Sistema de detección de anomalías de rendimiento y seguridad en redes de datos con inteligencia artificial

Abstract

This research project aims to implement a system for detecting performance and security anomalies in data networks, based on artificial intelligence. The increasing complexity of network infrastructures, the exponential growth of traffic, and the sophistication of cyber threats require advanced technological solutions that enable continuous, accurate, and proactive monitoring. The methodology is based on deep learning techniques, using models such as Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), and Gated Recurrent Units (GRU), applied to the analysis of data collected through network monitoring protocols such as SNMP, NetFlow, and Syslog. The information is gathered from multiple networks interconnected via ZeroTier, allowing secure and efficient integration of distributed environments. Data processing, storage, and visualization are performed using tools such as Logstash, Elasticsearch, Kibana, and Streamlit, within a modular and scalable architecture deployed with Docker and Kubernetes. This framework enables real-time management of large volumes of data and facilitates the implementation of anomaly detection models. The developed system identifies significant deviations from normal network behavior in a timely manner, supporting informed decision-making and reducing the risk of failures or intrusions. This solution contributes to strengthening cybersecurity and optimizing performance in enterprise networks.