Model access methods, network device and computer program product
By receiving and processing model subscription requests through a second network function, and only performing actual updates when model files are updated, the problems of low model access efficiency and waste of signaling resources are solved, achieving more efficient model access and resource utilization.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- ZTE CORP
- Filing Date
- 2025-11-25
- Publication Date
- 2026-07-30
AI Technical Summary
In existing technologies, model access is inefficient and signaling resources are wasted. In particular, when multiple model consumers subscribe to model updates, NWDAF needs to notify each consumer individually, and each consumer needs to re-acquire the model, resulting in wasted resources.
A model access method is provided, which receives model subscription requests from a first network function through a second network function, carries multiple model identifiers, and sends model notification messages. The method only updates the model files when they are updated, thus avoiding invalid resource downloads.
It improves model access efficiency, reduces the waste of signaling resources, and optimizes the utilization of network resources.
Smart Images
Figure CN2025137580_30072026_PF_FP_ABST
Abstract
Description
Model access methods, network devices and computer program products
[0001] Cross-references to related applications
[0002] This application is based on and claims priority to Chinese patent application CN202510113642.0, filed on January 23, 2025, entitled “Model Access Method, Network Device and Computer Program Product”, the entire contents of which are incorporated herein by reference. Technical Field
[0003] This disclosure relates to the field of communications, and more specifically, to a model access method, a network device, and a computer program product. Background Technology
[0004] The Network Data Analytics Function (NWDAF) trains the model, and the trained model is stored in the Analytics Data Repository Function (ADRF). The model consumer requests a model from the NWDAF (Model Training Logical Function, MTLF). The NWDAF (MTLF) can either retrieve the model from the ADRF and send it to the model consumer, or authorize the model consumer to retrieve the model directly from the ADRF by sending the ADRF(set)id to the model consumer, who then retrieves the model from the ADRF. If the NWDAF (MTLF) retrains the model (i.e., the model is updated), the NWDAF (MTLF) will similarly send the updated model to the model consumer or notify the model consumer to retrieve the updated model from the ADRF.
[0005] In scenarios where NWDAF (MTLF) authorizes model consumers to directly obtain models from ADRF, if multiple model consumers have subscribed to the model and these models undergo retraining, NWDAF (MTLF) needs to notify each model consumer individually, and then each model consumer must retrieve the updated model from ADRF. Furthermore, when NWDAF (MTLF) stores the trained model in ADRF, it carries the model's unique identifier, model access permissions, and the model file / address. If only the model address is carried, ADRF downloads the model file based on that address and stores it. However, if NWDAF (MTLF) has not retrained the model, but only the model access permissions have been updated, NWDAF (MTLF) will still send an update message to ADRF. This update message carries the model's unique identifier, model access permissions, and the model file / address. If only the model address is carried, even if the model file has not been re-updated, ADRF will still download the model file again and store it, resulting in a waste of signaling resources.
[0006] In summary, the relevant technologies suffer from low model access efficiency and wasted signaling resources. Summary of the Invention
[0007] This disclosure provides a model access method, a network device, and a computer program product.
[0008] According to one embodiment of this disclosure, a model access method is provided, comprising: a second network function receiving a first model subscription request from a first network function, the first model subscription request including a plurality of model identifiers; the second network function sending a model notification message to the first network function, the model notification message carrying corresponding model information.
[0009] According to another embodiment of this disclosure, a model access method is provided, comprising: a first network function sending a first model subscription request to a second network function, the first model subscription request including a plurality of model identifiers; the first network function receiving a model notification message from the second network function, the model notification message carrying corresponding model information.
[0010] According to another embodiment of this disclosure, a model access method is provided, comprising: a second network function receiving a third model update storage request from a third network function; and the second network function updating corresponding model information based on the third model update storage request.
[0011] According to yet another embodiment of this disclosure, a network device is also provided, the network device including a receiver, a transmitter, and a processor, the network device being configured to perform the steps of any of the above method embodiments via at least one of the receiver, the transmitter, and the processor.
