ML model providing method and apparatus, communication system, storage medium and program product

Abstract

The present disclosure provides an ML model providing method and device, a communication system, a storage medium and a program product. The ML model providing method comprises: receiving a model query request for querying an ML model sent by a V-RE-NWDAF network element with an MTLF function; in the case that the ML model exists locally, sending a model query request response to the V-RE-NWDAF network element; in the case that a first message for requesting to provide the ML model sent by the V-RE-NWDAF network element is received, determining a model transmission mode of the ML model; sending a second message corresponding to the model transmission mode to the V-RE-NWDAF network element, so that the V-RE-NWDAF network element obtains the ML model by using the second message.

Description

Technical Field

[0001] This disclosure relates to the field of communications, and in particular to a method and apparatus for providing ML models, a communication system, a storage medium, and a program product. Background Technology

[0002] The existing protocol specifications define the capability of RE-NWDAF (Roaming Exchange Network Data Analytics Function). As an NWDAF with roaming exchange capability, RE-NWDAF is used to complete the exchange of network data and network analysis between the visited network operator and the home network operator in roaming scenarios.

[0003] For example, regarding the Analytics Exposure scenario, 3GPP (3rd Generation Partnership Project) defines specific procedures and methods for VPLMN (Visited Public Land Mobile Network) to provide analytics access to HPLMMN (Home Public Land Mobile Network), and for HPLMMN to provide analytics access to VPLMN. Regarding the Data Collection scenario, 3GPP defines specific procedures and methods for VPLMN to collect data from HPLMMN, and for HPLMMN to collect data from VPLMN.

[0004] As a key network element for network intelligence, NWDAF includes various modules such as AnLF (Analytics Logical Function), MTLF (Model Training Logical Function), ADRF (Analytics Data Repository Function), DCCF (Data Collection Coordination Function), and MFAF (Messaging Framework Adaptor Function). NWDAF with MTLF functionality can provide training and provisioning capabilities for ML (Machine Learning) models for NWDAF analysis. Summary of the Invention

[0005] The inventors noted that while data and analytics can be exchanged between RE-NWDAFs in related technologies, the methods for exchanging ML models have not yet been defined. Because the AI ​​(Artificial Intelligence) capabilities of NWDAFs are typically limited to within a single network, cross-domain ML models cannot be obtained, thus restricting the improvement of end-to-end service quality and the development of innovative businesses.

[0006] Accordingly, this disclosure provides a method for providing ML models, enabling RE-NWDAF to exchange ML models, thereby achieving the sharing of ML models between the home network and the visited network.

[0007] In a first aspect of this disclosure, a method for providing an ML model is provided, executed by a network data analysis function H-RE-NWDAF (Host Data Analysis Function with Roaming Switching Capability) in a home network with Model Training Logic Function (MTLF), comprising: receiving a model query request for querying a machine learning (ML) model from a network data analysis function V-RE-NWDAF (Visitor Data Analysis Function with Roaming Switching Capability) in a visited network with MTLF functionality; if the ML model exists locally, sending a model query request response to the V-RE-NWDAF; upon receiving a first message from the V-RE-NWDAF requesting the provision of the ML model, determining a model transmission mode for the ML model; and sending a second message corresponding to the model transmission mode to the V-RE-NWDAF, so that the V-RE-NWDAF uses the second message to obtain the ML model.

[0008] In some embodiments, the model transfer mode of the ML model includes: direct embedding mode or pre-signed Uniform Resource Locator (URL) mode.

[0009] In some embodiments, when the model transmission mode is the direct embedding mode, the second message includes the model encoding of the ML model, the model acquisition method of the ML model is obtained through the direct embedding mode, and the model additional information; when the model transmission mode is the pre-signed URL mode, the second message includes the pre-signed URL corresponding to the ML model, the model acquisition method of the ML model is obtained through the pre-signed URL mode, and the model additional information.

[0010] In some embodiments, sending a second message corresponding to the model transmission mode to the V-RE-NWDAF network element includes: detecting whether the state of the ML model is normal; if the state of the ML model is normal, sending a second message corresponding to the model transmission mode to the V-RE-NWDAF network element; if the state of the ML model is abnormal, sending an error response code to the V-RE-NWDAF network element to identify the abnormal situation.

[0011] In some embodiments, determining the model transmission mode of the ML model includes: upon receiving the first message, verifying the first message; and if the verification passes, determining the model transmission mode of the ML model based on the size of the ML model and the current load.

[0012] In some embodiments, the model query request response includes first model information and first model acquisition verification information of the ML model. Verifying the first message includes: extracting second model acquisition verification information and second model information from the first message; determining whether the second model acquisition verification information has been tampered with based on the first model acquisition verification information; if the second model acquisition verification information has not been tampered with, determining whether the validity period of the second model acquisition verification information has expired; if the validity period of the second model acquisition verification information has not expired, determining whether the second model information and the first model information are consistent; if the second model information and the first model information are consistent, then the verification is determined to be successful.

[0013] In some embodiments, if the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the second model information is inconsistent with the first model information, a corresponding error response code is sent to the V-RE-NWDAF network element.

[0014] In some embodiments, receiving a model query request for querying an ML model sent by a V-RE-NWDAF network element with MTLF functionality includes: receiving the model query request sent by the V-RE-NWDAF network element according to the roaming protocol between the H-RE-NWDAF network element and the V-RE-NWDAF network element.

[0015] In some embodiments, the first message includes a model provision request and the second message includes a model provision request response; or the first message includes a model provision subscription request and the second message includes a model provision notification.

[0016] In some embodiments, if the ML model does not exist locally, a model acquisition request for obtaining the ML model is sent to other NWDAF network elements with MTLF functionality in the current PLMN; the ML model is obtained according to the model acquisition request response sent by the other NWDAF network elements; and the model query request response is sent to the V-RE-NWDAF network element.

[0017] In a second aspect of this disclosure, an H-RE-NWDAF network element with MTLF functionality is provided, comprising: a memory; and a processor coupled to the memory, the processor being configured to execute instructions stored in the memory to implement the method as described in any of the above embodiments.

[0018] In a third aspect of this disclosure, a method for providing an ML model is provided, executed by a V-RE-NWDAF network element with MTLF functionality, comprising: upon receiving a model acquisition request from an NF consumer in a VPLMN network for acquiring an ML model, determining whether the ML model exists locally; if the ML model does not exist locally and the cross-PLMN request identifier in the model acquisition request is true, sending a model query request to an H-RE-NWDAF network element with MTLF functionality for querying the ML model; upon receiving a model query request response from the H-RE-NWDAF network element, sending a first message to the H-RE-NWDAF network element requesting the provision of the ML model; acquiring the ML model according to a second message sent by the H-RE-NWDAF network element; and sending a model acquisition request response to the NF consumer so that the NF consumer can acquire the ML model using the model acquisition request response.

[0019] In some embodiments, obtaining the ML model includes: obtaining the ML model according to the model acquisition method included in the second message.

[0020] In some embodiments, sending a model acquisition request response to the NF consumer includes: performing an integrity check on the ML model; and if the integrity check passes, sending the model acquisition request response to the NF consumer.

[0021] In some embodiments, the model query request response includes model information of the ML model and model acquisition verification information; the first message includes the model information and the model acquisition verification information.

[0022] In some embodiments, the first message includes a model provision request and the second message includes a model provision request response; or the first message includes a model provision subscription request and the second message includes a model provision notification.

[0023] In a fourth aspect of this disclosure, a V-RE-NWDAF network element with MTLF functionality is provided, comprising: a memory; and a processor coupled to the memory, the processor being configured to execute instructions stored in the memory to implement the method as described in any of the above embodiments.

[0024] In a fifth aspect of this disclosure, a communication system is provided, comprising: an H-RE-NWDAF network element having MTLF functionality as described in any of the preceding embodiments; and a V-RE-NWDAF network element having MTLF functionality as described in any of the preceding embodiments.

[0025] In a sixth aspect of this disclosure, a method for providing an ML model is provided, executed by a V-RE-NWDAF network element with MTLF functionality, comprising: receiving a model query request for querying an ML model sent by an H-RE-NWDAF network element with MTLF functionality; if the ML model exists locally, sending a model query request response to the H-RE-NWDAF network element; if a first message for requesting the provision of the ML model is received from the H-RE-NWDAF network element, determining a model transmission mode of the ML model; and sending a second message corresponding to the model transmission mode to the H-RE-NWDAF network element, so that the H-RE-NWDAF network element obtains the ML model using the second message.

[0026] In some embodiments, the model transmission mode of the ML model includes: direct embedding mode or pre-signed URL mode.

[0027] In some embodiments, when the model transmission mode is the direct embedding mode, the second message includes the model encoding of the ML model, the model acquisition method of the ML model is obtained through the direct embedding mode, and the model additional information; when the model transmission mode is the pre-signed URL mode, the second message includes the pre-signed URL corresponding to the ML model, the model acquisition method of the ML model is obtained through the pre-signed URL mode, and the model additional information.

[0028] In some embodiments, sending a second message corresponding to the model transmission mode to the H-RE-NWDAF network element includes: detecting whether the state of the ML model is normal; if the state of the ML model is normal, sending a second message corresponding to the model transmission mode to the H-RE-NWDAF network element; if the state of the ML model is abnormal, sending an error response code to the H-RE-NWDAF network element to identify the abnormal situation.

