Regarding the Feasibility Study on traffic characteristics and performance requirements for AI/ML Model Transfer via direct device connection, the TR analyses use cases of AIML-Ph2 as follows:

• Use cases on split AI/ML operation between AI/ML endpoints for AI inference by leveraging direct device connection:
  • Proximity based work task offloading for AI/ML inference
  • Local AI/ML model split on factory robots.
• Use cases on AI/ML model/data distribution and sharing by leveraging direct device connection:
  • AI Model Transfer Management through Direct Device Connection;
  • 5GS assisted transfer learning for trajectory prediction.
• Use cases on Distributed/Federated Learning by leveraging direct device connection:
  • Direct device connection assisted Federated Learning;
  • Asynchronous FL via direct device connection;
  • 5GS assisted distributed joint inference for 3D object detection;

It is recommended to proceed with normative work including the potential new requirements identified by this TR. The consolidated potential requirements in Clause 8 are the baseline for the subsequent normative work.