[0012] According to yet another embodiment of this disclosure, a computer-readable storage medium is also provided, wherein a computer program is stored therein, wherein the computer program is configured to perform the steps in any of the above method embodiments when it is run.
[0013] According to yet another embodiment of this disclosure, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.
[0014] According to yet another embodiment of this disclosure, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps in any of the above method embodiments. Attached Figure Description
[0015] Figure 1 is a flowchart of a model access method according to an embodiment of this disclosure;
[0016] Figure 2 is a flowchart of another model access method according to an embodiment of this disclosure;
[0017] Figure 3 is a flowchart of yet another model access method according to an embodiment of this disclosure;
[0018] Figure 4 is a structural block diagram of a network device according to an embodiment of the present disclosure;
[0019] Figure 5 is a flowchart illustrating the model access method based on model update according to an embodiment of this disclosure;
[0020] Figure 6 is a flowchart illustrating the model access method based on model deletion according to an embodiment of this disclosure;
[0021] Figure 7 is a flowchart illustrating the model access method based on an immediate reporting instruction according to an embodiment of this disclosure;
[0022] Figure 8 is a schematic diagram of the process by which the MTLF instructs the ADRF to perform a model update according to an embodiment of this disclosure. Detailed Implementation
[0023] The embodiments of this disclosure will be described in detail below with reference to the accompanying drawings and examples.
[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of the embodiments of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.
[0025] In related technologies, the Network Data Analysis Function (NWDAF) provides analysis services for 5GC Network Functions (NF) and Operations, Administration, and Maintenance (OAM). The NWDAF includes the following logical functions: an Analytics logical function (AnLF) for performing inference and deriving analytical information (i.e., deriving statistical and / or predictive information based on analysis consumer requests); and a Model Training Logical Function (MTLF) for training machine learning (ML) models. An NWDAF network element can contain only MTLF or AnLF, or both. The Data Repository Function (ADRF) is used to store machine learning (ML) models.
[0026] This embodiment provides a model access method. Figure 1 is a flowchart of the model access method according to this embodiment. As shown in Figure 1, the process includes the following steps:
[0027] In step S102, the second network function receives a first model subscription request from the first network function, the first model subscription request including multiple model identifiers.
[0028] In this embodiment of the disclosure, the first network function can be a model consumer, the second network function can be an analytical data repository function (ADRF), and the model identifier can be a unique model identifier (ML model id).
[0029] In this embodiment of the disclosure, before the second network function receives the first model subscription request from the first network function, the third network function performs model training and stores the trained model in the second network function.
[0030] In this embodiment of the disclosure, before the second network function receives a first model subscription request from the first network function, the first network function subscribes to a certain model from the third network function, and the third network function authorizes the first network function to directly obtain the model from the second network function.
[0031] In this embodiment, the first model subscription request may further include a one-time reporting instruction, which is to prevent the first network function from simultaneously receiving model notification messages from both the second and third network functions. When the first model subscription request includes a one-time reporting instruction, the third network function will only report the initial model; subsequent model updates will not trigger reporting by the third network function. That is, the first network function will only receive model notification messages from the second network function.
[0032] In step S104, the second network function sends a model notification message to the first network function, and the model notification message carries the corresponding model information.
[0033] In one exemplary embodiment, before the second network function sends a model notification message to the first network function, the method further includes: the second network function receiving a first model update storage request from the third network function, and updating the model based on the first model update storage request, wherein the first model update storage request carries a model identifier of the updated model, the model file or model address of the updated model, and a model file update indication.
[0034] In an exemplary embodiment, in response to the second network function receiving a first model update storage request from the third network function, the corresponding model information carried in the model notification message sent by the second network function to the first network function includes: the model identifier of the updated model; and the storage address of the updated model in the second network function.
[0035] In one exemplary embodiment, the second network function receives a second model update storage request from the third network function and updates the model access permissions based on the second model update storage request. The second model update storage request carries the model identifier of the updated access permission model, the model file or model address of the updated access permission model, and the model access permissions of the updated access permission model.