[0029] In some embodiments, determining the model transmission mode of the ML model includes: upon receiving the first message, verifying the first message; and if the verification passes, determining the model transmission mode of the ML model based on the size of the ML model and the current load.

[0030] In some embodiments, the model query request response includes first model information and first model acquisition verification information of the ML model. Verifying the first message includes: extracting second model acquisition verification information and second model information from the first message; determining whether the second model acquisition verification information has been tampered with based on the first model acquisition verification information; if the second model acquisition verification information has not been tampered with, determining whether the validity period of the second model acquisition verification information has expired; if the validity period of the second model acquisition verification information has not expired, determining whether the second model information and the first model information are consistent; if the second model information and the first model information are consistent, then the verification is determined to be successful.

[0031] In some embodiments, if the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the second model information is inconsistent with the first model information, a corresponding error response code is sent to the H-RE-NWDAF network element.

[0032] In some embodiments, receiving a model query request for querying an ML model sent by an H-RE-NWDAF network element with MTLF functionality includes: receiving the model query request sent by the H-RE-NWDAF network element according to the roaming protocol between the H-RE-NWDAF network element and the V-RE-NWDAF network element.

[0033] In some embodiments, if the ML model does not exist locally, a model acquisition request for obtaining the ML model is sent to other NWDAF network elements with MTLF functionality in the current PLMN; the ML model is obtained according to the model acquisition request response sent by the other NWDAF network elements; and the model query request response is sent to the H-RE-NWDAF network element.

[0034] In some embodiments, the first message includes a model provision request and the second message includes a model provision request response; or the first message includes a model provision subscription request and the second message includes a model provision notification.

[0035] In some embodiments, when the ML model is updated, a model provisioning notification for indicating the model update is sent to the H-RE-NWDAF network element, wherein the model provisioning notification for indicating the model update includes the method for obtaining the updated ML model and additional model information.

[0036] In a seventh aspect of this disclosure, a V-RE-NWDAF network element with MTLF functionality is provided, comprising: a memory; and a processor coupled to the memory, the processor being configured to execute instructions stored in the memory to implement the method as described in any of the above embodiments.

[0037] In an eighth aspect of this disclosure, a method for providing an ML model is provided, executed by an H-RE-NWDAF network element with MTLF functionality, comprising: upon receiving a model acquisition request from an NF consumer in an HPLMN network for acquiring an ML model, determining whether the ML model exists locally; if the ML model does not exist locally and the cross-PLMN request identifier in the model acquisition request is true, sending a model query request to a V-RE-NWDAF network element with MTLF functionality for querying the ML model; upon receiving a model query request response from the V-RE-NWDAF network element, sending a first message to the V-RE-NWDAF network element requesting the provision of the ML model; acquiring the ML model according to a second message sent by the V-RE-NWDAF network element; and sending a model acquisition request response to the NF consumer so that the NF consumer can acquire the ML model using the model acquisition notification.

[0038] In some embodiments, obtaining the ML model includes: obtaining the ML model according to the model acquisition method included in the second message.

[0039] In some embodiments, sending a model acquisition request response to the NF consumer includes: performing an integrity check on the ML model; and if the integrity check passes, sending the model acquisition request response to the NF consumer.

[0040] In some embodiments, the model query request response includes model information of the ML model and model acquisition verification information; the first message includes the model information and the model acquisition verification information.

[0041] In some embodiments, the first message includes a model provision request and the second message includes a model provision request response; or the first message includes a model provision subscription request and the second message includes a model provision notification.

[0042] In a ninth aspect of this disclosure, an H-RE-NWDAF network element with MTLF functionality is provided, comprising: a memory; and a processor coupled to the memory, the processor being configured to execute instructions stored in the memory to implement the method as described in any of the above embodiments.

[0043] In a tenth aspect of this disclosure, a communication system is provided, comprising: a V-RE-NWDAF network element having MTLF functionality as described in any of the above embodiments; and an H-RE-NWDAF network element having MTLF functionality as described in any of the above embodiments.

[0044] In the eleventh aspect of this disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method as described in any of the above embodiments.

[0045] In a twelfth aspect of this disclosure, a computer program product is provided, including computer instructions, wherein the computer instructions, when executed by a processor, implement the method as described in any of the above embodiments.

[0046] Other features and advantages of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating a method for providing an ML model according to an embodiment of this disclosure; Figure 2 This is a schematic diagram of the structure of an H-RE-NWDAF network element with MTLF function according to an embodiment of this disclosure; Figure 3 A flowchart illustrating a method for providing an ML model according to another embodiment of this disclosure; Figure 4 This is a schematic diagram of the structure of a V-RE-NWDAF network element with MTLF function according to an embodiment of this disclosure; Figure 5 This is a schematic diagram of the structure of a communication system according to an embodiment of the present disclosure; Figure 6 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure; Figure 7A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure; Figure 8 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure; Figure 9 This is a schematic diagram of the structure of a V-RE-NWDAF network element with MTLF function according to another embodiment of this disclosure; Figure 10 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure; Figure 11 This is a schematic diagram of the structure of an H-RE-NWDAF network element with MTLF function according to another embodiment of this disclosure; Figure 12 This is a schematic diagram of the structure of a communication system according to another embodiment of the present disclosure; Figure 13 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure; Figure 14 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure. Detailed Implementation

[0049] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit this disclosure or its application or use. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0050] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of this disclosure.

[0051] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0052] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0053] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0054] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0055] Figure 1 This is a flowchart illustrating a method for providing an ML model according to an embodiment of the present disclosure. In some embodiments, the following ML model providing method is performed by an H-RE-NWDAF (Home-RE-NWDAF, Network Data Analysis Function with Roaming Switching Capability in the Home Network) network element with MTLF (hereinafter referred to as H-RE-NWDAF network element in this embodiment), including steps 11-14.

[0056] In step 11, a model query request for querying the ML model is received from a V-RE-NWDAF (Visited-RE-NWDAF, Network Data Analysis Function with Roaming Switching Capability in Visited Network) network element with MTLF function (hereinafter referred to as V-RE-NWDAF network element in this embodiment).

[0057] For example, the model query request used to query ML models is the Nnwdaf_RoamingML ModelInfo_Request request, which carries information such as ML model ID, ML model accuracy requirements, ML model format, PLMN (Public Land Mobile Network) to which it belongs, and S-NSSAI (Single Network Slice Selection Assistance Information) to which it belongs.

[0058] In some embodiments, a model query request sent by a V-RE-NWDAF network element is received according to the roaming protocol between the H-RE-NWDAF network element and the V-RE-NWDAF network element.

[0059] In other words, based on the roaming protocol between the H-RE-NWDAF network element and the V-RE-NWDAF network element, it is determined whether to accept the model query request sent by the V-RE-NWDAF network element.

[0060] In step 12, if the ML model exists locally, a model query request response is sent to the V-RE-NWDAF network element.

[0061] In some embodiments, the presence of the required ML model can be determined locally by parsing the model query request sent by the V-RE-NWDAF network element.

[0062] For example, the model query request response is Nnwdaf_RoamingML ModelInfo_Request response. The model query request response includes the first model information of the ML model, the vendor information of the model, and the first model retrieval verification information, etc.

[0063] In some embodiments, if the ML model does not exist locally, a model acquisition request for obtaining the ML model is sent to other NWDAF network elements with MTLF functionality in the current PLMN. Based on the model acquisition request responses sent by other NWDAF network elements, the ML model is obtained, and a model query request response is sent to the V-RE-NWDAF network element.

[0064] In other words, if an H-RE-NWDAF network element does not have the required ML model locally, it will obtain the required ML model from other NWDAF network elements with MTLF functionality in the current PLMN, thereby effectively increasing the ML model provision capability of the H-RE-NWDAF network element.

[0065] In step 13, upon receiving the first message from the V-RE-NWDAF network element requesting the provision of an ML model, the model transmission mode of the ML model is determined.

[0066] In some embodiments, the step of determining the model transfer mode of the ML model includes steps S101-S102.

[0067] S101. Upon receiving the first message, verify the first message.

[0068] In some embodiments, the step of verifying the first message includes steps S201-S205.

[0069] S201. Extract the second model from the first message to obtain verification information and second model information.

[0070] S202. Based on the verification information obtained from the first model included in the model query request response, determine whether the verification information obtained from the second model has been tampered with.

[0071] S203. If the verification information obtained by the second model has not been tampered with, determine whether the validity period of the verification information obtained by the second model has expired.

[0072] S204. If the validity period of the verification information obtained by the second model has not expired, determine whether the second model information and the first model information included in the model query request response are consistent.

[0073] S205. If the information of the second model is consistent with the information of the first model, then the verification is confirmed to be successful.

[0074] It should be noted that if the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the information of the second model is inconsistent with the information of the first model, then the verification will be deemed unsuccessful.

[0075] S102. If the verification is successful, determine the model transmission mode of the ML model based on the size of the ML model and the current load.

[0076] In some embodiments, the model transport mode of the ML model includes: direct embedding mode or pre-signed URL (Uniform Resource Locator) mode.

[0077] In step 14, a second message corresponding to the model transmission mode is sent to the V-RE-NWDAF network element so that the V-RE-NWDAF network element can use the second message to obtain the ML model.

[0078] For example, when the model transmission mode is direct embedding mode, the second message includes the model code of the ML model, the model acquisition method of the ML model obtained through direct embedding mode, and additional model information (such as input / output format requirements, model version, model description, etc.).

