests; however, certain aspects that prove critical in field liquefaction (e.g., soil fabric, dynamic loading, drainage conditions and ageing effects) are challenging to reproduce in the laboratory. This has prompted researchers to investigate liquefaction directly in the field by means of blast tests, i.e. experiments in which liquefaction is induced by controlled explosive detonations.Numerical simulation is nowadays frequently employed to simulate the onset and development of soil liquefaction. In this context, the primary objective of the present study is the numerical simulation of a controlled blast-induced liquefaction test conducted at a silty sand site in Bondeno, Ferrara, Italy. To this end, a constitutive model named CASM-SM has been implemented to reproduce soil behaviour during cyclic loading, having been recast within a subloading plasticity framework. The modelling of soil behaviour under blast loading is challenging due to several factors, including the characteristics of soil deformation close to the explosive, soil stiffness degradation, high frequencies in acceleration records (exceeding 100 Hz), and a significant increase of excess pore water pressure over a short timeframe (on the order of milliseconds). The Bondeno blast test simulation served as an excellent platform for validating the performance of the CASM-SM model. Particular attention was paid to the reproduction of the acceleration records, as well as velocities, Arias intensity, Fourier spectra, and spectral acceleration. Moreover, excess pore water pressures were compared against pore pressure transducer data. The results obtained from the numerical simulation provide relevant insights into blast-induced liquefaction within the silty sand layer.
  

DOCTORAL DEGREE IN MECHANICAL, FLUIDS AND AEROSPACE ENGINEERING

• KHOSRAVI, HAMID: Enhancing microfluidic and electrochemical sensors for biological and environmental analysis
  
  Author: KHOSRAVI, HAMID
  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 MECHANICAL, FLUIDS AND AEROSPACE ENGINEERING
  Department: Department of Mechanical Engineering (EM)
  Mode: Normal
  Deposit date: 27/02/2026
  Reading date: 13/04/2026
  Reading time: 15:30
  Reading place: Sala de conferències del TR5, ESEIAAT.
  Thesis director: CASALS TERRE, JASMINA
  Thesis abstract: The transition toward low-cost, portable, and environmentally conscious analytical technologies has intensified the pursuit of sustainable alternatives to conventional laboratory instrumentation. This thesis develops paper-based and electrochemical sensing platforms that prioritize circular-economy principles by employing renewable and waste-derived materials. Non-wood cellulose fibers were selected as substrates for microfluidic paper-based analytical devices (μPADs), while industrial mill scale was valorized to synthesize magnetite nanoparticles for electrode modification, demonstrating that sustainability and high analytical performance can be synergistic.In the first study, μPADs fabricated from alternative cellulose sources were evaluated. Their fiber morphology and porosity strongly influenced capillary flow and colorimetric responses. Compared to commercial cellulose papers, non-wood substrates enabled substantially faster wicking and significantly reduced detection time, underscoring their suitability for rapid, low-resource diagnostics.The second study focused on lactate detection using magnetite-modified electrodes. Wastederived Fe₃O₄ nanoparticles enhanced electron transfer and enzyme immobilization, enabling an exceptionally broad detection range alongside high sensitivity and a low detection limit. To the best of our knowledge, this work represents the first demonstration of a lactate biosensing platform that simultaneously achieves such a wide dynamic range while retaining high analytical sensitivity, making it suitable for applications from trace physiological monitoring to highly concentrated food and fermentation environments.Finally, a novel electrochemical strategy was developed for polyethylene terephthalate (PET) microplastic quantification in water. Leveraging the natural affinity between PET and magnetite nanoparticles, the approach transitions from the traditional use of magnetite for magnetic pre-concentration toward direct and quantitative electrochemical measurement, successfully validated in synthetic and real water matrices.Overall, this thesis demonstrates that renewable and waste-derived materials from non-wood cellulose to mill-scale-derived magnetite can serve as functional components in advanced sensing platforms, advancing sustainable analytical technologies for biomedical and environmental applications.
  
• SABÁN FOSCH, ALEJANDRO: End-to-end design and development of an Autonomous Flight Safety System enabling reusable space missions in Europe
  
  Author: SABÁN FOSCH, 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 MECHANICAL, FLUIDS AND AEROSPACE ENGINEERING
  Department: Department of Mechanical Engineering (EM)
  Mode: Normal
  Deposit date: 05/03/2026
  Reading date: 14/04/2026
  Reading time: 10:00
  Reading place: ESEIAAT - Aula 3.6
  Thesis director: SORIA GUERRERO, MANUEL | DIEZ LLEDO, EDUARD | SUREDA ANFRES, MIQUEL
  Thesis abstract: This dissertation presents the design, development, and validation of an Autonomous Flight Safety System (AFSS) tailored to the operational and regulatory needs of reusable launch vehicles in Europe. Motivated by the shift from expendable rockets to reusable systems and the consequent need for autonomous range safety, the research situates itself at the intersection of technology, safety assurance, and certification.A requirement-driven approach, grounded in European and international regulations, ensured alignment with certification pathways. A review of current FSS and regulatory frameworks established the baseline from which requirements were derived. These were structured through a model-based systems engineering methodology, implemented in ARCADIA and SysML, guiding functional decomposition and definition of a three-layered architecture. The AFSS design comprises four application modules: navigation, flight dynamics assessment, decision-making, and Integrated Vehicle Health Management (IVHM).Each module was independently implemented and validated. The navigation subsystem met outage-handling requirements, reliably bridging data gaps. The flight dynamics assessment integrated 3D flight corridor checks, and impact prediction with aerodynamic effects and dispersion evaluation at low operational cost. For reusable launchers, the IVHM subsystem is essential, as safe operation requires monitoring systems for re-entry. This module classified anomalies accurately, highlighting the trade-off between expert-tuned and data-driven approaches due to sensitivity to membership function parametrisation. The decision-making logic consistently executed termination rules under nominal and degraded conditions, confirming robustness.A RAMS (Reliability, Availability, Maintainability, and Safety) analysis critically assessed maturity. Navigation and decision-making were identified as the most safety-critical functions, with redundancy mitigating risks but leaving common-mode vulnerabilities. Prototype hardware (HW) was selected according to Technology Readiness Level (TRL) criteria, suitable for ground validation at TRL 7 system level. This reflected a focus on validating software and architecture, while dedicated space-qualified HW -required for certification under harsher conditions such as radiation and vibration- lay beyond scope.Integration testing guaranteed the correctness of the AFSS prototype before the ground campaign at the Kiruna spaceport to achieve TRL 7, a milestone in European AFSS development. The prototype demonstrated coherent behaviour across processors, reliable synchronization between redundant chains, and real-time telemetry from the Real-time target machine. Although processing loads neared the limits of the selected low-end HW, it met its main objective: validating the complete AFSS software chain. Nonetheless, borderline safety decisions under certain conditions showed that resilience depends on algorithmic choices, parametrisation, and execution margins.The research shows AFSS architectures are technically feasible, regulation-aware, and progressing towards operational use, though challenges remain. Future work should address processor scalability with multi-core, space-qualified platforms; enhance navigation robustness against GNSS jamming and spoofing; extend IVHM towards prognostics; and evaluate navigation architectures (IMU-only versus integrated IMU/GNSS) once launcher avionics are defined. Equally critical is institutional progress: certifying AFSS will require new regulatory frameworks and joint experimental programmes aligning technical validation with policy evolution.By combining regulatory awareness, rigorous engineering, and validation to TRL 7, this dissertation contributes not only a prototype but also a roadmap. It demonstrates feasibility while clarifying remaining challenges, providing a foundation for the safe deployment of autonomous flight safety in Europe's reusable launchers.
  

