Conference Agenda

This paper presents Twin5G, a proposed intelligent digital twin for heterogeneous 5G networks (HetNet) designed for advanced monitoring, prediction, operational automation, and evolution toward production environments.

The system integrates initial network simulation, cloud-based analytical storage, and artificial intelligence to represent, analyze, and anticipate infrastructure behavior before critical events occur in the real network.

The proposed architecture considers a heterogeneous network composed of macrocells, microcells, picocells, and femtocells, along with an intelligent processing layer capable of supporting tasks such as congestion prediction, latency estimation, anomaly detection, decision automation, and resource self-management. In its current state, Twin5G implements
an initial simulation phase in MATLAB Online to generate
performance metrics such as latency, throughput, received power, and link conditions, which are then exported for
processing in the cloud.
As an architectural evolution, the use of Google
Colab for data preparation, BigQuery as the analytical foundation
of the digital twin, Vertex AI as an intelligent layer for inference
and predictive analytics, and Cloud Run as the deployment environment for
an interactive web interface is proposed. Additionally, Looker
Studio can be used for executive visualization of historical
metrics and results.

Twin5G is conceived as a scalable foundation for building intelligent management systems in 5G networks, capable of supporting advanced monitoring functions, operational optimization, and proactive decision-making in heterogeneous, high-demand scenarios.