[0079] When the model transmission mode is the pre-signed URL mode, the second message includes the pre-signed URL corresponding to the ML model, the model acquisition method of the ML model is obtained through the pre-signed URL mode, and additional model information (such as input and output format requirements, model version, model description, etc.).

[0080] In some embodiments, before sending the second message corresponding to the model transmission mode to the V-RE-NWDAF network element, the status of the ML model is checked. If the ML model is in a normal state, the second message corresponding to the model transmission mode is sent to the V-RE-NWDAF network element. If the ML model is in an abnormal state, an error response code indicating the abnormal situation is sent to the V-RE-NWDAF network element.

[0081] For example, when the ML model's state becomes abnormal, a second message is sent to the V-RE-NWDAF network element. This second message includes the corresponding error response code. For example, the error response code may include MODEL_STATUS_INVALID.

[0082] In some embodiments, during the process of determining whether the verification passes, if the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the information of the second model is inconsistent with the information of the first model, a second message is sent to the V-RE-NWDAF network element. The second message includes a corresponding error response code. For example, the error response code may include INVALID or EXPIRED_TOKEN.

[0083] In some embodiments, V-RE-NWDAF network elements and H-RE-NWDAF network elements can interact using a request / response mode or a subscription / notification mode.

[0084] In some embodiments, the first message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a request to the model, and the second message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a response to the request to the model.

[0085] For example, a model provision request can be named Nnwdaf_RoamingML ModelProvision_Request, and a model provision request response can be named Nnwdaf_RoamingML ModelProvision_Request response.

[0086] In some embodiments, the first message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a subscription request for the model, and the second message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a notification for the model.

[0087] For example, a model can provide a subscription request as Nnwdaf_RoamingML ModelProvision_Subscribe, and a model can provide a notification as Nnwdaf_RoamingML ModelProvision_Notify.

[0088] In the ML model provision method provided in the above embodiments of this disclosure, the H-RE-NWDAF network element receives a model query request sent by the V-RE-NWDAF network element for querying an ML model. If an ML model exists locally, the H-RE-NWDAF network element sends a model query request response to the V-RE-NWDAF network element. Upon receiving a first message from the V-RE-NWDAF network element requesting the provision of an ML model, the H-RE-NWDAF network element determines the model transmission mode of the ML model and sends a second message corresponding to the model transmission mode to the V-RE-NWDAF network element, so that the V-RE-NWDAF network element can obtain the ML model using the second message. This enables the RE-NWDAF to have the capability to exchange ML models, realizing the sharing of ML models between the home network and the visited network.

[0089] Figure 2 This is a schematic diagram of the structure of an H-RE-NWDAF network element with MTLF function (hereinafter referred to as H-RE-NWDAF network element in the following description of this embodiment) according to an embodiment of the present disclosure.

[0090] like Figure 2 As shown, the H-RE-NWDAF network element 20 can be represented in the form of a general computing device. The H-RE-NWDAF network element 20 includes a memory 21, a processor 22, and a bus 23 that connects different system components.

[0091] The memory 21 may include, for example, system memory, non-volatile storage media, etc. System memory may store, for example, an operating system, applications, a boot loader, and other programs. System memory may include volatile storage media, such as random access memory (RAM) and / or cache memory. Non-volatile storage media may store, for example, instructions for a corresponding embodiment of an executing ML model providing a method. Non-volatile storage media include, but are not limited to, disk storage, optical storage, flash memory, etc.

[0092] Processor 22 can be implemented using a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA) or other programmable logic devices, discrete hardware components such as discrete gates or transistors. Accordingly, each module, such as the acquisition module, calculation module, and adjustment module, can be implemented by executing instructions in the central processing unit (CPU) running memory to perform the corresponding steps, or by implementing dedicated circuits that perform the corresponding steps.

[0093] For example, processor 22 is configured to execute instructions stored in memory 21, such as Figure 1 The method involved in any of the embodiments.

[0094] Bus 23 can use any of the various bus architectures. For example, bus architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MCA) bus, and the Peripheral Component Interconnect (PCI) bus.

[0095] The interfaces 24, 25, and 26 of the H-RE-NWDAF network element 20, as well as the memory 21 and processor 22, can be connected via bus 23. Input / output interface 24 provides connection interfaces for input / output devices such as monitors, mice, and keyboards. Network interface 25 provides connection interfaces for various networked devices. Storage interface 26 provides connection interfaces for external storage devices such as floppy disks, USB flash drives, and SD cards.

[0096] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations thereof, can be implemented by computer-readable program instructions.

[0097] These computer-readable program instructions are provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable device to produce a machine, such that execution of the instructions by the processor produces means for implementing the functions specified in one or more boxes of the flowchart and / or block diagram.

[0098] These computer-readable program instructions may also be stored in a computer-readable storage medium. These instructions cause a computer to work in a particular manner to produce an article of manufacture, including instructions that implement the functions specified in one or more boxes in a flowchart and / or block diagram.

[0099] This disclosure may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects.

[0100] This disclosure also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement... Figure 1 The method involved in any of the embodiments.

[0101] This disclosure also provides a computer program product, including computer instructions, wherein the computer instructions, when executed by a processor, implement as follows: Figure 1 The method involved in any of the embodiments.

[0102] Figure 3 This is a flowchart illustrating a method for providing an ML model according to another embodiment of the present disclosure. In some embodiments, the following ML model providing method is performed by a V-RE-NWDAF network element having an MTLF (hereinafter referred to as V-RE-NWDAF network element in the description of this embodiment), including steps 31-35.

[0103] In step 31, upon receiving a model retrieval request from an NF consumer in the VPLMN network for obtaining an ML model, it is determined whether an ML model exists locally.

[0104] In some embodiments, NF consumers in a VPLMN discover the V-RE-NWDAF within that VPLMN through a discovery mechanism. If the V-RE-NWDAF does not have MTLF functionality, it cannot provide services. If the V-RE-NWDAF does have MTLF functionality, it sends a model acquisition request to obtain the ML model. This model acquisition request includes information such as the ML model ID, ML model accuracy requirements, ML model format, and cross-PLMN request identifier.

[0105] In step 32, if the ML model does not exist locally and the cross-PLMN (Public Land Mobile Network) request identifier in the model acquisition request is true, a model query request for querying the ML model is sent to the H-RE-NWDAF network element with MTLF function (hereinafter referred to as H-RE-NWDAF network element in this embodiment).

[0106] For example, a model query request used to query an ML model includes Nnwdaf_RoamingML ModelInfo_Request. This model query request includes information such as the ML model ID, ML model accuracy requirements, ML model format, the PLMN to which it belongs, and the S-NSSAI to which it belongs.

[0107] It should be noted that if a local ML model exists, the V-RE-NWDAF network element will directly provide the ML model to the NF consumer.

[0108] In step 33, upon receiving the model query request response from the H-RE-NWDAF network element, a first message is sent to the H-RE-NWDAF network element to request the provision of an ML model.

[0109] In some embodiments, the model query request response includes model information of the ML model and model acquisition verification information. The first message includes model information and model acquisition verification information. Therefore, the H-RE-NWDAF network element can use the model information and model acquisition verification information included in the first message to verify the first message.

[0110] In step 34, the ML model is obtained based on the second message sent by the H-RE-NWDAF network element.

[0111] In some embodiments, an ML model is obtained according to the model acquisition method included in the second message.

[0112] In step 35, a model retrieval request response is sent to the NF consumer so that the NF consumer can retrieve the ML model using the model retrieval request response.

[0113] In some embodiments, the ML model is first subjected to an integrity check. If the integrity check passes, the model retrieval request response is sent to the NF consumer. This improves the user experience of the NF consumer when using the ML model.

[0114] In some embodiments, V-RE-NWDAF network elements and H-RE-NWDAF network elements can interact using a request / response mode or a subscription / notification mode.

[0115] In some embodiments, the first message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a request to the model, and the second message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a response to the request to the model.

[0116] For example, a model provision request can be named Nnwdaf_RoamingML ModelProvision_Request, and a model provision request response can be named Nnwdaf_RoamingML ModelProvision_Request response.

[0117] In some embodiments, the first message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a subscription request for the model, and the second message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a notification for the model.

[0118] For example, a model can provide a subscription request as Nnwdaf_RoamingML ModelProvision_Subscribe, and a model can provide a notification as Nnwdaf_RoamingML ModelProvision_Notify.

[0119] In the ML model provision method provided in the above embodiments of this disclosure, when a V-RE-NWDAF network element receives a model acquisition request from an NF consumer in a VPLMN network for obtaining an ML model, if the ML model does not exist locally and the cross-PLMN request identifier in the model acquisition request is true, it sends a model query request to the H-RE-NWDAF network element for querying the ML model. Upon receiving a model query request response from the H-RE-NWDAF network element, the V-RE-NWDAF network element sends a first message to the H-RE-NWDAF network element requesting the provision of an ML model. Based on a second message sent by the H-RE-NWDAF network element, it obtains the ML model and sends a model acquisition request response to the NF consumer, enabling the NF consumer to obtain the ML model using the model acquisition request response. This enables the RE-NWDAF to have the capability to exchange ML models, realizing the sharing of ML models between the home network and the visited network.

[0120] Figure 4 This is a schematic diagram of the structure of a V-RE-NWDAF network element with MTLF function (hereinafter referred to as V-RE-NWDAF network element in the following description of this embodiment) according to an embodiment of the present disclosure.