[0036] In this embodiment, the third network function sends a model update storage request to the second network function in two situations: first, when the model file is updated after retraining, corresponding to the first model update storage request; and second, when the model access permissions are updated, corresponding to the second model update storage request. If only the model access permissions are updated, the second network function does not need to update the already stored model file, so even if the second model update storage request carries the model address, the second network function will not update the model file.
[0037] In one exemplary embodiment, before the second network function sends a model notification message to the first network function, the method further includes: the second network function receiving a model deletion request from the third network function and deleting the corresponding model file based on the model deletion request.
[0038] In an exemplary embodiment, in response to the second network function receiving a model deletion request from the third network function, the corresponding model information carried in the model notification message sent by the second network function to the first network function includes: the model identifier of the deleted model.
[0039] This embodiment also provides a model access method. Figure 2 is a flowchart of another model access method according to an embodiment of this disclosure. As shown in Figure 2, the process includes the following steps:
[0040] In step S202, the first network function sends a first model subscription request to the second network function, and the first model subscription request includes multiple model identifiers.
[0041] In one exemplary embodiment, before the first network function sends a first model subscription request to the second network function, the method further includes: the first network function receiving a first message from a third network function, the first message carrying a model identifier of the model to be accessed and a second network function identifier.
[0042] In this embodiment of the disclosure, the first message is used to notify or authorize the first network function to obtain the model file from the second network function.
[0043] In this embodiment, after the first network function obtains the first message from the third network function, it can obtain the model file corresponding to the model to be accessed from the second network function, or it can directly send a first model subscription request to the second network function to subscribe to model update events. The first model subscription request may carry an immediate reporting instruction; in this case, the second network function will immediately send the initial model to the first network function upon receiving the first model subscription request. Alternatively, the first model subscription request may not carry an immediate reporting instruction; in this case, the second network function does not need to immediately send the initial model to the first network function.
[0044] In step S204, the first network function receives a model notification message from the second network function, and the model notification message carries the corresponding model information.
[0045] In one exemplary embodiment, in response to a model update performed by a second network function, the first network function receives model information carried in a model notification message from the second network function, including: the model identifier of the updated model; and the storage address of the updated model in the second network function.
[0046] In one exemplary embodiment, in response to a second network function performing model deletion, the first network function receives model information carried in a model notification message from the second network function, including: the model identifier of the deleted model.
[0047] This embodiment also provides a model access method. Figure 3 is a flowchart of another model access method according to an embodiment of this disclosure. In this embodiment, the first network function is not involved, but only the interaction between the second network function and the third network function is included.
[0048] As shown in Figure 3, the process includes the following steps:
[0049] In step S302, the second network function receives a third model update storage request from the third network function.
[0050] Step S304: The second network function updates the corresponding model information based on the third model update storage request.
[0051] In an exemplary embodiment, in response to the third model update storage request carrying a model file update instruction, the second network function updates the corresponding model information based on the third model update storage request, including: the second network function updates and stores the corresponding model file according to the third model update storage request, wherein the third model update storage request also carries the model identifier of the updated model, the model file of the updated model, or the model address.
[0052] In an exemplary embodiment, in response to a third model update storage request carrying updated model access permissions but not a model file update indication, the second network function updates the corresponding model information based on the third model update storage request, including: the second network function updates the model access permissions of the corresponding model according to the third model update storage request, wherein the third model update storage request also carries a model identifier of the model with updated access permissions, and the model file or model address of the model with updated access permissions.
[0053] This disclosure provides a model access method, network device, and computer program product. The method includes: a second network function receiving a first model subscription request from a first network function, the first model subscription request including multiple model identifiers; and the second network function sending a model notification message to the first network function, the model notification message carrying corresponding model information. This disclosure solves the problems of low model access efficiency and wasted signaling resources in related technologies, thereby improving model access efficiency and signaling resource utilization.
[0054] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solutions of the embodiments of this disclosure, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the embodiments of this disclosure.
[0055] This embodiment also provides a model access device for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.
[0056] The model access device provided in this embodiment is configured in a second network function and includes a first receiving module and a first sending module. The first receiving module is used to receive a first model subscription request from the first network function, the first model subscription request including multiple model identifiers. The first sending module is used to send a model notification message to the first network function, the model notification message carrying corresponding model information.
