Method, device and readable storage medium for analyzing model transmission status in a subscription network

Abstract

The present disclosure provides a method, device, and storage medium for analyzing a model transmission state in a subscription network. The method includes an application function AF sending a first message to a network data analysis function NWDAF directly or via a network capability opening function NEF, and the AF receiving analysis information of an AI / ML model transmission state determined by the NWDAF based on data of an AI / ML model transmission state sent from another network function 5GC NF(s) of a 5G core network sent directly or via the NEF, the analysis information being used to adjust a network policy parameter and / or information of an application layer model. The present disclosure can effectively analyze the AI / ML model transmission state, and further allows the network to effectively adjust a network transmission policy based on the AI / ML model transmission state, and allows a third party to obtain the analysis of the AI / ML model transmission state to adjust application layer information.

Description

[Technical Field]

[0001] The present disclosure relates to the field of communication technology, and more particularly to a method, an apparatus, and a readable storage medium for analyzing a model transmission state in a subscription network. [Background technology]

[0002] In recent years, technological breakthroughs in artificial intelligence have led to increasingly widespread applications of artificial intelligence. However, due to strict energy consumption, computational, and memory cost limitations on mobile devices, heavyweight artificial intelligence (AI) / machine learning (ML) models (hereafter referred to as AI / ML models) cannot be executed on the devices. Therefore, the current approach is to transmit the inferences of many AI / ML models from mobile devices to the cloud or other devices, which means that the AI / ML models need to be transmitted to the cloud or other devices.

[0003] In addition, the requirements for transmitting AI / ML models are increasing, taking into consideration the privacy protection of transmitted data and the stress reduction of network transmitted data. As a channel for transmitting AI / ML models, the 5G system must support the disclosure of monitoring and status information about AI-ML sessions to third parties in order to meet the requirements for transmitting AI / ML models in TS 22.261, which SA1 #93e passes, and to enhance the intelligent capabilities of 5G networks.

[0004] However, in the prior art, it is not possible to effectively analyze the AI / ML model transmission status, and furthermore, it is not possible for the network to effectively adjust the network transmission policy based on the AI / ML model transmission status, and it is not possible for a third party to obtain an analysis of the AI / ML model transmission status for adjusting application layer information. Summary of the Invention [Problem to be solved by the invention]

[0005] The present disclosure provides a method, device, and readable storage medium for analyzing model transmission status in a subscription network, which solves the technical problems that the AI / ML model transmission status cannot be effectively analyzed, the network cannot effectively adjust network transmission policies based on the AI / ML model transmission status, and a third party cannot obtain the analysis of the AI / ML model transmission status, making it impossible to adjust application layer information. [Means for solving the problem]

[0006] In a first aspect, the present disclosure provides a method for analyzing a model transmission state in a subscription network, applied to an application function AF, the method comprising: Sending a first message to a Network Data Analysis Function (NWDAF), directly or via a Network Capabilities Opening Function (NEF), to request analysis information of the artificial intelligence / machine learning (AI / ML) model transmission status in the subscription network; and receiving, directly or via the NEF, analysis information of the AI / ML model transmission state determined by the NWDAF based on data of the AI / ML model transmission state received from other network functions (5GC NF(s)) of the 5G core network, transmitted from the NWDAF; The analyzed information is used to adjust network policy parameters and / or application layer model information.

[0007] In a possible design, the AI / ML model transmission status data is obtained by the NWDAF sending a second message to the 5GC NF(s) based on parameters requested in the received first message, and the second message is used to collect data for analyzing the AI / ML model transmission status in the network.

[0008] In a possible design, the parameters requested in the first message include a network data analysis identifier, an identifier of one user equipment (UE) or a set of UEs that will receive the AI / ML model, or any UE that meets the analysis criteria. Identifier , including at least one of an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a protocol data unit (PDU) session that transmits the quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of the UE that uses the AI / ML model; an identifier of an application that uses the AI / ML model; an identifier of a quality of service stream that transmits the AI / ML model; an uplink bit rate for transmitting the AI / ML model and a downlink bit rate for transmitting the AI / ML model; an uplink packet delay for the AI / ML model and a downlink packet delay for the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; the number of packet retransmissions of the AI / ML model; a data collection time; a time for the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; a data network of a PDU session for transmitting the quality of service stream of the AI / ML model; and a service flow used for the AF; the analysis information includes at least one of: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information using the AI / ML model; a validity period of the analysis result; a user plane function (UPF) that provides AI / ML model transmission; a data network name of the PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; an identifier of the quality of service stream for transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; and the number of packet retransmissions of the AI / ML model. If the AI / ML model performs federated learning, the parameters requested in the first message further include federated learning group information, and the federated learning group information includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that it provides an AI / ML model or participates in the federated learning.

[0009] In an embodiment of the present disclosure, a request for analysis information of the AI / ML model transmission state in a subscription network is sent to an NWDAF, analysis information determined based on the collected 5GC NF(s) data is received from the NWDAF, and network policy parameters and / or application layer model information are adjusted based on the analysis information, thereby achieving effective analysis of the AI / ML model transmission state, and further allowing the network to effectively adjust network transmission policies based on the AI / ML model transmission state, and allowing a third party to obtain the analysis of the AI / ML model transmission state and adjust application layer information.

[0010] In one possible design, after receiving the analysis information, the method further comprises: sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analyzed information; The first request is used to request an update of a network policy parameter for AI / ML model transmission, and the network policy parameter is used to optimize the AI / ML model transmission state.

[0011] In one possible design, the step of sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analyzed information comprises: determining new quality of service parameters, including at least one of a 5G quality of service identifier for transmitting the AI / ML model, a reflective quality of service control, an uplink direction maximum bitrate for transmitting the AI / ML model, a downlink direction bitrate for transmitting the AI / ML model, a minimum uplink direction bitrate for transmitting the AI / ML model, a downlink direction bitrate for transmitting the AI / ML model, and a priority of the quality of service stream, based on at least one of the uplink direction bitrate for transmitting the AI / ML model and the downlink direction packet delay of the AI / ML model, the uplink direction packet delay of the AI / ML model and the downlink direction packet delay of the AI / ML model in the analysis information, the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model, the number of packet transmissions of the AI / ML model, the number of packet retransmissions of the AI / ML model, and the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; The AI / ML model in the analysis information Use determining area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an application identifier, area information using the AI / ML model, IP address information of the application service using the AI / ML model, a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model, and a data network name of a PDU session for transmitting the quality of service stream of the AI / ML model; or, if the AI / ML model performs federated learning, determining area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an identifier of a federated learning group that instructs analysis in the analysis information, an identifier of a UE or UE(s) participating in federated learning, and an identifier of each application that provides the AI / ML model or that participates in federated learning; Based on the area information and address information of the UE(s) and each AF transmitting the AI / ML model, determining the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the data network access identifier DNAI and the AI / ML model transmission state; The new quality of service parameters, the DNAI, and the area information and address information of the UE(s) and each AF corresponding to the DNAI in the first request Request and sending the parameter to the PCF directly or via the NEF.

[0012] In a possible design, the step of determining a data network access identifier DNAI and area information and address information of the UE(s) and each AF corresponding to the DNAI based on area information and address information of the UE(s) and each AF transmitting the AI ​​ / ML model includes: determining whether the current routing path is bad based on the area information and address information of the UE(s) transmitting the AI / ML model and each AF; If the current routing path is bad, determining the destination addresses of both the UE(s) transmitting the AI / ML model and the AFs based on the address information and area information of the UE(s); determining a closest route based on the destination address; and determining, based on the closest route, the DNAI and the area information and address information of the UE(s) and each AF corresponding to the DNAI.

[0013] In a possible design, the first requirement may specifically be: The 5G quality of service identifier in the PCC rules, the reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream are requested to be adjusted based on the new quality of service parameters, and the first update result is fed back to the PCF directly or via the NEF; The first update result is determined based on a result of the PCF adjusting the PCC rules based on the new quality of service parameters; In this case, the method further comprises: receiving the first update result sent from the PCF directly or via the NEF, wherein the first update result includes that the first request is accepted or that the first request is rejected.

[0014] In a possible design, the first requirement may specifically be: Session management function SMF Network required The base The method is used to request the PCF to determine whether a session management policy needs to be updated, and if the SMF determines that the session management policy needs to be updated, to determine that the PCF will send a second request to the SMF, wherein parameters requested in the second request include at least one of a DNAI, a traffic-oriented policy identifier, and traffic path information, and the second request is used by the SMF to determine a user plane function (UPF) to be selected based on the new session management policy and provide the corresponding DNAI, traffic-oriented policy identifier, and traffic path information; In this case, the method further comprises: receiving a second update result sent from the PCF directly or via the NEF, the second update result being determined by the PCF based on whether a new session management policy sent from the SMF updates the UPF path; The second update result is 2 The first request is accepted or the first request is rejected.

[0015] In an embodiment of the present disclosure, new quality of service parameters are determined based on the analysis information, and the new quality of service parameters are sent to the PCF, so that the PCF can adjust PCC rules accordingly based on the new quality of service parameters, or update SM policies through the SMF to provide DNAI and area information and address information of the UE(s) and each AF corresponding to the DNAI, and receive a notification indicating that the first request sent from the PCF is accepted or rejected, thereby requesting to adjust network policies based on the NWDAF analysis results to optimize the AI / ML model transmission state.

[0016] In one possible design, after receiving the analysis information, the method further comprises: adjusting, based on the analysis information, information of the application layer model used to update quality of service parameters, including at least one of model compression, model size, model transmission time slot, and model encoding and decoding; determining new quality of service parameters, including a 5G quality of service identifier, a reflective quality of service control, a maximum uplink bit rate for transmitting the AI / ML model, a maximum downlink bit rate for transmitting the AI / ML model, a minimum uplink bit rate for transmitting the AI / ML model, a minimum downlink bit rate for transmitting the AI / ML model, and a priority of the quality of service stream, based on the information of the application layer model to be adjusted; sending a third request to a Policy Control Function (PCF) directly or via the NEF; The parameters requested in the third request include the new quality of service parameters, and the third request is used to request an update of the quality of service parameters.

[0017] In a possible design, the third requirement may be specifically: Based on the new quality of service parameters, a 5G quality of service identifier in the PCC rules, a reflective quality of service control, a maximum uplink bit rate for transmitting an AI / ML model, a maximum downlink bit rate for transmitting an AI / ML model, a minimum uplink bit rate for transmitting an AI / ML model, and a minimum downlink bit rate for transmitting an AI / ML model; Quality of Service Stream Used to request the PCF to adjust the priority, In this case, the method further comprises: receiving a third update result sent from the PCF directly or via the NEF, the third update result being determined by the PCF based on the result of adjusting the PCC rules, and the third update result including that the third request is accepted or that the third request is rejected.

[0018] In a possible design, after adjusting the information of the application layer model, the method further comprises: directly sending information about the application layer model to be adjusted, including model compression, model size, and encoding and decoding of the model, to the PCF, wherein the information about the application layer model to be adjusted is used by the PCF to support adjusting the 5G quality of service identifier, reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream in the PCC rules; In this case, the method further comprises: receiving a fourth update result sent from the PCF, the fourth update result being determined by the PCF based on the result of adjusting the PCC rules based on information of the application layer model to be adjusted, and the fourth update result including that the third request is accepted or that the third request is rejected.

[0019] In a possible design, after adjusting the information of the application layer model, the method further comprises: Model transmission time for information on the application layer model to be adjusted band directly to the PCF, and a model transmission time in the information of the adjusted application layer model is included. band is used to support the PCF to adjust a gate state parameter in the PCC rule, and the gate state parameter is used to support the SMF to update a session management policy based on a transmission start time and a transmission end time in the gate state; In this case, the method further comprises: receiving a fifth update result sent from the PCF, the fifth update result being determined by the PCF based on the result of the new session management policy sent from the SMF, and the fifth update result including that the third request is accepted or that the third request is rejected.

[0020] In the embodiments of the present disclosure, the information of the application layer model is adjusted based on the analysis information, and new quality of service parameters are determined based on the information of the adjusted application layer model, and the new quality of service parameters are sent to the PCF, so that the PCF adjusts the PCC rules accordingly based on the new quality of service parameters; or the information of the application layer model to be adjusted is sent directly to the PCF, so that the PCF adjusts the above quality of service parameters in the PCC rules based on the model compression, model size, and model encoding and decoding; or the PCF adjusts the model transmission time. band By adjusting the gate state parameters in the PCC rules based on the above, the SMF updates the session management policy based on the transmission start time and transmission end time in the gate state. By receiving a notification sent from the PCF indicating that the third request is accepted or rejected, the information of the application layer model can be adjusted based on the NWDAF analysis result, and the QoS request can be updated to optimize the AI / ML model transmission state.

[0021] In a second aspect, the present disclosure provides a method for analyzing a model transmission state in a subscription network, applied to a network data analysis function NWDAF, the method comprising: receiving, directly or via a network capability opening function NEF, a first message sent from an application function AF, the first message being used to request analytical information on the state of transmission of an artificial intelligence / machine learning AI / ML model in a subscription network; Sending a second message to other network functions (5GC NF(s)) of the 5G core network based on the parameters requested in the first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network; receiving AI / ML model transmission state data transmitted from other network functions (5GC NF(s)) of the 5G core network, analyzing the AI / ML model transmission state data, and obtaining analysis information of the AI / ML model transmission state; The analyzed information is used to adjust network policy parameters and / or application layer model information via the AF.

[0022] In a possible design, the parameters requested in the first message include a network data analysis identifier, an identifier of one user equipment (UE) or a set of UEs that will receive the AI / ML model, or any UE that meets the analysis criteria. Identifier , including at least one of an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a protocol data unit (PDU) session that transmits the quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of the UE that uses the AI / ML model; an identifier of an application that uses the AI / ML model; an identifier of a quality of service stream that transmits the AI / ML model; an uplink bit rate for transmitting the AI / ML model and a downlink bit rate for transmitting the AI / ML model; an uplink packet delay for the AI / ML model and a downlink packet delay for the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; the number of packet retransmissions of the AI / ML model; a data collection time; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; a data network of a PDU session for transmitting the quality of service stream of the AI / ML model; and a service flow used for the AF; the analysis information includes at least one of: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information using the AI / ML model; a validity period of the analysis result; a user plane function (UPF) that provides AI / ML model transmission; a data network name of the PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; an identifier of the quality of service stream for transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; and the number of packet retransmissions of the AI / ML model. If the AI / ML model performs federated learning, the parameters requested in the first message further include federated learning group information, and the federated learning group information includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that it provides an AI / ML model or participates in the federated learning.