[0121] like Figure 4 As shown, the V-RE-NWDAF network element 40 includes a memory 41, a processor 42, a bus 43, an input / output interface 44, a network interface 45, and a storage interface 46.

[0122] Figure 4 and Figure 2 The difference is that, in Figure 4 In the illustrated embodiment, processor 42 is configured to execute instructions stored in memory 41 as follows: Figure 3 The method involved in any of the embodiments.

[0123] This disclosure also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement... Figure 3 The method involved in any of the embodiments.

[0124] This disclosure also provides a computer program product, including computer instructions, wherein the computer instructions, when executed by a processor, implement as follows: Figure 3 The method involved in any of the embodiments.

[0125] Figure 5 This is a schematic diagram of the structure of a communication system according to an embodiment of this disclosure. Figure 5 As shown, the communication system includes a V-RE-NWDAF network element 51 with MTLF function and an H-RE-NWDAF network element 52 with MTLF function.

[0126] It should be noted that the V-RE-NWDAF network element 51 is... Figure 4 The V-RE-NWDAF network element involved in any of the embodiments, H-RE-NWDAF network element 52 is Figure 2 The H-RE-NWDAF network element involved in any of the embodiments.

[0127] The following examples illustrate... Figure 5 The communication system shown is described.

[0128] Figure 6 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure includes steps 61-611.

[0129] In step 61, the NF consumer in the VPLMN discovers the V-RE-NWDAF in the VPLMN through the discovery mechanism. If the V-RE-NWDAF does not have MTLF functionality, it cannot provide services. If the V-RE-NWDAF has MTLF functionality, it sends an ML model acquisition request, which includes information such as ML model ID, ML model accuracy requirements, ML model format, and cross-PLMN request identifier.

[0130] In step 62, after receiving the ML model retrieval request, V-RE-NWDAF determines whether the required ML model exists locally.

[0131] In step 63, if the required ML model does not exist locally in V-RE-NWDAF and the cross-PLMN request flag is true, V-RE-NWDAF sends a model query request (Nnwdaf_RoamingML ModelInfo_Request) to H-RE-NWDAF with MTLF functionality to query the ML model. This request carries information such as ML model ID, ML model accuracy requirements, ML model format, PLMN, and S-NSSAI.

[0132] In step 64, H-RE-NWDAF checks the roaming protocol between H-RE-NWDAF and V-RE-NWDAF to determine whether it can receive the model query request.

[0133] In step 65, if H-RE-NWDAF is able to receive the model query request, H-RE-NWDAF parses the model query request and determines whether the required ML model exists locally. If the required ML model exists locally, H-RE-NWDAF sends a model query request response (Nnwdaf_RoamingML ModelInfo_Request response) to V-RE-NWDAF to inform V-RE-NWDAF of ML model information, the vendor information to which the model belongs, and the model's verification information.

[0134] In step 66, V-RE-NWDAF sends a model provision request (Nnwdaf_RoamingMLModelProvision_Request) to H-RE-NWDAF in order to request the provision of the ML model from H-RE-NWDAF. The model provision request includes information such as model acquisition and verification information, ML model ID, ML model accuracy requirements, ML model format, PLMN, and S-NSSAI.

[0135] In step 67, after receiving the model provision request, H-RE-NWDAF verifies the model provision request. For example, it extracts the model acquisition verification information, determines whether the model acquisition verification information has been tampered with, determines whether the validity period has expired, and determines whether information such as the ML model ID and the PLMN to which it belongs is consistent.

[0136] In step 671, if the verification fails, H-RE-NWDAF performs exception handling. H-RE-NWDAF sends a model provisioning request response (Nnwdaf_RoamingML ModelProvision_Request response) to V-RE-NWDAF, which includes the corresponding error response code.

[0137] For example, if the verification information obtained by the model has been tampered with, the error response code included in the model's request response will be INVALID. If the validity period of the verification information obtained by the model has expired, the error response code included in the model's request response will be EXPIRED_TOKEN.

[0138] In this case, V-RE-NWDAF needs to re-execute the above process to resend the model provisioning request.

[0139] In step 68, if the verification is successful, H-RE-NWDAF determines the model transmission mode based on the model size, load conditions, etc.

[0140] For example, the model transmission mode can be either direct embedding mode or pre-signed URL mode.

[0141] In step 681, H-RE-NWDAF sends a model provisioning request response (Nnwdaf_RoamingML ModelProvision_Request response) to V-RE-NWDAF corresponding to the model transmission mode.

[0142] For example, when the model delivery mode is direct embedding mode, H-RE-NWDAF sends the model provision request response to V-RE-NWDAF. The model provision request response includes the model code of the ML model, the model acquisition method for obtaining the ML model through direct embedding mode, and additional model information (e.g., input / output format requirements, model version, model description, etc.).

[0143] For example, when the model transmission mode is the pre-signed URL mode, H-RE-NWDAF sends the model provision request response to V-RE-NWDAF. The model provision request response includes the pre-signed URL corresponding to the ML model, the model acquisition method for obtaining the ML model through the pre-signed URL mode, and additional model information (e.g., input / output format requirements, model version, model description).

[0144] In step 682, when H-RE-NWDAF is preparing to send the ML model, if the ML model is in an abnormal state, it will also send the Model Provision Request response (Nnwdaf_RoamingML ModelProvision_Request response). The Model Provision Request response carries an error code identifier to explain the abnormal state of the model, such as MODEL_STATUS_INVALID.

[0145] In step 69, V-RE-NWDAF obtains the ML model according to the model acquisition method included in the model provision request response and performs model integrity verification.

[0146] In step 610, if the model integrity check passes, V-RE-NWDAF sends an ML model retrieval request response to the NF consumer in VPLMN.

[0147] In step 611, the NF consumer obtains the request response based on the ML model and obtains the ML model.

[0148] Figure 7 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure includes steps 71-713.

[0149] In step 71, the NF consumer in the VPLMN discovers the V-RE-NWDAF in the VPLMN through the discovery mechanism. If the V-RE-NWDAF does not have MTLF functionality, it cannot provide services. If the V-RE-NWDAF has MTLF functionality, it sends an ML model acquisition request, which includes information such as ML model ID, ML model accuracy requirements, ML model format, and cross-PLMN request identifier.

[0150] In step 72, after receiving the ML model acquisition request, V-RE-NWDAF determines whether the required ML model exists locally.

[0151] In step 73, if the required ML model does not exist locally in V-RE-NWDAF and the cross-PLMN request flag is true, V-RE-NWDAF sends a model query request (Nnwdaf_RoamingML ModelInfo_Request) to H-RE-NWDAF with MTLF functionality to query the ML model. This request carries information such as ML model ID, ML model accuracy requirements, ML model format, PLMN to which it belongs, and S-NSSAI to which it belongs.

[0152] In step 74, H-RE-NWDAF checks the roaming protocol between H-RE-NWDAF and V-RE-NWDAF to determine whether it can receive the model query request.

[0153] In step 75, if the H-RE-NWDAF is able to receive the model query request, the H-RE-NWDAF parses the model query request and determines whether the required ML model exists locally. If the required ML model does not exist locally, a model acquisition request can be sent to other NWDAFs (with MTLF functionality) in this PLMN, carrying the relevant information from the aforementioned model query request.

[0154] In step 76, if other NWDAFs can provide the model, a model retrieval request response is sent to H-RE-NWDAF so that H-RE-NWDAF can retrieve the ML model through the model retrieval request response.

[0155] In step 77, H-RE-NWDAF sends a model query request response (Nnwdaf_RoamingMLModelInfo_Request response) to V-RE-NWDAF to inform V-RE-NWDAF of ML model information, the vendor information to which the model belongs, and the model's verification information.

[0156] In step 78, V-RE-NWDAF sends a model provision request (Nnwdaf_RoamingMLModelProvision_Request) to H-RE-NWDAF in order to request the provision of the ML model from H-RE-NWDAF. The model provision request includes information such as model acquisition and verification information, ML model ID, ML model accuracy requirements, ML model format, PLMN, and S-NSSAI.

[0157] In step 79, after receiving the model provision request, H-RE-NWDAF verifies the model provision request. For example, it extracts the model acquisition verification information, determines whether the model acquisition verification information has been tampered with, determines whether the validity period has expired, and determines whether information such as the ML model ID and the PLMN to which it belongs is consistent.

[0158] In step 791, if the verification fails, H-RE-NWDAF performs exception handling. H-RE-NWDAF sends a model provisioning request response (Nnwdaf_RoamingML ModelProvision_Request response) to V-RE-NWDAF, which includes the corresponding error response code.

[0159] For example, if the verification information obtained by the model has been tampered with, the error response code included in the model's request response will be INVALID. If the validity period of the verification information obtained by the model has expired, the error response code included in the model's request response will be EXPIRED_TOKEN.

[0160] In this case, V-RE-NWDAF needs to re-execute the above process to resend the model provisioning request.

[0161] In step 710, if the verification is successful, H-RE-NWDAF determines the model transmission mode based on the model size, load conditions, etc.

[0162] For example, the model transmission mode can be either direct embedding mode or pre-signed URL mode.

[0163] In step 7101, H-RE-NWDAF sends a model provisioning request response (Nnwdaf_RoamingML ModelProvision_Request response) to V-RE-NWDAF corresponding to the model transmission mode.