[0057] The model access device provided in this embodiment is configured in a first network function and includes a second receiving module and a second sending module. The second sending module is used to send a first model subscription request to the second network function, the first model subscription request including multiple model identifiers. The second receiving module is used to receive model notification messages from the second network function, the model notification messages carrying corresponding model information.
[0058] The model access device provided in this embodiment is configured in a second network function and includes a third receiving module and an updating module. The third receiving module is used to receive a third model update storage request from the third network function. The updating module is used to update the corresponding model information based on the third model update storage request.
[0059] In this embodiment of the disclosure, the above-mentioned model access device may also include different modules, and the naming and functional division of the modules may be selected in different ways according to the actual situation, without specific limitations.
[0060] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.
[0061] This disclosure also provides a network device. FIG4 is a structural block diagram of the network device according to an embodiment of this disclosure. As shown in FIG4, the network device 400 includes a receiver 401, a transmitter 402 and a processor 403. The network device 400 is used to execute the steps of the above-described model access method embodiment through at least one of the receiver 401, transmitter 402 and processor 403.
[0062] This disclosure also provides a computer-readable storage medium storing a computer program configured to perform the steps in any of the above method embodiments when executed.
[0063] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.
[0064] This disclosure also provides an electronic device including a memory and a processor, the memory storing a computer program and the processor being configured to run the computer program to perform the steps in any of the above method embodiments.
[0065] In one exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.
[0066] This disclosure also provides a computer program product, including a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.
[0067] In one exemplary embodiment, the computer program product described above includes a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the methods described in various embodiments of this application.
[0068] Specific examples in this embodiment can be found in the examples described in the above embodiments and exemplary implementations, and will not be repeated here.
[0069] Obviously, those skilled in the art should understand that the modules or steps of the embodiments of this disclosure described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the embodiments of this disclosure are not limited to any particular combination of hardware and software.
[0070] To enable those skilled in the art to better understand the technical solutions of the embodiments of this disclosure, the following description is provided in conjunction with different embodiments.
[0071] Example 1
[0072] This embodiment provides a model update access method. The MTLF updates the model information on the ADRF and carries a model file update indication. The ADRF reports the model update event to the Model Consumer, and the reported message contains the updated model file. In this embodiment, the first network function is the Model Consumer, the second network function is the ADRF, and the third network function is the NWDAF, which includes the MTLF function.
[0073] Figure 5 is a flowchart illustrating the model access method based on model update according to an embodiment of this disclosure. As shown in Figure 5, the model update access method includes:
[0074] Step S500: MTLF performs model training and stores the trained model in ADRF.
[0075] Step S501: The MTLF sends a model storage request to the ADRF to store one or more ML models. The model storage request carries the model's unique identifier (ML model id), model access permissions (allowed NF instance ids), and the model file or model address.
[0076] Step S502: If ADRF receives a model address, ADRF downloads the model file based on that address and then stores it.
[0077] Step S503: ADRF sends a model storage response to MTLF, which carries the results of the model storage.
[0078] Step S504: The model Consumer subscribes to the ML model associated with a certain Analytics ID from the MTLF.
[0079] In this embodiment, the subscription request may include a one-time reporting instruction to instruct the MTLF to only report once.
[0080] In this embodiment, to avoid the model consumer receiving change notifications from both MTLF and ADRF entities simultaneously, the model consumer instructs to report only once in step S504. This way, MTLF will only report the initial model, and subsequent model updates will not trigger MTLF reporting. In other words, the model consumer will only receive model update reporting messages from ADRF.
[0081] Step S505: The MTLF authorizes the model consumer to directly obtain the ML model associated with the Analytics ID from the ADRF. The MTLF then sends the requested model information to the model consumer. The model information includes the ML model ID and the ADRF(set) ID.
[0082] In this embodiment, MTLF can send a first message to the model Consumer, wherein the first message is used to authorize the model Consumer to directly obtain the ML model associated with the Analytics ID from ADRF.
