Granted Patent B2

US 12,388,717 · App. 17/904,938 · Granted Aug 12, 2025

Inventors: Yoonseon Han (Suwon-si, KR); Jungshin Park (Suwon-si, KR)

Assignee: Samsung Electronics Co., Ltd.

H04L41/145H04L41/16H04L41/34H04L41/5058H04L43/06

Abstract

Claims (23)

1. A method of a network data analytics function (NWDAF) in a wireless communication system, the method comprising:

receiving, from a network function (NF), a request message for analytics information;

identifying that federated learning (FL) is required;

transmitting, to a network repository function (NRF), a discovery message for discovering at least another NWDAF for FL that supports service related FL, wherein the discovery message comprises information about FL capability;

receiving, from the NRF, a response message with respect to the discovery message, wherein the response message with respect to the discovery message comprises information about the at least another NWDAF for FL;

selecting a target NWDAF for FL;

transmitting, to the target NWDAF for FL, a request message comprising analytics identifier (ID), machine learning (ML) model ID, and a ML model;

receiving, from the target NWDAF for FL, information about the ML model which is trained by the target NWDAF for FL; and

obtaining an integrated model based on the information about the model which is trained by the target NWDAF for FL.

2. The method of claim 1 , wherein the discovery message further comprises an analytics identifier (ID) and a target NF type information corresponding to NWDAF.

3. A network data analytics function (NWDAF) in a wireless communication system, the NWDAF comprising:

a transceiver within programmable data processing equipment; and

at least one processor within the programmable data processing equipment, the at least one processor configured to:

receive, from a network function (NF), a request message for analytics information, through the transceiver;

identify that federated learning, FL, is required;

transmit, through the transceiver to a network repository function (NRF), a discovery message for discovering at least another NWDAF for FL that supports service related FL, wherein the discovery message comprises information about FL capability;

receive, from the NRF, a response message with respect to the discovery message, through the transceiver, wherein the response message with respect to the discovery message comprises information about the at least another NWDAF for FL;

based on the information about the at least another NWDAF, transmit, to the at least another NWDAF, a request message for processing the analytics information, through the transceiver;

select a target NWDAF for FL;

transmit, through the transceiver to the target NWDAF for FL, a request message comprising an analytics identifier (ID), a machine learning (ML) model ID, and a ML model;

receive, through the transceiver from the target NWDAF for FL, information about the ML model which is trained by the target NWDAF for FL; and

obtain an integrated model based on the information about the model which is trained by the target NWDAF for FL.

4. The NWDAF of claim 3 , wherein the discovery message further comprises an analytics identifier (ID) and a target NF type information corresponding to NWDAF.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (36)