[0164] For example, when the model delivery mode is direct embedding mode, H-RE-NWDAF sends the model provision request response to V-RE-NWDAF. The model provision request response includes the model code of the ML model, the model acquisition method for obtaining the ML model through direct embedding mode, and additional model information (e.g., input / output format requirements, model version, model description, etc.).

[0165] For example, when the model transmission mode is the pre-signed URL mode, H-RE-NWDAF sends the model provision request response to V-RE-NWDAF. The model provision request response includes the pre-signed URL corresponding to the ML model, the model acquisition method for obtaining the ML model through the pre-signed URL mode, and additional model information (e.g., input / output format requirements, model version, model description).

[0166] In step 7102, when H-RE-NWDAF is preparing to send the ML model, if the ML model is in an abnormal state, it will also send the Model Provide Request Response, which carries an error code identifier to explain the abnormal state of the model, such as MODEL_STATUS_INVALID.

[0167] In step 711, V-RE-NWDAF obtains the ML model according to the model acquisition method included in the model provision request response and performs model integrity verification.

[0168] In step 712, if the model integrity check passes, V-RE-NWDAF sends an ML model retrieval request response to the NF consumer in VPLMN.

[0169] In step 713, the NF consumer obtains the request response based on the ML model and obtains the ML model.

[0170] Figure 8 This is a flowchart illustrating a method for providing an ML model according to one embodiment of the present disclosure. In some embodiments, the following ML model providing method is performed by a V-RE-NWDAF network element having an MTLF (hereinafter referred to as V-RE-NWDAF network element in the description of this embodiment), including steps 81-84.

[0171] In step 81, a model query request for querying an ML model is received from an H-RE-NWDAF network element with MTLF functionality (hereinafter referred to as H-RE-NWDAF network element in this embodiment).

[0172] For example, the model query request used to query ML models is the Nnwdaf_RoamingML ModelInfo_Request request, which carries information such as ML model ID, ML model accuracy requirements, ML model format, PLMN to which it belongs, and S-NSSAI to which it belongs.

[0173] In some embodiments, a model query request sent by an H-RE-NWDAF network element is received according to the roaming protocol between the V-RE-NWDAF network element and the H-RE-NWDAF network element.

[0174] In other words, based on the roaming protocol between the V-RE-NWDAF network element and the H-RE-NWDAF network element, it is determined whether to accept the model query request sent by the H-RE-NWDAF network element.

[0175] In step 82, if the ML model exists locally, a model query request response is sent to the H-RE-NWDAF network element.

[0176] In some embodiments, the presence of the required ML model can be determined locally by parsing the model query request sent by the H-RE-NWDAF network element.

[0177] For example, the model query request response is Nnwdaf_RoamingML ModelInfo_Request response. The model query request response includes the first model information of the ML model, the vendor information of the model, and the first model retrieval verification information, etc.

[0178] In some embodiments, if the ML model does not exist locally, a model acquisition request for obtaining the ML model is sent to other NWDAF network elements with MTLF functionality in the current PLMN. Based on the model acquisition request responses sent by other NWDAF network elements, the ML model is obtained, and a model query request response is sent to the H-RE-NWDAF network element.

[0179] In other words, if a V-RE-NWDAF network element does not have the required ML model locally, it will obtain the required ML model from other NWDAF network elements with MTLF functionality in the current PLMN, thereby effectively increasing the ML model provision capability of the V-RE-NWDAF network element.

[0180] In step 83, upon receiving the first message from the H-RE-NWDAF network element requesting the provision of an ML model, the model transmission mode of the ML model is determined.

[0181] In some embodiments, the step of determining the model transfer mode of the ML model includes steps S301-S302.

[0182] S301. Upon receiving the first message, verify the first message.

[0183] In some embodiments, the step of verifying the first message includes steps S401-S405.

[0184] S401. Extract the second model from the first message to obtain verification information and second model information.

[0185] S402. Based on the verification information obtained from the first model included in the model query request response, determine whether the verification information obtained from the second model has been tampered with.

[0186] S403. If the verification information obtained by the second model has not been tampered with, determine whether the validity period of the verification information obtained by the second model has expired.

[0187] S404. If the validity period of the verification information obtained by the second model has not expired, determine whether the information of the second model is consistent with the information of the first model included in the model query request response.

[0188] S405. If the information of the second model is consistent with the information of the first model, then the verification is confirmed to be successful.

[0189] It should be noted that if the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the information of the second model is inconsistent with the information of the first model, then the verification will be deemed unsuccessful.

[0190] S302. If the verification is successful, determine the model transmission mode of the ML model based on the size of the ML model and the current load.

[0191] In some embodiments, the model transfer mode of the ML model includes: direct embedding mode or pre-signed URL mode.

[0192] In step 84, a second message corresponding to the model transmission mode is sent to the H-RE-NWDAF network element so that the H-RE-NWDAF network element can use the second message to obtain the ML model.

[0193] For example, when the model transmission mode is direct embedding mode, the second message includes the model code of the ML model, the model acquisition method of the ML model obtained through direct embedding mode, and additional model information (such as input / output format requirements, model version, model description, etc.).

[0194] When the model transmission mode is the pre-signed URL mode, the second message includes the pre-signed URL corresponding to the ML model, the model acquisition method of the ML model is obtained through the pre-signed URL mode, and additional model information (such as input and output format requirements, model version, model description, etc.).

[0195] In some embodiments, before sending the second message corresponding to the model transmission mode to the H-RE-NWDAF network element, the status of the ML model is checked. If the ML model is in a normal state, the second message corresponding to the model transmission mode is sent to the H-RE-NWDAF network element. If the ML model is in an abnormal state, an error response code indicating the abnormal situation is sent to the H-RE-NWDAF network element.

[0196] For example, when the ML model's state becomes abnormal, a second message is sent to the H-RE-NWDAF network element. This second message includes the corresponding error response code. For example, the error response code may include MODEL_STATUS_INVALID.

[0197] In some embodiments, during the process of determining whether the verification passes, if the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the information of the second model is inconsistent with the information of the first model, a second message is sent to the H-RE-NWDAF network element. The second message includes a corresponding error response code. For example, the error response code may include INVALID or EXPIRED_TOKEN.

[0198] In some embodiments, H-RE-NWDAF network elements and V-RE-NWDAF network elements can interact using a request / response mode or a subscription / notification mode.

[0199] In some embodiments, the first message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a request to the model, and the second message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a response to the request to the model.

[0200] For example, a model provision request can be named Nnwdaf_RoamingML ModelProvision_Request, and a model provision request response can be named Nnwdaf_RoamingML ModelProvision_Request response.

[0201] In some embodiments, the first message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a subscription request for the model, and the second message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a notification for the model.

[0202] For example, a model can provide a subscription request as Nnwdaf_RoamingML ModelProvision_Subscribe, and a model can provide a notification as Nnwdaf_RoamingML ModelProvision_Notify.

[0203] In some embodiments, when the H-RE-NWDAF network element and the V-RE-NWDAF network element interact using a subscription / notification mode, if the ML model is updated, the V-RE-NWDAF network element sends a model provisioning notification to the H-RE-NWDAF network element to indicate the model update, wherein the model provisioning notification to indicate the model update includes the method for obtaining the updated ML model and additional model information.

[0204] For example, when the ML model is updated, the V-RE-NWDAF network element sends Nnwdaf_RoamingML ModelProvision_Notify to the H-RE-NWDAF network element. This notification includes the method for obtaining the updated ML model (e.g., the model file location), additional model information (e.g., updated input / output format requirements, model version, model description), etc.

[0205] In the ML model provision method provided in the above embodiments of this disclosure, the V-RE-NWDAF network element receives a model query request sent by the H-RE-NWDAF network element for querying an ML model. If an ML model exists locally, the V-RE-NWDAF network element sends a model query request response to the H-RE-NWDAF network element. Upon receiving a first message from the H-RE-NWDAF network element requesting the provision of an ML model, the V-RE-NWDAF network element determines the model transmission mode of the ML model and sends a second message corresponding to the model transmission mode to the H-RE-NWDAF network element, so that the H-RE-NWDAF network element can obtain the ML model using the second message. This enables the RE-NWDAF to have the capability to exchange ML models, realizing the sharing of ML models between the home network and the visited network.

[0206] Figure 9 This is a schematic diagram of the structure of a V-RE-NWDAF network element with MTLF function (hereinafter referred to as V-RE-NWDAF network element in the following description of this embodiment) according to another embodiment of the present disclosure.

[0207] like Figure 9 As shown, the V-RE-NWDAF network element 90 includes a memory 91, a processor 92, a bus 93, an input / output interface 94, a network interface 95, and a storage interface 96.

[0208] Figure 9 and Figure 4 The difference is that, in Figure 9 In the illustrated embodiment, processor 92 is configured to execute instructions stored in memory 91 as follows: Figure 8 The method involved in any of the embodiments.

[0209] This disclosure also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement... Figure 8 The method involved in any of the embodiments.

[0210] This disclosure also provides a computer program product, including computer instructions, wherein the computer instructions, when executed by a processor, implement as follows: Figure 8 The method involved in any of the embodiments.

[0211] Figure 10 This is a flowchart illustrating a method for providing an ML model according to another embodiment of the present disclosure. In some embodiments, the following ML model providing method is performed by an H-RE-NWDAF network element having an MTLF (hereinafter referred to as H-RE-NWDAF network element in the description of this embodiment), including steps 101-105.