DOCTORAL DEGREE IN NATURAL RESOURCES AND THE ENVIRONMENT

• MIRFALLAH LIALESTANI, SEYED POORYA: Geothermal Potential Assessment in Catalonia, Spain: Innovative AI-Powered 3D Mapping using Advanced Hybrid Multitask and Semi-Supervised Learning Frameworks
  
  Author: MIRFALLAH LIALESTANI, SEYED POORYA
  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 NATURAL RESOURCES AND THE ENVIRONMENT
  Department: Department of Mining, Industrial and ICT Engineering (EMIT)
  Mode: Normal
  Deposit date: 12/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: HIMI BENOMAR, MAHJOUB
  Thesis abstract: This dissertation introduces an innovative interdisciplinary framework that merges artificial intelligence with metaheuristic optimization to advance geothermal energy exploration in Catalonia, Spain. By integrating machine learning techniques such as deep learning, neural networks, and decision trees with advanced optimization methods including the Firefly Algorithm and a newly developed Modified Multi-Objective Bat Algorithm (MoBA), the study significantly improves subsurface temperature prediction accuracy, reduces exploration costs, and enhances geothermal resource assessment and management.The research generates high-resolution 3D geothermal maps at depths of 50, 120, 150, and 180 m, across an extensive area of approximately 7,942 km². The MoBA demonstrated exceptional performance in forecasting subsurface temperatures, outperforming conventional methods by adapting to complex and variable geological conditions, minimizing manual intervention, and balancing multiple optimization objectives. This approach enabled the identification of high-potential zones, particularly in Vallès and Penedès, La Selva, Empordà, Fossa d’Olot, Plana de Vic, and the Central Depression.Key findings reveal that geothermal potential increases significantly with depth, with average estimated capacities rising from 15,408.5 MWh/year at 50 m to 19,260.6 MWh/year at 150 m. Approximately 24% of surveyed locations were classified as “very good,” indicating strong suitability for small to medium-scale geothermal systems. The results also confirmed a weak correlation between thermal conductivity and geothermal gradient (r² = 0.02), highlighting the geological heterogeneity of the region and the importance of site-specific evaluations.The AI-driven models aligned closely with prior research (e.g., Colmenar-Santos et al., 2016), validating the reliability of this approach while offering greater spatial detail and resolution. MoBA-based predictions and spatial distribution maps offer actionable insights for policymakers and energy planners, identifying zones suitable for district heating, hybrid systems, and decentralized renewable energy integration.Despite challenges such as data availability and computational demands, the integration of AI with metaheuristic algorithms presents a scalable, cost-effective, and precise method for geothermal exploration. This work not only contributes a powerful tool for sustainable resource exploitation in Catalonia but also provides a transferable model for enhancing renewable energy strategies in other geologically diverse regions.
  

DOCTORAL DEGREE IN PHOTONICS

• PUJOL CLOSA, MARIA DEL PILAR: Wave Propagation in Hyperbolic Metamaterial Waveguides
  
  Author: PUJOL CLOSA, MARIA DEL PILAR
  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 PHOTONICS
  Department: Institute of Photonic Sciences (ICFO)
  Mode: Normal
  Deposit date: 11/03/2026
  Reading date: 15/04/2026
  Reading time: 10:00
  Reading place: ICFO Auditorium
  Thesis director: ARTIGAS GARCIA, DAVID
  Thesis abstract: Low-loss waveguides are essential for energy-efficient photonic circuits, optical communications, and sensing applications. Over the past century, two lossless phenomena—Dyakonov modes and Bound States in the Continuum (BICs)—have been discovered in anisotropic waveguides, where permittivities differ but share the same sign. Hyperbolic metamaterials (HMMs) exhibit extreme anisotropy, with ordinary and extraordinary permittivities of opposite signs, enabling unconventional light manipulation. Their unique properties have attracted broad interest for applications including subdiffraction imaging, spontaneous emission control, and enhanced light-matter interactions. This raises a fundamental question: can extreme hyperbolic anisotropy support novel confinement mechanisms or new regimes of lossless propagation? Prior research on HMM waveguides has been constrained to simplified models or propagation along principal axes, leaving systematic exploration of arbitrary propagation directions, and the phenomena they may reveal, as a critical gap.To address this gap, this thesis develops a semi-analytical computational framework that combines a transfer-matrix formulation with a complex-plane Newton-Raphson root finder, enabling stable tracking of guided and leaky modes for arbitrary propagation directions. This tool allows systematic exploration of a wide range of parameters and configurations previously difficult to study.This thesis provides the most comprehensive investigation to date of light propagation in planar HMM waveguides. For the first time, the work analyzes both type I and type II HMM waveguides across all in-plane propagation directions and with arbitrary optic axis orientations. The analysis reveals how hyperbolic anisotropy fundamentally influences polarization, confinement, polarization exchange between modes, mode ordering, radiation mechanisms, and slow light arising from topological transitions. This establishes general trends, identifies new guiding regimes, and maps the landscape of wave phenomena in these extreme anisotropic systems.The exploration of leaky modes enabled a key discovery: Dirac points embedded in the Continuum (DECs), a novel class of topological degeneracy in non-Hermitian systems. DECs emerge when a symmetry-protected BIC and an interferometric BIC intersect linearly. At this intersection, the system exhibits a real eigenvalue, two orthogonal modes, and zero radiation loss—locally Hermitian behavior despite being embedded in a non-Hermitian system. The presence of both BICs suppresses Exceptional Points (EPs) and collapses the Fermi arc to a single point. Because DECs arise from universal BIC interactions rather than material-specific properties, this phenomenon extends beyond hyperbolic media, with implications in the fields of topological photonics and non-Hermitian physics.This thesis demonstrates the framework’s generality and reliability through application to anisotropic liquid-crystal waveguides, where predicted BIC trajectories match experimental observations, and to $\sigma$-near-zero metasurfaces, where the framework accurately reproduces published dispersion diagrams. These validations confirm its applicability beyond hyperbolic systems.This thesis establishes a comprehensive theoretical and computational understanding of wave propagation in planar HMM waveguides for both type I and type II configurations and discovers DECs as a novel physical phenomenon with implications beyond hyperbolic media. By revealing how extreme anisotropy enables new guiding regimes and loss suppression, this work advances the understanding of light confinement in open, strongly anisotropic systems and provides new routes for designing low-loss photonic devices.
  