[0023] In a third aspect, the present disclosure provides an apparatus for analyzing a model transmission state in a subscription network, the apparatus including: a memory, a transceiver, and a processor; The memory is used to store a computer program, the transceiver is used to transmit and receive data under the control of the processor, and the processor reads the computer program stored in the memory; Sending a first message to a Network Data Analysis Function (NWDAF), directly or via a Network Capabilities Opening Function (NEF), to request analysis information of an artificial intelligence / machine learning (AI / ML) model transmission status in a subscription network; Receive, directly or via the NEF, analysis information of the AI / ML model transmission state determined by the NWDAF based on data of the AI / ML model transmission state received from other network functions (5GC NF(s)) of the 5G core network; The analyzed information is used to adjust network policy parameters and / or application layer model information.

[0024] In a fourth aspect, the present disclosure provides an apparatus for analyzing a model transmission state in a subscription network, the apparatus including: a memory, a transceiver, and a processor; The memory is used to store a computer program, the transceiver is used to transmit and receive data under the control of the processor, and the processor reads the computer program stored in the memory; Receiving a first message sent from an application function AF directly or via a network capability opening function NEF, the first message being used to request analysis information of an artificial intelligence / machine learning AI / ML model transmission status in a subscription network; Sending a second message to other network functions (5GC NF(s)) of the 5G core network based on the parameters requested in the first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network; Receives AI / ML model transmission status data sent from other network functions (5GC NF(s)) of the 5G core network, analyzes the AI / ML model transmission status data, and obtains analysis information of the AI / ML model transmission status; The analyzed information is used to adjust network policy parameters and / or application layer model information via the AF.

[0025] In a fifth aspect, the present disclosure provides an apparatus for analyzing a model transmission state in a subscription network, the apparatus comprising: A sending unit used to send a first message to a network data analysis function (NWDAF) directly or via a network capability opening function (NEF) to request analysis information of an artificial intelligence / machine learning (AI / ML) model transmission status in a subscription network; An analysis unit used to receive, directly or via the NEF, analysis information of the AI / ML model transmission state determined by the NWDAF based on data of the AI / ML model transmission state received from other network functions (5GC NF(s)) of the 5G core network; The analyzed information is used to adjust network policy parameters and / or application layer model information.

[0026] In a sixth aspect, the present disclosure provides an apparatus for analyzing a model transmission state in a subscription network, the apparatus comprising: A receiving unit used to receive a first message sent from an application function AF directly or via a network capability opening function NEF, the first message being used to request analysis information of an artificial intelligence / machine learning AI / ML model transmission status in a subscription network; A transmitting unit used to transmit a second message to other network functions (5GC NF(s)) of a 5G core network based on the parameters requested in the first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network; An analysis unit used to receive AI / ML model transmission state data transmitted from other network functions (5GC NF(s)) of the 5G core network, analyze the AI / ML model transmission state data, and obtain analysis information of the AI / ML model transmission state; The analyzed information is used to adjust network policy parameters and / or application layer model information via the AF.

[0027] In a seventh aspect, the present disclosure provides a processor-readable storage medium having stored thereon a computer program, the computer program being used to cause the processor to perform a method according to any of the first and / or second aspects.

[0028] The present disclosure provides a method, device, and storage medium for analyzing a model transmission state in a subscription network, comprising: transmitting, directly or via a network capability opening function (NEF), to a network data analysis function (NWDAF), a first message used to request analysis information of an artificial intelligence / machine learning (AI / ML) model transmission state in a subscription network; receiving, directly or via the NEF, analysis information of the AI / ML model transmission state determined by the NWDAF based on data of the AI / ML model transmission state received from other network functions (5GC NF(s)) of a 5G core network; the data of the AI / ML model transmission state obtained by the NWDAF transmitting a second message to the 5GC NF(s) based on parameters requested in the received first message; the second message being used to collect data for analyzing the AI / ML model transmission state in the network; and the analysis information being used to adjust network policy parameters and / or application layer model information. A request for analysis information of the AI / ML model transmission state in the subscription network is sent to the NWDAF, and the analysis information sent from the NWDAF is determined based on the collected 5GC NF(s) data. Based on the analysis information, the network policy parameters and / or application layer model information are adjusted, thereby effectively analyzing the AI / ML model transmission state. Furthermore, the network effectively adjusts the network transmission policy based on the AI / ML model transmission state, and a third party can obtain the analysis of the AI / ML model transmission state and adjust the application layer information.

[0029] The contents described in this section are as follows: This disclosure It is not intended to identify key or important features of the embodiments, This disclosure It is not intended to limit the scope of This disclosure Other features of the present invention will be readily understood from the following specification.

[0030] In order to more clearly explain the technical solutions in the present disclosure or the prior art, the following briefly introduces the drawings that need to be used in the description of the embodiments or the prior art. This disclosure , and a person skilled in the art can derive other drawings based on these drawings without performing any creative work. [Brief explanation of the drawings]

[0031] [Figure 1] FIG. 1 is a network architecture diagram of a method for analyzing a model transmission state in a subscription network provided by an embodiment of the present disclosure. [Figure 2] FIG. 1 is a network architecture diagram of 5GC that supports network data analysis provided by an embodiment of the present disclosure. [Figure 3] 1 is a flowchart of a method for analyzing a model transmission state in a subscription network provided by a first embodiment of the present disclosure; [Figure 4] 1 is a signaling flowchart of a method for analyzing a model transmission state in a subscription network provided by a first embodiment of the present disclosure; [Figure 5] 10 is a signaling flowchart of a method for analyzing a model transmission state in a subscription network provided by a second embodiment of the present disclosure; [Figure 6] 10 is a signaling flowchart of a method for analyzing a model transmission state in a subscription network provided by a third embodiment of the present disclosure; [Figure 7] 10 is a flowchart of a method for analyzing a model transmission state in a subscription network provided by a fourth embodiment of the present disclosure; [Figure 8] FIG. 1 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by an embodiment of the present disclosure. [Figure 9]FIG. 10 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by another embodiment of the present disclosure. [Figure 10] FIG. 10 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by yet another embodiment of the present disclosure. [Figure 11] FIG. 10 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by yet another embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0032] In the present disclosure, the term "and / or" describing an association relationship between related objects means that three relationships can exist, for example, A and / or B can represent three cases: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally indicates that the related objects before and after it are in an "or" relationship.

[0033] Hereinafter, the technical solutions in the embodiments of the present disclosure will be clearly and completely described with reference to the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are only some embodiments of the present disclosure, and not all embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without any creative work fall within the scope of protection of the present disclosure.

[0034] To clearly understand the technical solution of this disclosure, we first introduce the prior art solution in detail. In the prior art, among the demands of SA1 R18 through which SA#93e passes, there are at least the following scenarios that require the transmission of AI / ML models:

[0035] Scene 1: Distribution and sharing of AI / ML models. Due to changes in tasks and environments, mobile devices have memory limitations and cannot pre-install all models. Therefore, mobile devices need to download new AI / ML models in real time from the network via the 5G system.

[0036] Scene 2: Federated learning algorithm using 5GS. When the cloud server trains a global model, it needs to aggregate the models trained locally by each terminal device. During each training iteration, one terminal device downloads the global model from the cloud server and trains it with local data. The terminal reports the intermediate training results to the cloud server. The cloud server aggregates the intermediate training results from all terminals, updates the global model, and redistributes the global model to the terminals, which then perform the next iteration.

[0037] Scene 3: Splitting an AI / ML model between AI / ML endpoints. A single AI / ML model can be split into multiple parts based on the current task or environment. The trend is to infer computationally complex and energy-intensive parts on the network, while inferring privacy-sensitive or latency-sensitive parts on the device. For example, the device downloads / loads the model, infers specific layers / parts, and then sends the intermediate results to the network. The network then executes the remaining layers / parts and feeds the inference results back to the device. This scenario may involve model transmission, as it transmits parts of the model at the initial step or intermediate stages.

[0038] Therefore, as a channel for transmitting AI / ML models, the 5G system must enhance the intelligent capabilities of the 5G network and meet the requirements for the AI / ML models in TS 22.261, which SA1 #93e passes, to be transmitted by the 5G system. In order to do so, the 5G system must support the monitoring and disclosure of status information about AI-ML sessions to third parties. However, currently, there is no analysis of the AI / ML model transmission status, and third parties cannot effectively adjust their own behavior based on the AI / ML model transmission status, and the network cannot effectively adjust the network status based on the AI / ML model transmission status.

[0039] The inventors further discovered that effective analysis of the AI / ML model transmission status requires interactions between an application function (AF), a network exposure function (NEF), a network data analytic function (NWDAF), and each network function (NF). As shown in Figure 1, the AF can directly or via the NEF send a request to the NWDAF indicating an analysis of the AI / ML model transmission status in the subscription network. The NWDAF can then collect data from each network function (NF) (i.e., NF(s)) in the 5G core network (5GC) to analyze and provide feedback on the AI / ML model transmission status in the network, effectively analyzing the AI / ML model transmission status. Furthermore, the network can effectively adjust the network status based on the AI / ML model transmission status, and a third party can obtain the analysis of the AI / ML model transmission status and adjust their own operational data.

[0040] Therefore, based on the creative research of the above inventors, the present disclosure proposes a method for analyzing a model transmission state in a subscription network, in which the present disclosure includes transmitting, directly or via a network capability opening function (NEF), to a network data analysis function (NWDAF), a first message used to request analysis information of an artificial intelligence / machine learning (AI / ML) model transmission state in a subscription network, and receiving, directly or via the NEF, analysis information of the AI / ML model transmission state determined by the NWDAF based on data of the AI / ML model transmission state transmitted from other network functions (i.e., 5GC NF(s)) of the 5G core network, the data of the AI / ML model transmission state being obtained by the NWDAF transmitting a second message to the 5GC NF(s) based on parameters requested in the received first message, the second message being used to collect data for analyzing the AI / ML model transmission state in the network, and the analysis information being used to adjust network policy parameters and / or application layer model information. A request for analysis information of the AI / ML model transmission state in the subscription network is sent to the NWDAF, and the analysis information sent from the NWDAF is determined based on the collected 5GC NF(s) data. Based on the analysis information, the network policy parameters and / or application layer model information are adjusted, thereby effectively analyzing the AI / ML model transmission state. Furthermore, the network effectively adjusts the network transmission policy based on the AI / ML model transmission state, and a third party can obtain the analysis of the AI / ML model transmission state and adjust the application layer information.

[0041] FIG. 2 is a network architecture diagram of 5GC that supports network data analysis provided by an embodiment of the present disclosure. As shown in FIG. 2, in an embodiment of the present disclosure, the NWDAF is a network analysis function managed by a carrier. The NWDAF can provide data analysis services to 5GC network functions, application functions (AFs), and operation administration and maintenance (OAMs). Here, the analysis results may be historical statistical information or predictive information. The NWDAF can serve one or more network slices.

[0042] Here, 5GC includes many other functions, such as the User Plane Function (UPF), Session Management Function (SMF), Access and Mobility Management Function (AMF), Unified Data Repository (UDR), Network Exposure Function (NEF), AF, Policy Control Function (PCF), and Online Charging System (OCS). These other functions can be collectively referred to as NFs. The NWDAF communicates with other functional entities (5GC NFs) and OAMs in the 5G core network based on service-oriented interfaces.

[0043] A 5GC has different NWDAF instances, which can provide different types of specialized analytics. To enable a consumer NF to find the appropriate NWDAF instance to provide a specific type of analytics, the NWDAF instance must support Analytic ID when registering with the Network Repository Function (NRF), where the Analytic ID represents the analytics type (or analytics identifier). Therefore, when querying the NRF for an NWDAF instance, the consumer NF can provide the Analytic ID to indicate which type of analytics is required. The 5GC network function and OAM then determine how to use the data analytics provided by the Network Data Analysis Function (NWDAF) to improve network performance.

[0044] In an embodiment of the present disclosure, in an application scenario, the AF requests the NWDAF to analyze the transmission status of an AI / ML model, and the analysis result (or analysis information) includes an identifier of the application using the AI / ML model (i.e., Application ID), area information using the AI / ML model, a time period during which the AI / ML model is transmitted, AI / ML transmission The information includes the model size, quality of service (i.e., QoS)-related information for transmitting the AI / ML model, the network slice used to transmit the AI / ML model, and the data network name (DNN) information. If federated learning is used, the information further includes a group identifier (i.e., federated learning group ID), the UE ID or UE group ID participating in federated learning, and the address information of the application server that provides the model or participates in federated learning. Based on the data analysis provided by the NWDAF, the AF requests that the 5GS adjust the network policy or adjust the application layer AI / ML model parameters to optimize the AI / ML model transmission state. The 5GC NF(s) adjust the network policy based on the request, or the AF adjusts the application layer AI / ML model parameters based on the data analysis.

[0045] Here, when AF sends an analysis request to NWDAF, if AF is in the receiving area, AF can directly send the request to NWDAF, and if AF is in the receiving area, do not have In this case, the AF can send a request to the NWDAF via the NEF, i.e., the AF can send a request to the NEF and the NEF can send a request to the NWDAF.

[0046] Therefore, a request for analysis information of the AI / ML model transmission state in the subscription network is sent to the NWDAF directly or via the NEF, and the analysis information determined based on the collected 5GC NF(s) data is received from the NWDAF, and the network policy parameters and / or application layer model information are adjusted based on the analysis information, thereby effectively analyzing the AI / ML model transmission state, and the network effectively adjusts the network transmission policy based on the AI / ML model transmission state, and a third party can obtain the analysis of the AI / ML model transmission state and adjust the application layer information.

