R1-2410049
discussion
Discussion on specification support for AIML-based positioning accuracy enhancement
From Fujitsu
Summary
This Fujitsu contribution discusses AI/ML-based positioning accuracy enhancement for NR air interface, providing 12 proposals and 7 observations across model input/output, training-inference consistency, and performance monitoring aspects for various positioning cases.
Position
Fujitsu advocates FOR: sample-based measurement with configured offset (Option D), deprioritizing phase information study due to negligible gains vs overhead, using LOS/NLOS indicator as confidence level rather than LOS likelihood, supporting multiple timing information reporting, and gNB/TRP-based model control. They push AGAINST: associated ID mechanisms for multi-TRP scenarios due to impractical training burden, explicit AI/ML indicators in positioning reports, and fixed offset approaches that may miss significant samples or create overhead.
Key proposals
- Proposal 1 (Sec 2.1.1): For sample-based method, Option D is suggested where the starting point of Nt consecutive samples is determined by a configured offset relative to reference time, with offset decided by LMF based on potential UE distance to TRPs
- Proposal 2 (Sec 2.1.1): Revise candidate values for sample-based measurement - Nt include at least 16, 32, 64, 128; Nt' include at least 9, 16, 32 (Note: Nt'<= Nt); k include at least 0…5
- Proposal 3 (Sec 2.1.2): Study on phase measurement and reporting for model input of Case2b/3b is deprioritized in Rel-19 normative work
- Proposal 4 (Sec 2.2.1): For AI/ML assisted positioning Case3a, LOS/NLOS indicator meaning is changed to confidence level of timing information corresponding to (virtual) LOS path generated from model output
- Proposal 5 (Sec 2.2.2): For AI/ML assisted positioning (Case2a/Case3a), there is no need to introduce additional indication to indicate an AI/ML-based positioning report
- Proposal 6 (Sec 2.2.2): For AI/ML assisted positioning Case 2a/3a, multiple timing information reporting for (virtual) LOS path can be supported in addition to the LOS/NLOS indicator
- Proposal 7 (Sec 2.3): The study on associated ID to indicate NW-side additional conditions involving multiple TRPs is suggested to be deprioritized in AI/ML-based positioning
- Proposal 8 (Sec 2.3): Existing assistance data IEs defined for UE-based DL-AoD but not for UE-based DL-TDOA can be provided to UE-based DL-TDOA with explicit indication if justified necessary for consistency
- Proposal 9 (Sec 2.4.1): Regarding UE-side monitoring metrics calculation of Case 1, Option A-1/A-2 is supportive if high-quality ground truth label availability can be confirmed; Option A-3 is supportive if PRU measurement input format matches UE model input format
- Proposal 10 (Sec 2.4.1): Regarding LMF-side monitoring metrics calculation of Case 1, Option B-1 is supportive if high-quality ground truth label availability can be confirmed; Option B-2 is supportive if PRU measurement input format matches UE model input format
- Proposal 11 (Sec 2.4.2): Regarding performance monitoring metric calculation in label-based monitoring of Case 3a, model LCM control/decision is suggested to be done by gNB/TRP for both Option A and Option B
- Proposal 12 (Sec 2.4.3): Performance monitoring for case 3b can be realized by implementation manner and no need for further study