DOCTORAL DEGREE IN POLYMERS AND BIOPOLYMERS

• COLOMBI, SAMUELE: Soft and Conductive Material Architectures for Flexible Electronics: from Hydrogels to Nanomembranes
  
  Author: COLOMBI, SAMUELE
  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 POLYMERS AND BIOPOLYMERS
  Department: Department of Chemical Engineering (EQ)
  Mode: Normal
  Deposit date: 20/02/2026
  Reading date: 14/04/2026
  Reading time: 10:30
  Reading place: EEBE - UPCSala Polivalent, Ed. Ahttps://eebe.upc.edu/ca/lescola/com-arribar
  Thesis director: ALEMAN LLANSO, CARLOS ENRIQUE | GARCÍA TORRES, JOSÉ MANUEL
  Thesis abstract: Soft materials are key enablers of the next generation of flexible and multifunctional systems for applications spanning biomedicine, energy, and environmental technologies. This thesis focuses on the design, fabrication, and implementation of (nano)engineered polymeric materials with tailored properties for use in soft electronic devices, drug delivery systems, solar-driven water evaporators, and functional scaffolds for tissue engineering. To this end, three main families of materials were developed: composite alginate (Alg)-based hydrogels, composite gelatin methacrylate (GelMA)-based hydrogels, and polylactic acid (PLA)-based nanomembranes. For each material platform, structure–property–function relationships were systematically investigated across distinct biotechnological scenarios to achieve enhanced performance. Alg-based composite hydrogels were engineered as versatile, water-rich platforms through the incorporation of functional nanomaterials (such as PLA nanofibers and gold nanoparticles), enzymes, and conducting polymers (e.g., PEDOT:PSS). These systems were designed to exhibit improved mechanical robustness, controlled porosity, and tunable physicochemical and functional properties while maintaining their intrinsic biocompatibility. By modulating composition, the Alg-based hydrogels were successfully applied as drug delivery matrices for the sustained release of lactate, as soft electronic platforms for temperature and H₂O₂ sensing, and as flexible energy-storage devices. In addition, the introduction of secondary covalent crosslinking was explored as a strategy to enhance operational stability in H₂O₂ biosensing and solar-driven steam generation applications. In parallel, bilayered PANI–PLA nanomembranes incorporating aligned gold nanopillars were fabricated as free-standing, lightweight, conformable, and mechanically stable platforms for skin electronics. These nanomembranes enabled simultaneous pH and non-enzymatic NADH sensing, demonstrating their suitability for monitoring skin physiology and infection-related biomarkers. Finally, GelMA hydrogels were nanoengineered with magnetoelectric and/or graphene-based nanomaterials to develop bioactive scaffolds for cardiac tissue regeneration. The incorporation of these nanomaterials allowed precise tuning of the mechanical, electrical, magnetic, and biological properties of the hydrogels. Their ability to support cell adhesion and proliferation, combined with their capacity to sense cellular activity, highlights their potential as multifunctional scaffolds for engineered tissues and bioelectronic interfaces. Overall, this thesis demonstrates that rational materials design across multiple length scales enables the development of soft matter systems with tailored and synergistic functionalities, paving the way toward next-generation flexible electronic devices, controlled therapeutic platforms, and sustainable water-treatment technologies.
  

DOCTORAL DEGREE IN SIGNAL THEORY AND COMMUNICATIONS

• GORT JELMER DIRK, BEREND: AI-Driven Zero-Touch Orchestration of Edge-Cloud Services
  
  Author: GORT JELMER DIRK, BEREND
  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 SIGNAL THEORY AND COMMUNICATIONS
  Department: Department of Signal Theory and Communications (TSC)
  Mode: Normal
  Deposit date: 10/03/2026
  Reading date: 24/04/2026
  Reading time: 10:00
  Reading place: ETSETB B3 Teleensenyament, Campus Nord UPC, Barcelona
  Thesis director: ANTONOPOULOS, ANGELOS | UMBERT JULIANA, ANNA
  Thesis abstract: 6G networks demand orchestration systems capable of managing thousands of distributed microservices under sub-millisecond latency constraints. Traditional centralized approaches introduce unacceptable delays, create single points of failure in heterogeneous edge-cloud infrastructures, and require constant attention from human operators. This dissertation addresses three critical challenges: (1) computational constraints that prevent the deployment of predictive models on edge devices, (2) lack of generalization of models across diverse types of applications, and (3) lack of validated autonomous orchestration without human intervention.To address these challenges, this dissertation develops three complementary frameworks that combine lightweight machine learning, attention-based deep learning, and agentic artificial intelligence for zero-touch service management in distributed 6G edge-cloud environments.The first contribution, AERO (Adaptive Edge-cloud Resource Orchestration), addresses the challenge of running predictions on resource-constrained edge devices. Current transformer models require millions of parameters (e.g., Pathformer: 2.4M), making them impractical for edge deployment. AERO achieves competitive accuracy with only 599 parameters, making edge deployment feasible and reducing reliance on cloud round-trips when local inference is preferred. Evaluations demonstrate sub-millisecond inference (0.38ms), 13% energy savings, and 99% fewer SLA violations compared to reactive scheduling, which allocates resources only after demand changes occur.The second contribution, OmniFORE (Framework for Optimization of Resource Forecasts in Edge-cloud networks), addresses the operational challenge of maintaining separate models per application. A single OmniFORE model generalizes across heterogeneous workloads without retraining, replacing the need for dedicated per-application models. Cross-dataset evaluation on industry-standard benchmarks (Google and Alibaba production traces) demonstrates 30% better accuracy than ModernTCN while maintaining 15× faster inference than AGCRN.The third contribution, AgentEdge, addresses the challenge of agentic orchestration in distributed edge-cloud environments. Existing agent frameworks target generic domains or centralized cloud infrastructures, leaving distributed 6G environments without autonomous management solutions. AgentEdge introduces multi-agent orchestration to this domain, translating natural language intent (e.g., "deploy with low latency") into validated orchestration actions across heterogeneous infrastructure. Evaluations demonstrate 78.3% success rate (2.76× higher than single-agent baselines), 10× reduction in API call variability, and power savings up to 300.8W across deployments scaling from 8 to 35 nodes.The research has produced 5 journal publications, 3 international conference papers in IEEE venues, and 1 Elsevier book chapter.
  
• KASULURU, VAISHNAVI: AI-Driven Network Service Management for Efficient and Sustainable Open-RAN Systems in 6G and Beyond
  
