R1-2410468
discussion
Specification support for CSI prediction
From Qualcomm
Summary
This Qualcomm document proposes introducing associated IDs to ensure consistency between training and inference phases for AI/ML CSI prediction with UE-side models. The document contains 2 formal proposals and 4 observations focused on addressing how network-side additional conditions affect channel distribution and model performance.
Position
Qualcomm advocates FOR introducing associated IDs to categorize network-side additional conditions for CSI prediction, arguing that cell ID alone is insufficient due to multiple TRPs/RRUs within cells and potential field deployment variations not captured in RAN1 models. They push AGAINST relying solely on generalized models without categorization mechanisms, emphasizing that associated IDs provide flexibility while enabling unified design across use cases with minimal specification effort.
Key proposals
- Proposal 1 (Sec 2): For AI/ML CSI prediction use case with AI/ML model at UE side, associated ID should be introduced to ensure consistency between training and inference
- Proposal 2 (Sec 2): The associated ID should represent NW-side additional conditions. Given the same associated ID, UE assumes the same NW-side additional conditions, i.e., spatial relationship (scenario / configuration / antenna layout / beamforming, etc.), for data collection, model training and inference