[0212] In step 101, upon receiving a model retrieval request from an NF consumer in the HPLMN network for obtaining an ML model, it is determined whether an ML model exists locally.

[0213] In some embodiments, NF consumers in an HPLMN discover the H-RE-NWDAF within that HPLMN through a discovery mechanism. If the H-RE-NWDAF does not have MTLF functionality, it cannot provide services. If the H-RE-NWDAF does have MTLF functionality, it sends a model acquisition request to obtain the ML model. This model acquisition request includes information such as the ML model ID, ML model accuracy requirements, ML model format, and cross-PLMN request identifier.

[0214] In step 102, if the ML model does not exist locally and the cross-PLMN request identifier in the model acquisition request is true, a model query request for querying the ML model is sent to the V-RE-NWDAF network element with MTLF function (hereinafter referred to as V-RE-NWDAF network element in this embodiment).

[0215] For example, a model query request used to query an ML model includes Nnwdaf_RoamingML ModelInfo_Request. This model query request includes information such as the ML model ID, ML model accuracy requirements, ML model format, the PLMN to which it belongs, and the S-NSSAI to which it belongs.

[0216] It should be noted that if a local ML model exists, the H-RE-NWDAF network element will directly provide the ML model to the NF consumer.

[0217] In step 103, upon receiving the model query request response from the V-RE-NWDAF network element, a first message is sent to the V-RE-NWDAF network element to request the provision of an ML model.

[0218] In some embodiments, the model query request response includes model information of the ML model and model acquisition verification information. The first message includes model information and model acquisition verification information. Therefore, the V-RE-NWDAF network element can use the model information and model acquisition verification information included in the first message to verify the first message.

[0219] In step 104, the ML model is obtained according to the second message sent by the V-RE-NWDAF network element.

[0220] In some embodiments, an ML model is obtained according to the model acquisition method included in the second message.

[0221] In step 105, a model retrieval request response is sent to the NF consumer so that the NF consumer can retrieve the ML model using the model retrieval request response.

[0222] In some embodiments, the ML model is first subjected to an integrity check. If the integrity check passes, the model retrieval request response is sent to the NF consumer. This improves the user experience of the NF consumer when using the ML model.

[0223] In some embodiments, H-RE-NWDAF network elements and V-RE-NWDAF network elements can interact using a request / response mode or a subscription / notification mode.

[0224] In some embodiments, the first message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a request to the model, and the second message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a response to the request to the model.

[0225] For example, a model provision request can be named Nnwdaf_RoamingML ModelProvision_Request, and a model provision request response can be named Nnwdaf_RoamingML ModelProvision_Request response.

[0226] In some embodiments, the first message sent by the H-RE-NWDAF network element to the V-RE-NWDAF network element can provide a subscription request for the model, and the second message sent by the V-RE-NWDAF network element to the H-RE-NWDAF network element can provide a notification for the model.

[0227] For example, a model can provide a subscription request as Nnwdaf_RoamingML ModelProvision_Subscribe, and a model can provide a notification as Nnwdaf_RoamingML ModelProvision_Notify.

[0228] In the ML model provision method provided in the above embodiments of this disclosure, when the H-RE-NWDAF network element receives a model acquisition request from an NF consumer in the HPLMN network for obtaining an ML model, if the ML model does not exist locally and the cross-PLMN request identifier in the model acquisition request is true, it sends a model query request to the V-RE-NWDAF network element for querying the ML model. Upon receiving a model query request response from the V-RE-NWDAF network element, the H-RE-NWDAF network element sends a first message to the V-RE-NWDAF network element requesting the provision of an ML model. Based on a second message sent by the V-RE-NWDAF network element, it obtains the ML model and sends a model acquisition request response to the NF consumer, enabling the NF consumer to obtain the ML model using the model acquisition request response. This enables the RE-NWDAF to have the capability to exchange ML models, realizing the sharing of ML models between the home network and the visited network.

[0229] Figure 11 This is a schematic diagram of the structure of an H-RE-NWDAF network element with MTLF function (hereinafter referred to as H-RE-NWDAF network element in the following description of this embodiment) according to another embodiment of the present disclosure.

[0230] like Figure 11 As shown, the H-RE-NWDAF network element 110 includes a memory 111, a processor 112, a bus 113, an input / output interface 114, a network interface 115, and a storage interface 116.

[0231] Figure 11 and Figure 9 The difference is that, in Figure 11In the illustrated embodiment, processor 112 is configured to execute instructions stored in memory 111 as follows: Figure 10 The method involved in any of the embodiments.

[0232] This disclosure also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement... Figure 10 The method involved in any of the embodiments.

[0233] This disclosure also provides a computer program product, including computer instructions, wherein the computer instructions, when executed by a processor, implement as follows: Figure 10 The method involved in any of the embodiments.

[0234] Figure 12 This is a schematic diagram of the structure of a communication system according to another embodiment of this disclosure. Figure 12 As shown, the communication system includes an H-RE-NWDAF network element 121 with MTLF function and a V-RE-NWDAF network element 122 with MTLF function.

[0235] It should be noted that H-RE-NWDAF network element 121 is... Figure 10 The H-RE-NWDAF network element involved in any embodiment, V-RE-NWDAF network element 122 is Figure 8 The V-RE-NWDAF network element involved in any of the embodiments.

[0236] The following examples illustrate... Figure 12 The communication system shown is described below.

[0237] Figure 13 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure includes steps 131-1314.

[0238] In step 131, the NF consumer in the HPLMN discovers the H-RE-NWDAF in the HPLMN through the discovery mechanism. If the H-RE-NWDAF does not have MTLF functionality, it cannot provide services. If the H-RE-NWDAF has MTLF functionality, it sends an ML model acquisition request, which includes information such as ML model ID, ML model accuracy requirements, ML model format, and cross-PLMN request identifier.

[0239] In step 132, after receiving the ML model retrieval request, H-RE-NWDAF determines whether the required ML model exists locally.

[0240] In step 133, if the required ML model does not exist locally and the cross-PLMN request flag is true, H-RE-NWDAF sends a model query request (Nnwdaf_RoamingML ModelInfo_Request) for the ML model to V-RE-NWDAF that has MTLF functionality. This model query request carries information such as ML model ID, ML model accuracy requirements, ML model format, PLMN to which it belongs, and S-NSSAI to which it belongs.

[0241] In step 134, V-RE-NWDAF checks the roaming protocol between V-RE-NWDAF and H-RE-NWDAF to determine whether it can receive the model query request.

[0242] In step 135, if V-RE-NWDAF is capable of receiving the model query request, V-RE-NWDAF receives the model query request. V-RE-NWDAF parses the model query request and determines whether the required ML model exists locally. If the required ML model exists, V-RE-NWDAF sends a model query request response (Nnwdaf_RoamingMLModelInfo_Request response) to H-RE-NWDAF to inform H-RE-NWDAF about ML model information, the vendor information to which the model belongs, and the model's verification information.

[0243] In step 136, H-RE-NWDAF sends a model provisioning subscription request (Nnwdaf_RoamingMLModelProvision_Subscribe) to V-RE-NWDAF in order to request the provision of the ML model from V-RE-NWDAF, carrying information such as model acquisition and verification information, ML model ID, ML model accuracy requirements, ML model format, PLMN, and S-NSSAI.

[0244] In step 137, after receiving the model subscription request, V-RE-NWDAF verifies the request. For example, it extracts the model acquisition verification information, determines whether the information has been tampered with, whether the validity period has expired, and whether information such as the ML model ID and PLMN is consistent.

[0245] In step 1371, if the verification fails, V-RE-NWDAF performs exception handling. V-RE-NWDAF sends a model provisioning subscription request response (Nnwdaf_RoamingML ModelProvision_Subscriberesponse) to H-RE-NWDAF, which includes the corresponding error response code.

[0246] For example, if the verification information obtained by the model has been tampered with, the error response code included in the model's subscription request response will be INVALID. If the validity period of the verification information obtained by the model has expired, the error response code included in the model's subscription request response will be EXPIRED_TOKEN.

[0247] In this case, H-RE-NWDAF needs to re-execute the above process to resend the model delivery subscription request.

[0248] In step 138, if the verification is successful, V-RE-NWDAF determines the model transmission mode based on the model size, load conditions, etc.

[0249] For example, the model transmission mode can be either direct embedding mode or pre-signed URL mode.

[0250] In step 1381, V-RE-NWDAF sends a model provisioning notification (Nnwdaf_RoamingML ModelProvision_Notify) to H-RE-NWDAF corresponding to the model transmission mode.

[0251] For example, when the model delivery mode is direct embedding mode, V-RE-NWDAF sends the model provisioning notification to H-RE-NWDAF. The model provisioning notification includes the model code of the ML model, the model acquisition method for obtaining the ML model through direct embedding mode, and additional model information (e.g., input / output format requirements, model version, model description, etc.).

[0252] For example, when the model delivery mode is the pre-signed URL mode, V-RE-NWDAF sends the model provisioning notification to H-RE-NWDAF. The model provisioning notification includes the pre-signed URL corresponding to the ML model, the model acquisition method for obtaining the ML model through the pre-signed URL mode, and additional model information (e.g., input / output format requirements, model version, model description).