  Author: KASULURU, VAISHNAVI
  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 SIGNAL THEORY AND COMMUNICATIONS
  Department: Department of Signal Theory and Communications (TSC)
  Mode: Normal
  Deposit date: 24/02/2026
  Reading date: 15/05/2026
  Reading time: 14:00
  Reading place: ETSETB B3 Teleensenyament, Campus Nord UPC, Barcelona
  Thesis director: BLANCO BOTANA, LUIS | ZEYDAN, ENGIN
  Thesis abstract: This thesis presents an efficient and sustainable AI-driven resource management framework for next-generation Open Radio Access Network (O-RAN) in the context of the emerging 6G era. The framework operates in a cloud-native 6G environment and translates predictive intelligence into reliable, energy-aware orchestration. It combines advanced predictive modeling with optimization-based control to address challenges, like stochastic demand, multitenancy, and computational complexity in O-RAN. The predictive forecasting architecture is the core of the framework, quantifying uncertainty and interdependencies among network resources across multiple tenants. Probabilistic forecasting models generate distributions of future resource demands, enabling service providers to perform more informed and risk-aware resource orchestration in complex multitenant environments.Initially, the framework considered in this thesis considers univariate probabilistic estimators, including Simple-Feed-Forward (SFF), Deep Autoregressive Recurrent network (DeepAR), and Transformers, to predict individual resource demands and support effective provisioning in O-RAN. These models deliver efficient, agile, and uncertainty-aware resource predictions, which are integrated into a novel percentile-based orchestration strategy, Dynamic Percentile Adjustment Approach (DYNp). The proposed method dynamically adjusts the percentage to ensure efficient resource utilization in O-RAN systems. Selecting an appropriate percentile is critical for balancing resource waste and service reliability. However, univariate probabilistic estimates do not capture cross-resource interdependencies, leading to suboptimal decision-making. To address this limitation, the framework incorporates state-of-the-art multivariate probabilistic forecasting models such as Gaussian Process Vector Autoregression (GPVAR) and the Temporal Fusion Transformer (TFT). They jointly process multiple time series and provide robust estimates of future resource demands. These models effectively learn complex interdependencies among different resources and key parameter indicators across network slices and tenants. Furthermore, we have evaluated how low-rank approximation in GPVAR estimator enhances scalability and robustness by reducing the algorithm's training time. One of the main goals of this thesis is to achieve energy efficiency and effective resource management and sharing. By considering predictive intelligence together with power consumption, the proposed techniques proactively optimize the activation and deactivation of radio resources or radio units. This strategy significantly reduces power consumption while maintaining user experience and adhering to Service Level Agreement (SLA) guarantees. Furthermore, another relevant contribution of the thesis is the extension of the traditional cellular O-RAN architecture to include Cell-Free massive Multiple-Input and Multiple-Output (CF-mMIMO) networks, reflecting the architectural evolution for beyond 6G systems. This provides a scalable approach to ultra-dense, energy-efficient O-RAN deployments. Finally, the algorithm tools considered in the dissertation are implemented as modular applications to facilitate deployment across O-RAN. The cloud-native implementation of the forecasting and orchestration pipeline is a notable achievement. Each module has been containerized using Docker, and its functionality is exposed via Representational State Transfer (REST) APIs, such as Swagger. This enables the pipeline to operate as microservices, supporting flexible deployment, scalable execution and seamless integration within O-RAN. The thesis establishes a mathematical and architectural foundation for deploying AI-driven, sustainable, and energy-optimized O-RAN with uncertainty adaptation. It provides a basis for realizing intelligent, autonomous, and stable 6G networks and supports future research and industrial implementation of AI-powered O-RAN ecosystems.
  

DOCTORAL DEGREE IN STATISTICS AND OPERATIONS RESEARCH

• BORJA ROBALINO, RICARDO STALIN: Optimización bayesiana en técnicas machine Learning clásicas: redes neuronales y XGBoost y su aplicación como modelos predictores de diabetes en pacientes ecuatorianos
  
  Author: BORJA ROBALINO, RICARDO STALIN
  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 STATISTICS AND OPERATIONS RESEARCH
  Department: Department of Statistics and Operations Research (EIO)
  Mode: Normal
  Deposit date: 22/01/2026
  Reading date: 10/04/2026
  Reading time: 11:00
  Reading place: Dia 10 d’ abril de 2026 a les 11h del matí a la Sala de Juntes de la FIB
  Thesis director: MONLEON GETINO, ANTONIO | GIBERT OLIVERAS, CARINA
  Thesis abstract: Machine learning (ML) is a branch of artificial intelligence that allows human capabilities to be imitated through various algorithms and techniques that learn from data using learning processes (supervised, unsupervised, or reinforcement) for decision-making with minimal human intervention. Classic ML models have generated great results in the automation of classification and regression processes in various areas. Within classification, artificial neural networks (ANN) have gained relevance due to their ability to learn and model complex nonlinear relationships. Similarly, the XGBoost model based on decision trees has demonstrated great efficiency, speed, scalability, and performance, winning several competitions. On the other hand, Bayesian inference has provided a probabilistic and revolutionary framework for optimizing machine learning models, with the implementation of uncertainty in the estimation process, combining evidence with prior beliefs, in order to reduce overfitting and improve predictions by adjusting parameters and hyperparameters.This research aims to optimize two classic machine learning techniques (artificial neural networks and XGBoost) for classification using Bayesian inference and to build a diabetes prevention model for the Ecuadorian population. The study begins with a theoretical and mathematical conceptualization of each algorithm, followed by an analysis of the various points of intervention, programming, and implementation of Bayesian models using Markov chain Monte Carlo (MCMC) estimation techniques and variational inference (VI), validation using public databases, implementation of a client-server system with multiple specialized backends, and, finally, the development of a real application as predictors of type 1, type 2, and gestational diabetes.As a result, a Bayesian model was implemented in artificial neural networks (ANN) at two inference points. The first adjusted the parameters at each backpropagation step; however, it presented itself as an option with a prohibitive computational overhead. As a second intervention, an adjustment was made to the activation function in the final layer, obtaining positive and computationally viable results. In the case of XGBoost, the predictions were adjusted at each boosting step before vectorization, demonstrating high predictive power in both the use of the MCMC technique and IV. Validation with the Pima Indians Diabetes database and the use of various distribution functions demonstrated the robustness and sensitivity of the implemented models, while generalization and consistency were verified through application to various databases. In all cases, results superior to or equal to those obtained using the traditional model were obtained, depending on the characteristics of the data.In addition, a web application (client-server) was implemented with Bayesian proposals, allowing users to interact with the models in an easy and intuitive way, with options for data loading, parameter configuration and probability distributions, estimation techniques (MCMC or IV), training-validation process or use of cross-validation, real-time results, and model download options. The application of the Bayesian proposal to a real case, such as the prediction of type 1, type 2, and gestational diabetes, with data from Ecuadorian patients, presented encouraging results (accuracy = 99.47%), becoming the first predictive model for the three types of diabetes at the regional and national level, confirming that the use of this approach is an excellent alternative for the optimization of machine learning models.
  
• ROGNON, PAUL JORIS DENIS: Improving variable selection properties by leveraging external data.
  
  Author: ROGNON, PAUL JORIS DENIS
  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 STATISTICS AND OPERATIONS RESEARCH
  Department: Department of Statistics and Operations Research (EIO)
  Mode: Normal
  Deposit date: 09/03/2026
  Reading date: 20/04/2026
  Reading time: 14:00
  Reading place: Sala d'actes de la FMECampus Diagonal Sud, Edifici U. C. Pau Gargallo, 14 08028 Barcelona
  Thesis director: ROSSELL RIBERA, DAVID | ZWIERNIK, PÌOTR
  Thesis abstract: Sparse high-dimensional signal recovery is only possible under certain conditions on the number of parameters, sample size, signal strength, and underlying sparsity. I show that leveraging external information, as possible with data integration or transfer learning, allows pushing these mathematical limits. Specifically, I consider external information-dependent l0 penalties and Bayesian variable selection methods, show that they attain model selection consistency under milder conditions than standard methods, and that they also attain faster model recovery rates. First, I obtain results for oracle-based penalties and prior inclusion probabilities that have access to perfect sparsity and signal strength information. Those results provide an understanding of how and when external information helps variable selection. They also provide a theoretical benchmark to evaluate practical non-oracle selection methods using external information. Subsequently, I propose data-based procedures grounded in empirical Bayes methods that leverage external information to ease variable selection and do not require an oracle. I derive their properties in the particular case where external information partitions the set of variables in blocks with potentially distinct characteristics. Finally, I discuss a computational framework for the incorporation of external information in Bayesian variable selection through empirical Bayes in the general case.
  

