PRS/SRS Measurement Compression for Low-Overhead AI/ML Positioning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing wireless communication systems face inefficiencies in reporting channel impulse response (CIR) and channel frequency response (CFR) measurements due to high over-the-air and wired reporting overhead, particularly in AI/ML positioning, which affects the accuracy and efficiency of positioning processes.

Innovation Solution

Implementing AI/ML models at user equipment (UE) and network nodes to compress CIR/CFR measurements, allowing for efficient reporting of compressed measurements to reduce overhead and enhance AI/ML inference accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If CIR/CFR measurements are reported in full detail for AI/ML positioning, then positioning accuracy is improved, but reporting overhead increases significantly

Engineering Contradiction:

Improvepositioning accuracyVSAvoidreporting overhead

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts and reports only the most significant components of CIR/CFR measurements rather than full detailed measurements. By identifying and reporting key features that contribute most to positioning accuracy, the system achieves good positioning performance with reduced reporting overhead.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms the measurement data by changing its representation parameters - converting full CIR/CFR measurements into compressed forms such as peak indices, amplitudes, and selected channel parameters. This parameter transformation maintains essential positioning information while significantly reducing the quantity of data to be reported.

Inventive Principle:

Principle #35Parameter changes

2Quantity of substance

If compressed CIR/CFR measurements are reported, then reporting overhead is reduced, but positioning accuracy may deteriorate

Engineering Contradiction:

Improvereporting overheadVSAvoidpositioning accuracy

Core Design Contradiction:

Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent employs feedback mechanisms where the system evaluates the quality and sufficiency of compressed measurement data. Based on feedback from positioning performance evaluation, the system can adjust compression parameters, select different feature sets, or request additional measurements to maintain positioning accuracy while keeping overhead low.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The compression strategy is made dynamic and adaptive rather than static. The system adjusts the level of compression and selection of reported parameters based on current positioning requirements, channel conditions, and AI/ML model needs, optimizing the balance between overhead reduction and accuracy maintenance.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS12531765B2Compressing and reporting PRS/SRS measurements for LMF-sided AI/ML positioning

Publication Date: 2026.01.20 QUALCOMM INC

AI summary

Aspects presented herein may improve the efficiency and performance of artificial intelligence (AI)/machine learning (ML) (AI/ML) positioning by enabling a user equipment (UE) to compress downlink (DL) reference signal measurements to reduce reporting overhead for the DL reference signal measurements. In one aspect, a UE performs at least one channel impulse response (CIR) measurement or at least one channel frequency response (CFR) measurement for a set of positioning reference signals (PRSs). The UE compresses the at least one CIR measurement or the at least one CFR measurement for the set of PRSs. The UE reports, for a network entity, one or more of the at least one compressed CIR measurement or the at least one compressed CFR measurement for the set of PRSs.