Pasar al contenido principal

REMANDO2030-UPV: MEDIDAS Y MODELADO AVANZADO DEL CANAL RADIO PARA EL DESARROLLO DE LOS FUTUROS SISTEMAS DE COMUNICACIONES MOVILES IMT-2020

PID2024-158965OB-C21

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para la Investigación y el Desarrollo Experimental
Subprograma Subprograma Estatal de Generación de Conocimiento Científico-Técnico y Desarrollo Experimental
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2024
Plan Estatal 2024-2027 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

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

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Machine Learning-Based Method for LOS/NLOS Identification in V2V Environment

  • Chakkour, Yousra
  • Fernández, Herman
  • Rodrigo Peñarrocha, Vicent Miquel|||0000-0002-8075-4851
  • Rubio Arjona, Lorenzo|||0000-0003-3882-4673
  • Reig, Juan|||0000-0003-4541-9326
[EN] Accurate identification of line-of-sight (LOS) and non-line-of-sight (NLOS) conditions is a critical challenge in wireless communication, particularly the vehicle-to-vehicle (V2V) channels, owing to the high dynamics inherent to V2V environments. This study explores the use of machine learning (ML) techniques for LOS/NLOS classification based on wireless channel features extracted from both the measured received power level and the difference between the measured attenuation and free space loss (FSL). Feature correlation, importance analysis, and cumulative scoring are applied to derive a compact and informative feature set. A performance analysis with varying numbers of input features, along with an ablation study, showed that reliable classification can be maintained even with a reduced feature set. In addition, per-class evaluation is performed to capture distinct LOS and NLOS behaviors, which is particularly relevant when prioritizing specific outcomes such as minimizing false LOS or false NLOS detection. Balanced datasets were created through random sampling from the LOS and NLOS classes, with training and testing performed using K-fold cross-validation (CV). Three models, random forest (RF), support vector machines (SVM), and LogitBoost, were evaluated based on recall, specificity, precision, and the F1-score metric. The results indicate that the RF model outperforms the others and that a minimal yet balanced training dataset from both the LOS and NLOS classes is sufficient to achieve high classification reliability. The proposed method offers interpretability, a low computational load framework, and flexibility to address specific classification needs. This aspect provides valuable insights into improving the performance of vehicular communication systems., This work was supported in part by the Vicerrectorado de Investigación de la Universitat Politècnica de València under Grant PAID-11-24, and in part by MCIN/AEI/10.13039/501100011033 through the I+D+i Projects under Grant PID2024-158965OB-C21.