Propuesta de modelos estadísticos para conexiones inalámbricas en el marco de la ley 21.046.
Loading...
Date
2026
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
La presente memoria tiene como objetivo proponer un método estadístico para determinar la velocidad mínima garantizada en redes móviles 4G, en el contexto de la Ley 21.046, mediante el uso de modelos de distribución de velocidad y detectores de Tasa de Falsa Alarma Constante CFAR aplicados a mediciones en terreno. Para ello, se realizó una campaña de recolección de datos de velocidad de subida y bajada utilizando la aplicación Drive Test UdeC y un dispositivo fijo en la ciudad de Concepción, ejecutando pruebas en protocolo TCP y UDP estandarizadas en redes 4G y 5G. Tras un proceso de depuración y clasificación de los registros, se evaluó el ajuste de distribuciones paramétricas clásicas mediante la prueba de Kolmogorov-Smirnov en distintos niveles de desagregación. Adicionalmente, se diseñó un experimento controlado con un dispositivo fijo en 4G para aislar la variabilidad del canal, modelando la velocidad como una señal determinista con ruido aditivo y aplicando un algoritmo CFAR sobre los histogramas resultantes para estimar probabilidades de detección y definir umbrales operativos. Los resultados evidenciaron que el throughput móvil presenta un comportamiento multimodal y una alta variabilidad que impide su modelado mediante una distribución única simple, rechazándose sistemáticamente las hipótesis paramétricas tradicionales. No obstante, en el escenario controlado se constató que velocidades objetivo del orden de 10 Mbps en subida y bajada en servidores nacionales se sostienen con alta probabilidad, mientras que en el enlace de subida surgen limitaciones a partir de los 40-45 Mbps, lo que refleja un diseño de red asimétrico favorable al enlace descendente. Se concluye que el enfoque basado en detectores CFAR permite definir umbrales de velocidad mínima garantizada estadísticamente robustos y adaptados a las condiciones del experimento, constituyendo una contribución metodológica significativa para la fiscalización técnica de la calidad de servicio móvil bajo la normativa vigente, al ofrecer una alternativa objetiva frente a la evaluación basada únicamente en promedios temporales.
The objective of this thesis is to propose a statistical method for determining the guaranteed minimum speed in 4G mobile networks, within the context of Law 21.046, through the use of speed distribution models and Constant False Alarm Rate (CFAR) detectors applied to field measurements. To this end, a data collection campaign for upload and download speeds was conducted using the Drive Test UdeC application and a fixed device in the city of Concepción, executing standardized TCP and UDP protocol tests on 4G and 5G networks. Following a process of data cleaning and classification, the fit of classical parametric distributions was evaluated using the Kolmogorov-Smirnov test at various levels of disaggregation. Additionally, a controlled experiment was designed with a fixed 4G device to isolate channel variability, modeling speed as a deterministic signal with additive noise and applying a CFAR algorithm to the resulting histograms to estimate detection probabilities and define operational thresholds. The results evidenced that mobile throughput presents multimodal behavior and high variability, preventing its modeling through a single simple distribution, with traditional parametric hypotheses being systematically rejected. However, in the controlled scenario, it was found that target speeds of around 10 Mbps in uplink and downlink on national servers are sustained with high probability, whereas limitations arise in the uplink starting at 40-45 Mbps, reflecting an asymmetric network design favoring the downlink. It is concluded that the approach based on CFAR detectors allows for the definition of statistically robust guaranteed minimum speed thresholds adapted to the radio conditions of the experiment, constituting a significant methodological contribution to the technical supervision of mobile service quality under current regulations, by offering an objective alternative to evaluation based solely on temporal averages.
The objective of this thesis is to propose a statistical method for determining the guaranteed minimum speed in 4G mobile networks, within the context of Law 21.046, through the use of speed distribution models and Constant False Alarm Rate (CFAR) detectors applied to field measurements. To this end, a data collection campaign for upload and download speeds was conducted using the Drive Test UdeC application and a fixed device in the city of Concepción, executing standardized TCP and UDP protocol tests on 4G and 5G networks. Following a process of data cleaning and classification, the fit of classical parametric distributions was evaluated using the Kolmogorov-Smirnov test at various levels of disaggregation. Additionally, a controlled experiment was designed with a fixed 4G device to isolate channel variability, modeling speed as a deterministic signal with additive noise and applying a CFAR algorithm to the resulting histograms to estimate detection probabilities and define operational thresholds. The results evidenced that mobile throughput presents multimodal behavior and high variability, preventing its modeling through a single simple distribution, with traditional parametric hypotheses being systematically rejected. However, in the controlled scenario, it was found that target speeds of around 10 Mbps in uplink and downlink on national servers are sustained with high probability, whereas limitations arise in the uplink starting at 40-45 Mbps, reflecting an asymmetric network design favoring the downlink. It is concluded that the approach based on CFAR detectors allows for the definition of statistically robust guaranteed minimum speed thresholds adapted to the radio conditions of the experiment, constituting a significant methodological contribution to the technical supervision of mobile service quality under current regulations, by offering an objective alternative to evaluation based solely on temporal averages.
Description
Tesis presentada para optar al título de Ingeniero/a Civil en Telecomunicaciones.
Keywords
5G sistemas de comunicación móvil, Histogramas, Teléfonos móviles