Method and apparatus for supporting federated learning in wireless communication system

1 . A method for supporting federated learning (FL) by an application federated learning support function (AFLSF) in a wireless communication system, the method comprising:

receiving, from an application function (AF), a message for requesting assistance for an FL operation;

obtaining, from a network data analytics function (NWDAF), analytics to get assistance on determination of an FL configuration recommendation;

obtaining information related to at least one user equipment (UE) from a session management function (SMF) and an access and mobility management function (AMF);

deriving a FL configuration recommendation based on the analytics and the obtained information related to the at least one UE;

transmitting, to the AF, a response message including at least one parameter related to the FL configuration recommendation; and

providing an updated recommendation to the AF to select a UE for the FL operation, based on a validation request message for an artificial intelligence/machine learning (AI/ML) session identifier (AMSID) from the SMF.

2 . The method of claim 1 , wherein the message for requesting the assistance for the FL operation comprises at least one of:

a list or group of UEs participating in the FL operation;

a maximum number of learning iterations related to model exchanges between UEs and an application server;

a time window for model aggregation indicating maximum time that the application server can wait to receive local models from UEs to provide an update of a global model;

a minimum number of awaited models indicating number of local UE models which the application server requires to receive within the time window for the model aggregation to provide the update of the global model;

continuous training flag indicating whether UEs are allowed to continue performing training once the UEs have shared a local model and before the UE receive the update of the global model;

quality of service (QoS) requirements requested by the AF for the FL operation;

the AMSID to be used for the FL operation;

at least one target areas where UEs should be located to be selected for the FL operation; and

operation time validity indicating time during which the FL operation may be performed.

3 . The method of claim 1 , wherein obtaining the information related to the at least one UE from the SMF and the AMF comprise consuming an event exposure service from the SMF and the AMF to obtain the information related to the at least one UE.

4 . The method of claim 3 , further comprising:

obtaining a user plane status from the SMF by consuming the event exposure service from the SMF,

wherein the FL configuration recommendation is derived further based on the user plane status.

5 . The method of claim 1 , wherein the response message comprises at least one of:

a list or group of UEs participating in the FL operation;

a maximum number of learning iterations related to model exchanges between UEs and an application server;

a time window for model aggregation indicating maximum time that the application server can wait to receive local models from UEs to provide an update of a global model;

a minimum number of awaited models indicating number of local UE models which the application server requires to receive within the time window for the model aggregation to provide the update of the global model;

continuous training flag indicating whether UEs are allowed to continue performing training once the UEs have shared a local model and before the UE receive the update of the global model;

quality of service (QoS) requirements requested by the AF for the FL operation;

the AMSID to be used for the FL operation;

at least one target areas where UEs should be located to be selected for the FL operation; and

operation time validity indicating time during which the FL operation may be performed.

6 . The method of claim 1 , further comprising:

receiving, from the AF, information indicating the UE to which the AMSID should be provided, and information indicating conditions that need to be met before the UE is allowed to engage in the FL; and

allocating the AMSID to the UE based on determining that the conditions are met before the UE is allowed to engage in the FL.

7 . The method of claim 6 , wherein the conditions that need to be met before the UE is allowed to engage in the FL comprises at least one of:

a location of the UE where the FL is allowed;

time of a day at which the UE connect to be part of a FL procedure;

duration and frequency of the FL operation; and

whether a certain number of UEs need to be available at a same time.

8 . The method of claim 6 , further comprising:

receiving, from the SMF, the validation request message for the AMSID to be used by the UE to request to use the FL based on the SMF receiving a session management message including the AMSID;

validating whether the AMSID can be used for the UE; and

transmitting, to the SMF, a validation response message indicating that the UE is authorized to join the FL operation using the AMSID.

9 . The method of claim 8 , wherein validating whether the AMSID can be used for the UE is based on previous interactions of the AFLSF with the AF.

10 . The method of claim 8 , wherein the validation response message comprises the AMSID and required QoS for the FL operation.

11 . An apparatus of an application federated learning support function (AFLSF) for supporting federated learning (FL) in a wireless communication system, the apparatus comprising:

a transceiver; and

a controller coupled to the transceiver, wherein the controller is configured to:

receive, from an application function (AF), a message for requesting assistance for an FL operation;

obtain, from a network data analytics function (NWDAF), analytics to get assistance on determination of an FL configuration recommendation;

obtain information related to at least one user equipment (UE) from a session management function (SMF) and an access and mobility management function (AMF);

derive a FL configuration recommendation based on the analytics and the obtained information related to the at least one UE;

transmit, to the AF, a response message including at least one parameter related to the FL configuration recommendation; and

provide an updated recommendation to the AF to select a UE for the FL operation, based on a validation request message for an artificial intelligence/machine learning (AI/ML) session identifier (AMSID) from the SMF.