[0047] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

[0048] 3 is a flowchart of the method for analyzing the model transmission state in a subscription network provided by the first embodiment of the present disclosure. As shown in FIG. 3, when the execution body of the method for analyzing the model transmission state in a subscription network provided by the first embodiment of the present disclosure is an AF, the method for analyzing the model transmission state in a subscription network provided by the first embodiment of the present disclosure includes the following steps:

[0049] Step 101: The AF sends a first message to the Network Data Analysis Function NWDAF directly or via the Network Capabilities Opening Function NEF.

[0050] Here, the first message is used to request analytical information on the transmission status of an artificial intelligence / machine learning AI / ML model in a subscription network.

[0051] In this embodiment, the parameters requested in the first message include a network data analysis identifier (i.e., Analytics ID), an identifier of one user equipment (UE) or a set of UEs that will receive the AI / ML model, or any UEs that meet the analysis criteria (i.e., Target of Analytics Reporting). Identifier , an identifier of the application using the AI / ML model (i.e., Application ID), an area of ​​the AI / ML model transmission (i.e., AoI (Area of ​​Interest)), a network slice (i.e., S-NSSAI) indicating a protocol data unit PDU session transmitting the quality of service stream of the AI / ML model, a data network (i.e., DNN) indicating a PDU session transmitting the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission (i.e., Model transmission duration), a start timestamp of the AI / ML model transmission (i.e., Model transmission start), an end timestamp of the AI / ML model transmission (i.e., Model transmission stop), a size of the AI / ML transmission model (i.e., Model size), a quality of service requirement indicating the quality of service stream for transmitting the AI / ML model (i.e., 5QI (5G QoS Identifier)), and / or a specific quality of service requirement for indicating the transmission of the AI / ML model (i.e., QoS Characteristics).

[0052] Here, the specific quality of service requirement may be, for example, packet transmission delay, packet error rate, etc. See Table 1 below for an example of parameters required in the first message.

[0053] [Table 1-1] [Table 1-2]

[0054] If federated learning exists, the parameters requested in the first message may further include federated learning group information (i.e., Federated Learning (FL) group information), which includes at least one of an identifier of a federated learning group for instructing analysis (i.e., Federated Learning (FL) group ID), an identifier of a UE or UEs participating in federated learning (i.e., Federated Learning (FL) UE ID or UE group ID), and an identifier of an application participating in federated learning (i.e., Federated Learning (FL) Application ID).

[0055] [Table 2-1] [Table 2-2]

[0056] In this embodiment, if the AF is not in a reception area (i.e., the AF is not in a reception area), the AF sends an AI / ML model transmission status subscription request, such as an Nnef_AnalyticsExposure_Subscribe (i.e., analytics release subscription) or Nnef_AnalyticsExposure_Fetch (i.e., analytics release fetch) request, to the NEF. The NEF then sends a first message to the NWDAF. This first message may be an AI / ML model transmission status subscription request, such as an Nnwdaf_AnalyticsSubscription_Subscribe (i.e., analytics subscription subscription) or an Nnwdaf_AnalyticsInfo_Request (i.e., analytics information) request, which may include parameters as shown in the table, to request analytics information about the AI / ML model transmission status in the subscription network. If the AF is in a reception area (i.e., the AF is in a reception area), the AF sends the first message directly to the NWDAF.

[0057] Here, the AI / ML model transmission status subscription request may include parameters such as those shown in Table 1 or Table 2, and requests analysis information on the AI / ML model transmission status in the subscription network.

[0058] Step 102: The AF receives, directly or via the NEF, the analysis information of the AI / ML model transmission status sent from the NWDAF.

[0059] Here, the analysis information is determined by the NWDAF based on AI / ML model transmission status data received from other network functions (5GC NF(s)) of the 5G core network.

[0060] Optionally, the AI / ML model transmission status data is obtained by the NWDAF sending a second message to the 5GC NF(s) based on parameters requested in the received first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network.

[0061] In this embodiment, the second message includes the current location of the UE that uses the AI / ML model (i.e., UE location), an identifier of the application that uses the AI / ML model (i.e., Application ID, which may be a server ID or an AF ID), a quality of service stream identifier (i.e., QFI) that transmits the AI / ML model, an uplink bit rate (i.e., bit rate for UL direction) that transmits the AI / ML model and a downlink bit rate (i.e., bit rate for DL ​​direction) that transmits the AI / ML model, an uplink packet delay (i.e., packet delay for UL direction) that transmits the AI / ML model and a downlink packet delay (i.e., packet delay for the DL direction) that transmits the AI / ML model, the number of abnormal releases of the quality of service stream (QoS Sustainability) in the time period that transmits the AI / ML model, the number of packet transmissions of the AI / ML model, the number of packet retransmissions of the AI / ML model (i.e., packet retransmission), data collection time (i.e., Timestamp), time of AI / ML model transmission (i.e., time zone of AI / ML model transmission), start timestamp of AI / ML model transmission, end timestamp of AI / ML model transmission, size of AI / ML transmission model, network slice of PDU session for transmitting quality of service stream of AI / ML model, data network of PDU session for transmitting quality of service stream of AI / ML model, and service flow (i.e., IP filter information) used for the AF.

[0062] See Table 3 below for an example of the second message.

[0063] [Table 3-1] [Table 3-2]

[0064] If federated learning exists, the second message further includes at least one of a federated learning group identifier (i.e., Federated Learning (FL) group ID) for instructing the analysis, a UE identifier or UEs participating in federated learning (i.e., Federated Learning (FL) UE ID or UE group ID), and an application identifier (i.e., Federated Learning (FL) Application ID) participating in federated learning. See Table 4 below for an example of the second message.

[0065] [Table 4-1] [Table 4-2]

[0066] In this embodiment, if the AF is unable to receive data, it receives the analysis information of the AI / ML model transmission status transmitted from the NEF, and this analysis information is transmitted from the NWDAF to the NEF. If the AF is able to receive data, it receives the analysis information of the AI / ML model transmission status transmitted directly from the NWDAF.

[0067] Here, the analysis information includes at least one of the following: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information for using the AI / ML model; a validity period of the analysis result (i.e., a validity period); a user plane function UPF (i.e., UPF Info) that provides the AI / ML model transmission; a data network name of a PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; and quality of service requirements (i.e., QoS requirements). The quality of service requirements include the quality of service stream identifier (i.e., QFI) for transmitting the AI / ML model, the uplink bit rate for transmitting the AI / ML model and the downlink bit rate for transmitting the AI / ML model, the uplink packet delay for the AI / ML model and the downlink packet delay for the AI / ML model, the number of abnormal releases of the quality of service stream during the time period for transmitting the AI / ML model, the number of times the reporting threshold for abnormal releases of the quality of service stream during the time period for transmitting the AI / ML model is reached, the number of packet transmissions of the AI / ML model, and the number of packet retransmissions of the AI / ML model. See Table 5 below for an example of analysis information.

[0068] [Table 5-1] [Table 5-2]

[0069] If federated learning is present, the analysis information further includes at least one of the following: a federated learning group identifier for instructing the analysis, a UE identifier or UE(s) identifiers participating in the federated learning, and each application identifier (i.e., Application Server Instance Address) indicating that the application provides an AI / ML model or participates in the federated learning. See Table 6 below for an example of analysis information.

[0070] [Table 6-1] [Table 6-2] [Table 6-3]

[0071] In this embodiment, a request for analysis information of the AI / ML model transmission state in the subscription network is sent to the NWDAF, and analysis information determined based on the collected 5GC NF(s) data is received from the NWDAF. Network policy parameters and / or application layer model information are adjusted based on the analysis information, thereby achieving effective analysis of the AI / ML model transmission state, and the network effectively adjusts the network transmission policy based on the AI / ML model transmission state, allowing a third party to obtain the analysis of the AI / ML model transmission state and adjust the application layer information.

[0072] Illustratively, refer to FIG. 4, which is a signaling flowchart of a method for analyzing a model transmission state in a subscription network provided by Example 1 of the present disclosure, and FIG. 4 is a signaling interaction diagram between an AF, an NWDAF, an NEF, and an NF in the method for analyzing a model transmission state in a subscription network provided by this example. The method for analyzing a model transmission state in a subscription network provided by this example includes the following steps (i.e., the corresponding signaling interaction process in Example 1: the AF requests the NWDAF to provide an AI / ML model transmission state analysis) (wherein, in steps 4011 to 4016, the AF is in an area where reception is unavailable, and in steps 4021 to 4024, the AF is in an area where reception is available).

[0073] Step 4011: If the AF is not capable of receiving, the AF sends an AI / ML model transmission status subscription Nnef_AnalyticsExposure_Subscribe (i.e., analytics exposure subscription) or Nnef_AnalyticsExposure_Fetch (i.e., analytics exposure fetch) request to the NEF.

[0074] In this embodiment, the request may include parameters such as those shown in Table 1 or Table 2, and requests analysis information on the AI / ML model transmission status in the subscription network.

[0075] In step 4012, the NEF sends an AI / ML model open transmission status subscription Nnwdaf_AnalyticsSubscription_Subscribe (i.e., analytics subscription) or Nnwdaf_AnalyticsInfo_Request (i.e., analytics information) request to the NWDAF.

[0076] Here, the AI / ML model open transmission status subscription Nnwdaf_AnalyticsSubscription_Subscriber Nnwdaf_AnalyticsInfo_Request request can be the first message.

[0077] In this embodiment, the request may include parameters such as those shown in Table 1 or Table 2, and requests analysis information on the AI / ML model transmission status in the subscription network.

[0078] In step 4013, the NWDAF calls Nnf_EventExposure_Subscribe (event exposure subscription) to collect data in the 5GC NF(s).

[0079] Here, the collected data is used to analyze the AI / ML model transmission status in the network as shown in Table 3 or Table 4. The method by which the NWDAF sends the second message to the 5GC NF(s) is for the NWDAF to call Nnf_EventExposure_Subscribe.

[0080] In step 4014, the 5GC NF(s) invokes Nnf_EventExposure_Notify (i.e., event release notification) and feeds back the required data to the NWDAF.

[0081] In step 4015, the NWDAF calls Nnwdaf_AnalyticsSubscription_Notify (i.e., analytics subscription notification) or Nnwdaf_AnalyticsInfo_Request response (i.e., analytics information request response) to send analytics information of the AI / ML model transmission status to the NEF.

[0082] In step 4016, the NEF calls Nnef_AnalyticsExposure_Notify (i.e., analysis release notification) or Nnef_AnalyticsExposure_Fetch response (i.e., analysis release fetch response) to send the analysis information of the AI / ML model transmission status to the AF.

[0083] Here, the analysis information is shown in Table 5 or Table 6.

[0084] In step 4021, if the AF is capable of receiving, the AF sends an AI / ML model transmission status subscription request directly to the NWDAF to perform the operations described in step 4012. The consumer may also be a PCF or an SMF.

[0085] Step 4022 executes the operations described in step 4013, i.e., step 4013.

[0086] Step 4023 executes the operations described in step 4014, i.e., step 4014.

[0087] In step 4024, the NWDAF sends the analysis information of the AI / ML model transmission status directly to the AF, as described in step 4016.

[0088] Embodiment 2, after receiving the analysis information, the method further comprises: Sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analyzed information.

[0089] Here, the first request is used to request an update of a network policy parameter for AI / ML model transmission, and the network policy parameter is used to optimize the AI / ML model transmission state.

[0090] Specifically, the AF requests network policy adjustment to optimize the AI / ML model transmission state based on the NWDAF analysis result. Specifically, if the AF is in the reception area, the AF sends a first request directly to the PCF to request the PCF to update the network policy parameters for AI / ML model transmission. If the AF is not in the reception area, the AF sends a first request to the PCF via the NEF to request the PCF to update the network policy parameters for AI / ML model transmission.

[0091] Optionally, sending a first request to a Policy Control Function PCF directly or via the NEF based on the analysis information can be realized by the following steps:

[0092] Step a1: based on at least one of the uplink direction bitrate for transmitting the AI / ML model and the downlink direction bitrate for transmitting the AI / ML model, the uplink direction packet delay of the AI / ML model and the downlink direction packet delay of the AI / ML model, the number of abnormal releases of the quality of service stream during the time period for transmitting the AI / ML model, the number of packet transmissions of the AI / ML model, the number of packet retransmissions of the AI / ML model, and the number of times a reporting threshold for abnormal release of the quality of service stream is reached during the time period for transmitting the AI / ML model, in the analysis information, determine new quality of service parameters for transmitting the AI / ML model, including at least one of a 5G quality of service identifier, reflective quality of service control, a maximum uplink direction bitrate for transmitting the AI / ML model, a maximum downlink direction bitrate for transmitting the AI / ML model, a minimum uplink direction bitrate for transmitting the AI / ML model, a minimum downlink direction bitrate for transmitting the AI / ML model, and a priority of the quality of service stream.

[0093] Step a2: The AI / ML model in the analysis information UseA step of determining area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an application identifier, area information using the AI / ML model, IP address information of the application service using the AI / ML model, a network slice of a PDU session for transmitting the service quality stream of the AI / ML model, and a data network name of a PDU session for transmitting the service quality stream of the AI / ML model; or, if the AI / ML model performs federated learning, determining area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an identifier of a federated learning group that instructs analysis in the analysis information, an identifier of a UE or UE(s) participating in federated learning, and an identifier of each application that provides the AI / ML model or that indicates participation in federated learning.

[0094] Step a3: Based on the area information and address information of the UE(s) transmitting the AI / ML model and each AF, determine the data network access identifier DNAI and the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the AI / ML model transmission state.

[0095] Step a4: The new quality of service parameters, the DNAI, and the area information and address information of the UE(s) and each AF corresponding to the DNAI are input to the first request. Request It is sent as a parameter to the PCF directly or via the NEF.

[0096] Specifically, the AF determines new QoS parameters for transmitting the AI / ML model based on analytical information related to transmitting the AI / ML model obtained from the NWDAF, such as the uplink bit rate for transmitting the AI / ML model and the downlink bit rate for transmitting the AI / ML model, the uplink packet delay of the AI / ML model and the downlink packet delay of the AI / ML model, the number of abnormal releases of the quality of service stream during the time period for transmitting the AI / ML model, the number of packet transmissions of the AI / ML model, the number of packet retransmissions of the AI / ML model, and the number of times the reporting threshold for abnormal releases of the quality of service stream during the time period for transmitting the AI / ML model is reached, and provides the new QoS parameters to the PCF.

