Desarrollo de una aplicación para analizar indicadores de desempeño en equipos de cuarta generación con dashboards en Grafana

Abstract

This research develops an application for the automated analysis of key performance indicators (KPIs) in 4G LTE core network equipment, with the aim of optimizing the monitoring and diagnosis of its behavior through interactive dashboards in Grafana. The project integrates data processing and visualization tools, using Python, SQLite, and the STUMPY library for the detection of anomalies in time series. The methodology applied, which is descriptive and applied in nature, was structured in four iterative phases: training, definition of requirements, development, and final implementation. In the first phase, the KPIs of the Session Border Controller (SBC) element were identified and classified. Subsequently, an anomaly detection algorithm based on the Matrix Profile was implemented, which allowed significant deviations in the behavior of network metrics to be identified. The designed system processes data from CSV or ZIP files and generates databases in SQLite format along with Grafana-compatible JSON files, enabling dynamic visualization of the results. The graphical interface, developed in CustomTkinter, facilitates user interaction with the data and the automation of dashboards. The results show that the application improves data interpretation and anomaly detection accuracy, and provides an adaptable tool for monitoring performance in 4G networks, with potential for extension to 5G environments.