12 . The apparatus of claim 11 , wherein the message for requesting the assistance for the FL operation comprises at least one of:

a list or group of UEs participating in the FL operation;

a maximum number of learning iterations related to model exchanges between UEs and an application server;

a time window for model aggregation indicating maximum time that the application server can wait to receive local models from UEs to provide an update of a global model;

a minimum number of awaited models indicating number of local UE models which the application server requires to receive within the time window for the model aggregation to provide the update of the global model;

continuous training flag indicating whether UEs are allowed to continue performing training once the UEs have shared a local model and before the UE receive the update of the global model;

quality of service (QoS) requirements requested by the AF for the FL operation;

the AMSID to be used for the FL operation;

at least one target areas where UEs should be located to be selected for the FL operation; and

operation time validity indicating time during which the FL operation may be performed.

13 . The apparatus of claim 11 , wherein for obtaining the information related to the at least one UE from the SMF and the AMF, the controller is configured to consume an event exposure service from the SMF and the AMF to obtain the information related to the at least one UE.

14 . The apparatus of claim 11 , wherein the response message comprises at least one of:

a list or group of UEs participating in the FL operation;

a maximum number of learning iterations related to model exchanges between UEs and an application server;

a time window for model aggregation indicating maximum time that the application server can wait to receive local models from UEs to provide an update of a global model;

a minimum number of awaited models indicating number of local UE models which the application server requires to receive within the time window for the model aggregation to provide the update of the global model;

continuous training flag indicating whether UEs are allowed to continue performing training once the UEs have shared a local model and before the UE receive the update of the global model;

quality of service (QoS) requirements requested by the AF for the FL operation;

the AMSID to be used for the FL operation;

at least one target areas where UEs should be located to be selected for the FL operation; and

operation time validity indicating time during which the FL operation may be performed.

15 . The apparatus of claim 11 , wherein the controller is further configured to:

receive, from the AF, information indicating the UE to which the AMSID should be provided, and information indicating conditions that need to be met before the UE is allowed to engage in the FL; and

allocate the AMSID to the UE based on determining that the conditions are met before the UE is allowed to engage in the FL.

16 . The apparatus of claim 15 , wherein the controller is further configured to:

receive, from the SMF, the validation request message for the AMSID to be used by the UE to request to use the FL based on the SMF receiving a session management message including the AMSID;

validate whether the AMSID can be used for the UE; and

transmit, to the SMF, a validation response message indicating that the UE is authorized to join the FL operation using the AMSID.

17 . The apparatus of claim 16 , wherein the controller is configured to validate whether the AMSID can be used for the UE based on previous interactions of the AFLSF with the AF.

18 . The apparatus of claim 16 , wherein the validation response message comprises the AMSID and required QoS for the FL operation.

19 . A non-transitory computer readable storage medium storing instructions, which, executed by a controller of an apparatus of an application federated learning support function (AFLSF), causes the apparatus to perform operations, the operations comprising:

receiving, from an application function (AF), a message for requesting assistance for an FL operation;

obtaining, from a network data analytics function (NWDAF), analytics to get assistance on determination of an FL configuration recommendation;

obtaining information related to at least one user equipment (UE) from a session management function (SMF) and an access and mobility management function (AMF);

deriving a FL configuration recommendation based on the analytics and the obtained information related to the at least one UE;

transmitting, to the AF, a response message including at least one parameter related to the FL configuration recommendation; and

providing an updated recommendation to the AF to select a UE for the FL operation, based on a validation request message for an artificial intelligence/machine learning (AI/ML) session identifier (AMSID) from the SMF.

20 . The non-transitory computer readable storage medium of claim 19 , wherein the message for requesting the assistance for the FL operation comprises at least one of:

a list or group of UEs participating in the FL operation;

a maximum number of learning iterations related to model exchanges between UEs and an application server;

a time window for model aggregation indicating maximum time that the application server can wait to receive local models from UEs to provide an update of a global model;

a minimum number of awaited models indicating number of local UE models which the application server requires to receive within the time window for the model aggregation to provide the update of the global model;

continuous training flag indicating whether UEs are allowed to continue performing training once the UEs have shared a local model and before the UE receive the update of the global model;

quality of service (QoS) requirements requested by the AF for the FL operation;

the AMSID to be used for the FL operation;

at least one target areas where UEs should be located to be selected for the FL operation; and

operation time validity indicating time during which the FL operation may be performed.