R1-2410470
discussion
Other aspects of AI/ML model and data
From Qualcomm
Summary
Qualcomm's contribution to RAN1 meeting #119 discusses AI/ML model identification and transfer/delivery aspects for NR air interface. The document contains 2 main proposals focused on concluding current model identification work and deprioritizing certain model transfer cases.
Position
Qualcomm advocates for concluding the current model identification work as sufficient and moving forward, while pushing against pursuing certain model transfer/delivery cases (specifically Case z4 for UE-sided models) due to engineering challenges and UE capability requirements. They support a pragmatic approach that focuses on feasible solutions rather than continuing extensive studies on complex model identification procedures.
Key proposals
- Proposal 1: Conclude that the progress made so far including identifying different model identification options, and further details and procedures associated with those options are sufficient as far as the scope of this agenda item is concerned.
- Proposal 2: Conclude that model transfer/delivery case z4 is deprioritized for UE-sided models.
- Agreement (RAN1 #116): To facilitate the discussion, RAN1 studies the model identification type A with more details related to use cases.
- Agreement (RAN1 #116): RAN1 studies MI-Option 1 (Model identification with data collection related configuration(s) and/or indication(s)), MI-Option 2 (Model identification with dataset transfer), and MI-Option 3 (Model identification in model transfer from NW to UE) as starting points for model identification type B.
- Agreement (RAN1 #116-bis): From RAN1 perspective, for UE-sided model(s) developed at UE side, following procedure is an example (AI-Example1) of MI-Option1 including NW signaling data collection configurations with associated IDs, UE data collection, AI/ML model development at UE side, and UE reporting model information to NW.
- Agreement (RAN1 #117): From RAN1 perspective, for UE part of two-sided model, further study MI-Option2 example (AI-Example2-1) including dataset transfer from NW to UE via standardized signaling, UE model development based on dataset, and UE reporting model information to NW.
- Conclusion (RAN1 #118): From RAN1 perspective, model identification is at least applicable to some of inter-vendor training collaboration option(s) of CSI compression using two-sided model (if supported).
- Conclusion (RAN1 #118): The model identification procedure dedicated to MI-Option5 is not pursued for Rel-19 normative work.
- Conclusion (RAN1 #118): The model identification procedure dedicated to MI-Option2 for one-sided model is not pursued for Rel-19 normative work.
- Agreement (RAN1 #118-bis): Regarding the study of MI-Option2 for the two-sided model, ID (denoted as ID-X) can be transmitted from network/network-side to UE/UE-side for the dataset.