R1-2410153
discussion
AI/ML Model and Data
From Google
Summary
Google presents a technical document on AI/ML model identification and data collection for NR air interface, proposing 8 specific proposals covering model identification types, UE data collection procedures, and ID configuration schemes. The document advocates for simplifying model identification approaches by deprioritizing certain options and clarifying the relationships between different identification schemes.
Position
Google advocates FOR simplifying AI/ML model identification by assuming model IDs are known after UE connection (MI type A) and prioritizing MI-Option 1 for maintaining understanding of model properties. They push AGAINST the complexity of MI-Option 2 and MI-Option 3, arguing for deprioritization of dataset transfer and model transfer methods. Google specifically supports consolidating ID-X with model ID to avoid redundant identification schemes and differentiates between two-side and one-side model configuration requirements.
Key proposals
- Proposal 1 (Model Identification): For MI type A, it is assumed that the indication of a model ID is known by the NW and UE after UE connected to the NW. No additional specification work is required to maintain the same communication between the NW and UE on the indication of a model ID.
- Proposal 2 (Model Identification): MI-Option 1 is necessary to assist the NW and UE to maintain the same understanding for the property of model input and model output, so that the NW can configure corresponding DL RS for the UE to identify the model input and configure corresponding UL resource for model output report.
- Proposal 3 (Model Identification): Deprioritize MI-Option 2 and MI-Option 3.
- Proposal 4 (UE Data Collection): Support the NW and UE to maintain the same understanding on when the UE can perform data collection.
- Proposal 5 (UE Data Collection): Support the NW to configure whether the associated ID in different cells indicating the same additional conditions or not.
- Proposal 6 (ID Connections): The ID-X for MI-Option2 is model ID.
- Proposal 7 (ID Connections): For two-side model, support to configure the model ID in CSI report configuration for inference and configure the associated ID for CSI-RS resource set for CSI acquisition for UE-side data collection.
- Proposal 8 (ID Connections): For one-side model, support to configure the associated ID in CSI report configuration for inference and for SSB/CSI-RS resource set for BM or CSI-RS resource set for CSI acquisition for UE-side data collection. Model ID is not applicable for one-side model.