Granted Patent B2

US 12,574,300 · App. 18/692,769 · Granted Mar 10, 2026

Inventor: Xiaoyan Duan (Beijing, CN)

Assignee: DATANG MOBILE COMMUNICATIONS EQUIPMENT CO., LTD.

H04L41/16G06N3/098H04L41/0893

Abstract

Claims (96)

1 . A federated learning group processing method, applied to a first functional entity and comprising:

determining characteristic information of a federated learning (FL) group;

determining a second functional entity according to the characteristic information of the FL group; and

adding the second functional entity to the FL group;

wherein the method further comprises:

determining whether the second functional entity meets conditions for leaving the FL group;

if the second functional entity meets the conditions for leaving the FL group, removing the second functional entity from the FL group.

2 . The federated learning group processing method according to claim 1 , wherein the determining the characteristic information of the FL group comprises:

receiving a first request sent by an application functional entity or a network exposure functional entity, and determining the characteristic information of the FL group according to application information or service information which is carried in the first request and related to the FL group;

or,

receiving a second request sent by a policy control functional entity, and determining the characteristic information of the FL group according to a control policy which is carried in the second request and related to the FL group.

3 . The federated learning group processing method according to claim 1 , wherein the characteristic information of the FL group comprises first information, and/or identification information of functional entities allowed to join the FL group;

the first information comprises any one or more of the following:

information of an application or a service corresponding to the FL group;

area information corresponding to the FL group;

network slice information corresponding to the FL group;

data network information corresponding to the FL group.

4 . The federated learning group processing method according to claim 3 , wherein the determining the second functional entity according to the characteristic information of the FL group comprises:

determining the second functional entity that meets a first condition, according to the characteristic information of the FL group;

wherein the first condition comprises one or a combination of the following conditions:

supporting the application or service corresponding to the FL group;

serving the data network, network slice or area corresponding to the FL group;

accessing the data network or network slice corresponding to the FL group;

located in the area corresponding to the FL group; and

the identification information being comprised in the identification information of the functional entities allowed to join the FL group.

5 . The federated learning group processing method according to claim 3 , wherein the adding the second functional entity to the FL group comprises:

sending an FL request to the second functional entity, wherein the FL request is configured to request the second functional entity to join the FL group, the FL request carries information of the first functional information and the first information of the FL group;

receiving a first FL response of joining the FL group sent by the second functional entity.

6 . The federated learning group processing method according to claim 3 , wherein the determining the second functional entity according to the characteristic information of the FL group specifically comprises:

receiving an FL request sent by the second functional entity, wherein the FL request is configured to request to join the FL group, the FL request carries the first information of the FL group and identification information of the second functional entity;

determining the FL group to which the second functional entity is to join, according to the first information;

determining whether the identification information of the second functional entity is comprised in the identification information of the functional entities allowed to join the FL group;

if so, determining that the second functional entity is the second functional entity corresponding to the characteristic information of the FL group.

7 . The federated learning group processing method according to claim 1 , further comprising:

storing information of the FL group, wherein the information comprises one or more of the following:

information of the first functional entity;

information of the second functional entity;

identification information of the FL group;

information of an application or a service corresponding to the FL group;

network slice information corresponding to the FL group;

data network information corresponding to the FL group; and

information of a machine learning (ML) model of the FL group.

8 . The federated learning group processing method according to claim 7 , wherein the removing the second functional entity from the FL group comprises:

sending an FL request to the second functional entity, wherein the FL request is configured to request the second functional entity to leave the FL group;

receiving a FL response of leaving the FL group sent by the second functional entity;

or

the removing the second functional entity from the FL group comprises:

removing information of the second functional entity from the information of the FL group.

9 . The federated learning group processing method according to claim 1 , further comprising:

receiving updated characteristic information of the FL group sent by a third functional entity;

the determining whether the second functional entity meets the conditions for leaving the FL group comprises:

according to the updated characteristic information of the FL group, determining whether the second functional entity meets the conditions for leaving the FL group.

