Method and apparatus for processing traffic by using artificial intelligence model in wireless communication system

1 . A method performed by a user plane function (UPF) entity in a wireless communication system, the method comprising:

receiving multiple traffic packets;

identifying an encrypted traffic packet among the multiple traffic packets;

identifying frame type information of the encrypted traffic packet, based on an artificial intelligence (AI) model embedded in the UPF;

identifying a quality of service (QoS) flow corresponding to the encrypted traffic packet, based on the frame type information; and

transmitting the encrypted traffic packet, based on the QoS flow,

wherein the frame type information includes information indicating whether the encrypted traffic packet includes at least one of an intra frame (I-frame), a predictive frame (P-frame), or a bi-directional frame (B-frame).

2 . The method of claim 1 , further comprising:

receiving, from an artificial intelligence management function (AIMF) entity, information indicating an activation of online training of the AI model; and

performing the online training of the AI model, based on the frame type information.

3 . The method of claim 2 , further comprising:

Receiving, from the AIMF entity, a request for deployment of the AI model;

deploying a version of the AI model, based on the request; and

notifying the AIMF entity of a completion of deployment of the AI model.

4 . The method of claim 2 , further comprising:

receiving, from the AIMF entity, a request for backup of the AI model;

backing up a current version of the AI model, based on the request; and

notifying the AIMF of a completion of backup of the AI model.

5 . The method of claim 2 , further comprising:

receiving, from the AIMF entity, a request for monitoring of the AI model;

determining a performance of the AI model; and

receiving, from the AIMF entity, an indication determined based on the the performance,

wherein the indication includes at least one of an indication to stop using the AI model, an indication to roll back the AI model, or an indication to back up the AI model.

6 . The method of claim 5 , wherein the performance of the AI model is determined based on a classification ratio of traffic packets, or feedback information from at least one entity receiving the encrypted traffic packet.

7 . The method of claim 6 , wherein, in case that the indication to roll back the AI model is received from the AIMF entity, the method further comprises:

identifying a version of another AI model to be changed, based on at least one of a rollback condition included in the indication to roll back the AI model or information on a backed-up AI model;

deactivating the online training of the AI model;

changing a deactivated AI model to the identified version of the another AI model to be changed;

activating an online training of the another AI model;

identifying another frame type information of at least one encrypted traffic packet, based on the another AI model and the online training; and

notifying the AIMF of a completion of an AI model rollback request.

8 . The method of claim 1 , further comprising:

identifying an unencrypted traffic packet among the multiple traffic packets; and

identifying frame type information of the unencrypted traffic packet, based on header information of the unencrypted traffic packet and payload header information.

9 . The method of claim 1 ,

wherein the frame type information is acquired based on the AI model, by using header information of the encrypted traffic packet as an input value, and

wherein the header information includes at least one of a header size, a header reception interval, or an encoding algorithm.

10 . The method of claim 1 , wherein the encrypted traffic packet among the multiple traffic packets is encrypted using a secure real-time transport protocol (SRTP).

11 . A user plane function (UPF) entity in a wireless communication system, the UPF entity comprising:

at least one processor; and

at least one memory, communicatively coupled to the at least one processor, storing instructions executable by the at least one processor individually or in any combination to cause the UPF entity to:

receive multiple traffic packets,

identify an encrypted traffic packet among the multiple traffic packets,

identify frame type information of the encrypted traffic packet, based on an artificial intelligence (AI) model embedded in the UPF,

identify a quality of service (QoS) flow corresponding to the encrypted traffic packet, based on the frame type information, and

transmit the encrypted traffic packet, based on the identified QoS flow,

wherein the frame type information includes information indicating whether the encrypted traffic packet includes at least one of an intra frame (I-frame), a predictive frame (P-frame), or a bi-directional frame (B-frame).

12 . The UPF entity of claim 11 , wherein the instructions further cause the UPF entity to:

receive, from an artificial intelligence management function (AIMF) entity, information indicating an activation of online training of the AI model, and

perform the online training of the AI model, based on the frame type information.

13 . The UPF entity of claim 12 , wherein the instructions further cause the UPF entity to:

receive, from the AIMF entity, a request for deployment of the AI model,

deploy a version of the AI model, based on the request, and

notify the AIMF entity of a completion of deployment of the AI model.

14 . The UPF entity of claim 12 , wherein the instructions further cause the UPF entity to:

receive, from the AIMF entity, a request for backup of the AI model,

back up a current version of the AI model, based on the request, and

notify the AIMF of a completion of backup of the AI model.

15 . The UPF entity of claim 12 , wherein the instructions further cause the UPF entity to:

receive, from the AIMF entity, a request for monitoring of the AI model,

determining a performance of the AI model, and

receive, from the AIMF entity, an indication determined based on the performance,

wherein the indication includes at least one of an indication to stop using the AI model, an indication to roll back the AI model, or an indication to back up the AI model.

16 . The UPF entity of claim 15 , wherein the performance of the AI model is determined based on a classification ratio of traffic packets, or feedback information from at least one entity receiving the encrypted traffic packet.

17 . The UPF entity of claim 16 , wherein, in case that the indication to roll back the AI model is received from the AIMF entity, the instructions further cause the UPF entity to:

identify a version of another AI model to be changed, based on at least one of a rollback condition included in the indication to roll back the AI model, or information on a backed-up AI model,

deactivate the online training of the AI model,

change a deactivated AI model to the identified version of the another AI model to be changed,

activate an online training of the another AI model, identify another frame type information of the at least one encrypted traffic packet, based on the another AI model and the online training, and

notify the AIMF of a completion of an AI model rollback request.

18 . The UPF entity of claim 11 , wherein the instructions further cause the UPF entity:

identify an unencrypted traffic packet among the multiple traffic packets, and

identify frame type information of the unencrypted traffic packet, based on header information of the unencrypted traffic packet and payload header information.

19 . The UPF entity of claim 11 , wherein, the frame type information is acquired based on the AI model, by using header information of the encrypted traffic packet as an input value, and

wherein the header information includes at least one of a header size, a header reception interval, or an encoding algorithm.

20 . The UPF entity of claim 11 , wherein the encrypted traffic packet among the multiple traffic packets is encrypted using a secure real-time transport protocol (SRTP).