DOCTORAL DEGREE IN STRUCTURAL ANALYSIS

• BORGES CAVALCANTI, DANILO: Finite Element Method with Embedded Strong Discontinuities for Coupled Hydro-Mechanical Problems
  
  Author: BORGES CAVALCANTI, DANILO
  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: Department of Civil and Environmental Engineering (DECA)
  Mode: Change of supervisor + Article-based thesis
  Deposit date: 09/03/2026
  Reading date: 20/04/2026
  Reading time: 14:00
  Reading place: Sala Zienkiewich (CIMNE) Building C1, UPC - Campus North Gran Capitan S/N 08034 Barcelona
  Thesis director: DE POUPLANA SARDÀ, IGNASI | CAMPOS RAMOS MARTHA, LUIZ FERNANDO | DE MESQUITA ROEHL, DEANE
  Thesis abstract: Coupled hydro–mechanical (HM) processes in fractured porous media govern the performance and safety of several subsurface engineering applications, where pressure-driven changes in stress and permeability can control injectivity, leakage pathways, and fault reactivation potential. This thesis develops a robust and versatile finite element formulation for transient HM problems in the presence of pre-existing strong discontinuities that remains practical for integration into standard finite element workflows. The proposed approach is formulated within the Embedded Finite Element Method (E-FEM) and grounded on the Strong Discontinuity Approach (SDA), enabling an implicit representation of fractures and faults while circumventing mesh conformity constraints. A unified description is introduced to model discontinuities acting either as preferential flow paths or as hydraulic barriers, capturing the longitudinal flow along the discontinuity and the transversal exchange with the porous matrix in steady-state and transient settings. The formulation is systematically verified against discrete fracture models with interface elements and applied to benchmark problems representative of fractured-reservoir conditions, including a coupled fault reactivation scenario. In addition, the thesis investigates the occurrence of spurious oscillations in cohesive traction fields along embedded discontinuities and demonstrates that the choice of an SDA-based embedded formulation can markedly improve traction smoothness. These improvements strengthen the use of E-FEM for HM assessments involving pre-existing fractures and faults.
  

DOCTORAL DEGREE IN THEORY AND HISTORY OF ARCHITECTURE

• DE SANTIAGO GARCIA, JAVIER NICOLAS: La imagen poética en los procesos de producción, proyectuales y habitar de la vivienda de autoproducción en Lomas del Centinela.
  
  Author: DE SANTIAGO GARCIA, JAVIER NICOLAS
  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 THEORY AND HISTORY OF ARCHITECTURE
  Department: Department of History and Theory of Architecture and Communication Techniques (THATC)
  Mode: Normal
  Deposit date: 09/02/2026
  Reading date: 08/04/2026
  Reading time: 16:00
  Reading place: ETSAB (Esc.Técn.Sup.Arquitect.Bcn)-Sala GradosAv. Diagonal, 649-BCNVideoconfer.: https://meet.google.com/qfv-wpcj-hnoInicio conexión: 15:30h local
  Thesis director: USANDIZAGA CALPARSORO, MIGUEL M. | SERRA PERMANYER, MARTA
  Thesis abstract: This thesis addresses the search for the poetic image in self-built housing in the Lomas del Centinela neighborhood, in Zapopan, Jalisco, Mexico. This search focuses on some of the processes identified with self-built housing, particularly in the production processes, design processes, and the dwelling process.The concept of poetic image is constructed through the discourses of authors such as Gaston Bachelard, Carlos González y Lobo, Steven Holl, Manuel Martín Hernández, Juhanni Pallasmaa, and Alberto Pérez-Gómez. It can be defined as the element that results from having imagined something with a specific purpose in mind. It is referred to as an element because it can manifest as a thought, a written text, a goal, a conceptual drawing, a method to achieve an end, a way of experiencing a place, of appropriating it, and of recognizing oneself in that journey.This study examines women from Lomas del Centinela who have self-built their homes, approaching their history and the forms of production they engaged in during the years they, along with their families, pursued the dream of building their own homes.During this research, they were interviewed, seeking in their discourses elements that could be considered poetic images, with the aim of exploring their self-building processes through the lens of the poetic image. Through this search, not only is the importance of imagination emphasized, but also the fact that self-building processes are not spontaneous. Rather, in much of their conception, construction, and inhabitation, they respond—beyond the management of minimal resources—to a strong imaginative component. This entails a profound reflection on the potential of self-built housing in terms of creative achievement.
  
• ZIAIEBIGDELI, MOHAMMADAMIN: Computing quality in housing: examining the evolution and systematization in computational design in architectural plan design and analysis
  
  Author: ZIAIEBIGDELI, MOHAMMADAMIN
  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 THEORY AND HISTORY OF ARCHITECTURE
  Department: Department of History and Theory of Architecture and Communication Techniques (THATC)
  Mode: Normal
  Deposit date: 19/02/2026
  Reading date: 10/04/2026
  Reading time: 12:00
  Reading place: ETSAB (Esc. Técn. Sup. Arquitectura Bcn)-Sala GradosAv. Diagonal, 649-BCNVideoconfer.: https://meet.google.com/uoh-cdjb-mdpInicio conexión: 11:30h
  Thesis director: ROSSELLO NICOLAU, MARÍA ISABEL | HERNÁNDEZ FALAGÁN, DAVID
  Thesis abstract: This research undertakes a meticulous exploration into the systematization and analysis of architectural plans, tracing their evolution through the lens of computational design thinking from the 20th century to the present. The specific context is living spaces, where the quality and functionality are intrinsically tied to the systematic extraction and application of design parameters. This dissertation explores the evolution of computational thinking in architecture by bridging the theoretical paradigms introduced by pioneers Alexander Klein, Cedric Price, Christopher Alexander, and Nicholas Negroponte with contemporary design practices. Through an inductive methodology, three case studies are analyzed to demonstrate how computational tools and technologies enhance spatial quality, adaptability, and user engagement. Each case study corresponds to the key indicators established by these pioneers—spatial geometry, temporal adaptability, participatory design, and digital computation—illustrating their enduring influence on modern architecture. By mapping these foundational ideas to advanced practices such as parametric modeling, smart systems, and digital fabrication, this research highlights the practical relevance of these paradigms in addressing contemporary architectural challenges and shaping innovative, human-centered spaces.The journey begins with Alexander Klein’s Existenzminimum concept in the 1920s-1930s. His scientific approach to design emphasized space optimization and the systematization of plans using both qualitative and quantitative parameters, laying the groundwork for future computational design methods. As we move into the 1960s, we encounter Christopher Alexander’s ‘Pattern Language.’ Alexander brought a new perspective to plan systematization, employing rules and systems to extract design parameters and analyze architectural planning. His methodologies form a vital cornerstone of computational design as we know it today.In the 1970s, Cedric Price introduced a more dynamic approach. His vision of architecture, as exemplified in the 24 hour living toy project, accommodated changing user needs and behaviors. His focus on user-centric parameters marked a shift in computational design thinking towards adaptability and flexibility. Presently, computational design thinking in the context of living spaces aims to synthesize the lessons from these key figures, emphasizing the analysis and extraction of design parameters. These parameters, enriched by the advancements in technology and the increasingly user-centric approach, are utilized to optimize and improve the quality of living spaces This research’s trajectory, from past methodologies to contemporary practices, provides an in-depth understanding of how computational thinking and plan analysis have evolved. By extracting and studying the diverse parameters used across time, this research aims to enrich the computational design discourse and advance the design of living spaces.
  