[0083] Step S506: The model consumer sends a model retrieval request to ADRF, requesting the model file corresponding to the ML model id.
[0084] Step S507: ADRF verifies whether the model consumer has model access permissions, i.e., whether the model consumer is included in the allowed NF instance ids or whether the model consumer is the MTLF where the model was originally stored. If the verification is successful, ADRF sends a model retrieval response to the model consumer, i.e., it sends the storage address of the model file corresponding to the ML model id (the address where ADRF stores the model file).
[0085] Step S508: After obtaining the model file from ADRF, the model Consumer immediately sends a model subscription request to ADRF. The model subscription request carries several ML model IDs.
[0086] In this embodiment, after the model Consumer obtains the model file from ADRF, the model subscription request sent to ADRF is the first model subscription request in the above embodiment.
[0087] Step S509: ADRF returns the model subscription response.
[0088] Step S510: MTLF retrained the model.
[0089] Step S511: MTLF sends a model storage request to ADRF to update one or more ML models that have already been stored. The request carries the model's unique identifier (ML model id), the model file / model address, and a model file update indication.
[0090] In this embodiment, the model storage request sent by MTLF in step S511 is the first model update storage request in the above embodiment.
[0091] In this embodiment, if the MTLF does not retrain the model, but the model's access permissions change, the MTLF will also trigger step S511, sending a second model update storage request to the ADRF. This request includes the model's unique identifier, model access permissions, and the model file / model address, but will not include a model file update indication. Since the ADRF does not receive a model file update indication, it will not trigger step S514.
[0092] Step S512: If ADRF receives a model address, ADRF downloads the model file based on that address and then stores it.
[0093] Step S513: ADRF sends a model storage response to MTLF, which carries the results of the model storage.
[0094] Step S514: After receiving a model storage request with a model file update indication, ADRF sends a model notification message to the model consumers that have subscribed to the model update event. The message carries the ML model id of the updated model and the storage address of the corresponding model file (the address where ADRF stores the model file).
[0095] Example 2
[0096] This embodiment provides a model deletion access method. MTLF deletes the model stored on ADRF, and ADRF sends a model update notification to the model Consumer, reporting the deleted model ID. In this embodiment, the first network function is the model Consumer, the second network function is ADRF, and the third network function is NWDAF, which includes the MTLF function.
[0097] Figure 6 is a flowchart illustrating the model access method based on model deletion according to an embodiment of this disclosure. As shown in Figure 6, the model deletion access method includes:
[0098] S600: MTLF performs model training and stores the trained model in ADRF.
[0099] S601: MTLF sends a model storage request to ADRF to store one or more ML models. The request carries the model's unique identifier (ML model id), model access permissions (allowed NF instance ids), and the model file or model address.
[0100] S602: If ADRF receives a model address, ADRF downloads the model file based on that address and then stores it.
[0101] S603: ADRF sends a model store response to MTLF, which carries the results of the model store.
[0102] S604: The model Consumer subscribes to the ML model associated with a certain Analytics ID from the MTLF.
[0103] S605: The MTLF-authorized model consumer directly retrieves the ML model associated with the Analytics ID from the ADRF and sends the requested model information to the model consumer. The model information includes the ML model ID and the ADRF(set)ID.
[0104] In this embodiment, MTLF can send a first message to the model Consumer, wherein the first message is used to authorize the model Consumer to directly obtain the ML model associated with the Analytics ID from ADRF.
[0105] S606: The model consumer sends a model retrieval request to ADRF, requesting the model file corresponding to the ML model id.
[0106] S607: ADRF verifies whether the model consumer has model access permissions, i.e., whether the model consumer is included in the allowed NF instance ids or whether the model consumer is the MTLF where the model was originally stored. If the verification is successful, ADRF sends a model retrieval response to the model consumer, which carries the storage address of the model file corresponding to the ML model id (the address where ADRF stores the model file).
[0107] S608: After obtaining the model file from ADRF, the model Consumer immediately sends a model subscription request to ADRF. The model subscription request carries several ML model IDs.
