Estudio de la calidad de la conexión en redes de telefonía celular en base a mediciones de campo y técnicas de Machine Learning : estudio del comportamiento de los parámetros de RF en modo inactivo y estático en redes de telefonía celular en base a mediciones de campo y técnicas de Machine Learning.

Abstract

Nowadays, signal quality has become an extremely important issue due to the need to stay connected through our mobile phones. For this reason, it is important to understand the conditions under which signal quality can be affected. Hence, this study takes into consideration RF parameters that allow the identification of conditions categorized as good, moderate, or poor quality. To achieve this, the RSRQ parameter (Reference Signal Received Quality), measured in dB, is used to establish ranges for these quality labels. In addition to this RF parameter, two others are included: RSSI (Received Signal Strength Indicator) and RSSNR (Reference Signal Signal to Noise Ratio). These parameters will serve as predictive variables for analyzing the quality of the signal received by the mobile phone. Alongside these RF parameters, environmental factors are also included as predictive variables to determine whether any of them have a negative impact on received signal quality. The considered factors include temperature, atmospheric pressure, ultraviolet radiation, and wind speed. To conduct this study, two machine learning classification techniques were employed: decision trees and k-nearest neighbors. The obtained results have accuracies exceeding 70%, indicating the effective performance of these algorithms.