NAUTICAL ENGINEERING, MARINE AND NAVAL RADIOELECTRONICS

• ORTEGA OLIVA, NARCISO: FRONTEX Y LA FRONTERA MEDITERRÁNEA: UN NUEVO MODELO PARA RECUPERAR LA HUMANIDAD EN EL MAR
  
  Author: ORTEGA OLIVA, NARCISO
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: NAUTICAL ENGINEERING, MARINE AND NAVAL RADIOELECTRONICS
  Department: Department of Nautical Sciences and Engineering (CEN)
  Mode: Normal
  Deposit date: 13/03/2026
  Reading date: pending
  Reading time: pending
  Reading place: pending
  Thesis director: RODRIGO DE LARRUCEA, JAIME
  Thesis abstract: The report provides a critical and reasoned analysis of the evolution of FRONTEX and European border management, combining legal, political, sociological and operational aspects. It uses a robust methodology with empirical data, fieldwork, international comparative analyses, advanced statistical tools such as multiple correspondence analysis and Chi-square tests, as well as graphs and tables that facilitate understanding of the results. It identifies structural problems such as covert militarisation, the externalisation of responsibilities and the crisis of legitimacy of FRONTEX, and proposes concrete and viable solutions for structural reform towards a more ethical, transparent and human rights-friendly model. It aims to contribute to the academic and legal debate on European border management with scientific rigour.
  

DOCTORAL DEGREE IN ARCHITECTURAL, BUILDING CONSTRUCTION AND URBANISM TECHNOLOGY

• JAMSHIDZADEH, AMIN: Multiobjective optimization and parametric design strategies for steel modular structures in high-rise buildings: a computational approach
  
  Author: JAMSHIDZADEH, AMIN
  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 ARCHITECTURAL, BUILDING CONSTRUCTION AND URBANISM TECHNOLOGY
  Department: Department of Architectural Technology (TA)
  Mode: Normal
  Deposit date: 25/03/2026
  Deposit END date: 09/04/2026
  Thesis director: ALBAREDA VALLS, ALBERT
  Thesis abstract: The rapid urbanization and increasing demand for high-rise buildings have necessitated advancements in construction methods to meet the challenges of vertical urban development. Modular construction emerges as a promising solution, offering efficiency, speed, and sustainability in high-rise building projects. However, structural integrity remains a critical concern, particularly in tall structures where lateral stability is paramount. This research investigates the optimization of bracing frame analysis in modular high-rise buildings, aiming to enhance structural performance while maintaining efficiency and cost-effectiveness.The study begins with a comprehensive review of modular construction techniques, theoretical frameworks in high-rise building design, and structural analysis methods for bracing frames. Theoretical frameworks such as systems theory, prefabrication theory, and lean construction principles provide the conceptual basis for understanding the complexities of high-rise modular construction and the optimization of structural systems.Methodologically, the research adopts a multi-objective optimization approach, leveraging the NSGA-II algorithm to explore the trade-offs between conflicting objectives in bracing frame design, such as structural stability, material efficiency, and construction cost. Case studies of selected high-rise modular building projects are analyzed to validate the optimization framework and evaluate its effectiveness in improving structural performance.The findings reveal significant insights into the integration of bracing frame systems in modular high-rise buildings, highlighting the potential for optimization to enhance structural efficiency and resilience. The application of the NSGA-II algorithm enables the identification of Pareto-optimal solutions, offering designers and engineers a range of optimized design alternatives to consider during the planning and construction phases.In conclusion, this research contributes to advancing the knowledge and practice of modular construction in high-rise buildings by proposing an innovative optimization framework for bracing frame analysis. The findings have implications for improving structural design efficiency, reducing construction costs, and enhancing the sustainability of tall modular structures in urban environments. Further research directions are suggested to explore the scalability and applicability of the optimization framework in real-world construction projects.
  
• MARTIN FANJUL, VALENTIN: Arquitectura y autismo: consideraciones acústicas para el desarrollo de espacios específicos
  
  Author: MARTIN FANJUL, VALENTIN
  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 ARCHITECTURAL, BUILDING CONSTRUCTION AND URBANISM TECHNOLOGY
  Department: Department of Architectural Technology (TA)
  Mode: Normal
  Deposit date: 25/03/2026
  Deposit END date: 09/04/2026
  Thesis director: LACASTA PALACIO, ANA MARIA | DAUMAL DOMENECH, FRANCESC DE PAULA
  Thesis abstract: This doctoral thesis combines acoustics, autism and architecture with an unequivocal practical focus, so that the conclusions and possible future avenues of research it may generate can be used in specific spaces for people on the autism spectrum.Existing knowledge is recent and scarce, it is mainly based on the experience of technicians who have created specific spaces, whether newly built or adapted. This fact is further emphasised when the disorder itself adds the concept of spectrum, which implies very broad and diverse conditions that further hinder the ability to obtain generalised solutions.Another factor that has influenced this search is the difficulty for potential users to convey their acoustic needs, especially those with more severe disorders, which is why the research has been structured in two ways.The first is a quantitative survey aimed at technical managers and carers of people with ASD, consisting of a set of 10 questions, which was sent to 715 valid email addresses in 30 countries and in three languages: English, French and Spanish.The second is qualitative in nature and is organised through the study of scientific texts, visits to specific centres, relationships with autistic people and their families, as well as professional experience with this group.It has been argued that venues should be designed primarily for those who suffer from hyperacusis, but studies have not been conclusive about how this condition evolves over the course of their lives. For this reason, it has been recommended that acoustic properties be maximised during childhood and youth, when there is no doubt about its prevalence.A significant factor found is that acoustic interventions should prioritise high-pitched sounds over low-pitched ones, as they are more annoying. For this reason, acoustic solutions were presented using existing construction systems based on the data obtained.The state of the art recommends solutions to improve sound environments, including the use of carpets, but after studying other disciplines, it has been found that they should not be used despite their acoustic properties. During visits to specific centres, those in charge reported acoustic problems relating to activities in greenhouses and swimming pools, which is why interventions based on existing construction solutions have been developed.The studies carried out have detected an anomaly between whether high-pitched or low-pitched sounds, depending on the language of the speakers, cause more discomfort. While for English and Spanish it was the high-pitched sounds, for French it was both, but the limitations of the sample obtained do not allow these findings to be irrefutable, although they do generate a line of research to consider the language of people with ASD as a factor in acoustic design.The complexity of acoustic design for autistic people has been demonstrated, as evidenced by the modifications made in the specific centres consulted. Collaboration in its design is necessary between all those involved, including family members, technicians and the staff responsible for their care. Acoustic quality is not exclusive to autistic people; research that improves their sound environments will also improve those of neurotypical people.KeywordsAcoustics, autism, architecture, ASD, acoustic quality, sound environments, design and construction solutions.
  

