Systems and methods for label generation for online learning of channel sounding

The method addresses the challenge of generating accurate CSI labels by using L-RS and CSI-RS measurements to train ML models for improved CSI estimation and adaptation in wireless communications.

WO2025064477A1 Publication Date: 2025-03-27INTERDIGITAL PATENT HOLDINGS INC
6 Cites 1 Cited by

Patent Information

Application Number
PCT/US2024/047171
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-09-21
Filing Date
2024-09-18
Publication Date
2025-03-27

AI Technical Summary

Technical Problem

Current systems face challenges in efficiently generating accurate channel state information (CSI) labels for online learning of channel sounding in wireless communications.

Method used

The method involves a wireless transmit/receive unit (WTRU) receiving learning reference signals (L-RS) and channel state information reference signals (CSI-RS), determining measurements from these signals, and generating CSI labels. These labels are used to train a machine learning (ML) model, which applies to CSI-RS measurements to produce output CSI measurements. The method also includes calculating a loss function and determining model drift to adjust the training process.

Benefits of technology

This approach enables the generation of high-quality CSI labels for training ML models, improving the accuracy of CSI estimation and adapting to changes in the channel environment through online learning and model drift detection.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

A method implemented by a wireless transmit / receive unit (WTRU), including receiving one or more learning reference signals (L-RSs) and receiving one or more channel state information (CSI) reference signals (CSI-RSs). L-RS measurements are determined based on the one or more received L-RSs, and CSI-RS measurements are determined based on the one or more received CSI-RSs. One or more CSI labels are generated based on the L-RS measurements and the CSI-RS measurements. One or more output CSI measurements are generated by applying a machine learning (ML) model to the CSI-RS measurements. The ML model is trained with the one or more generated CSI labels and the one or more generated output CSI measurements.
Need to check novelty before this filing date? Find Prior Art