[0097] The AF will provide analytical information on the transmission of AI / ML models obtained from the NWDAF, e.g., UseBased on the identifier of the application, the area information using the AI / ML model, the IP address information of the application service using the AI / ML model, the network slice of the PDU session for transmitting the quality of service stream of the AI / ML model, and the data network name of the PDU session for transmitting the quality of service stream of the AI / ML model, or based on analysis information related to transmitting the AI / ML model obtained from the NWDAF, such as an identifier of a federated learning group instructing the analysis, an identifier of a UE or UE(s) participating in federated learning, and an identifier of each application indicating that the AI / ML model is provided or that participates in federated learning, the AF determines the area information and address information of the UE(s) and each AF that transmit the AI / ML model, and further determines a data network access identifier DNAI and the area information and address information of the UE(s) and each AF corresponding to the DNAI. The AF sends a first request to the PCF directly or via the NEF, the first request including new QoS parameters, a DNAI, and area information and address information of the UE(s) corresponding to the DNAI and each AF, and requests the PCF to update related policy parameters for AI / ML model transmission based on the request parameters included in the first request and optimize the AI / ML model transmission state.

[0098] Optionally, step a3 can be realized by the following steps: Step a31: Determine whether the current routing path is bad based on the area information and address information of the UE(s) transmitting the AI / ML model and each AF. Step a32: if the current routing path is bad, determine the destination addresses of both the UE(s) transmitting the AI / ML model and the AFs based on the address information of the UE(s) and the area information of the UE(s). Step a33: determining the closest route based on the destination address. Step a34: Determine the DNAI, the area information and address information of the UE(s) and each AF corresponding to the DNAI based on the closest route.

[0099] Specifically, it determines that the current routing path is poor, selects a DNAI that can provide a better service experience and performance, and provides the DNAI that transmits the AI / ML model, as well as the area information and address information of the UE(s) corresponding to the DNAI and each AF(s) to the PCF.

[0100] Optionally, the first request specifically includes: The 5G quality of service identifier in the PCC rules, the reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream are requested to be adjusted based on the new quality of service parameters, and the first update result is instructed to be fed back to the PCF directly or via the NEF; The first update result adjusts the PCC rules based on the new quality of service parameters. result It was decided by the PCF by In this case, the method further comprises: receiving, directly or via an NEF, the first update result sent from the PCF, wherein the first update result includes that the first request is accepted or that the first request is rejected.

[0101] Specifically, the AF will determine the new QoS parameters provided by the PCF and the PCC rules. 5G quality of service identifier, reflective quality of service control, maximum uplink bit rate for transmitting the AI / ML model, maximum downlink bit rate for transmitting the AI / ML model, minimum uplink bit rate for transmitting the AI / ML model, minimum downlink bit rate for transmitting the AI / ML model, adjust the priority of the quality of service stream, and notify the first update result directly or via the NEF, i.e., request to notify the AF that the first request is accepted or that the first request is rejected.

[0102] Optionally, the first request specifically includes: Session management function SMF Network required The base Request the PCF to determine whether the session management policy needs to be updated, and if the SMF determines that the session management policy needs to be updated, the PCF is used to decide to send a second request to the SMF, where parameters requested in the second request include at least one of a DNAI, a traffic-oriented policy identifier, and traffic path information, and the second request is used by the SMF to determine a user plane function (UPF) to be selected based on the new session management policy and provide the corresponding DNAI, traffic-oriented policy identifier, and traffic path information; In this case, the method further comprises: receiving a second update result sent from the PCF directly or via the NEF, the second update result being determined by the PCF based on whether the new session management policy sent from the SMF updates the UPF path; The second update result is 2 The first request is accepted or the first request is rejected.

[0103] Specifically, AF provides session management functionality. SMF Network required The baseThe PCF requests the PCF to determine whether the session management policy needs to be updated. If the PCF determines that the SMF needs to update the session management policy, the PCF sends the SMF a second request including request parameters such as a DNAI, a traffic-oriented policy identifier, and traffic path information. The SMF determines a user plane function (UPF) to be selected based on the new session management policy, provides the corresponding DNAI, traffic-oriented policy identifier, and traffic path information, and updates the session management policy, i.e., updates the SM policy. Then, the PCF notifies the AF of the second update result directly or via the NEF, i.e., notifies the AF that the second request is accepted or rejected.

[0104] Illustratively, refer to FIG. 5, which is a signaling flowchart of a method for analyzing a model transmission state in a subscription network provided by Embodiment 2 of the present disclosure, and FIG. 5 is a signaling interaction diagram between an AF, an NEF, and a PCF in the method for analyzing a model transmission state in a subscription network provided by this embodiment. The method for analyzing a model transmission state in a subscription network provided by this embodiment includes the following steps (i.e., the corresponding signaling interaction process of Embodiment 2: based on the NWDAF analysis result, the AF requests adjusting a network policy to optimize the AI / ML model transmission state) (wherein, in steps 5011-5014, the AF is in an area with no coverage, and in steps 5021-5022, the AF is in an area with coverage).

[0105] Step 5010, the AF contracts with the NWDAF and retrieves the AI / ML model transmission status analysis.

[0106] In this embodiment, as described in steps 4011 to 4024 above, the AF subscribes to the NWDAF via the NEF or directly to obtain AI / ML model transmission status analysis.

[0107] Step 5011: The AF sends a Nnef_AFsessionWithQoS_Update (ie, AF session update based on quality of service) request to the NEF.

[0108] In this embodiment, if the AF is not receivable, for an AF session to be established (i.e., an AF session) and an AF session with a QoS requirement, the AF can send an Nnef_AFsessionWithQoS_Update (i.e., an AF session update based on quality of service) request to the NEF to update related policy parameters for AI / ML model transmission, thereby optimizing the AI / ML model transmission state.

[0109] Specifically, the AF receives AI / ML model transmission analysis information (QoS flow Bit Rate (i.e., quality of service stream bit rate, e.g., uplink direction bit rate for transmitting the AI / ML model and downlink direction bit rate for transmitting the AI / ML model), QoS flow Packet Delay (i.e., quality of service stream packet delay, e.g., uplink direction packet delay of the AI / ML model and downlink direction packet delay of the AI / ML model), QoS Sustainability, Packet transmission, Packet retransmission) acquired from the NWDAF. 、QBased on the 5G Sustainability Reporting Threshold(s), new QoS parameters for transmitting the AI / ML model (5G QoS Identifier (5QI), Reflective QoS Control, UL-maximum bitrate (i.e., maximum uplink bitrate for transmitting the AI / ML model), DL-maximum bitrate (i.e., maximum downlink bitrate for transmitting the AI / ML model), UL-guaranteed bitrate (i.e., minimum uplink bitrate for transmitting the AI / ML model), DL-guaranteed bitrate (i.e., minimum downlink bitrate for transmitting the AI / ML model), Priority Level) are determined, and the new QoS parameters are provided to the PCF, i.e., the new QoS parameters increase or decrease as the prediction results increase or decrease so that 5GS can meet the QoS requirements for model transmission.

[0110] The AF also determines area and address information of the UE(s) and AF(s) that will transmit the AI / ML model based on analysis information related to AI / ML model transmission acquired from the NWDAF (AF ID transmitting the AI / ML model, UE area information using the AI / ML model, IP address information of the application service using the AI / ML model, network slice for transmitting the AI / ML model, and DNN information). Alternatively, the analysis information may also include information related to federated learning group information (Federated Learning (FL) group ID, Federated Learning (FL) UE ID or UE group ID, and Federated Learning (FL) Application ID) and determines area and address information of the UE(s) and AF(s) that will transmit the AI / ML model. If the AF determines that the current routing route is poor, it selects a DNAI that can provide a better service experience and performance, and provides the DNAI that transmits the AI / ML model and the area and address information of the UE(s) and AF(s) corresponding to the DNAI to the PCF.

[0111] Here, the process of determining that the current routing path is bad is as follows: the AF determines the current routing path based on the AF's IP address information and the UE(s)'s area information (which can correspond to IDs such as AMF, SMF, UPF, etc., and the N6 interface of the UPF can connect to DN). For example, if a UE / server joins / exits (a federated learning group) before the next transmission, the path is bad, or if the number of hops is too large, the path is bad.

[0112] The process of selecting a DNAI that can provide a better service experience and performance can determine the destination address for both transmissions based on the received IP address information and the area information of the UE(s), find a better (recent) routing path, and correspond to the DNAI.

[0113] In step 5012, the NEF sends an Npcf_PolicyAuthorization_Update request (ie, a policy authorization update request) to the PCF.

[0114] In this embodiment, the NEF sends the above information to the PCF via an Npcf_PolicyAuthorization_Update request (i.e., a policy authorization update request) to update related policy parameters for AI / ML model transmission and optimize the AI / ML model transmission status.

[0115] Here, the Npcf_PolicyAuthorization_Update request can be the second request.

[0116] In step 5013, the PCF notifies the NEF of the result (ie, the PCF sends an Npcf_PolicyAuthorization_Update response to the NEF).

[0117] In this embodiment, based on the information provided by the NEF, specifically, based on the new QoS parameters provided, the PCF adjusts the 5G QoS Identifier (5QI), Reflective QoS Control, UL-maximum bitrate, DL-maximum bitrate, UL-guaranteed bitrate, DL-guaranteed bitrate, and Priority Level in the PCC rules accordingly, and sends an Npcf_PolicyAuthorization_Update response to notify the NEF of the result.

[0118] If the PCF determines that the SMF needs to update the policy information, the PCF sends an Npcf_SMPolicyControl_UpdateNotify request (i.e., session policy control update notify request) (DNAI, Per DNAI: Traffic steering policy identifier, Per DNAI: N6 traffic routing information) to the SMF to update the SM policy, and the SMF selects a UPF based on this policy and provides the DNAI, Per DNAI: Traffic steering policy identifier, Per DNAI: N6 traffic routing information. Here, the Npcf_SMPolicyControl_UpdateNotify request may be a second request.

[0119] In step 5014, the NEF sends an Nnef_AFsessionWithQoS_Update response to the AF.

[0120] In this embodiment, the NEF sends a Nnef_AFsessionWithQoS_Update response (i.e., an AF session update response with quality of service) to inform the AF that the request is accepted or rejected.

[0121] Step 5021: The AF sends an Npcf_PolicyAuthorization_Update request directly to the PCF.

[0122] In this embodiment, if the AF is receivable, the AF directly sends an Npcf_PolicyAuthorization_Update request to the PCF to update related policy parameters for AI / ML model transmission, optimize the AI / ML model transmission state, and perform the operations described in step 5012.

[0123] Step 5022, the PCF notifies the AF of the result directly (ie, the PCF sends an Npcf_PolicyAuthorization_Update response (ie, a policy authorization update response)) to the AF directly).

[0124] In this embodiment, first, the PCF performs the operations described in step 5012 based on the information provided by the AF. The PCF directly notifies the AF that the request is accepted or rejected.

[0125] In embodiment 3, after receiving the analysis information, the method can further include the following steps:

[0126] Step b1: based on the analysis information, adjust the information of the application layer model used to update the quality of service parameters, including at least one of model compression, model size, model transmission time period, and model encoding and decoding.

[0127] Step b2: Based on the information of the application layer model to be adjusted, new quality of service parameters are determined, including a 5G quality of service identifier, a reflective quality of service control, a maximum uplink bit rate for transmitting the AI / ML model, a maximum downlink bit rate for transmitting the AI / ML model, a minimum uplink bit rate for transmitting the AI / ML model, a minimum downlink bit rate for transmitting the AI / ML model, and a priority of the quality of service stream.

[0128] Step b3: Sending a third request to the Policy Control Function PCF directly or via the NEF.

[0129] Here, the parameters requested in the third request include the new quality of service parameters, and the third request is used to request an update of the quality of service parameters.

[0130] Specifically, the AF updates the quality of service parameters by adjusting the information of the application layer model, such as model compression, model size, model transmission time slot, and model encoding and decoding, based on the QoS request information provided by the NWDAF. Based on the information of the application layer model to be adjusted, the AF determines the new quality of service parameters and sends them to the PCF directly or via the NEF. to A third request is sent that includes the new quality of service parameters.

[0131] Optionally, the third request specifically includes: Based on the new quality of service parameters, a 5G quality of service identifier in the PCC rules, a reflective quality of service control, a maximum uplink bit rate for transmitting an AI / ML model, a maximum downlink bit rate for transmitting an AI / ML model, a minimum uplink bit rate for transmitting an AI / ML model, and a minimum downlink bit rate for transmitting an AI / ML model; Quality of Service Stream Used to request the PCF to adjust the priority, In this case, the method further comprises: receiving, directly or via an NEF, a third update result sent from the PCF, the third update result being determined by the PCF based on the result of adjusting the PCC rules, and the third update result including that the third request is accepted or that the third request is rejected.

[0132] Specifically, based on the new QoS parameters provided, the AF determines the 5G quality of service identifier in the PCC (i.e., policy and charging control) rules, the reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, Quality of Service StreamsRequest the PCF to adjust the priority accordingly. The PCF will adjust the PCC rules and notify the AF, either directly or via the NEF, that the request is accepted or rejected.

[0133] Optionally, after adjusting the information of the application layer model, the method may further include the steps of: directly sending information about the application layer model to be adjusted, including model compression, model size, and encoding and decoding of the model, to the PCF, wherein the information about the application layer model to be adjusted is used by the PCF to support adjusting the 5G quality of service identifier, reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream in the PCC rules; In this case, the method further comprises: receiving a fourth update result sent from the PCF, the fourth update result being determined by the PCF based on a result of adjusting the PCC rules based on information of the application layer model to be adjusted, and the fourth update result including that the third request is accepted or that the third request is rejected.

[0134] Specifically, the PCF adjusts the above QoS parameters in the PCC rules based on the model compression, model size, and model encoding and decoding, such as the 5G quality of service identifier, reflective quality of service control, maximum uplink bitrate for transmitting the AI / ML model, maximum downlink bitrate for transmitting the AI / ML model, minimum uplink bitrate for transmitting the AI / ML model, minimum downlink bitrate for transmitting the AI / ML model, and quality of service stream priority. The PCF adjusts the PCC rules and notifies the AF directly or via the NEF that the request is accepted or rejected.

