Optimizing model performance of user equipments (UES) based on collective intelligence at base station

By reconstructing and optimizing CSI feedback at the base station and sharing model parameters among UEs, the variability in CSI reporting among UEs with similar capabilities is addressed, leading to improved CSI prediction and compression KPIs and enhanced communication performance.

WO2025116946A1 Publication Date: 2025-06-05RAKUTEN SYMPHONY INC +1
5 Cites 1 Cited by

Patent Information

Application Number
PCT/US2024/017212
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-11-30
Filing Date
2024-02-26
Publication Date
2025-06-05

AI Technical Summary

Technical Problem

UEs with similar capabilities and reporting similar Reference Signal Received Power (RSRP) and ground truth in the same cell often provide varying CSI feedback due to differences in training and/or inference algorithms used, impacting CSI prediction Key Performance Indicators (KPIs) and CSI compression KPIs.

Method used

A base station is configured to transmit a first segment of CSI-RS to multiple UEs, receive compressed CSI feedback predicted using a model associated with each UE, reconstruct CSI-RS for each UE, identify UEs providing optimized CSI based on a predefined threshold, and transmit assistance information and model parameters to other UEs for training.

Benefits of technology

This approach optimizes model performance of UEs by improving CSI prediction and compression KPIs, enhancing overall communication reliability and efficiency by identifying and sharing best practices in CSI feedback among UEs.

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

Abstract

Embodiments of present disclosure disclose optimizing model performance of UEs based on collective intelligence at base-station. The base-station transmits first segment CSI-RS to plurality of UEs in an area via DL channel corresponding to each of the plurality of UEs. The base-station receives compressed CSI feedback comprising second segment of predicted CSI-RS and one or more CSI parameters from each of the plurality of UEs. The base-station reconstructs CSI-RS associated with each of the plurality of UEs based on CSI feedback received from corresponding UE and first segment. The base-station identifies one or more UEs from plurality of UEs providing an optimized CSI based on predefined threshold-value, using reconstructed CSI-RS. Thereafter, base-station transmits, assistance-information and model-parameters associated with one or more UEs to each of the plurality of UEs, for training corresponding model. Thus, present disclosure improves KPIs of UEs, based on model-parameters and assistance-information of best performing UE(s).
Need to check novelty before this filing date? Find Prior Art