Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 3
Encontrada(s) 1 página(s)

e_Buah Biblioteca Digital Universidad de Alcalá
Publicación Contribución de congreso (ConferenceOutput)

Convolutional Neural Network-Based DOA Estimation for Multiple Incident Signals in Noisy Environments

  • Gutiérrez Blanco, Óscar|||0000-0001-6153-7125
  • Cabornero Pascual, David|||0009-0004-9284-8451
  • Lozano Plata, Lorena|||0000-0002-0137-7868
2026 20th European Conference on Antennas and Propagation (EuCAP), , 19/04/2026-25/04/2026, Dublín, Irlanda, This paper presents a deep learning approach for direction-of-arrival (DOA) estimation using convolutional neural networks (CNNs). Building upon recent advances in neural architectures for signal processing, we propose a CNN model that directly estimates incident angles from raw temporal samples collected by non-uniform linear antenna arrays. The network is trained using synthetic data generated under varying conditions, including different numbers of signal sources and signal-to-noise ratios (SNRs). The proposed architecture demonstrates high accuracy in both single-source and multi-source scenarios, with robustness to noise and array geometry variations. Experimental results show that increasing the training dataset size significantly improves performance in complex environments. The model maintains reliable performance across SNR levels ranging from 30dB to 0dB. These findings highlight the potential of CNN-based methods for scalable and accurate DOA estimation in real-world applications., Ministerio para la Transformación Digital y de la Función Pública
e_Buah Biblioteca Digital Universidad de Alcalá
Publicación Contribución de congreso (ConferenceOutput)

Acceleration of physical optics integral with relational databases

  • Cabornero Pascual, David|||0009-0004-9284-8451
  • Paramés Martínez, Diego
  • González Diego, Iván|||0000-0002-3252-199X
  • Lozano Plata, Lorena|||0000-0002-0137-7868
  • Cátedra Pérez, Manuel Felipe|||0000-0003-0914-6391
European Conference on Antennas and Propagation, EuCAP, 19/04/2026-25/04/2026, Dublín, Irlanda, This work presents a set of algorithmic and architectural improvements to the Physical Optics (PO) method, aimed at enhancing its applicability to industrial-scale electromagnetic simulations. Building on previous research in diffraction modeling, relational databases?specifically PostgreSQL with its geospatial extension PostGIS?have been adopted to manage large volumes of geometric and intermediate data efficiently. The Gordon integral has been implemented natively in C within the database engine, significantly reducing computation time. Additionally, native spatial indexing and memory-aware data structures have enabled the simulation of complex scenarios on standard desktop hardware. An interpolation strategy has been introduced to accelerate ISAR image generation, allowing the reuse of previously computed integrals under small variations in frequency and scattering direction. The proposed framework eliminates memory bottlenecks and improves scalability, demonstrating its effectiveness through radar cross-section (RCS) and ISAR benchmarks. These results confirm the viability of database-driven PO simulations for real-world applications., Junta de Comunidades de Castilla-La Mancha, Ministerio para la Transformación Digital y de la Función Pública
e_Buah Biblioteca Digital Universidad de Alcalá
Publicación Artículo científico (JournalArticle)

Context-aware few-shot learning SPARQL query generation from natural language on an aviation knowledge graph

  • Hernández Camero, Inés Virginia
  • García López, Eva|||0000-0002-7598-3289
  • García Cabot, Antonio|||0000-0002-0298-3237
  • Caro Álvaro, Sergio|||0000-0002-3192-8499
Question answering over domain-specific knowledge graphs implies several challenges. It requires sufficient knowledge of the world and the domain to understand what is being asked, familiarity with the knowledge graph?s structure to build a correct query, and knowledge of the query language. However, mastering all of these is a time-consuming task. This work proposes a prompt-based approach that enables natural language to generate SPARQL queries. By leveraging the advanced language capabilities of large language models (LLMs), we constructed prompts that include a natural-language question, relevant contextual information from the domain-specific knowledge graph, and several examples of how the task should be executed. To evaluate our method, we applied it to an aviation knowledge graph containing accident report data. Our approach improved the results of the original work?in which the aviation knowledge graph was first introduced?by 6%, demonstrating its potential for enhancing SPARQL query generation for domain-specific knowledge graphs., Agencia Estatal de Investigación, Ministerio de Economía, Industria y Competitividad