6G-DALI

The increasing demand for developing 6G models is constrained by the lack of representative datasets, particularly for the RAN and Core Network (CN). This challenge arises not only from the limited number of available datasets but also from the restricted scenarios under which they have been generated. Most existing datasets fail to capture large-scale deployment conditions, such as high UE density per cell, multiple interconnected cells enabling mobility and handover scenarios, and other complex network configurations.
Although SNS-C platforms support experimentation across various 6G RAN and CN scenarios, the limited number of UEs that can be tested restricts the ability to emulate realistic, large-scale 6G environments. Consequently, critical functionalities (such as Xn and F1 handovers) are not thoroughly validated, nor is the scalability of CN functions and algorithms designed to handle high UE densities. In addition, many platforms do not support the execution of ML models and RL agents, which require validation in real, scalable, and realistic environments.
To address this gap, this use case highlights the capabilities of 6G-DALI to conduct large-scale experiments using the AI framework and the Digital Twin (DT), with two main objectives. First, it generates datasets and integrates them into the 6G Data Space through the 6G-DALI ETL pipeline. Second, it implements RLOps processes to test and validate trained RL agents