DOCTORAL DEGREE IN ARTIFICIAL INTELLIGENCE

• OLIVA FELIPE, LUIS JAVIER: Artificial Virtuous Agents in water management scenarios
  
  Author: OLIVA FELIPE, LUIS JAVIER
  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 ARTIFICIAL INTELLIGENCE
  Department: Department of Computer Science (CS)
  Mode: Normal
  Deposit date: 24/03/2026
  Deposit END date: 08/04/2026
  Thesis director: CORTÉS GARCÍA, CLAUDIO ULISES | CORTÉS MARTÍNEZ, ATIA
  Thesis abstract: As human populations continue to grow, resource consumption rises inexorably, leading to eventual depletion, as stated by Malthus and especially by Hardin's Tragedy of the commons. To maintain resource sustainability, coordinating the collective action of those who seek benefit from consuming that resource is necessary. However, there is a small part of the whole set of participants which may act selfishly and ignore any norm or social convention that constrains overconsumption or regulates the way it is consumed. This occurs because collective action, as explained by Olson in Logics of Collective Action (Olson, 1971), can face two different social dilemmas: The prisoner's dilemma and the free-rider dilemma. The former explains how individual choice leads to a local optimal (i.e., personal benefit) but at the cost of a bad global result (i.e., common good gets depleted and, finally, nobody benefits any longer). The latter explains how free-riders, i.e., selfish or rogue agents, can burden the maintenance of the common good as the group of consumers grow since avoiding paying the cost to preserve it may not be perceived (e.g., reducing the consumption rate or paying taxes) while the common good is still provided and consumers cannot be excluded.This thesis explores the use of virtue ethics in the decision-making process of agents. Virtues are usually associated with moral excellence or a conformity to a standard of right. In our case, the goal is to represent, to some extent, virtuosity according to classic ethics, concretely, Plato’s and Aristotle’s ethical schools of thought. In Plato and Aristotle's writings, virtues are usually defined as character traits that drive how an agent should behave in the world. In both approaches the goal of the agent is to act in the right way, according to virtues, to achieve a state of eudaimonia (having a plentiful and fulfilling life). We understand that the rightness of an agent is not only determined by what they want to do but also by how they balance their actions to perform them (i.e., acting according to your values and virtues). We would like to analyse how to model values and virtues to represent, to some extend, a Platonic and Eudaimonistic form of virtue ethics. Our model of values and virtues uses Swanton’s target-centered approach to virtue ethics.In both ethical approaches, it is not only necessary to consider the individual but also the society these individuals live within. These societies should foster their individuals to be virtuous. Since values and virtues are considered incommensurable, we decided to focus on a specific context to which is water consumption. For this reason, the thesis is divided into two main chapters: one focusing on the individual consumption of fresh water in households in a mid-size Mediterranean town, and another at the societal level, which uses a model of industrial wastewater management in the besos river basin, whose catchment is located in Barcelona, and is a typical example of a Mediterranean complex catchment. The catchment area is one of the most populated catchments in Catalonia, having more than 2M people connected, whose working population is mainly occupied in industrial activities and the services sector. These industries are of various activities with an important presence of the chemical, metallurgical, plastics, textile, paper and food sectors. Each of these sectors has different needs with regard to water usage.We raise the hypothesis that artificial virtue agents may perform better than classic utilitarian agents with regards to common-pool resource sustainable consumption. We think that both approaches (Platonic and Aristotelian) are not better to each other but depending on the context it is possible that significant differences become relevant. That is, both are two different ways to act rightfully.
  
• VILALTA ARIAS, ARMAND: Semantic Embeddings in Deep Convolutional Neural Networks
  
  Author: VILALTA ARIAS, ARMAND
  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 ARTIFICIAL INTELLIGENCE
  Department: Department of Computer Science (CS)
  Mode: Normal
  Deposit date: 26/03/2026
  Deposit END date: 10/04/2026
  Thesis director: GARCÍA GASULLA, DARIO | CORTÉS GARCÍA, CLAUDIO ULISES
  Thesis abstract: One of the fundamental questions in Artificial Intelligence (AI) is how to represent knowledge effectively. The main challenge lies in devising a representation that can be exploited for intelligent computations. Since 2012, when Deep Neural Networks (DNNs) established themselves as the state of the art in Computer Vision, their use has rapidly expanded to a wide range of AI domains, including text and speech recognition, image generation, gaming, and robotics. This widespread success stems from their nature as representation learning techniques. However, DNNs require large volumes of data and considerable computational resources to learn such representations, which significantly limits the range of problems to which these models can be directly applied.In this thesis, we explore the possibility of reusing the knowledge learned by a DNN for a specific problem to solve a different one, which is commonly known as transfer learning. We focus on the representations learned by a DNN, codified in the activations of its neurons in response to an input, namely embeddings. All this work stems from a first analysis to understand where valuable information is encoded in Convolutional Neural Networks (CNNs) within the context of transfer learning. When used for characterizing every class of eleven datasets, we statistically measure the discriminative power of every feature found within a deep CNN. We seek to provide new insights into the behaviour of CNN features, particularly the ones from convolutional layers, which they had not used in previous literature. Our results confirm that low and middle level features may behave differently from high-level features, but only under certain conditions. We find that all CNN features can be used for knowledge representation purposes both by their presence or by their absence, doubling the information a single CNN feature may provide. We also study how much noise these features may include and propose a thresholding approach to discard most of it.This enables the definition of a methodology to improve the generalization capabilities of CNN representations, which we refer to as the Full-Network Embedding (FNE), which successfully integrates convolutional and fully connected features. To do so, the embedding normalizes features in the context of the problem and discretizes their values to reduce noise and regularize the embedding space. Significantly, this also reduces the computational cost of processing the resultant representations. The proposed method outperforms single layer embeddings on several image classification tasks while also being more robust to the choice of the pre-trained model used for obtaining the initial features. The performance gap in classification accuracy between thoroughly tuned solutions and the full-network embedding is also reduced, making the proposed approach a competitive solution for a large set of applications. We consider FNE more semantic as it understands CNN activations as semantic concepts. Similarly to many human languages, it considers if a feature of the input is significant in a given context and if it is because of its presence or absence.A first stem is the use of the FNE as the basis for a network representation of concepts where complex network techniques are applied. We propose the construction of a graph embedding space, introducing a methodology to transform the knowledge coded within a deep convolutional network into a topological space.The second stem, uses the same techniques in the context of multimodal embeddings representing text and images. The FNE provides a multi-scale representation of images, which results in richer characterizations while focusing on the relevant information through its semantic discretization. Results for image annotation and image retrieval tasks show a constant improvement when applied to different existing methodologies.
  

