RT Dissertation/Thesis
T1 Quality-driven management of next-generation mobile networks for advanced multimedia services.
A1 Peñaherrera-Pulla, Oswaldo Sebastián
K1 Aprendizaje automático (Inteligencia artificial) - Tesis doctorales
K1 Sistemas de comunicaciones inalámbricos
AB The rapid emergence of next-generation use cases, including metaverse applications, has redefined the landscape of services and applications, enabling immersive experiences across sectors such as education, industry, and entertainment. These innovations demand advanced mobile network capabilities, including high data rates, reduced latency, and enhanced reliability. While 5G technologies address some of these requirements, they also introduce complexities in managing diverse, resource-intensive services. Traditional network management strategies, reliant on human expertise and Operations Support Systems (OSS), have evolved toward automated solutions like Self-Organizing Networks (SON), yet challenges persist.This thesis addresses these challenges by integrating Artificial Intelligence (AI) and Machine Learning (ML) techniques into next-generation network management. Specifically, it focuses on the end-to-end (E2E) optimization of service performance through Key Quality Indicators (KQIs), which capture both network and user-centric metrics. A comprehensive literature review identifies KQIs critical for advanced services such as Cloud Gaming, 360-degree video, and Cloud Virtual Reality. Empirical analysis is conducted using testbeds under controlled conditions to assess service performance across various wireless technologies, mobility scenarios, and radio conditions.The research emphasizes the 360-degree video service to demonstrate the efficacy of ML in network optimization. A dataset generated from these studies enables the development of an ML-based framework to predict KQIs using radio and network data. The framework incorporates the proposed PET$_{score}$ metric, which balances prediction accuracy and computational efficiency. Additionally, a resource optimization mechanism integrates ML-based resource models and numerical optimization to ensure Service-Level Agreement (SLA) compliance while balancing Quality of Experience (QoE) and resource usage.
PB UMA Editorial
YR 2025
FD 2025
LK https://hdl.handle.net/10630/38241
UL https://hdl.handle.net/10630/38241
LA eng
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 4 mar 2026