Optimización de redes UMTS soportada en machine learning

Abstract

This investigation is carried out the analysis, the validation, and the implementation of one algorithm took from wide Machine Learning world, to generate one or several models that can support the optimization process of UMTS (Universal Mobile Telecommunications System - Sistema universal de telecomunicaciones móviles) networks. Some statistical counters and KPIs (Key Performance Indicators – Indicadores claves de desempeño) from UMTS cells are presented that normally are used by optimizing engineers to diagnostic the status of the cells and that they will be useful as input information to create a data set for training several Machine Learning techniques. For this is done the analysis of several Machine Learning techniques and depending on the data set characteristics, it is chosen two techniques, which will be trained and evaluated to finally select the technic with the best performance, the final model of this technic is implemented over some programming language that permits the diagnostic of future cases.