R1-2407649
discussion
AI/ML for Positioning Accuracy Enhancement
From Ericsson
Summary
This Ericsson document presents a comprehensive analysis of AI/ML for positioning accuracy enhancement, providing 66 technical proposals across protocol integration, signaling enhancements, model outputs, training data collection, performance monitoring, and model inference. The document strongly advocates for sample-based measurements over path-based measurements and opposes the inclusion of phase information as model input.
Position
Ericsson strongly advocates for sample-based measurements over legacy path-based measurements, demonstrating superior performance with lower complexity and signaling overhead. They firmly oppose including phase information as model input due to complexity, deployment costs, and minimal accuracy gains. Ericsson pushes for self-contained model monitoring without external label assistance and emphasizes total-power PDP over multi-port complex measurements to reduce signaling overhead while maintaining positioning accuracy.
Key proposals
- Proposal 1 (Sec 2.1): RAN1 assumes Case 1 is supported using UE based DL-TDOA procedures as a foundation. It is up to RAN2 to decide whether a set of new procedure is to be defined for Case 1, or extension to DL-TDOA suffices.
- Proposal 2 (Sec 2.2): RAN1 assumes Case 3a is supported using UL-TDOA or multi RTT procedure as a foundation. It is up to RAN3 to decide whether a set of new procedure is to be defined for Case 3a, or extension to existing procedure suffices.
- Proposal 4 (Sec 3.1): For Rel-19 AI/ML based positioning Case 3b/2b, the measurement report of (PDP, DP) support sample-based measurement.
- Proposal 7 (Sec 3.1.1.1): The starting time of the Nt consecutive samples is the timing of the first detected path, quantized to the timing grid.
- Proposal 8 (Sec 3.1.1.2): For Case 3b (and 2a if needed), support {Nt, Nt'} parameter combinations: {9, 9}; {16, 16}, {16, 9}; {32, 16}, {32, 9}; {64, 32}, {64, 16}, {64, 9}.
- Proposal 10 (Sec 3.1.1.5): For the model input types for Case 3b and 3a (if applicable), consider input based on DP or PDP samples containing sample powers summed over all receive antenna ports, i.e., total-power PDP.
- Proposal 12 (Sec 3.2): Do not support phase information for determining model input, including CIR and single phase value for first path/sample.
- Proposal 16 (Sec 3.2.4): RAN1 to down-prioritize the signaling approach(es) and/or measurement definitions to support CIR model input types for Case 3b (1st priority) and Case 2b (2nd priority).
- Proposal 25 (Sec 4.1.3): For AI/ML assisted positioning at gNB (Case 3a), the model output is uplink relative time of arrival (TUL-RTOA).
- Proposal 30 (Sec 5): For Rel-19 AI/ML based positioning, support collecting labelled and optionally un-labelled training data samples.
- Proposal 31 (Sec 5.1.1): For training data collection of AI/ML based positioning, for both Part A and Part B, time stamp is mandatorily provided by the training data generation entity.
- Proposal 50 (Sec 6.1): For AI/ML based positioning, support self-monitoring by model inference entity as a baseline. Furthermore, self-contained monitoring is preferred, where no estimated label needs to be signaled from external entities.
- Proposal 63 (Sec 7.3): For Case 1, to ensure the consistency of NW-side additional condition (e.g., beam configuration of DL PRS) across training and inference, support associated ID for indicating NW conditions/configurations.