[0135] Optionally, after adjusting the information in the application layer model, the method further comprises: Model transmission time for information on the application layer model to be adjusted band directly to the PCF, and a model transmission time in the information of the adjusted application layer model is included. band is used to support the PCF to adjust a gate state parameter in the PCC rule, and the gate state parameter is used to support the SMF to update a session management policy based on a transmission start time and a transmission end time in the gate state; In this case, the method further comprises: receiving a fifth update result sent from the PCF, the fifth update result being determined by the PCF based on the result of the new session management policy sent from the SMF, and the fifth update result including that the third request is accepted or that the third request is rejected.

[0136] Specifically, PCF is a model transmission time bandBased on this, the SMF adjusts the Gate status (i.e., gate state parameter) in the PCC rules and updates the SM policy, which affects the transmission start and end times of the stream and feeds back to the PCF, which notifies the AF directly or via the NEF that the request is accepted or rejected.

[0137] Illustratively, refer to FIG. 6, which is a signaling flowchart of a method for analyzing a model transmission state in a subscription network provided by a third embodiment of the present disclosure, and a signaling interaction diagram between an AF, an NEF, and a PCF in the method for analyzing a model transmission state in a subscription network provided by this embodiment. The method for analyzing a model transmission state in a subscription network provided by this embodiment includes the following steps (i.e., the corresponding signaling interaction process of the third embodiment: the AF adjusts application layer model information such as model compression, model size, model transmission time slot, and model encoding and decoding based on the analysis information provided by the NWDAF, and further updates QoS requirements, which is similar to the step flow of the second embodiment): (wherein, in steps 6012 to 6014, the AF is in an area where reception is unavailable, and in steps 6021 to 6022, the AF is in an area where reception is available).

[0138] Step 6010, the AF contracts with the NWDAF and retrieves the AI / ML model transmission status analysis.

[0139] Specifically, as described above in steps 4011 to 4024, the AF subscribes to the NWDAF via the NEF or directly to obtain AI / ML model transmission status analysis.

[0140] In step 6011, the AF adjusts the application layer behavior based on the analysis information.

[0141] In this embodiment, the AF adjusts the information of the application layer model, such as model compression, model size, model transmission time period, and model encoding and decoding, based on the QoS requirement information provided by the NWDAF.

[0142] Step 6012: The AF sends a Nnef_AFsessionWithQoS_Update request to the NEF.

[0143] In step 6013, the NEF sends an Npcf_PolicyAuthorization_Update request to the PCF.

[0144] Here, the Npcf_PolicyAuthorization_Update request can be the third request.

[0145] In step 6014, the PCF notifies the NEF of the result (i.e., the PCF sends an Npcf_PolicyAuthorization_Update response to the NEF). ) .

[0146] In step 6015, the NEF sends an Nnef_AFsessionWithQoS_Update response to the AF.

[0147] Step 6021: The AF sends an Npcf_PolicyAuthorization_Update request directly to the PCF.

[0148] In step 6022, the PCF notifies the AF of the result directly (ie, the PCF sends an Npcf_PolicyAuthorization_Update response (ie, a policy authorization update response)) directly to the AF).

[0149] Specifically, the AF requests a session to perform a QoS update. Specifically, the AF determines new QoS parameters for transmitting the AI / ML model based on the adjusted model compression, model size, model transmission time slot, model encoding and decoding, etc., and provides the new QoS parameters, including the 5G QoS Identifier (5QI), Reflective QoS Control, UL-maximum bitrate, DL-maximum bitrate, UL-guaranteed bitrate, DL-guaranteed bitrate, and Priority Level, to the PCF. Alternatively, the AF directly sends information adjusting the model compression, model size, model transmission time slot, and model encoding and decoding to the PCF.

[0150] In step 6013, the PCF adjusts the 5G QoS Identifier (5QI), Reflective QoS Control, UL-maximum bitrate, DL-maximum bitrate, UL-guaranteed bitrate, DL-guaranteed bitrate, and Priority Level in the PCC rule accordingly based on the provided new QoS parameters. Alternatively, the PCF adjusts the above QoS parameters in the PCC rule based on the model compression, model size, and model encoding and decoding. Alternatively, the PCF adjusts the model transmission time. band Based on this, the SMF adjusts the Gate status in the PCC rules and updates the SM policy, which affects the start and end times of the stream transmission. and provide feedback to PCF. Finally, the AF is notified that the request is accepted or rejected in the manner of step 5014 or 5022 in the second embodiment.

[0151] In the present disclosure, the NWDAF receives an AI / ML model transmission status analysis request and the parameters included therein sent by the AF, and the input data collected by the NWDAF from the 5GC NF(s) is used to analyze the AI / ML model transmission status in the network. The NWDAF performs the analysis and sends AI / ML model transmission status analysis information to the AF. The AF requests the AF session to perform QoS update based on the AI / ML model transmission status analysis information, and the corresponding policy control function (PCF) and session management function (SMF) perform QoS update. SMF Network required Moto etc. The AF adjusts the policies of the network elements, and the AF adjusts the relevant parameters of the application layer model information based on the AI / ML model transmission state analysis information, and further adjusts the QoS, thereby optimizing the AI / ML model transmission state. This allows a third party to obtain the AI / ML model transmission state, and the network further adjusts its own behavior to meet the AI / ML model transmission requirements based on the model transmission analysis results. The third party can also adjust the behavior of the model application layer to achieve efficient transmission of the AI / ML model based on the model transmission analysis results, thereby ensuring the traffic experience and traffic performance of the AI / ML model transmission.

[0152] Figure 7 is a flowchart of the method for analyzing the model transmission state in a subscription network provided by Example 4 of the present disclosure. As shown in Figure 7, when the entity that performs the method for analyzing the model transmission state in a subscription network provided by this example is an NWDAF, the method for analyzing the model transmission state in a subscription network provided by this example includes the following steps:

[0153] Step 701: The NWDAF receives a first message sent from the application function AF directly or via the network capability opening function NEF, which is used to request analytical information on the artificial intelligence / machine learning AI / ML model transmission status in the subscription network.

[0154] In step 702, the NWDAF sends a second message to other network functions (5GC NF(s)) in the 5G core network based on the parameters requested in the first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network.

[0155] Step 703: The NWDAF receives AI / ML model transmission state data sent from other network functions 5GC NF(s) in the 5G core network, analyzes the AI / ML model transmission state data, and obtains analysis information of the AI / ML model transmission state.

[0156] The analyzed information is used to adjust network policy parameters and / or application layer model information via the AF.

[0157] Optionally, the parameters requested in the first message include a network data analysis identifier, an identifier of one user equipment (UE) or set of UEs that will receive the AI / ML model, or any UE that satisfies the analysis criteria. Identifier , including at least one of an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a protocol data unit (PDU) session that transmits the quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of the UE that uses the AI / ML model; an identifier of an application that uses the AI / ML model; an identifier of a quality of service stream that transmits the AI / ML model; an uplink bit rate for transmitting the AI / ML model and a downlink bit rate for transmitting the AI / ML model; an uplink packet delay for the AI / ML model and a downlink packet delay for the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; the number of packet retransmissions of the AI / ML model; a data collection time; a time for the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; a data network of a PDU session for transmitting the quality of service stream of the AI / ML model; and a service flow used for the AF; the analysis information includes at least one of: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information using the AI / ML model; a validity period of the analysis result; a user plane function (UPF) that provides AI / ML model transmission; a data network name of the PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; an identifier of the quality of service stream for transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; and the number of packet retransmissions of the AI / ML model. If the AI / ML model performs federated learning, the parameters requested in the first message further include federated learning group information, and the federated learning group information includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that it provides an AI / ML model or participates in the federated learning.

[0158] In this embodiment, the AF receives a request for analysis information of the AI / ML model transmission state in the subscription network sent by the AF, collects data from other network functions 5GC NF(s) in the 5G core network based on the parameters requested in the first message, receives the AI / ML model transmission state data sent from the other network functions 5GC NF(s) in the 5G core network, analyzes the AI / ML model transmission state data, and obtains analysis information of the AI / ML model transmission state, thereby effectively analyzing the AI / ML model transmission state. The AF then adjusts network policy parameters and / or application layer model information based on the analysis information, and the network effectively adjusts network transmission policies based on the AI / ML model transmission state. A third party can obtain the analysis of the AI / ML model transmission state and adjust the application layer information.

[0159] It should be noted that the method for analyzing the model transmission state in a subscription network provided by the embodiment of the present disclosure can realize all the method steps of the method embodiment shown in FIG. 4 and can achieve similar technical effects, and detailed descriptions of the parts and effects similar to those of the method embodiment of this embodiment will not be repeated here.

[0160] FIG. 8 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by an embodiment of the present disclosure. As shown in FIG. 8, when the analysis device for a model transmission state in a subscription network provided by this embodiment is applied to AF, the analysis device for a model transmission state in a subscription network provided by this embodiment includes a transceiver 800 used to transmit and receive data under the control of a processor 810.

[0161] 8, the bus architecture may include any number of interconnected buses and bridges, linking various circuits, such as one or more processors represented by processor 810 and memory represented by memory 820. The bus architecture may further link various other circuits, such as peripherals, regulators, and power management circuits, which are known in the art and will not be further described herein. The bus interface provides an interface. The transceiver 800 may be multiple elements, including a transmitter and a receiver, providing a unit for communicating with various other devices over transmission media, including wireless channels, wired channels, optical cables, and the like. The processor 810 is responsible for managing the bus architecture and normal processing, and the memory 820 may store data used by the processor 810 in performing operations.

[0162] The processor 810 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD), 810 may employ a multi-core architecture.

[0163] In this embodiment, the memory 820 is used to store computer programs, and the transceiver 800 is used to transmit and receive data under the control of the processor 810. The processor 810 controls the memory 820 to store computer programs. 820 reading a computer program stored in Sending a first message to a Network Data Analysis Function (NWDAF), directly or via a Network Capabilities Opening Function (NEF), to request analysis information of an artificial intelligence / machine learning (AI / ML) model transmission status in a subscription network; receiving, directly or via the NEF, analysis information of the AI / ML model transmission state determined by the NWDAF based on data of the AI / ML model transmission state received from other network functions (5GC NF(s)) of the 5G core network, wherein the data of the AI / ML model transmission state is obtained by the NWDAF sending a second message to the 5GC NF(s) based on parameters requested in the received first message, and the second message is used to collect data for analyzing the AI / ML model transmission state in the network; The analyzed information is used to adjust network policy parameters and / or application layer model information.

[0164] Optionally, the parameters requested in the first message include a network data analysis identifier, an identifier of one user equipment (UE) or set of UEs that will receive the AI / ML model, or any UE that satisfies the analysis criteria. Identifier , including at least one of an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a protocol data unit (PDU) session that transmits the quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of the UE that uses the AI / ML model; an identifier of an application that uses the AI / ML model; an identifier of a quality of service stream that transmits the AI / ML model; an uplink bit rate for transmitting the AI / ML model and a downlink bit rate for transmitting the AI / ML model; an uplink packet delay for the AI / ML model and a downlink packet delay for the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; the number of packet retransmissions of the AI / ML model; a data collection time; a time for the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; a data network of a PDU session for transmitting the quality of service stream of the AI / ML model; and a service flow used for the AF; the analysis information includes at least one of: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information using the AI / ML model; a validity period of the analysis result; a user plane function (UPF) that provides AI / ML model transmission; a data network name of the PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; an identifier of the quality of service stream for transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; and the number of packet retransmissions of the AI / ML model. If the AI / ML model performs federated learning, the parameters requested in the first message further include federated learning group information, and the federated learning group information includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that it provides an AI / ML model or participates in the federated learning.

[0165] Optionally, the processor 810 may further After receiving the analysis information, for sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analysis information; The first request is used to request an update of a network policy parameter for AI / ML model transmission, and the network policy parameter is used to optimize the AI / ML model transmission state.

[0166] Optionally, when the processor 810 is used to send a first request to a Policy Control Function PCF directly or via the NEF based on the analysis information, the processor 810 specifically determine new quality of service parameters for transmitting the AI / ML model, including at least one of a 5G quality of service identifier, a reflective quality of service control, an uplink direction bitrate for transmitting the AI / ML model, a downlink direction bitrate for transmitting the AI / ML model, an uplink direction packet delay of the AI / ML model, a downlink direction packet delay of the AI / ML model, a number of abnormal releases of the quality of service stream during a time period for transmitting the AI / ML model, a number of packet transmissions of the AI / ML model, a number of packet retransmissions of the AI / ML model, and a number of times a reporting threshold for abnormal release of the quality of service stream is reached during a time period for transmitting the AI / ML model, based on at least one of the analysis information; The AI / ML model in the analysis information UseDetermine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an application identifier, area information using the AI / ML model, IP address information of the application service using the AI / ML model, network slice of the PDU session for transmitting the service quality stream of the AI / ML model, and data network name of the PDU session for transmitting the service quality stream of the AI / ML model; or, if the AI / ML model performs federated learning, determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an identifier of a federated learning group that instructs analysis in the analysis information, an identifier of a UE or UE(s) participating in federated learning, and an identifier of each application that provides the AI / ML model or that participates in federated learning; Based on the area information and address information of the UE(s) transmitting the AI / ML model and each AF, determine the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the data network access identifier DNAI and the AI / ML model transmission state; The new quality of service parameters, the DNAI, and the area information and address information of the UE(s) and each AF corresponding to the DNAI in the first request Request It is used as a parameter to send to the PCF directly or via the NEF.

[0167] Optionally, when the processor 810 is used to determine a data network access identifier DNAI and area information and address information of the UE(s) and each AF corresponding to the DNAI based on area information and address information of the UE(s) and each AF transmitting the AI / ML model, specifically: Determine whether the current routing path is bad based on the area information and address information of the UE(s) transmitting the AI / ML model and each AF, If the current routing path is bad, determine the destination addresses of both the UE(s) transmitting the AI / ML model and the AF(s) based on the address information and the area information of the UE(s), determining a closest route based on the destination address; Based on the closest route, the DNAI and the area information and address information of the UE(s) and each AF corresponding to the DNAI are used to determine the DNAI.