DOCTORAL DEGREE IN MARINE SCIENCES

• PAVO FERNÁNDEZ, EVA: Coastal dynamics and resilience in the low-engineered coast of Mozambique
  
  Author: PAVO FERNÁNDEZ, EVA
  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 MARINE SCIENCES
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Normal
  Deposit date: 23/03/2026
  Deposit END date: 07/04/2026
  Thesis director: GRACIA GARCIA, VICENTE | GRIFOLL COLLS, MANUEL
  Thesis abstract: Coastal areas are crucial and sensitive environments that are increasingly at risk due to climate change. Yet, a significant research gap persists in integrating the various scales of coastal analysis, from long-term geomorphological settings to short-term (extreme events), in regions with limited information, as is the case of Mozambique.This thesis tackles this gap by utilizing a comprehensive, multiscale approach to study the dynamic and under-researched coastline of Mozambique. The region serves as a compelling case study, presenting a diverse geomorphology and high vulnerability to tropical cyclones within a data-scarce context. The research uses high-resolution satellite imagery, hydrodynamic data, geomorphological mapping, and process-based numerical modelling to provide a holistic assessment of coastal processes.By bridging the scales between remote sensing, geomorphological context, and process-based modelling, this thesis aims to advance the predictive understanding of coastal change in Mozambique. The outcomes provide a scientific foundation for informed coastal management, support coastal resilience, and offer a transferable template for studying other vulnerable, data-scarce coastlines globally.
  
• SÁNCHEZ-ARCILLA ROSANAS, AGUSTÍN: PREDICTIVE MAINTENANCE OF HARBOUR STRUCTURES AND COASTAL SYSTEMS
  
  Author: SÁNCHEZ-ARCILLA ROSANAS, AGUSTÍN
  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 MARINE SCIENCES
  Department: Department of Civil and Environmental Engineering (DECA)
  Mode: Article-based thesis
  Deposit date: 26/03/2026
  Deposit END date: 10/04/2026
  Thesis director: SANCHEZ-ARCILLA CONEJO, AGUSTIN | ALTOMARE, CORRADO
  Thesis abstract: This PhD thesis contributes to the understanding and maintenance of Mediterranean and Atlantic coastal systems under accelerating climatic and human pressures. For that purpose a new methodology has been developed, combining beach morphodynamics, adaptation pathways with tipping points, and considering relevant institutional barriers to coastal restoration. The developed methodology focuses on combining process-based coastal engineering, machine learning techniques and adaptation governance frameworks to get a systemic approach for an adaptation based on restoration.Such an approach considers the barriers to coastal restoration, along with the necessary “enablers” to overcome them, always supported by coastal engineering techniques, process-based models, predictions using machine learning, and adaptation governance frameworks. In this way, a systemic approach to coastal dynamics is obtained, making it possible to make decisions based on predicting and analyzing how a typical beach may evolve under future climate scenarios and what the most appropriate responses are.Mediterranean beaches are highly dynamic albeit constrained systems often with scarcity of sediments, encroachment of urban areas, and limited accommodation space. These factors, together with growing human pressures, are expected to increase the risk of erosion and reduce the resilience of the considered beaches. The first contribution of this thesis is developed in the paper “Coastal Adaptation Pathways and Tipping Points for Typical Mediterranean Beaches under Future Scenarios”. This paper identifies adaptation tipping points for typical Mediterranean beach types, subject to 50- and 100-year storms, combined with the projected sea-level rise. The research quantifies when conventional measures such as shoreface nourishment, dune reinforcement, or structural protection are bound to lose effectiveness, suggesting the application of adaptation pathways. The results present a time-sequenced approach for proactive decisions, enabling a flexible coastal management and avoiding reactive protection with high impacts and carbon footprint.Based on the improved systemic understanding of beach systems, the second contribution of the thesis is presented in the article “Morphodynamic Predictions Based on Machine Learning: Performance and Limits for Pocket Beaches near the Bilbao Port.” This work analyzes the use of artificial intelligence (machine learning or ML) in coastal morphodynamics. Using high-resolution datasets associated with pocket beaches affected by port interventions, the predictive performance of machine learning models has been compared with conventional methodologies that apply process-based models or traditional statistical techniques. The results show that ML-based models exhibit a high short- and medium-term predictive capacity for capturing complex coastal responses, although their performance is more limited under extrapolated or extreme conditions. This study contributes to integrating coastal engineering with artificial intelligence, providing scalable tools for early warning systems and adaptive coastal monitoring.The cornerstone for sustainable coastal restoration is presented in the third contribution of the Thesis, developed in the article “Coastal Restoration Agreements under Climate Change: Barriers and Enablers.” This work links the analysis of physical processes with governance mechanisms to address the adaptation deficit present in coastal areas. Even when the technical or physical adaptation pathways are known, their implementation faces institutional fragmentation, regulatory rigidity, and social perception. Through an analysis based on application cases, this research identifies the main barriers and enabling factors for coastal restoration under climate uncertainty.
  

DOCTORAL DEGREE IN SIGNAL THEORY AND COMMUNICATIONS

• NOURMOHAMMADI, FARZANEH: Deep Learning Driven Blocking Prediction in Elastic Optical Networks using Spatio-Temporal Hybrid Neural Network
  
  Author: NOURMOHAMMADI, FARZANEH
  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 SIGNAL THEORY AND COMMUNICATIONS
  Department: Department of Signal Theory and Communications (TSC)
  Mode: Normal
  Deposit date: 24/03/2026
  Deposit END date: 08/04/2026
  Thesis director: COMELLAS COLOME, JAUME
  Thesis abstract: The explosive growth of bandwidth-intensive services such as ultra-high-definition video streaming, cloud gaming, and virtual reality is pushing optical transport networks toward greater flexibility and efficiency. Elastic Optical Networks (EONs) address these demands by allocating spectrum in variable-sized frequency slots rather than fixed-grid channels. However, the dynamic setup and teardown of heterogeneous lightpaths fragment the optical spectrum, often leading to connection blocking even when sufficient total capacity exists. Accurate early prediction of blocking events enables network controllers to trigger proactive spectrum defragmentation and resource reallocation, improving service continuity and network utilization.This dissertation investigates deep learning-based approaches for blocking prediction in EONs using network state information represented as spectrum occupancy matrices. First, we introduce one-dimensional and two-dimensional Convolutional Neural Networks (1D-CNN and 2D-CNN) to automatically extract spatial fragmentation patterns from simulated EON topologies. The proposed 2D-CNN achieves a prediction accuracy of 92.17\% and outperforms conventional machine learning baselines such as Support Vector Machines and k-Nearest Neighbors. Building on this, we propose hybrid spatio-temporal architectures that integrate convolutional feature extraction with Long Short-Term Memory (LSTM) and Bidirectional LSTM (BiLSTM) units to capture both spatial and temporal evolution of spectrum fragmentation. The CNN–BiLSTM model achieves 94.1\% prediction accuracy while maintaining reasonable computational complexity, and the CNN–LSTM model offers a favorable trade-off between accuracy and training speed.By unifying spatial and temporal deep learning techniques, this work provides a comprehensive framework for early and reliable blocking prediction in dynamic EON environments. The proposed models can be integrated into elastic optical network controllers to enable timely defragmentation actions, reducing blocking probability and improving overall network efficiency. The findings advance the application of artificial intelligence to optical networking and lay the foundation for future integration with intelligent routing, spectrum assignment, and network automation strategies.