10 . The federated learning group processing method according to claim 1 , wherein the second functional entity meeting the conditions for leaving the FL group comprises one or more of the following situations:

the second functional entity no longer supports the application or service corresponding to the FL group;

the second functional entity no longer serves the data network, the network slice or the area corresponding to the FL group;

the second functional entity no longer accesses the data network or the network slice corresponding to the FL group;

the second functional entity leaves the area corresponding to the FL group;

the identification information of the second functional entity is no longer comprised in the identification information of the functional entities allowed to join the FL group;

an FL request sent by the second functional entity is received, wherein the FL request is configured to request to leave the FL group;

the second functional entity has an abnormal behavior, wherein the abnormal behavior comprises: a frequency of establishing connections between the second functional entity and the first functional entity is higher than a first threshold, or an amount of data transmitted between the second functional entity and the first functional entity exceeds a second threshold.

11 . A functional entity, wherein the functional entity is a first functional entity and comprising a memory, a transceiver and a processor; wherein

the memory is configured to store computer programs; the transceiver is configured to send and receive data under a control of the processor; the processor is configured to read the computer programs in the memory to perform the federated learning group processing method according to claim 1 .

12 . A federated learning group processing method, wherein a federated learning (FL) group comprises a first functional entity, and the method is applied to a second functional entity and comprises:

obtaining characteristic information of the FL group;

determining whether a condition for joining the FL group is met, based on the characteristic information of the FL group;

when the condition for joining the FL group is met, joining the FL group;

wherein the method further comprises:

determining whether the second functional entity meets conditions for leaving the FL group;

when the second functional entity meets the conditions for leaving the FL group, sending an FL request to the first functional entity, wherein the FL request is configured to request to leave the FL group.

13 . The federated learning group processing method according to claim 12 , wherein the obtaining the characteristic information of the FL group comprises:

receiving a first request sent by an application functional entity or a network exposure functional entity, and obtaining the characteristic information of the FL group according to application information or service information which is carried in the fifth-first request and related to the FL group;

or,

receiving a second request sent by a policy control functional entity, and obtaining the characteristic information of the FL group according to a control policy which is carried in the second request and related to the FL group.

14 . The federated learning group processing method according to claim 12 , wherein prior to the joining the FL group, the method further comprises:

sending a first query request to a third functional entity, wherein the first query request carries information of the FL group;

receiving a first query response sent by the third functional entity, wherein the first query response carries information of at least one first functional entity that satisfies the information of the FL group;

based on the information of the at least one first functional entity, determining the first functional entity that is to establish the FL group with the second functional entity;

wherein the information of the FL group comprises at least one of the following:

information of an application or a service corresponding to the FL group;

network slice information corresponding to the FL group;

data network information corresponding to the FL group;

information of a machine learning (ML) model of the FL group.

15 . The federated learning group processing method according to claim 14 ,

wherein based on the information of the at least one first functional entity, determining the first functional entity that is to establish the FL group with the second functional entity comprises:

based on the information of the at least one first functional entity, selecting the first functional entity that satisfies a condition from the at least one first functional entity as the first functional entity to establish the FL group with the second functional entity;

wherein the condition comprises at least one of the following:

belonging to a same local network as the second functional entity;

serving an area where the second functional entity is located; and

having a shortest distance to the second functional entity among the at least one first functional entity.

16 . The federated learning group processing method according to claim 12 , wherein the second functional entity meeting the conditions for leaving the FL group comprises one or more of the following situations:

the second functional entity no longer supports the application or service corresponding to the FL group;

the second functional entity no longer serves the data network, the network slice or the area corresponding to the FL group;

the second functional entity no longer accesses the data network or the network slice corresponding to the FL group; and

the second functional entity leaves the area corresponding to the FL group.

17 . A functional entity, wherein the functional entity is a second functional entity and comprising a memory, a transceiver and a processor; wherein

the memory is configured to store computer programs; the transceiver is configured to send and receive data under a control of the processor; the processor is configured to read the computer programs in the memory to perform the federated learning group processing method according to claim 12 .

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (10)