[0108] In this embodiment, after the model Consumer obtains the model file from ADRF, the model subscription request sent to ADRF is the first model subscription request in the above embodiment.
[0109] S609: ADRF returns a model subscription response.
[0110] S610: MTLF has deleted one or more models that were already stored in ADRF.
[0111] S611: ADRF returns a model deletion response, which contains the deletion result.
[0112] S612: ADRF sends a notification message to the Model Consumer that has subscribed to model events, reporting the ML model ID of the deleted model.
[0113] Example 3
[0114] This embodiment provides a model access method for a model subscription request that includes an immediate reporting instruction. In one embodiment, when the ADRF receives a model subscription request carrying an immediate reporting instruction, the ADRF reports the current model file in its response. In this embodiment, the first network function is the Model Consumer, the second network function is the ADRF, and the third network function is the NWDAF that includes the MTLF function.
[0115] Figure 7 is a flowchart illustrating the model access method based on an immediate reporting instruction according to an embodiment of this disclosure. As shown in Figure 7, the model access method including a model subscription request with an immediate reporting instruction includes:
[0116] S700: MTLF is used for model training, and the trained model is stored in ADRF.
[0117] S701: MTLF sends a model storage request to ADRF to store one or more ML models. The request carries the model's unique identifier (ML model id), model access permissions (allowed NF instance ids), and the model file or model address.
[0118] S702: If ADRF receives a model address, ADRF downloads the model file based on that address and then stores it.
[0119] S703: ADRF sends a model store response to MTLF, which carries the model store results.
[0120] S704: The model Consumer subscribes to the ML model associated with a certain Analytics ID from the MTLF.
[0121] S705: The MTLF-authorized model consumer directly obtains the ML model associated with the Analytics ID from the ADRF, and then sends the requested model information to the model consumer. The model information includes the ML model ID and the ADRF(set)ID.
[0122] In this embodiment, MTLF can send a first message to the model Consumer, wherein the first message is used to authorize the model Consumer to directly obtain the ML model associated with the Analytics ID from ADRF.
[0123] S706: The model consumer sends a model subscription request to the ADRF. The model subscription request carries the ML model ID and an immediate reporting instruction.
[0124] In this embodiment, in step S706, the model subscription request sent by the model Consumer to the ADRF is the first model subscription request in the above embodiment.
[0125] S707: ADRF returns a model subscription response. The model subscription response carries the address of the model file corresponding to the ML model ID (the address where ADRF stores the model file).
[0126] The steps for subsequent model updates can be referred to in Examples 1 and 2, and will not be repeated here.
[0127] Example 4
[0128] This embodiment provides a model update access method for ADRF and MTLF interaction. In one embodiment, ADRF receives a model update request from MTLF, which carries a model file update instruction and the model address. ADRF then re-downloads and stores the model. In this embodiment, the second network function is ADRF, the third network function is NWDAF which includes MTLF functionality, and the first network function is not involved.
[0129] Figure 8 is a schematic diagram of the process by which the MTLF instructs the ADRF to perform a model update according to an embodiment of this disclosure. As shown in Figure 8, the model update access method for the interaction between the ADRF and the MTLF includes:
[0130] S800: MTLF is used to train the model, and the trained model is stored in ADRF.
[0131] S801: MTLF sends a model storage request to ADRF to store one or more ML models. The request carries the model's unique identifier (ML model id), model access permissions (allowed NF instance ids), and the model file or model address.
[0132] S802: If ADRF receives a model address, ADRF downloads the model file based on that address and then stores it.
[0133] S803: ADRF sends a response message to MTLF, which carries the results stored in the model.
[0134] S804: MTLF retrained the model, or MTLF updated the model access permissions, or MTLF updated the model already stored in ADRF.
[0135] S805: MTLF sends a model storage request to ADRF. The request carries the model's unique identifier (ML model id), the model file / model address, and a model file update indication.
[0136] In this embodiment, the MTLF updates the model information on the ADRF side in two situations: first, the model file is updated after model retraining; second, the model access permissions are updated. If only the model access permissions are updated, the model storage request sent by the MTLF to the ADRF will not carry a model file update indication, and the ADRF does not need to update the already stored model file. Therefore, even if the MTLF carries the model address in the model storage request, the ADRF will not trigger step S806.
