Title:
DEVICE AND METHOD FOR PERFORMING, ON BASIS OF CHANNEL INFORMATION, DEVICE GROUPING FOR FEDERATED LEARNING-BASED AIRCOMP OF NON-IID DATA ENVIRONMENT IN COMMUNICATION SYSTEM
Document Type and Number:
WIPO Patent Application WO/2023/068714
Kind Code:
A1
Abstract:
According to various embodiments of the present disclosure, an operation method of a server in a communication system is provided, the method comprising the steps of: receiving location information and channel information from each of multiple terminals; with respect to first terminals corresponding to some of the multiple terminals, performing grouping into a first terminal group on the basis of first location information and first channel information of the first terminals; with respect to second terminals which correspond to some of the multiple terminals and do not belong to the first terminal group, performing grouping into a second terminal group on the basis of second location information and second channel information of the second terminals; transmitting, to the first terminals, first resource allocation information on the basis of the first channel information of the first terminals; transmitting, to the second terminals, second resource allocation information on the basis of the second channel information of the second terminals; transmitting the first resource allocation information and the first channel information corresponding to the first resource allocation information to at least one second terminal that is some of the second terminals; receiving a first local model parameter based on the first channel information from the first terminals; receiving a second local model parameter pre-processed based on the first channel information and the second channel information from at least one second terminal; performing post-processing using the sum of the first local parameter and the second local model parameter so as to learn a global model; and transmitting a report of the global model to the first terminals and at least one second terminal.
More Like This:
Inventors:
LEE TAEHYUN (KR)
LEE SANGRIM (KR)
JEON KIJUN (KR)
JUNG IKJOO (KR)
KIM YEONGJUN (KR)
LEE SANGRIM (KR)
JEON KIJUN (KR)
JUNG IKJOO (KR)
KIM YEONGJUN (KR)
Application Number:
PCT/KR2022/015779
Publication Date:
April 27, 2023
Filing Date:
October 17, 2022
Export Citation:
Assignee:
LG ELECTRONICS INC (KR)
International Classes:
G06N20/20; H04L67/1042; H04L67/1074
Domestic Patent References:
| WO2021064239A1 | 2021-04-08 |
Other References:
WADU MADHUSANKA MANIMEL; SAMARAKOON SUMUDU; BENNIS MEHDI: "Joint Client Scheduling and Resource Allocation Under Channel Uncertainty in Federated Learning", IEEE TRANSACTIONS ON COMMUNICATIONS, IEEE SERVICE CENTER, PISCATAWAY, NJ. USA., vol. 69, no. 9, 11 June 2021 (2021-06-11), PISCATAWAY, NJ. USA. , pages 5962 - 5974, XP011877772, ISSN: 0090-6778, DOI: 10.1109/TCOMM.2021.3088528
TOYOTA MOTOR CORPORATION: "Update to AMMT use case - Uncompressed Federated Learning for image recognition", 3GPP DRAFT; S1-211144, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG1, no. Electronic Meeting; 20210510 - 20210520, 30 April 2021 (2021-04-30), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France , XP051998633
VIVO: "Trained ML Model Sharing in Federated Learning", 3GPP DRAFT; S2-2100859, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG2, no. e-meeting; 20210224 - 20210309, 18 February 2021 (2021-02-18), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France, XP052173357
PASE FRANCESCO; GIORDANI MARCO; ZORZI MICHELE: "On the Convergence Time of Federated Learning Over Wireless Networks Under Imperfect CSI", 2021 IEEE INTERNATIONAL CONFERENCE ON COMMUNICATIONS WORKSHOPS (ICC WORKSHOPS), IEEE, 14 June 2021 (2021-06-14), pages 1 - 7, XP033938797, DOI: 10.1109/ICCWorkshops50388.2021.9473615
TOYOTA MOTOR CORPORATION: "Update to AMMT use case - Uncompressed Federated Learning for image recognition", 3GPP DRAFT; S1-211144, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG1, no. Electronic Meeting; 20210510 - 20210520, 30 April 2021 (2021-04-30), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France , XP051998633
VIVO: "Trained ML Model Sharing in Federated Learning", 3GPP DRAFT; S2-2100859, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG2, no. e-meeting; 20210224 - 20210309, 18 February 2021 (2021-02-18), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France, XP052173357
PASE FRANCESCO; GIORDANI MARCO; ZORZI MICHELE: "On the Convergence Time of Federated Learning Over Wireless Networks Under Imperfect CSI", 2021 IEEE INTERNATIONAL CONFERENCE ON COMMUNICATIONS WORKSHOPS (ICC WORKSHOPS), IEEE, 14 June 2021 (2021-06-14), pages 1 - 7, XP033938797, DOI: 10.1109/ICCWorkshops50388.2021.9473615
Attorney, Agent or Firm:
ROYAL PATENT & LAW OFFICE (KR)
Download PDF: