R1-2407747
discussion
Specification support for positioning accuracy enhancement
Summary
This Tejas Networks document addresses AI/ML enhancements for NR positioning accuracy with 15 formal proposals and 14 observations covering model input parameters, training data collection, and performance monitoring across various positioning use cases.
Position
Tejas Networks advocates FOR leveraging existing Release-17 path-based measurement frameworks and emphasizes the critical need to address receiver implementation dependencies in AI-ML model performance. They push AGAINST overly complex new measurement frameworks, instead favoring reuse of existing parameters and standardized approaches while ensuring consistency between training and inference phases through comprehensive assistance data collection.
Key proposals
- Proposal 1 (Sec 2.1): The effect of receiver implementation on the performance of AI-ML model should be discussed
- Proposal 2 (Sec 2.1.1): Based on different deployment scenarios, the Nt can be supported
- Proposal 4 (Sec 2.1.3): The value of Nt' should be a fraction of the window size (Nt) given by Nt'=α×Nt where α should be either selected by the TRP and signalled to the LMF or vice-versa
- Proposal 6 (Sec 2.1.4): In outdoor scenarios, the value of n0 may need to accommodate a significantly wide range, as 5G networks can connect with UEs located up to 100 km away from the TRPs
- Proposal 8 (Sec 2.2): For reporting path-based measurements, the existing framework from release-17 standards should be utilized. If required, the number of additional measurements should be increased from 8 to a higher value
- Proposal 9 (Sec 4.1): Option-2 should be used to define the quality of Part-A of the training data and Option-1 should be used to define the quality of Part-B of the training data
- Proposal 10 (Sec 4.2): The PRS/SRS configurations, TRP-ID, frequency layer ID, and spatial filter information should be captured as part of the assistance information
- Proposal 11 (Sec 5): The following parameters should be collected as assistance information to ensure consistency between training and inference: TRP-ID or Area-Cell ID, Frequency layer ID, PRS/SRS Configurations COMB Factor, Number of RBs, PRS Resource ID, PRS Resource Set ID
- Proposal 11 (Sec 5): The associated ID should be computed as a function of TRP-ID or Area-Cell ID, frequency layer ID, PRS/SRS Configurations COMB Factor, Number of RBs, PRS Resource ID, PRS Resource Set ID
- Proposal 12 (Sec 6.1): For supporting UE side MPM metric calculation, the assistance data should be provided by the LMF to the UE or TRP for Case-1 and Case-3a respectively
- Proposal 13 (Sec 6.2): The range error or the positioning error should be considered as model performance metric for decision-making
- Proposal 14 (Sec 6.3): Both the MPM metric calculation and the MPM decision should be taken at one entity which can be either LMF or UE/TRP for Case-1 and Case-3a respectively
- Proposal 15 (Sec 6.4): The performance of the AI-ML models should be monitored periodically or aperiodically
- Proposal 3 (Sec 2.1.2): The existing values of k supported for path-based measurements can be reused for sample-based measurements as well
- Proposal 7 (Sec 2.1.4): In indoor scenarios, the values of n0 can lie in the interval for m