Digital Twin For 5G Networks and Beyond: Domain Transfer and Network Behavior Classification for 6G Readiness

Fifth-generation (5G) networks support applications with stringent latency and reliability requirements, posing significant challenges for real time monitoring and traffic optimization. In this context, Digital Twin (DT) technology offers a promising approach for simulating and classifying network behavior. This work presents a DT system for 5G that analyzes core network metrics in real time and classifies traffic using Long Short Term Memory (LSTM) neural networks. Built entirely using open source tools, specifically Open5GS and UERANSIM, the system combines a simulation environment with a classification model trained on synthetic data and evaluated on both simulated and real network traffic. While the DT demonstrates successful real time operation and supports online model fine tuning, domain adaptation remains a critical challenge, as model accuracy drops significantly on real world data. These findings confirm the system’s feasibility and underline the need for more robust domain transfer strategies, providing foundational insights for future sixth generation (6G) network deployments.