2026-05-27T23:26:56Zhttps://riuma.uma.es/rest/oai/request

oai:riuma.uma.es:10630/249542026-02-03T11:47:35Zcom_10630_2254col_10630_37959

Villegas, Javier Fortes-Rodríguez, Sergio Cantizani-Estepa, Juan Rasines Suárez, Javier Martín Cuerdo, Raúl Barco-Moreno, Raquel 2022-09-12T11:55:14Z 2022-09-12T11:55:14Z 2022-09 https://hdl.handle.net/10630/24954 The increase in the size and complexity of the cellular network is progressively complicating the operation and maintenance activities, as well as rising its operation cost. The growing complexity of the networks makes them more prone to failures, which can degrade the quality of experience (QoE) of the network users. In this way, to prevent the degradation of QoE, network operators are focusing on creating networks with self-healing functions, which are capable of automatically troubleshooting problems, making them more reliable and reducing their operation costs. For this matter, unsupervised Machine Learning (ML) algorithms are deployed to detect anomalous network status, however, these frequently lack explanation and network experts are required for this step. For this matter, the proposed paper presents a method to determine the relevant Key-Performance Indicators for any unsupervised clustering to facilitate the explanation of the clusters. spa http://creativecommons.org/licenses/by-nc-nd/4.0/ open access Attribution-NonCommercial-NoDerivatives 4.0 Internacional Redes de banda ancha - Congresos Sistemas autoorganizativos - Congresos Identificación de la relevancia de métricas celulares en clústeres no supervisados conference output