KIOps6G – AI Operations in 6G Networks

The use of machine learning (ML) in 6G networks will become increasingly important in the upcoming years, both for optimising the network itself (ML for networks) and for supporting general applications (Network for ML). Previous approaches to operating ML solutions have either been described specifically for ML-for-networks use – such as Model Training Logical Functions or Analytics Logical Functions – or realised more generically via MLOps platforms.

The aim of our project is to develop a distributed MLOps platform that is suitable for 6G networks and can operate both ML for networks and network for ML applications in a scalable and reliable manner, while responding flexibly to their changing requirements. A description language for ML apps is being developed that makes non-functional aspects explicit, such as periodic computing power requirements for training on accelerator hardware.

The results of the project should help to accelerate the establishment of ML-supported telecommunications management in the European market.