R1-2410531
discussion
Design for AI/ML based positioning
From MediaTek
Summary
MediaTek's technical document presents 13 proposals across 5 main sections addressing AI/ML-based positioning design for NR air interface, covering sample-based measurement reporting, receiver diversity, LOS/NLOS indicators, training consistency, performance monitoring, and assistance data aspects.
Position
MediaTek advocates FOR implementation flexibility in sample-based measurements by avoiding strict specification of intermediate steps like Nt consecutive samples, and FOR protecting sensitive network deployment information by using implicit signaling through associated IDs rather than explicit TRP locations. They push AGAINST overly restrictive receiver implementation constraints and unnecessary disclosure of proprietary information like exact TRP coordinates and beam shapes, while supporting UE-centric performance monitoring and area-level rather than cell-level associated ID management.
Key proposals
- Proposal 2a-1 (Sec 2a): Define new measurement type for sample-based measurement as 'the channel response obtained from the resource elements that carry the reference signals configured for the measurement'
- Proposal 2a-2 (Sec 2a): Consider not configuring the value of Nt, allowing Nt to be implicitly determined based on applied IDFT size to provide implementation flexibility
- Proposal 2a-3 (Sec 2a): Define start time to search for reporting Nt' samples using expected propagation delay minus delay uncertainty, rather than defining start time of Nt consecutive samples
- Proposal 2b-1 (Sec 2b): RSRP with respect to PDP for reporting should not be lower than RSRP of PDP obtained in any individual receiver branches for diversity combining
- Proposal 2c-1 (Sec 2c): Change LOS/NLOS indicator definition for AI/ML case to 'the likelihood that the LOS delay could be derived' instead of traditional channel-based definition
- Proposal 2c-2 (Sec 2c): AI/ML model output reporting needs to be indicated to network to avoid confusion with legacy definitions
- Proposal 3a-1 (Sec 3a): Provide associated ID to UE at training data generation and inference to discriminate significant changes in network side implementation
- Proposal 3a-2 (Sec 3a): Associated ID for positioning should be considered at area level rather than cell level for efficiency
- Proposal 4a-1 (Sec 4a): For case 1 performance monitoring, metric calculation should be conducted by UE
- Proposal 4a-2 (Sec 4a): For case 1 performance monitoring, Option B is not considered while all items within Option A can be considered
- Proposal 4a-3 (Sec 4a): UE should send error cause to LMF when performance degradation is identified, allowing LMF to terminate AI/ML method and reconfigure alternative positioning method
- Proposal 5a-1 (Sec 5a): TRP location information can be delivered implicitly through associated ID rather than explicit coordinates to protect sensitive deployment information
- Proposal 5a-2 (Sec 5a): Beam response can be indicated explicitly for multiple beams or implicitly when only single beam per TRP to optimize signaling efficiency