[0253] In step 1382, when V-RE-NWDAF is preparing to send the ML model, if the ML model is in an abnormal state, it will also send the model provisioning subscription request response (Nnwdaf_RoamingML ModelProvision_Subscribe response). The model provisioning subscription request response carries an error code identifier to explain the abnormal state of the model, such as MODEL_STATUS_INVALID.

[0254] In step 139, when the ML model subscribed to by the NF consumer in HPLMN in V-RE-NWDAF is updated, V-RE-NWDAF sends a model provisioning notification (Nnwdaf_RoamingMLModelProvision_Notify) to H-RE-NWDAF. This model provisioning notification carries the method for obtaining the ML model (such as the model file location), additional model information (such as the updated input / output format requirements, model version, model description, etc.).

[0255] In step 1310, H-RE-NWDAF obtains the ML model according to the model acquisition method included in the received model provisioning notification, and performs model integrity verification.

[0256] In step 1311, if the model integrity verification passes, H-RE-NWDAF sends an ML model retrieval request response to the NF consumer in HPLMN.

[0257] In step 1312, the NF consumer obtains the request response based on the ML model and obtains the ML model.

[0258] Figure 14 A flowchart illustrating a method for providing an ML model according to yet another embodiment of this disclosure includes steps 141-1414.

[0259] In step 141, the NF consumer in the HPLMN discovers the H-RE-NWDAF in the HPLMN through the discovery mechanism. If the H-RE-NWDAF does not have MTLF functionality, it cannot provide services. If the H-RE-NWDAF has MTLF functionality, it sends an ML model acquisition request, which includes information such as ML model ID, ML model accuracy requirements, ML model format, and cross-PLMN request identifier.

[0260] In step 142, after receiving the ML model retrieval request, H-RE-NWDAF determines whether the required ML model exists locally.

[0261] In step 143, if the required ML model does not exist locally and the cross-PLMN request flag is true, H-RE-NWDAF sends a model query request (Nnwdaf_RoamingML ModelInfo_Request) for the ML model to V-RE-NWDAF that has MTLF functionality. This model query request carries information such as ML model ID, ML model accuracy requirements, ML model format, PLMN to which it belongs, and S-NSSAI to which it belongs.

[0262] In step 144, V-RE-NWDAF checks the roaming protocol between V-RE-NWDAF and H-RE-NWDAF to determine whether it can receive the model query request.

[0263] In step 145, if the V-RE-NWDAF is capable of receiving the model query request, the V-RE-NWDAF receives the model query request. The V-RE-NWDAF parses the model query request and determines whether the required ML model exists locally. If the required ML model does not exist, the V-RE-NWDAF sends a model retrieval request to other NWDAFs (including MTLF functions) in this PLMN, carrying the relevant information from the aforementioned model query request.

[0264] In step 146, if other NWDAFs can provide the required ML model, the other NWDAFs send a model retrieval request response so that V-RE-NWDAFs can use the model retrieval request response to retrieve the required ML model.

[0265] In step 147, V-RE-NWDAF sends a model query request response (Nnwdaf_RoamingMLModelInfo_Request response) to H-RE-NWDAF to inform H-RE-NWDAF about ML model information, the vendor information to which the model belongs, and the model's verification information.

[0266] In step 148, H-RE-NWDAF sends a model provisioning subscription request (Nnwdaf_RoamingMLModelProvision_Subscribe) to V-RE-NWDAF in order to request the provision of the ML model from V-RE-NWDAF, carrying information such as model acquisition and verification information, ML model ID, ML model accuracy requirements, ML model format, PLMN to which it belongs, and S-NSSAI.

[0267] In step 149, after receiving the model subscription request, V-RE-NWDAF verifies the request. For example, it extracts the model acquisition verification information, determines whether the information has been tampered with, whether the validity period has expired, and whether information such as the ML model ID and PLMN is consistent.

[0268] In step 1491, if the verification fails, V-RE-NWDAF performs exception handling. V-RE-NWDAF sends a model provisioning subscription request response (Nnwdaf_RoamingML ModelProvision_Subscriberesponse) to H-RE-NWDAF, which includes the corresponding error response code.

[0269] For example, if the verification information obtained by the model has been tampered with, the error response code included in the model's subscription request response will be INVALID. If the validity period of the verification information obtained by the model has expired, the error response code included in the model's subscription request response will be EXPIRED_TOKEN.

[0270] In this case, H-RE-NWDAF needs to re-execute the above process to resend the model delivery subscription request.

[0271] In step 1410, if the verification is successful, V-RE-NWDAF determines the model transmission mode based on the model size, load conditions, etc.

[0272] For example, the model transmission mode can be either direct embedding mode or pre-signed URL mode.

[0273] In step 14101, V-RE-NWDAF sends a model provisioning notification (Nnwdaf_RoamingML ModelProvision_Notify) to H-RE-NWDAF corresponding to the model transmission mode.

[0274] For example, when the model delivery mode is direct embedding mode, V-RE-NWDAF sends the model provisioning notification to H-RE-NWDAF. The model provisioning notification includes the model code of the ML model, the model acquisition method for obtaining the ML model through direct embedding mode, and additional model information (e.g., input / output format requirements, model version, model description, etc.).

[0275] For example, when the model delivery mode is the pre-signed URL mode, V-RE-NWDAF sends the model provisioning notification to H-RE-NWDAF. The model provisioning notification includes the pre-signed URL corresponding to the ML model, the model acquisition method for obtaining the ML model through the pre-signed URL mode, and additional model information (e.g., input / output format requirements, model version, model description).

[0276] In step 14102, when V-RE-NWDAF is preparing to send the ML model, if the ML model is in an abnormal state, it will also send the model provisioning subscription request response (Nnwdaf_RoamingML ModelProvision_Subscribe response). The model provisioning subscription request response carries an error code identifier to explain the abnormal state of the model, such as MODEL_STATUS_INVALID.

[0277] In step 1411, when the ML model subscribed to by the NF consumer in HPLMN in V-RE-NWDAF is updated, V-RE-NWDAF sends a model provisioning notification (Nnwdaf_RoamingMLModelProvision_Notify) to H-RE-NWDAF. This model provisioning notification carries the method for obtaining the ML model (such as the model file location), additional model information (such as the updated input / output format requirements, model version, model description, etc.).

[0278] In step 1412, H-RE-NWDAF obtains the ML model according to the model acquisition method included in the received model provisioning notification, and performs model integrity verification.

[0279] In step 1413, if the model integrity check passes, H-RE-NWDAF sends an ML model retrieval request response to the NF consumer in HPLMN.

[0280] In step 1414, the NF consumer obtains the request response based on the ML model and retrieves the ML model.

[0281] By implementing the embodiments of this disclosure, bidirectional model exchange can be achieved between H-RE-NWDAF network elements and V-RE-NWDAF network elements. On the one hand, this facilitates the realization of cross-PLMN full-domain intelligence, allowing NWDAFs from different operators to share key models and improve service quality. On the other hand, it transforms ML models into operable assets, providing operators with new value growth points by offering their own trained high-quality models to other PLMNs. Thus, it provides a path to achieve 6G cross-PLMN network intelligence.

[0282] In some embodiments, the functional units described above may be implemented as general-purpose processors, programmable logic controllers (PLCs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or any suitable combination thereof for performing the functions described herein.

[0283] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0284] The description in this disclosure is provided for illustrative and descriptive purposes only and is not intended to be exhaustive or to limit the disclosure to its forms. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of this disclosure and to enable those skilled in the art to understand this disclosure and to design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A method for providing an ML model, executed by a network data analysis function H-RE-NWDAF network element with roaming switching capability in a home network having model training logic function MTLF, comprising: Receive model query requests for querying machine learning (ML) models from V-RE-NWDAF network elements with roaming switching capabilities in visited networks with MTLF functionality; If the ML model exists locally, a model query request response is sent to the V-RE-NWDAF network element; Upon receiving a first message from the V-RE-NWDAF network element requesting the provision of the ML model, the model transmission mode of the ML model is determined; A second message corresponding to the model transmission mode is sent to the V-RE-NWDAF network element so that the V-RE-NWDAF network element can use the second message to obtain the ML model.

2. The ML model providing method according to claim 1, wherein, The model transmission modes of the ML model include: direct embedding mode or pre-signed Uniform Resource Locator (URL) mode.

3. The ML model providing method according to claim 2, wherein, When the model transmission mode is the direct embedding mode, the second message includes the model encoding of the ML model, the model acquisition method of the ML model is obtained through the direct embedding mode, and additional model information; When the model transmission mode is the pre-signed URL mode, the second message includes a pre-signed URL corresponding to the ML model, and obtains the model acquisition method of the ML model and the model additional information through the pre-signed URL mode.

4. The ML model providing method according to claim 1, wherein, Sending the second message corresponding to the model transmission mode to the V-RE-NWDAF network element includes: Check whether the state of the ML model is normal; If the ML model is in a normal state, send a second message corresponding to the model transmission mode to the V-RE-NWDAF network element; If the state of the ML model is abnormal, an error response code to identify the abnormal situation is sent to the V-RE-NWDAF network element.

5. The ML model providing method according to claim 1, wherein, The determination of the model transfer mode of the ML model includes: Upon receiving the first message, the first message is verified; If the verification is successful, the model transmission mode of the ML model is determined based on the size of the ML model and the current load.

