| One of the key enablers of 6G is the Native support of Artificial Intelligence (AI) and Machine Learning (ML) at all the system levels, components and mechanisms, from the orchestration and management levels to the low-level optimisation of the infrastructure resources including Cloud, Edge, RAN, Core Network, as well as a transport network. However, this integration presents significant challenges, primarily the need for relevant datasets to train AI models. The availability of high-quality 6G data is still limited, and even when new models are developed, testing and validation remain complex without adequate evaluation platforms. To address these challenges, the 6G-DALI project proposes a framework that harmonizes Data Management with AI development. Its approach is defined by two “key” pillars: (i) AI experimentation as a service via MLOps and; (ii) Data and analytics collection and storage via DataOps. The 6G-DALI DataOps pillar provides the mechanisms for preparing clean and processed data that are stored within a 6G Dataspace and are made available for training and validating machine learning models as a service, a part of the MLOps Pillar. The end-to-end framework also delivers continuous monitoring, drift detection and retraining of models. 6G-DALI revolutionizeσ next-generation 6G networks by addressing the critical challenges of data availability, artificial intelligence integration and energy efficiency. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.