Method for performing federated learning in wireless communication system, and apparatus therefor

Patent metadata

Field	Value
Publication number	US-12556941-B2
Application number	US-202018024889-A
Country	US
Kind code	B2
Filing date	Sep 7, 2020
Priority date	Sep 7, 2020
Publication date	Feb 17, 2026
Grant date	Feb 17, 2026

Abstract

Disclosed is a method by which a base station performs federated learning in a wireless communication system. A method, according to one embodiment of the present disclosure, comprises: receiving, by a base station, a plurality of local parameters from a plurality of terminals; obtaining an integrated parameter on the basis of the plurality of local parameters; and transmitting the integrated parameter to each of the plurality of terminals, wherein at least one bias included in the plurality of local parameters is removed by using the plurality of local parameters. The base station/terminals of the present disclosure may be linked to an artificial intelligence, module, an unmanned aerial vehicle (UAV), a robot, an augmented reality (AR) device, a virtual reality (VR) device, a device related to a 6G service, and the like.

First claim

The invention claimed is:

1 . A method, performed by a base station, comprising:
performing, with a plurality of terminals, a random access procedure for performing a federated learning; wherein the random access procedure comprises: receiving, from the plurality of terminals, a random access preamble; transmitting, to the plurality of terminals, a random access response; receiving, from the plurality of terminals, a connection request message based on the random access response; and transmitting, to the plurality of terminals, a contention resolution message; receiving, from the plurality of terminals, a plurality of local parameters; obtaining an integrated parameter based on the plurality of local parameters; and transmitting, to the plurality of terminals, the integrated parameter, wherein obtaining the integrated parameter includes: removing at least one bias included in the plurality of local parameters by using the plurality of local parameters.

2 . The method of claim 1 , wherein the at least one bias is determined based on a sign of a gradient of the local parameter in which each of the at least one bias is included,
wherein obtaining the integrated parameter includes: determining a sign value of each of a plurality of gradients included in the plurality of local parameters by using at least one removal parameter included in the plurality of local parameters, and wherein removing the at least one bias includes removing the at least one bias based on a sign value of each of the plurality of gradients.

3 . The method of claim 2 , wherein determining the sign value includes:
determining the sign value of each of the plurality of gradients based on a number of at least one removal parameters.

4 . The method of claim 3 , wherein determining the sign value includes:
determining the number of at least one removal parameters based on a number of local parameters having a positive value among the plurality of local parameters.

5 . The method of claim 4 , wherein the at least one parameter is determined based on at least one of a number of the plurality of terminals and a transmission power of the plurality of terminals.

6 . A non-transitory computer-readable media storing instructions that, based on being executed by a processor, cause a base station to perform operations comprising:
performing, with a plurality of terminals, a random access procedure for performing a federated learning; wherein the random access procedure comprises: receiving, from the plurality of terminals, a random access preamble; transmitting, to the plurality of terminals, a random access response; receiving, from the plurality of terminals, a connection request message based on the random access response; and transmitting, to the plurality of terminals, a contention resolution message; receiving, from the plurality of terminals, a plurality of local parameters; obtaining an integrated parameter based on the plurality of local parameters; and transmitting, to the plurality of terminals, the integrated parameter, wherein obtaining the integrated parameter includes: removing at least one bias included in the plurality of local parameters by using the plurality of local parameters.

7 . The non-transitory computer-readable media of claim 6 , wherein the at least one bias is determined based on a sign of a gradient of the local parameter in which each of the at least one bias is included,
wherein obtaining the integrated parameter includes: determining a sign value of each of a plurality of gradients included in the plurality of local parameters by using at least one removal parameter included in the plurality of local parameters, and wherein removing the at least one bias includes removing the at least one bias based on a sign value of each of the plurality of gradients.

8 . The non-transitory computer-readable media of claim 7 , wherein determining the sign value includes:
determining the sign value of each of the plurality of gradients based on a number of at least one removal parameters.

9 . The non-transitory computer-readable media of claim 8 , wherein determining the sign value includes:
determining the number of at least one removal parameters based on a number of local parameters having a positive value among the plurality of local parameters.

10 . The non-transitory computer-readable media of claim 9 , wherein the at least one parameter is determined based on at least one of a number of the plurality of terminals and a transmission power of the plurality of terminals.

11 . A base station comprising:
one or more transceivers; one or more processors; and one or more memories connected to the one or more processors and configured to store instructions, wherein the instructions, based on being executed by the one or more processors, cause the base station to perform operations, wherein the operations include: performing, with a plurality of terminals, a random access procedure for performing a federated learning; wherein the random access procedure comprises: receiving, from the plurality of terminals, a random access preamble; transmitting, to the plurality of terminals, a random access response; receiving, from the plurality of terminals, a connection request message based on the random access response; and transmitting, to the plurality of terminals, a contention resolution message; receiving, from the plurality of terminals, a plurality of local parameters; obtaining an integrated parameter based on the plurality of local parameters; and transmitting, to the plurality of terminals, the integrated parameter, wherein obtaining the integrated parameter includes: removing at least one bias included in the plurality of local parameters by using the plurality of local parameters.

Assignees

• Lg Electronics Inc

Inventors

• Jeon Kijun
• Lee Sangrim
• Lee Hojae
• Kim Yeongjun
• Kim Sungjin

Classifications

• H04W74/0833

  Random access procedures, e.g. with 4-step access · CPC title

• H04L67/10

  in which an application is distributed across nodes in the network (software deployment G06F8/60; multiprogramming arrangements G06F9/46) · CPC title

• H04W88/08

  Access point devices · CPC title

• G06N3/0985

  Hyperparameter optimisation; Meta-learning; Learning-to-learn · CPC title

• G06N3/098 Primary

  Distributed learning, e.g. federated learning · CPC title