Predicción de rendimiento de redes LTE con técnicas de Machine Learning

Abstract

Abstract: Predicting the performance of a mobile cellular network is one of the main challenges when evaluating the quality of service of any mobile operator. This prediction allows the operator to be aware of future network scenarios and thus take corrective actions before they occur. Machine learning algorithms supported by large amounts of data in real scenarios provide mobile cellular network operators with a guarantee of efficiency in network operation. The present workuses Machine Learning techniques for the prediction of LTE network performance, for this purpose a database of a cellular mobile network was built using an Android application. The CRISP-DM methodology was used to perform data mining and we proposed different scenarios where the Machine Learning models were trained, finally these models were evaluated with a 70:30 partition (training: testing), using statistical metrics that allowed to establish the model that best fits the proposed scenarios.