[0168] Optionally, the first request may specifically include: The 5G quality of service identifier in the PCC rules, the reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream are requested to be adjusted based on the new quality of service parameters, and the first update result is instructed to be fed back to the PCF directly or via the NEF; The first update result is that the PCF adjusts the PCC rules based on the new quality of service parameters. result is determined by In this case, the processor 810 specifically further Used to receive the first update result sent from the PCF directly or via the NEF, the first update result including that the first request is accepted or that the first request is rejected.

[0169] Optionally, the first request may specifically include: Session management function SMF network The elementRequest the PCF to determine whether the session management policy needs to be updated, and if the SMF determines that the session management policy needs to be updated, the PCF is used to decide to send a second request to the SMF, where parameters requested in the second request include at least one of a DNAI, a traffic-oriented policy identifier, and traffic path information, and the second request is used by the SMF to determine a user plane function (UPF) to be selected based on the new session management policy and provide the corresponding DNAI, traffic-oriented policy identifier, and traffic path information; In this case, the processor 810 specifically further used to receive a second update result sent from the PCF directly or via the NEF, the second update result being determined by the PCF based on whether a new session management policy sent from the SMF updates the UPF path; The second update result is 2 This includes whether the request is accepted or rejected.

[0170] Optionally, the processor 810 may further After receiving the analysis information, adjusting information of the application layer model used to update quality of service parameters based on the analysis information, including at least one of model compression, model size, model transmission time slot, and model encoding and decoding; Based on the information of the application layer model to be adjusted, determining new quality of service parameters including a 5G quality of service identifier, a reflective quality of service control, a maximum uplink bit rate for transmitting the AI / ML model, a maximum downlink bit rate for transmitting the AI / ML model, a minimum uplink bit rate for transmitting the AI / ML model, a minimum downlink bit rate for transmitting the AI / ML model, and a priority of the quality of service stream; sending a third request to a Policy Control Function (PCF) directly or via the NEF; The parameters requested in the third request include the new quality of service parameters, and the third request is used to request an update of the quality of service parameters.

[0171] Optionally, the third request may specifically include: Based on the new quality of service parameters, the 5G quality of service identifier in the PCC rules, reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, and the minimum downlink bit rate for transmitting the AI / ML model, Quality of Service Stream Used to request the PCF to adjust the priority, In this case, processor 810 further Used to receive a third update result sent from the PCF directly or via the NEF, the third update result being determined by the PCF based on the result of adjusting the PCC rules, and the third update result including that the third request is accepted or that the third request is rejected.

[0172] Optionally, the processor 810 may further After adjusting the application layer model information, the model compression, model size, and model encoding and decoding in the adjusted application layer model information are used to directly send to the PCF, and the adjusted application layer model information is used by the PCF to support the adjustment of the 5G quality of service identifier, reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream in the PCC rules; In this case, processor 810 further Used to receive a fourth update result sent from the PCF, the fourth update result being determined by the PCF based on the result of adjusting the PCC rules based on the information of the application layer model to be adjusted, and the fourth update result including that the third request is accepted or that the third request is rejected.

[0173] Optionally, the processor 810 may further After adjusting the information of the application layer model, a model transmission time for the information of the adjusted application layer model is band The model transmission time in the information of the application layer model to be adjusted is used to directly send the information to the PCF. band is used to support the PCF to adjust a gate state parameter in the PCC rule, and the gate state parameter is used to support the SMF to update a session management policy based on a transmission start time and a transmission end time in the gate state; In this case, processor 810 further Used to receive a fifth update result sent from the PCF, the fifth update result being determined by the PCF based on the result of the new session management policy sent from the SMF, and the fifth update result including that the third request is accepted or that the third request is rejected.

[0174] In addition, the analysis device for the model transmission state in a subscription network provided by the embodiment of the present disclosure can realize all of the method steps of the method embodiment shown in Figures 3 to 6 and can achieve similar technical effects, so detailed descriptions of the parts and effects that are similar to those of the method embodiment of this embodiment will not be repeated.

[0175] FIG. 9 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by another embodiment of the present disclosure. As shown in FIG. 9, when the analysis device for a model transmission state in a subscription network provided by this embodiment is applied to AF, the analysis device 900 for a model transmission state in a subscription network provided by this embodiment is: A sending unit 901 is used to send a first message to a network data analysis function NWDAF directly or via a network capability opening function NEF, the first message being used to request analysis information of an artificial intelligence / machine learning AI / ML model transmission status in a subscription network; A receiving unit 902 is used to receive analysis information of AI / ML model transmission status sent from the NWDAF directly or via the NEF; The second message is used to collect data for analyzing the AI / ML model transmission status in the network; The analyzed information is used to adjust network policy parameters and / or application layer model information.

[0176] Optionally, the parameters requested in the first message include a network data analysis identifier, an identifier of one user equipment (UE) or set of UEs that will receive the AI / ML model, or any UE that satisfies the analysis criteria. Identifier , including at least one of an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a protocol data unit (PDU) session that transmits the quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of the UE that uses the AI / ML model; an identifier of an application that uses the AI / ML model; an identifier of a quality of service stream that transmits the AI / ML model; an uplink bit rate for transmitting the AI / ML model and a downlink bit rate for transmitting the AI / ML model; an uplink packet delay for the AI / ML model and a downlink packet delay for the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; the number of packet retransmissions of the AI / ML model; a data collection time; a time for the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; a data network of a PDU session for transmitting the quality of service stream of the AI / ML model; and a service flow used for the AF; the analysis information includes at least one of: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information using the AI / ML model; a validity period of the analysis result; a user plane function (UPF) that provides AI / ML model transmission; a data network name of the PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a quality of service request; an identifier of the quality of service stream for transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; and the number of packet retransmissions of the AI / ML model. If the AI / ML model performs federated learning, the parameters requested in the first message further include federated learning group information, and the federated learning group information includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that it provides an AI / ML model or participates in the federated learning.

[0177] Optionally, the transmitting unit further comprises: After receiving the analysis information, for sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analysis information; The first request is used to request an update of a network policy parameter for AI / ML model transmission, and the network policy parameter is used to optimize the AI / ML model transmission state.

[0178] Optionally, the sending unit may specifically: determine new quality of service parameters for transmitting the AI / ML model, including at least one of a 5G quality of service identifier, a reflective quality of service control, an uplink direction bitrate for transmitting the AI / ML model, a downlink direction bitrate for transmitting the AI / ML model, an uplink direction packet delay for the AI / ML model, a downlink direction packet delay for the AI / ML model, a number of abnormal releases of the quality of service stream during a time period for transmitting the AI / ML model, a number of packet transmissions of the AI / ML model, a number of packet retransmissions of the AI / ML model, and a number of times a reporting threshold for abnormal release of the quality of service stream is reached during a time period for transmitting the AI / ML model, based on at least one of the analysis information; The AI / ML model in the analysis information UseDetermine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an application identifier, area information using the AI / ML model, IP address information of the application service using the AI / ML model, network slice of the PDU session for transmitting the service quality stream of the AI / ML model, and data network name of the PDU session for transmitting the service quality stream of the AI / ML model; or, if the AI / ML model performs federated learning, determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on an identifier of a federated learning group that instructs analysis in the analysis information, an identifier of a UE or UE(s) participating in federated learning, and an identifier of each application that provides the AI / ML model or that participates in federated learning; Based on the area information and address information of the UE(s) transmitting the AI / ML model and each AF, determine the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the data network access identifier DNAI and the AI / ML model transmission state; The new quality of service parameters, the DNAI, and the area information and address information of the UE(s) and each AF corresponding to the DNAI in the first request Request It is used as a parameter to send to the PCF directly or via the NEF.

[0179] Optionally, the sending unit specifically further comprises: Determine whether the current routing path is bad based on the area information and address information of the UE(s) transmitting the AI / ML model and each AF, If the current routing path is bad, determine the destination addresses of both the UE(s) transmitting the AI / ML model and the AF(s) based on the address information and the area information of the UE(s), determining a closest route based on the destination address; Based on the closest route, the DNAI and the area information and address information of the UE(s) and each AF corresponding to the DNAI are used to determine the DNAI.

[0180] Optionally, the first request may specifically include: The 5G quality of service identifier in the PCC rules, the reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream are requested to be adjusted based on the new quality of service parameters, and the first update result is fed back to the PCF directly or via the NEF; The first update result is that the PCF adjusts the PCC rules based on the new quality of service parameters. result is determined by In this case, the sending unit further comprises: Used to receive a first update result sent from a PCF, either directly or via an NEF, the first update result including that the first request is accepted or that the first request is rejected.

[0181] Optionally, the first request may specifically include: Session management function SMF Network required The baseRequest the PCF to determine whether the session management policy needs to be updated, and if the SMF determines that the session management policy needs to be updated, the PCF is used to decide to send a second request to the SMF, where parameters requested in the second request include at least one of a DNAI, a traffic-oriented policy identifier, and traffic path information, and the second request is used by the SMF to determine a user plane function (UPF) to be selected based on the new session management policy and provide the corresponding DNAI, traffic-oriented policy identifier, and traffic path information; In this case, the receiving unit further comprises: used to receive a second update result sent from the PCF directly or via the NEF, the second update result being determined by the PCF based on whether a new session management policy sent from the SMF updates the UPF path; The second update result is 2 This includes whether the request is accepted or rejected.

[0182] Optionally, the apparatus further comprises a determining unit, the determining unit being configured to: After receiving the analysis information, adjusting information of the application layer model used to update quality of service parameters based on the analysis information, including at least one of model compression, model size, model transmission time slot, and model encoding and decoding; Determining the new quality of service parameters, including a 5G quality of service identifier, a reflective quality of service control, a maximum uplink bit rate for transmitting the AI / ML model, a maximum downlink bit rate for transmitting the AI / ML model, a minimum uplink bit rate for transmitting the AI / ML model, a minimum downlink bit rate for transmitting the AI / ML model, and a priority of the quality of service stream, based on information of the application layer model to be adjusted; sending a third request to a Policy Control Function (PCF) directly or via the NEF; The parameters requested in the third request include the new quality of service parameters, and the third request is used to request an update of the quality of service parameters.

[0183] Optionally, the third request may specifically include: Based on the new quality of service parameters, a 5G quality of service identifier in the PCC rules, a reflective quality of service control, a maximum uplink bit rate for transmitting an AI / ML model, a maximum downlink bit rate for transmitting an AI / ML model, a minimum uplink bit rate for transmitting an AI / ML model, and a minimum downlink bit rate for transmitting an AI / ML model; Quality of Service Stream Used to request the PCF to adjust the priority, In this case, the receiving unit further comprises: Used to receive a third update result sent from the PCF directly or via the NEF, the third update result being determined by the PCF based on the result of adjusting the PCC rules, and the third update result including that the third request is accepted or that the third request is rejected.

[0184] Optionally, the transmitting unit further comprises: After adjusting the application layer model information, the model compression, model size, and model encoding and decoding in the adjusted application layer model information are used to directly send to the PCF, and the adjusted application layer model information is used by the PCF to support adjusting the 5G quality of service identifier, reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream in the PCC rules; In this case, the receiving unit further comprises: Used to receive a fourth update result sent from the PCF, the fourth update result being determined by the PCF based on the result of adjusting the PCC rules based on the information of the application layer model to be adjusted, and the fourth update result including that the third request is accepted or that the third request is rejected.

[0185] Optionally, the transmitting unit further comprises: After adjusting the information of the application layer model, a model transmission time for the information of the adjusted application layer model is band The model transmission time in the information of the application layer model to be adjusted is used to directly send the information to the PCF. band is used to support the PCF to adjust a gate state parameter in the PCC rule, and the gate state parameter is used to support the SMF to update a session management policy based on a transmission start time and a transmission end time in the gate state; In this case, the receiving unit further comprises: Used to receive a fifth update result sent from the PCF, the fifth update result being determined by the PCF based on the result of the new session management policy sent from the SMF, and the fifth update result including that the third request is accepted or that the third request is rejected.

[0186] In addition, the analysis device for the model transmission state in a subscription network provided by the embodiment of the present disclosure can realize all of the method steps of the method embodiment shown in Figures 3 to 6 and can achieve similar technical effects, so detailed descriptions of the parts and effects that are similar to those of the method embodiment of this embodiment will not be repeated.

[0187] Figure 10 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by another embodiment of the present disclosure. As shown in Figure 10, when the analysis device for a model transmission state in a subscription network provided by this embodiment is applied to NWDAF, the analysis device for a model transmission state in a subscription network provided by this embodiment includes a transceiver 1000 used to transmit and receive data under the control of a processor 1010.

[0188] 10, the bus architecture may include any number of interconnected buses and bridges, linking various circuits, such as one or more processors represented by processor 1010 and memory represented by memory 1020. The bus architecture may further link various other circuits, such as peripherals, regulators, and power management circuits, which are known in the art and will not be further described herein. The bus interface provides an interface. The transceiver 1000 may be multiple elements, including a transmitter and a receiver, providing a unit for communicating with various other devices over transmission media, including wireless channels, wired channels, optical cables, and the like. The processor 1010 is responsible for managing the bus architecture and normal processing, and the memory 1020 may store data used by the processor 1010 in performing operations.

[0189] The processor 1010 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD), 1010 may employ a multi-core architecture.

[0190] In this embodiment, the memory 1020 is used to store computer programs, the transceiver 1000 is used to transmit and receive data under the control of the processor, and the processor 1010 is used to store the memory 1020 reading a computer program stored in Receiving a first message sent from an application function AF directly or via a network capability opening function NEF, the first message being used to request analysis information of an artificial intelligence / machine learning AI / ML model transmission status in a subscription network; Sending a second message to other network functions (5GC NF(s)) of the 5G core network based on the parameters requested in the first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network; Receives AI / ML model transmission status data sent from other network functions (5GC NF(s)) of the 5G core network, analyzes the AI / ML model transmission status data, and obtains analysis information of the AI / ML model transmission status; The analyzed information is used to adjust network policy parameters and / or application layer model information via the AF.