6. The ML model providing method according to claim 5, wherein, The model query request response includes the first model information of the ML model and the first model acquisition verification information. Verification of the first message includes: Extract the second model from the first message to obtain verification information and second model information; Based on the verification information obtained from the first model, determine whether the verification information obtained from the second model has been tampered with; If the verification information obtained by the second model has not been tampered with, determine whether the validity period of the verification information obtained by the second model has expired; If the validity period of the verification information obtained by the second model has not expired, determine whether the information of the second model and the information of the first model are consistent; If the second model information is consistent with the first model information, then the verification is deemed successful.

7. The ML model providing method according to claim 6, further comprising: If the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the information of the second model is inconsistent with the information of the first model, the corresponding error response code is sent to the V-RE-NWDAF network element.

8. The ML model providing method according to claim 1, wherein, The process of receiving a model query request for querying an ML model from a V-RE-NWDAF network element with MTLF functionality includes: According to the roaming protocol between the H-RE-NWDAF network element and the V-RE-NWDAF network element, the model query request sent by the V-RE-NWDAF network element is received.

9. The ML model providing method according to claim 1, wherein, The first message includes a model provision request, and the second message includes a model provision request response; or The first message includes a model providing a subscription request, and the second message includes a model providing a notification.

10. The method for providing an ML model according to any one of claims 1-9, further comprising: If the ML model does not exist locally, a model acquisition request is sent to other NWDAF network elements with MTLF function in the current PLMN to obtain the ML model; Based on the model acquisition request response sent by the other NWDAF network elements, obtain the ML model; Send the model query request response to the V-RE-NWDAF network element.

11. An H-RE-NWDAF network element with MTLF function, comprising: Memory; A processor, coupled to a memory, configured to implement the method as described in any one of claims 1-10 based on memory-stored instruction execution.

12. An ML model providing method, performed by a V-RE-NWDAF network element with MTLF functionality, comprising: Upon receiving a model retrieval request from an NF consumer in the VPLMN network for obtaining an ML model, determine whether the ML model exists locally; If the ML model does not exist locally and the cross-PLMN request identifier in the model acquisition request is true, a model query request for querying the ML model is sent to the H-RE-NWDAF network element with MTLF function. Upon receiving a model query request response from the H-RE-NWDAF network element, a first message is sent to the H-RE-NWDAF network element requesting the provision of the ML model. The ML model is obtained based on the second message sent by the H-RE-NWDAF network element; A model retrieval request response is sent to the NF consumer so that the NF consumer can retrieve the ML model using the model retrieval request response.

13. The ML model providing method according to claim 12, wherein, The process of obtaining the ML model includes: The ML model is obtained according to the model acquisition method included in the second message.

14. The ML model providing method according to claim 12, wherein, The step of sending a model retrieval request response to the NF consumer includes: Perform an integrity check on the ML model; If the integrity check passes, the model acquisition request response is sent to the NF consumer.

15. The ML model providing method according to claim 12, wherein, The model query request response includes the model information of the ML model and the model acquisition verification information; The first message includes the model information and the model acquisition verification information.

16. The ML model providing method according to any one of the embodiments of claims 12-15, wherein, The first message includes a model provision request, and the second message includes a model provision request response; or The first message includes a model providing a subscription request, and the second message includes a model providing a notification.

17. A V-RE-NWDAF network element with MTLF function, comprising: Memory; A processor, coupled to a memory, configured to implement the method as described in any one of claims 12-16 based on memory-stored instruction execution.

18. A communication system, comprising: The H-RE-NWDAF network element with MTLF function as described in claim 11; The V-RE-NWDAF network element with MTLF function as described in claim 17.

19. An ML model providing method, performed by a V-RE-NWDAF network element with MTLF functionality, comprising: Receive model query requests for querying ML models sent by H-RE-NWDAF network elements with MTLF functionality; If the ML model exists locally, a model query request response is sent to the H-RE-NWDAF network element; Upon receiving a first message from the H-RE-NWDAF network element requesting the provision of the ML model, the model transmission mode of the ML model is determined; A second message corresponding to the model transmission mode is sent to the H-RE-NWDAF network element so that the H-RE-NWDAF network element can use the second message to obtain the ML model.

20. The ML model providing method according to claim 19, wherein, The model transmission modes of the ML model include: direct embedding mode or pre-signed URL mode.

21. The ML model providing method according to claim 20, wherein, When the model transmission mode is the direct embedding mode, the second message includes the model encoding of the ML model, the model acquisition method of the ML model is obtained through the direct embedding mode, and additional model information; When the model transmission mode is the pre-signed URL mode, the second message includes a pre-signed URL corresponding to the ML model, and obtains the model acquisition method of the ML model and the model additional information through the pre-signed URL mode.

22. The ML model providing method according to claim 19, wherein, Sending the second message corresponding to the model transmission mode to the H-RE-NWDAF network element includes: Check whether the state of the ML model is normal; If the ML model is in a normal state, send a second message corresponding to the model transmission mode to the H-RE-NWDAF network element; If the state of the ML model is abnormal, an error response code to identify the abnormal situation is sent to the H-RE-NWDAF network element.

23. The ML model providing method according to claim 19, wherein, The determination of the model transfer mode of the ML model includes: Upon receiving the first message, the first message is verified; If the verification is successful, the model transmission mode of the ML model is determined based on the size of the ML model and the current load.

24. The ML model providing method according to claim 235, wherein, The model query request response includes the first model information of the ML model and the first model acquisition verification information. Verification of the first message includes: Extract the second model from the first message to obtain verification information and second model information; Based on the verification information obtained from the first model, determine whether the verification information obtained from the second model has been tampered with; If the verification information obtained by the second model has not been tampered with, determine whether the validity period of the verification information obtained by the second model has expired; If the validity period of the verification information obtained by the second model has not expired, determine whether the information of the second model and the information of the first model are consistent; If the second model information is consistent with the first model information, then the verification is deemed successful.

25. The ML model providing method according to claim 24, further comprising: If the verification information obtained by the second model is tampered with, the validity period of the verification information obtained by the second model expires, or the information of the second model is inconsistent with the information of the first model, a corresponding error response code is sent to the H-RE-NWDAF network element.

26. The ML model providing method according to claim 19, wherein, The process of receiving a model query request for querying an ML model from an H-RE-NWDAF network element with MTLF functionality includes: According to the roaming protocol between the H-RE-NWDAF network element and the V-RE-NWDAF network element, the model query request sent by the H-RE-NWDAF network element is received.

27. The ML model providing method according to claim 19, further comprising: If the ML model does not exist locally, a model acquisition request is sent to other NWDAF network elements with MTLF function in the current PLMN to obtain the ML model; Based on the model acquisition request response sent by the other NWDAF network elements, obtain the ML model; Send the model query request response to the H-RE-NWDAF network element.

28. The ML model providing method according to claim 19, wherein, The first message includes a model provision request, and the second message includes a model provision request response; or The first message includes a model providing a subscription request, and the second message includes a model providing a notification.

29. The method for providing an ML model according to any one of claims 19-28, further comprising: In the event of an ML model update, a model provisioning notification instructing the model update is sent to the H-RE-NWDAF network element, wherein the model provisioning notification instructing the model update includes the method for obtaining the updated ML model and additional model information.

30. A V-RE-NWDAF network element with MTLF function, comprising: Memory; A processor, coupled to memory, configured to implement the method as described in any one of claims 19-29 based on memory-stored instruction execution.

31. A method for providing an ML model, executed by an H-RE-NWDAF network element with MTLF functionality, comprising: Upon receiving a model retrieval request from an NF consumer in the HPLMN network for obtaining an ML model, determine whether the ML model exists locally; If the ML model does not exist locally and the cross-PLMN request identifier in the model acquisition request is true, a model query request for querying the ML model is sent to the V-RE-NWDAF network element with MTLF function. Upon receiving a model query request response from the V-RE-NWDAF network element, a first message is sent to the V-RE-NWDAF network element requesting the provision of the ML model. The ML model is obtained based on the second message sent by the V-RE-NWDAF network element; A model retrieval request response is sent to the NF consumer so that the NF consumer can retrieve the ML model using the model retrieval notification.

32. The ML model providing method according to claim 31, wherein, The process of obtaining the ML model includes: The ML model is obtained according to the model acquisition method included in the second message.

33. The ML model providing method according to claim 31, wherein, The step of sending a model retrieval request response to the NF consumer includes: Perform an integrity check on the ML model; If the integrity check passes, the model acquisition request response is sent to the NF consumer.

34. The ML model providing method according to claim 31, wherein The model query request response includes the model information of the ML model and the model acquisition verification information; The first message includes the model information and the model acquisition verification information.

35. The method for providing an ML model according to any one of claims 31-34, wherein, The first message includes a model provision request, and the second message includes a model provision request response; or The first message includes a model providing a subscription request, and the second message includes a model providing a notification.

36. An H-RE-NWDAF network element with MTLF function, comprising: Memory; A processor, coupled to a memory, configured to implement the method as described in any one of claims 31-35 based on memory-stored instruction execution.

37. A communication system, comprising: The V-RE-NWDAF network element with MTLF function as described in claim 30; The H-RE-NWDAF network element with MTLF function as described in claim 36.

38. A computer-readable storage medium, wherein, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method as described in any one of claims 1-10, 12-16, 19-29, and 31-35.

39. A computer program product comprising computer instructions, wherein the computer instructions, when executed by a processor, implement the method as described in any one of claims 1-10, 12-16, 19-29, and 31-35.

Images