[0137] S806: If ADRF receives a model address, ADRF will download the model file based on that address and then store it.
[0138] S807: ADRF sends a model store response to MTLF, which carries the model store results.
[0139] The above description is merely a preferred embodiment of this disclosure and is not intended to limit the scope of this disclosure. Various modifications and variations can be made to this disclosure by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this disclosure should be included within the protection scope of this disclosure.
Claims
1. A model access method, comprising: The second network function receives a first model subscription request from the first network function, the first model subscription request including multiple model identifiers; The second network function sends a model notification message to the first network function, and the model notification message carries the corresponding model information.
2. The method according to claim 1, wherein, Before the second network function sends a model notification message to the first network function, the method further includes: The second network function receives a first model update storage request from the third network function and updates the model based on the first model update storage request. The first model update storage request carries a model identifier of the updated model, the model file or model address of the updated model, and a model file update indication.
3. The method according to claim 2, wherein, In response to the second network function receiving the first model update storage request from the third network function, the model notification message sent by the second network function to the first network function carries the corresponding model information including: The model identifier of the updated model; the storage address of the updated model in the second network function.
4. The method of claim 1, wherein, Also includes: The second network function receives a second model update storage request from the third network function and updates the model access permissions based on the second model update storage request. The second model update storage request carries the model identifier of the updated access permission model, the model file or model address of the updated access permission model, and the model access permissions of the updated access permission model.
5. The method according to claim 1, wherein, Before the second network function sends a model notification message to the first network function, the method further includes: The second network function receives a model deletion request from the third network function and deletes the corresponding model file based on the model deletion request.
6. The method of claim 5, wherein, In response to the second network function receiving the model deletion request from the third network function, the model notification message sent by the second network function to the first network function carries the corresponding model information, including: Delete the model identifier of the model.
7. A model access method, comprising: The first network function sends a first model subscription request to the second network function, the first model subscription request including multiple model identifiers; The first network function receives a model notification message from the second network function, the model notification message carrying corresponding model information.
8. The method according to claim 7, wherein, Before the first network function sends the first model subscription request to the second network function, the method further includes: The first network function receives a first message from the third network function, the first message carrying a model identifier of the model to be accessed and a second network function identifier.
9. The method according to claim 7, wherein, In response to a model update performed by the second network function, the first network function receives the model information carried in the model notification message from the second network function, including: The model identifier of the updated model; the storage address of the updated model in the second network function.
10. The method according to claim 7, wherein, In response to the second network function performing model deletion, the first network function receives the model information carried in the model notification message from the second network function, including: Delete the model identifier of the model.
11. A model access method, comprising: The second network function receives a third model update storage request from the third network function; The second network function updates the corresponding model information based on the third model update storage request.
12. The method according to claim 11, wherein, In response to the model file update indication carried in the third model update storage request, the second network function updates the corresponding model information based on the third model update storage request, including: The second network function updates and stores the corresponding model file according to the third model update storage request, wherein the third model update storage request also carries the model identifier of the updated model, and the model file or model address of the updated model.
13. The method according to claim 11, wherein, In response to the third model update storage request carrying updated model access permissions but not a model file update indication, the second network function updates the corresponding model information based on the third model update storage request, including: The second network function updates the model access permissions of the corresponding model according to the third model update storage request. The third model update storage request also carries the model identifier of the model with updated access permissions, and the model file or model address of the model with updated access permissions.
14. A network device comprising a receiver, a transmitter, and a processor, the network device being configured to perform the steps of the method of any one of claims 1 to 13 via at least one of the receiver, the transmitter, and the processor.
15. A computer-readable storage medium storing a computer program, wherein, When the computer program is executed by a processor, it implements the steps of the method described in any one of claims 1 to 13.
16. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, performs the steps of the method of any one of claims 1 to 13.
17. A computer program product comprising a computer program that, when executed by a processor, implements the steps of the method described in any one of claims 1 to 13.