R1-2410715
discussion
Summary #2 of specification support for positioning accuracy enhancement
From Ericsson
Ericsson's prior position on
9.1.2
at
RAN1#118bis
· AI-synthesized, paraphrased
verify sources →
Advocates for sample-based measurements over legacy path-based measurements for superior performance and lower complexity, while strongly opposing phase information inclusion in model inputs due to deployment costs and minimal accuracy gains.
Summary
This document contains approximately 15 proposals and 2 conclusions from Ericsson covering AI/ML positioning enhancements including sample-based vs path-based measurements, LOS/NLOS indicators for model output, training data collection procedures, and model inference consistency requirements.
Position
Ericsson advocates FOR: (1) supporting both sample-based and path-based measurements as a compromise solution to avoid blocking progress, (2) reusing existing legacy IEs and signaling frameworks wherever possible to minimize specification impact, (3) label-free monitoring methods for self-contained model performance assessment, and (4) explicit provision of assistance data from LMF to UE for consistency between training and inference. They are pushing AGAINST: (1) supporting CIR/phase information for model input due to implementation complexity and transmitter/receiver phase variations, (2) mandatory reporting of LOS/NLOS indicators when timing information is AI/ML generated, and (3) overly complex new signaling mechanisms when legacy procedures can be reused.
Key proposals
- Proposal 2.1.1.2-1 (Sec 2.1): For sample-based measurement, the starting time of Nt consecutive samples is the timing of the first detected sample within a search window, with the search window determined by a configured offset relative to reference time, reusing existing Search Window Information IE in 38.455
- Proposal 2.1.1.2-2 (Sec 2.1): For sample-based measurement parameters, candidate values for Nt include at least 16, 32, 64, 128; Nt' include at least 9, 16, 32 (where Nt'≤Nt); and k include at least 0…5
- Proposal 2.1.3.2-1 (Sec 2.1): Support both sample-based measurement report and path-based measurement report for AI/ML positioning Case 3b, with LMF signaling to gNB which type is expected to be reported
- Proposal 3.1.2-1 (Sec 3.1): For AI/ML assisted positioning Case 3a, when LOS/NLOS indicator is reported with legacy timing measurements, it provides likelihood of line-of-sight propagation path and can reuse existing LoS/NloS Information IE in 38.455
- Proposal 3.1.3-2 (Sec 3.1): For AI/ML assisted positioning Case 3a, when timing information is reported for Rel-19 Case 3a, LOS/NLOS indicator is not reported if obtained by legacy method
- Proposal 4.1.3-1 (Sec 4.1): For AI/ML positioning Case 3a, regarding time stamp of channel measurement generated by TRP/gNB, existing IE Time Stamp in TS 38.455 can be reused from RAN1 perspective if the channel measurement needs to be reported
- Proposal 4.1.3-2 (Sec 4.1): For AI/ML positioning Case 1, if channel measurement and/or label is reported, both NR-TimeStamp and UTCTime IEs from TS 37.355 are included in the time stamp, with UTCTime as optional field
- Proposal 4.1.3-3 (Sec 4.1): For AI/ML positioning, when label is UE location and reported, label and quality indicator are provided by reusing existing IE using one of the geographic shapes defined in TS 23.032
- Proposal 4.4.2-1 (Sec 4.4): For training data collection, if Part A and Part B are generated by the same entity, they are provided in one report with paired entries grouped together sharing the same time stamp
- Proposal 5.1.3-1 (Sec 5.1): For AI/ML positioning Case 1, at least specified assistance information from DL-TDOA are provided from LMF to UE explicitly, with FFS on whether/how to provide remaining information
- Proposal 5.1.3-2 (Sec 5.1): For AI/ML positioning Case 1, at least specified assistance information are not needed from LMF to UE, neither explicitly nor implicitly, with note that LMF may optionally provide them for supporting legacy UE-based DL-AoD method
- Proposal 6.1.2-1 (Sec 6.1): For AI/ML positioning Cases, support label-free monitoring methods where model inference entity performs self-monitoring without external ground truth label information
- Proposal 6.2.3-1 (Sec 6.2): For model performance monitoring of AI/ML positioning Case 1, support at least Option A-1 and A-2 for label-based monitoring metric calculation, with legacy framework and signaling reused
- Conclusion 2.4.2-1 (Sec 2.4): For Rel-19 AI/ML based positioning for all use cases, there is no consensus in RAN1 to support CIR for determining model input
- Conclusion 2.4.2-2 (Sec 2.4): For Rel-19 AI/ML based positioning for all use cases, there is no consensus in RAN1 to support using one phase value for the first path or first sample only for determining model input