Access storage and management for federated machine learning

A Permissioned Distributed Ledger network addresses the security concerns in federated learning by authenticating and authorizing network functions, ensuring secure ML model distribution and collection, thus enhancing the integrity and privacy of ML model data for UE involvement in wireless communications.

WO2024260587A1PCT designated stage expired Publication Date: 2024-12-26LENOVO (SINGAPORE) PTE LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
PCT/EP2024/054188
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2024-02-09
Filing Date
2024-02-19
Publication Date
2024-12-26

AI Technical Summary

Technical Problem

Current 3GPP specifications for federated learning in wireless communications lack secure mechanisms for the distribution and collection of Machine Learning (ML) model information, particularly for User Equipment (UE) involvement, which is crucial for ensuring privacy and security in the sensitive FL process.

Method used

The implementation of a Permissioned Distributed Ledger (PDL) network facilitates secure access, storage, and management of ML models and their parameters by authenticating and authorizing network functions, enabling secure distribution and collection of ML model information through smart contracts, thus ensuring integrity and privacy in the FL process.

Benefits of technology

This solution provides a secure and optimized framework for the FL process, ensuring the integrity and privacy of ML model data, enabling secure distribution and collection, and supporting secure UE involvement, which is particularly advantageous for future networks like 6G with distributed architectures.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure EP2024054188_26122024_PF_FP_ABST
    Figure EP2024054188_26122024_PF_FP_ABST
Patent Text Reader

Abstract

Various aspects of the present disclosure relate to a PDL application function for use in a wireless network, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the PDL application function to: receive from a server network function a first model storage request message comprising a ML model ID and at least one of (i) the ML model, (ii) a download address from which the ML model may be downloaded, and (iii) model parameters of the ML model; authenticate and authorize the server network function; send to a PDL node a second model storage request message comprising the ML model and the ML model ID; receive from the PDL node a model storage response message comprising a ML model block address of the ML model stored by the PDL node; and forward to the server network function the model storage response message comprising the ML model block address.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Methods, architectures, apparatuses and systems directed to blockchain-enabled model storage, sharing and deployment for supporting distrubuted learning

    US20240045851A1

  • Authorization and authentication of machine learning model transfer

    WO2023215720A1

  • Methods, architectures, apparatuses and systems enabling artificial intelligence applications in networks

    WO2024006178A1