[0191] Optionally, the parameters requested in the first message include a network data analysis identifier, an identifier of one user equipment (UE) or set of UEs that will receive the AI / ML model, or any UE that satisfies the analysis criteria. Identifier, including at least one of an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a protocol data unit (PDU) session that transmits the quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of the UE that uses the AI / ML model; an identifier of an application that uses the AI / ML model; an identifier of a quality of service stream that transmits the AI / ML model; an uplink bit rate for transmitting the AI / ML model and a downlink bit rate for transmitting the AI / ML model; an uplink packet delay for the AI / ML model and a downlink packet delay for the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; the number of packet retransmissions of the AI / ML model; a data collection time; a time for the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; a data network of a PDU session for transmitting the quality of service stream of the AI / ML model; and a service flow used for the AF; the analysis information includes at least one of: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information using the AI / ML model; a validity period of the analysis result; a user plane function (UPF) that provides AI / ML model transmission; a data network name of the PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a quality of service request; an identifier of the quality of service stream for transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; and the number of packet retransmissions of the AI / ML model. If the AI / ML model performs federated learning, the parameters requested in the first message further include federated learning group information, and the federated learning group information includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that it provides an AI / ML model or participates in the federated learning.

[0192] In addition, the analysis device for the model transmission state in a subscription network provided by the embodiment of the present disclosure can realize all the method steps of the method embodiments shown in Figures 4 and 7 and can achieve similar technical effects, so detailed descriptions of the parts and effects that are similar to those of the method embodiments of this embodiment will not be repeated.

[0193] FIG. 11 is a configuration diagram of an analysis device for a model transmission state in a subscription network provided by another embodiment of the present disclosure. As shown in FIG. 11, when the analysis device for a model transmission state in a subscription network provided by this embodiment is applied to NWDAF, the analysis device 1100 for a model transmission state in a subscription network provided by this embodiment is: A receiving unit 1101 is used to receive a first message sent from an application function AF directly or via a network capability opening function NEF, the first message being used to request analysis information of an artificial intelligence / machine learning AI / ML model transmission status in a subscription network; A sending unit 1102 is used to send a second message to other network functions (5GC NF(s)) of a 5G core network based on the parameters required in the first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network; An analysis unit 1103 is used to receive AI / ML model transmission state data sent from other network functions 5GC NF(s) of the 5G core network, analyze the AI / ML model transmission state data, and obtain analysis information of the AI / ML model transmission state; The analyzed information is used to adjust network policy parameters and / or application layer model information via the AF.

[0194] Optionally, the parameters requested in the first message include a network data analysis identifier, an identifier of one user equipment (UE) or set of UEs that will receive the AI / ML model, or any UE that satisfies the analysis criteria. Identifier , including at least one of an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a protocol data unit (PDU) session that transmits the quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits the quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of the UE that uses the AI / ML model; an identifier of an application that uses the AI / ML model; an identifier of a quality of service stream that transmits the AI / ML model; an uplink bit rate for transmitting the AI / ML model and a downlink bit rate for transmitting the AI / ML model; an uplink packet delay for the AI / ML model and a downlink packet delay for the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; the number of packet retransmissions of the AI / ML model; a data collection time; a time for the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; a data network of a PDU session for transmitting the quality of service stream of the AI / ML model; and a service flow used for the AF; the analysis information includes at least one of: a network slice of a PDU session for transmitting the quality of service stream of the AI / ML model; an identifier of an application using the AI / ML model; area information using the AI / ML model; a validity period of the analysis result; a user plane function (UPF) that provides AI / ML model transmission; a data network name of the PDU session for transmitting the quality of service stream of the AI / ML model; a size of the AI / ML transmission model; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; an identifier of the quality of service stream for transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; the number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; the number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; the number of packet transmissions of the AI / ML model; and the number of packet retransmissions of the AI / ML model. If the AI / ML model performs federated learning, the parameters requested in the first message further include federated learning group information, and the federated learning group information includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of an application participating in the federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that it provides an AI / ML model or participates in the federated learning.

[0195] In addition, the analysis device for the model transmission state in a subscription network provided by the embodiment of the present disclosure can realize all the method steps of the method embodiments shown in Figures 4 and 7 and can achieve similar technical effects, so detailed descriptions of the parts and effects that are similar to those of the method embodiments of this embodiment will not be repeated.

[0196] The division of units in the above embodiments of the present disclosure is a rough outline and merely represents a logical functional division, and different division methods may be used in actual implementation. Furthermore, the functional units in each embodiment of the present disclosure may be integrated into a single processing unit, each unit may exist physically independently, or two or more units may be integrated into a single unit. The integrated units may be implemented in the form of hardware or software functional units.

[0197] The integrated units may be implemented as software functional units and stored in a processor-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present disclosure may be embodied essentially, or in part contributing to the prior art, or in whole or in part, in the form of a software product, and the computer software product may be stored in a storage medium. The computer software product may include instructions for causing a computer device (such as a personal computer, a server, or a network device) or a processor to execute all or some of the steps of the methods described in the embodiments of the present disclosure. Meanwhile, the storage medium may include various media capable of storing program code, such as a U disk, a removable hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0198] An embodiment of the present disclosure further provides a processor-readable storage medium having a computer program stored therein, the computer program being used to cause a processor to perform any of the method embodiments described above.

[0199] Here, the processor-readable storage medium may be a magnetic memory (e.g., a floppy disk, a hard disk, a magnetic tape, a magneto-optical disk (MO), etc.), an optical memory (e.g., optical disk( CD ) , Digital general-purpose optical disc ( DVD ) , Blu-ray Disc ( BD ) , Holographic general-purpose optical disc ( HVD ) etc.), and semiconductor memory (e.g., ROM, Erasable programming read-only memory ( EPROM ) , Electrically erasable programmable read-only memory ( EEPROM ) It may be any available medium or data storage device accessible to a processor, including but not limited to, non-volatile memory (NAND FLASH), solid state hard disk (SSD), etc.

[0200] Those skilled in the art will appreciate that embodiments of the present disclosure may be provided as a method, a system, or a computer program product. Thus, the present disclosure may employ examples of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. The present disclosure may also employ examples of computer program products embodied in one or more computer-usable storage media (including, but not limited to, disk memory, optical memory, etc.) containing computer-usable program code.

[0201] The present disclosure will be described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer-executable instructions. These computer-executable instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine such that the instructions, executed by the processor of the computer or other programmable data processing device, generate means for implementing the functions specified in one or more flows in the flowcharts and / or one or more blocks in the block diagrams.

[0202] These processor-executable instructions may also be stored in a processor-readable memory that can cause a computer or other programmable data processing apparatus to operate in a particular manner to produce an article of manufacture that includes an instruction apparatus that implements the functions specified in one or more flows of the flowcharts and / or one or more blocks of the block diagrams.

[0203] These processor-executable instructions may also be loaded into a computer or other programmable data processing device to cause the computer or other programmable device to perform a series of operational steps to generate a computer-implemented process, and thus the instructions that execute on the computer or other programmable device provide steps for implementing the functions specified in one or more flows of the flowcharts and / or one or more blocks of the block diagrams.

[0204] Obviously, those skilled in the art can make various modifications and variations to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalent techniques, the present disclosure also intends to include these modifications and variations.

[0205] This disclosure claims priority to a Chinese patent application bearing application number 202111407051.2 and entitled "Method, apparatus and readable storage medium for analyzing model transmission state in subscription network," filed with the China Patent Office on November 24, 2021, the entire contents of which are incorporated herein by reference. [Explanation of symbols]

[0206] 800 Transceiver 810 processor 820 memory 900 Analyzer 901 Transmitting Unit 902 receiving unit 1000 Transceivers 1010 processor 1020 memory 1100 Analysis equipment 1101 receiving unit 1102 Transmitting unit 1103 Analysis Unit

Claims

1. A method for analyzing a model transmission state in a subscription network, applied to an application function AF, comprising: sending, directly or via a Network Capabilities Opening Function NEF, to a Network Data Analysis Function NWDAF, a first message used to request analytical information of an artificial intelligence / machine learning AI / ML model transmission status in a subscription network; and receiving, directly or via the NEF, analysis information of the AI / ML model transmission state determined by the NWDAF based on data of the AI / ML model transmission state received from other network functions (5GC NF(s)) of the 5G core network, transmitted from the NWDAF; the analyzed information is used to adjust network policy parameters and / or application layer model information; After receiving the analysis information, the method further comprises: sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analyzed information; the first request is used to request an update of a network policy parameter for AI / ML model transmission, the network policy parameter is used to optimize the AI / ML model transmission state; sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analyzed information, determining new quality of service parameters for transmitting the AI / ML model, including at least one of a 5G quality of service identifier, a reflective quality of service control, an uplink direction maximum bit rate for transmitting the AI / ML model, a downlink direction bit rate for transmitting the AI / ML model, an uplink direction packet delay for the AI / ML model, a downlink direction packet delay for the AI / ML model, a number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model, a number of packet transmissions of the AI / ML model, a number of packet retransmissions of the AI / ML model, and a number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model, based on at least one of the analysis information; determining area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the application using the AI / ML model, area information using the AI / ML model, IP address information of the application service using the AI / ML model, network slice of the PDU session for transmitting the quality of service stream of the AI / ML model, and data network name of the PDU session for transmitting the quality of service stream of the AI / ML model in the analysis information; or, if the AI / ML model performs federated learning, determining area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the federated learning group instructing analysis in the analysis information, the identifier of the UE or UE(s) participating in federated learning, and each application identifier indicating providing the AI / ML model or participating in federated learning; According to the area information and address information of the UE(s) and each AF transmitting the AI / ML model, determining the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the data network access identifier DNAI and the AI / ML model transmission state; sending the new quality of service parameters, the DNAI, and area information and address information of the UE(s) and each AF corresponding to the DNAI to the PCF directly or via the NEF as request parameters in the first request; A method for analyzing a model transmission state in a subscription network, characterized by:

2. The data of the AI / ML model transmission state is obtained by the NWDAF sending a second message to the 5GC NF(s) based on parameters requested in the received first message, and the second message is used to collect data for analyzing the AI / ML model transmission state in the network; The parameters required in the first message include at least one of a network data analysis identifier, an identifier of one user equipment UE or a set of UEs receiving the AI / ML model, or an identifier of any UE that satisfies an analysis condition, an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a PDU session that transmits a quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits a quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of a UE using the AI / ML model; an identifier of an application using the AI / ML model; an identifier of a quality of service stream transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; a number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; a number of packet transmissions of the AI / ML model; a number of packet retransmissions of the AI / ML model; a data collection time; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting a quality of service stream of the AI / ML model; a data network of a PDU session for transmitting a quality of service stream of the AI / ML model; and a service flow used for the AF; The analysis information includes a network slice of a PDU session for transmitting a quality of service stream of the AI / ML model, an identifier of an application using the AI / ML model, area information using the AI / ML model, a validity period of the analysis result, a user plane function UPF that provides the AI / ML model transmission, a data network name of a PDU session for transmitting a quality of service stream of the AI / ML model, a size of the AI / ML transmission model, a time of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, and an AI / ML model transmission timestamp. the quality of service stream identifier for transmitting the model, the uplink direction bit rate for transmitting the AI / ML model and the downlink direction bit rate for transmitting the AI / ML model, the uplink direction packet delay for the AI / ML model and the downlink direction packet delay for the AI / ML model, the number of abnormal releases of the quality of service stream in the time period for transmitting the AI / ML model, the number of times a reporting threshold for abnormal release of the quality of service stream is reached in the time period for transmitting the AI / ML model, the number of packet transmissions of the AI / ML model, and the number of packet retransmissions of the AI / ML model; If the AI / ML model performs federated learning, the parameters required in the first message further include federated learning group information, the federated learning group information including at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in federated learning, and an identifier of an application participating in federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in federated learning, and an identifier of an application participating in federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that the application provides an AI / ML model or participates in the federated learning.

2. The method of claim 1 .

3. The step of determining a data network access identifier DNAI and area information and address information of the UE(s) and each AF corresponding to the DNAI based on area information and address information of the UE(s) and each AF transmitting the AI / ML model includes: determining whether the current routing path is bad based on the area information and address information of the UE(s) and each AF transmitting the AI / ML model; If the current routing path is bad, determining the destination address of both the UE(s) transmitting the AI / ML model and the AF according to the address information of each AF and the area information of the UE(s); determining a closest route based on the destination address; and determining a DNAI and area information and address information of the UE(s) and each AF corresponding to the DNAI based on the closest route.

2. The method of claim 1 .

4. Specifically, the first requirement is: The 5G quality of service identifier in the PCC rule, the reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream are requested to be adjusted based on the new quality of service parameters, and the first update result is used to instruct the PCF to feed back the adjusted first update result directly or via the NEF; the first update result is determined based on a result of the PCF adjusting PCC rules based on the new quality of service parameters; In this case, the method further comprises: receiving the first update result sent from the PCF directly or via the NEF, wherein the first update result includes that the first request is accepted or that the first request is rejected; 2. The method of claim 1 .

5. Specifically, the first requirement is: A session management function (SMF) network element requests the PCF to determine whether a session management policy needs to be updated, and if the SMF determines that the session management policy needs to be updated, the PCF is used to determine to send a second request to the SMF, and parameters required in the second request include at least one of a DNAI, a traffic-oriented policy identifier, and traffic path information, and the second request is used by the SMF to determine a user plane function (UPF) to be selected based on the new session management policy and provide the corresponding DNAI, traffic-oriented policy identifier, and traffic path information; In this case, the method further comprises: receiving a second update result sent from the PCF directly or via the NEF, the second update result being determined by the PCF based on whether a new session management policy sent from the SMF updates a UPF path; the second update result includes the second request being accepted or the first request being rejected.

2. The method of claim 1 .

6. After receiving the analysis information, the method further comprises: adjusting, based on the analysis information, information of the application layer model used to update quality of service parameters, including at least one of model compression, model size, model transmission time slot, and model encoding and decoding; Determining new quality of service parameters based on the information of the application layer model to be adjusted, including a 5G quality of service identifier, a reflective quality of service control, a maximum uplink bit rate for transmitting the AI / ML model, a maximum downlink bit rate for transmitting the AI / ML model, a minimum uplink bit rate for transmitting the AI / ML model, a minimum downlink bit rate for transmitting the AI / ML model, and a priority of the quality of service stream; sending a third request to the PCF directly or via the NEF; the parameters requested in the third request include the new quality of service parameters, and the third request is used to request an update of the quality of service parameters.

