Ai / ML model performance management method and device therefor

The proposed method addresses the challenge of managing AI/ML model performance in wireless communication systems by using CFRA for resource allocation and reporting, ensuring efficient performance management and lifecycle updates of AI/ML models in 5G and beyond.

WO2026059418A1PCT designated stage Publication Date: 2026-03-19SOGANG UNIV RES & BUSINESS DEV FOUND
View PDF 5 Cites 0 Cited by

Patent Information

Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2025-09-11
Publication Date
2026-03-19

AI Technical Summary

Technical Problem

There is a lack of effective methods for managing and reporting AI/ML model performance metrics in wireless communication systems, particularly when they fall below standard levels, which hampers efficient performance management and lifecycle management (LCM) of AI/ML models in 5G and future mobile communication systems.

Method used

A method and apparatus for managing AI/ML model performance by enabling UE devices to receive monitoring configuration information, request resources via contention-free random access (CFRA) when performance indicators drop below standards, and report performance metrics through allocated resources, incorporating CSI-related and LCM configuration information for beam management and CSI feedback.

Benefits of technology

Enables efficient performance management of AI/ML models by allowing rapid reporting and updating/resetting of models, thereby providing an effective LCM environment in mobile communication systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure KR2025095564_19032026_PF_FP_ABST
    Figure KR2025095564_19032026_PF_FP_ABST
Patent Text Reader

Abstract

An artificial intelligence (AI) / machine learning (ML) model performance management method and a device therefor are described. To this end, a communication method of a user equipment (UE) comprises: receiving AI / ML-related monitoring configuration information from a network; monitoring an AI / ML model of the UE for at least one of beam management or channel status information (CSI) feedback on the basis of the monitoring configuration information; when a prescribed performance index of the AI / ML model of the UE does not meet a criterion, requesting a resource from the network by transmitting a specific sequence to the network; and reporting information about the prescribed performance index via the resource allocated from the network.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Methods, devices, and computer readable medium for communication

    US20240291612A1

  • Model management for channel state estimation and feedback

    WO2023206207A1

  • Method and procedure for ai based CSI feedback with CSI prediction

    WO2023236098A1

  • Model monitoring for ML-based CSI compression

    WO2024065833A1

  • Methods of life cycle management for artificial intelligence (AI)-based channel state information (CSI) prediction

    WO2024167797A1