2. The method of claim 1 .

7. Specifically, the third requirement is: The 5G quality of service identifier in the PCC rules, the reflective quality of service control, the maximum uplink bit rate for transmitting the AI / ML model, the maximum downlink bit rate for transmitting the AI / ML model, the minimum uplink bit rate for transmitting the AI / ML model, the minimum downlink bit rate for transmitting the AI / ML model, and the priority of the quality of service stream are used to request the PCF to adjust the priority of the quality of service stream based on the new quality of service parameters; In this case, the method further comprises: receiving a third update result sent from the PCF directly or via the NEF, the third update result being determined by the PCF based on the result of adjusting PCC rules, and the third update result including that the third request is accepted or that the third request is rejected; 7. The method of claim 6.

8. After adjusting the information of the application layer model, the method further comprises: directly transmitting information about the application layer model to be adjusted, including model compression, model size, and encoding and decoding of the model, to the PCF, wherein the information about the application layer model to be adjusted is used by the PCF to support adjusting a 5G quality of service identifier in PCC rules, a reflective quality of service control, a maximum uplink bit rate for transmitting the AI / ML model, a maximum downlink bit rate for transmitting the AI / ML model, a minimum uplink bit rate for transmitting the AI / ML model, a minimum downlink bit rate for transmitting the AI / ML model, and a priority of a quality of service stream; In this case, the method further comprises: receiving a fourth updated result sent from the PCF, the fourth updated result being determined by the PCF based on a result of adjusting a PCC rule based on information of an application layer model to be adjusted, the fourth updated result including that the third request is accepted or that the third request is rejected; Or, After adjusting the information of the application layer model, the method further comprises: directly sending a model transmission time slot in the information of the application layer model to be adjusted to the PCF, wherein the model transmission time slot in the information of the application layer model to be adjusted is used to support the PCF to adjust a gate state parameter in a PCC rule, and the gate state parameter is used to support the SMF to update a session management policy based on a transmission start time and a transmission end time in the gate state; In this case, the method further comprises: receiving a fifth update result sent from the PCF, the fifth update result being determined by the PCF based on the result of the new session management policy sent from the SMF, and the fifth update result including that the third request is accepted or that the third request is rejected; 7. The method of claim 6.

9. A method for analyzing a model transmission state in a subscription network, which is applied to a network data analysis function NWDAF, comprising: receiving a first message sent from an application function AF directly or via a network capability opening function NEF, the first message being used to request analytical information on the state of transmission of an artificial intelligence / machine learning AI / ML model in a subscription network; Sending a second message to other network functions (5GC NF(s)) of a 5G core network based on the parameters requested in the first message, the second message being used to collect data for analyzing the AI / ML model transmission status in the network; Receiving data of the AI / ML model transmission state sent from other network functions 5GC NF(s) of the 5G core network, analyzing the data of the AI / ML model transmission state, and obtaining analysis information of the AI / ML model transmission state; The analysis information is used to adjust network policy parameters and / or application layer model information via the AF; The analysis information is further used by the AF to send a first request to a Policy Control Function (PCF) directly or via the NEF based on the analysis information; the first request is used to request an update of a network policy parameter for AI / ML model transmission, the network policy parameter is used to optimize the AI / ML model transmission state; Specifically, the analysis information is determining new quality of service parameters for transmitting the AI / ML model, including at least one of a 5G quality of service identifier, a reflective quality of service control, an uplink direction maximum bit rate for transmitting the AI / ML model, a downlink direction bit rate for transmitting the AI / ML model, a minimum uplink direction bit rate for transmitting the AI / ML model, a downlink direction bit rate for transmitting the AI / ML model, and a priority of the quality of service stream, based on at least one of an uplink direction bit rate for transmitting the AI / ML model and a downlink direction packet delay for the AI / ML model, an uplink direction packet delay for the AI / ML model and a downlink direction packet delay for the AI / ML model in the analysis information, a number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model, a number of packet transmissions of the AI / ML model, a number of packet retransmissions of the AI / ML model, and a number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; Determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the application using the AI / ML model, area information using the AI / ML model, IP address information of the application service using the AI / ML model, network slice of the PDU session for transmitting the quality of service stream of the AI / ML model, and data network name of the PDU session for transmitting the quality of service stream of the AI / ML model in the analysis information; or, if the AI / ML model performs federated learning, determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the federated learning group that instructs analysis in the analysis information, the identifier of the UE or UE(s) participating in federated learning, and each application identifier that provides the AI / ML model or indicates participation in federated learning; Based on the area information and address information of the UE(s) and each AF transmitting the AI / ML model, determining the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the data network access identifier DNAI and the AI / ML model transmission state; and sending the new quality of service parameters, the DNAI, and area information and address information of the UE(s) and each AF corresponding to the DNAI to the PCF directly or via the NEF as request parameters in the first request; A method for analyzing a model transmission state in a subscription network, characterized by:

10. The parameters required in the first message include at least one of a network data analysis identifier, an identifier of one user equipment UE or a set of UEs receiving the AI / ML model, or an identifier of any UE that satisfies an analysis condition, an identifier of an application that uses the AI / ML model, an area of ​​the AI / ML model transmission, a network slice indicating a PDU session that transmits a quality of service stream of the AI / ML model, a data network indicating a PDU session that transmits a quality of service stream of the AI / ML model, a time period of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, a size of the AI / ML transmission model, a quality of service requirement indicating a quality of service stream for transmitting the AI / ML model, and / or a specific quality of service requirement for indicating the transmission of the AI / ML model; the second message includes at least one of: a current location of a UE using the AI / ML model; an identifier of an application using the AI / ML model; an identifier of a quality of service stream transmitting the AI / ML model; an uplink direction bit rate for transmitting the AI / ML model and a downlink direction bit rate for transmitting the AI / ML model; an uplink direction packet delay of the AI / ML model and a downlink direction packet delay of the AI / ML model; a number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model; a number of packet transmissions of the AI / ML model; a number of packet retransmissions of the AI / ML model; a data collection time; a time of the AI / ML model transmission; a start timestamp of the AI / ML model transmission; an end timestamp of the AI / ML model transmission; a size of the AI / ML transmission model; a network slice of a PDU session for transmitting a quality of service stream of the AI / ML model; a data network of a PDU session for transmitting a quality of service stream of the AI / ML model; and a service flow used for the AF; The analysis information includes a network slice of a PDU session for transmitting a quality of service stream of the AI / ML model, an identifier of an application using the AI / ML model, area information using the AI / ML model, a validity period of the analysis result, a user plane function UPF that provides the AI / ML model transmission, a data network name of a PDU session for transmitting a quality of service stream of the AI / ML model, a size of the AI / ML transmission model, a time of the AI / ML model transmission, a start timestamp of the AI / ML model transmission, an end timestamp of the AI / ML model transmission, and an AI / ML model transmission timestamp. the quality of service stream identifier for transmitting the model, the uplink direction bit rate for transmitting the AI / ML model and the downlink direction bit rate for transmitting the AI / ML model, the uplink direction packet delay for the AI / ML model and the downlink direction packet delay for the AI / ML model, the number of abnormal releases of the quality of service stream in the time period for transmitting the AI / ML model, the number of times a reporting threshold for abnormal release of the quality of service stream is reached in the time period for transmitting the AI / ML model, the number of packet transmissions of the AI / ML model, and the number of packet retransmissions of the AI / ML model; If the AI / ML model performs federated learning, the parameters required in the first message further include federated learning group information, the federated learning group information including at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in federated learning, and an identifier of an application participating in federated learning; In this case, the second message further includes at least one of an identifier of a federated learning group for instructing analysis, an identifier of a UE or UE(s) participating in federated learning, and an identifier of an application participating in federated learning; In this case, the analysis information further includes at least one of an identifier of a federated learning group for instructing the analysis, an identifier of a UE or UE(s) participating in the federated learning, and an identifier of each application indicating that the application provides an AI / ML model or participates in the federated learning.

10. The method of claim 9.

11. An apparatus for analyzing a model transmission state in a subscription network, comprising: a sending unit used for sending, directly or via a network capability opening function NEF, to a network data analysis function NWDAF, a first message used for requesting analysis information of an artificial intelligence / machine learning AI / ML model transmission state in a subscription network; An analysis unit used to receive, directly or via the NEF, analysis information of an AI / ML model transmission state determined by the NWDAF based on data of an AI / ML model transmission state received from other network functions (5GC NF(s)) of a 5G core network, transmitted from the NWDAF; the analyzed information is used to adjust network policy parameters and / or application layer model information; The transmitting unit further comprises: After receiving the analysis information, sending a first request to a Policy Control Function (PCF) directly or via the NEF based on the analysis information; the first request is used to request an update of a network policy parameter for AI / ML model transmission, the network policy parameter is used to optimize the AI / ML model transmission state; The transmitting unit specifically includes: determining new quality of service parameters for transmitting the AI / ML model, including at least one of a 5G quality of service identifier, a reflective quality of service control, an uplink direction maximum bit rate for transmitting the AI / ML model, a downlink direction bit rate for transmitting the AI / ML model, a minimum uplink direction bit rate for transmitting the AI / ML model, a downlink direction bit rate for transmitting the AI / ML model, and a priority of the quality of service stream, based on at least one of an uplink direction bit rate for transmitting the AI / ML model and a downlink direction packet delay for the AI / ML model, an uplink direction packet delay for the AI / ML model and a downlink direction packet delay for the AI / ML model in the analysis information, a number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model, a number of packet transmissions of the AI / ML model, a number of packet retransmissions of the AI / ML model, and a number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; Determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the application using the AI / ML model, area information using the AI / ML model, IP address information of the application service using the AI / ML model, network slice of the PDU session for transmitting the quality of service stream of the AI / ML model, and data network name of the PDU session for transmitting the quality of service stream of the AI / ML model in the analysis information; or, if the AI / ML model performs federated learning, determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the federated learning group that instructs analysis in the analysis information, the identifier of the UE or UE(s) participating in federated learning, and each application identifier that provides the AI / ML model or indicates participation in federated learning; Based on the area information and address information of the UE(s) and each AF transmitting the AI / ML model, determining the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the data network access identifier DNAI and the AI / ML model transmission state; and sending the new quality of service parameters, the DNAI, and area information and address information of the UE(s) and each AF corresponding to the DNAI to the PCF directly or via the NEF as request parameters in the first request; An apparatus for analyzing a model transmission state in a subscription network, comprising:

12. The data of the AI / ML model transmission state is obtained by the NWDAF sending a second message to the 5GC NF(s) based on the parameters requested in the received first message, and the second message is used to collect data for analyzing the AI / ML model transmission state in the network.

12. The device of claim 11 .

13. An apparatus for analyzing a model transmission state in a subscription network, comprising: a receiving unit used to receive a first message sent from an application function AF directly or via a network capability opening function NEF, the first message being used to request analytical information on the transmission status of an artificial intelligence / machine learning AI / ML model in a subscription network; A sending unit used to send a second message to other network functions (5GC NF(s)) of a 5G core network based on the parameters required in the first message, the second message being used to collect data for analyzing an AI / ML model transmission state in the network; An analysis unit used to receive data of the AI / ML model transmission state sent from other network functions 5GC NF(s) of the 5G core network, analyze the data of the AI / ML model transmission state, and obtain analysis information of the AI / ML model transmission state; The analysis information is used to adjust network policy parameters and / or application layer model information via the AF; The analysis information is further used by the AF to send a first request to a Policy Control Function (PCF) directly or via the NEF based on the analysis information; the first request is used to request an update of a network policy parameter for AI / ML model transmission, the network policy parameter is used to optimize the AI / ML model transmission state; Specifically, the analysis information is determining new quality of service parameters for transmitting the AI / ML model, including at least one of a 5G quality of service identifier, a reflective quality of service control, an uplink direction maximum bit rate for transmitting the AI / ML model, a downlink direction bit rate for transmitting the AI / ML model, a minimum uplink direction bit rate for transmitting the AI / ML model, a downlink direction bit rate for transmitting the AI / ML model, and a priority of the quality of service stream, based on at least one of an uplink direction bit rate for transmitting the AI / ML model and a downlink direction packet delay for the AI / ML model, an uplink direction packet delay for the AI / ML model and a downlink direction packet delay for the AI / ML model in the analysis information, a number of abnormal releases of the quality of service stream in a time period for transmitting the AI / ML model, a number of packet transmissions of the AI / ML model, a number of packet retransmissions of the AI / ML model, and a number of times a reporting threshold for abnormal release of the quality of service stream is reached in a time period for transmitting the AI / ML model; Determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the application using the AI / ML model, area information using the AI / ML model, IP address information of the application service using the AI / ML model, network slice of the PDU session for transmitting the quality of service stream of the AI / ML model, and data network name of the PDU session for transmitting the quality of service stream of the AI / ML model in the analysis information; or, if the AI / ML model performs federated learning, determine area information and address information of the UE(s) transmitting the AI / ML model and each AF based on the identifier of the federated learning group that instructs analysis in the analysis information, the identifier of the UE or UE(s) participating in federated learning, and each application identifier that provides the AI / ML model or indicates participation in federated learning; Based on the area information and address information of the UE(s) and each AF transmitting the AI / ML model, determining the area information and address information of the UE(s) and each AF corresponding to the DNAI used to provide a path for optimizing the data network access identifier DNAI and the AI / ML model transmission state; and sending the new quality of service parameters, the DNAI, and area information and address information of the UE(s) and each AF corresponding to the DNAI to the PCF directly or via the NEF as request parameters in the first request; An apparatus for analyzing a model transmission state in a subscription network, comprising:

14. A processor-readable storage medium having a computer program stored therein, the computer program being used to cause the processor to execute the method according to any one of claims 1 to 8. A processor-readable storage medium comprising:

15. A processor-readable storage medium, the processor-readable storage medium storing a computer program, the computer program being used to cause the processor to execute a method according to claim 9 or 10. A processor-readable storage medium comprising:

Images