Communication method and apparatus, and storage medium and program product

Abstract

Provided in the embodiments of the present application are a communication method and apparatus, and a storage medium and a program product. The method comprises: a core network device sending a first request, wherein the first request is used for requesting that an access network device trains a first model on an access network device side, or is used for requesting that the access network device reports training data used for training the first model; the core network device further receiving a first response, wherein the first response comprises training data for training a second model or a first model on a terminal device side, and the first response further comprises a first identifier, and the first identifier is used for identifying the first model; the core network device further sending the first identifier to an application device, wherein the first identifier is further used for identifying the second model, and the first model and the second model form a dual-end model; and associating the first model with the second model on the basis of the first identifier, such that during inference application, the access network device and / or the terminal device acquiring, on the basis of the first identifier, a corresponding model for inference. The method can be applied to the technical field of communications.

Description

Communication methods and devices, storage media and software products

[0001] This application claims priority to Chinese Patent Application No. 202411508518.6, filed with the China National Intellectual Property Administration on October 26, 2024, entitled "Communication Method and Apparatus, Storage Medium and Program Product", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of communication technology, and in particular to a communication method and apparatus, storage medium and program product. Background Technology

[0003] A dual-end model can be understood as dividing a complete artificial intelligence (AI) model into two parts and deploying them separately on the network side and the terminal device side. The models deployed on the network side and the terminal device side need to be trained and inferred jointly. Alternatively, it can be understood as deploying two AI models separately on the network side and the terminal device side, and these two models need to be trained and inferred jointly. For example, in a channel state information (CSI) compression application scenario, the model deployed on the terminal device side compresses the CSI, while the model deployed on the network side decompresses the CSI. The models deployed on the terminal device side and the network side work together to complete CSI feedback based on AI compression.

[0004] Multiple dual-end models may be deployed on both the network side and the terminal device side. During model inference, the network and terminal devices need to select a matching model from these multiple deployed dual-end models for inference. However, the network side or the terminal device side may be unable to identify a matching model from the multiple deployed dual-end models for inference, causing the dual-end model to fail to run. Summary of the Invention

[0005] This application discloses a communication method and apparatus, a storage medium and a program product, which can enable terminal devices and / or access network devices to obtain the required model for model inference during inference applications.

[0006] In a first aspect, embodiments of this application provide a communication method. This method can be applied to core network side equipment, such as core network elements or modules within core network elements (e.g., circuits, chips, or chip systems), or logical nodes, logical modules, or software capable of implementing all or part of the core network element functions. Taking the application of this method to core network equipment as an example, in this method, the core network equipment sends a first request to an access network equipment, the first request being used to request the access network equipment to train a first model; or, the first request being used to request the access network equipment to report training data for training the first model. The core network equipment also receives a first response from the access network equipment. The first response includes training data for training a second model or training data for training the first model, and the first response also includes a first identifier, the first identifier being used to identify the first model. The core network equipment also sends the aforementioned first identifier to an application device, the first identifier also being used to identify the second model. The application device is used to train the second model, the first model and the second model forming a dual-end model, the first model corresponding to the access network equipment, and the second model corresponding to the terminal device.

[0007] In this embodiment, the access network device assigns a first identifier, which is used to identify a first model and also to identify a second model. That is, the model identifier can be used to associate the dual-end models on the access network device and terminal device sides, realizing model association between the dual-end models. Thus, when the access network device or terminal device performs inference applications, it can obtain the model used by the inference application by acquiring the first identifier.

[0008] The first identifier generated by the access network device can identify both the first model on the access network device side and the second model on the terminal device side. Model association between the two models can be achieved based on the first identifier.

[0009] The first model corresponds to the access network device and can be understood as being used by the access network device. Optionally, the first model is deployed on the access network device. Correspondingly, the second model corresponds to the terminal device and can be understood as being used by the terminal device. Optionally, the second model is deployed on the terminal device.

[0010] In one possible implementation, the core network device sends a second request to the application device, the second request being used to request the application device to train a second model, the second request including training data for training the second model, and the second request also including a first identifier.

[0011] The application device is used to train the second model. By sending a first identifier to the application device, the application device knows that the second model corresponds to the first identifier.

[0012] In one possible implementation, the core network device stores association information between a first identifier, a first vendor identifier, and a second vendor identifier, wherein the first vendor identifier corresponds to the access network device and the second vendor identifier corresponds to the terminal device.

[0013] In this example, the core network device stores association information between a first identifier, a first vendor identifier, and a second vendor identifier. This allows the access network device or terminal device to determine the first identifier from the stored association information based on the first vendor identifier and the second vendor identifier during inference applications. Consequently, the access network device or terminal device can obtain the model used for the inference application based on the first identifier.

[0014] The first vendor could be, for example, a manufacturer of access network equipment or the owner of the access network equipment (such as an operator). The second vendor could be, for example, a manufacturer of terminal equipment or a chip manufacturer.

[0015] In another possible implementation, the core network device saves the aforementioned association information to a storage function network element. For example, the core network device sends this association information to the storage function network element, which then stores the association information.

[0016] In this way, core network devices can obtain this associated information based on storage function network elements.

[0017] In one possible implementation, the core network device also sends a candidate identifier to the access network device, which is used to determine a first identifier that is within the range of candidate identifiers.

[0018] Candidate identifiers are used to assist in generating model identifiers. Candidate identifiers can be a range of model identifiers or a list of model identifiers.

[0019] In this example, the core network device helps the access network device determine the model identifier by sending candidate identifiers to the access network device.

[0020] In one possible implementation, the core network device further receives a third request for an identifier of the model used to perform inference. This third request includes a first vendor identifier and a second vendor identifier, which are used to determine the model used to perform inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device. The core network device also sends a first identifier, whereby the first identifier identifies both the first and second models as the models used to perform inference.

[0021] In this example, when performing model inference, the access network device and the terminal device can obtain the model's identifier from the core network device by sending a first vendor identifier and a second vendor identifier, thereby determining the matching dual-end model to perform the inference task. Then, the matching model is obtained through the obtained model identifier for model inference.

[0022] In one possible implementation, the core network device further determines, based on the first vendor identifier and the second vendor identifier, the first model and the second model identified by the first identifier as the models for performing inference.

[0023] In one possible implementation, the core network device also receives a model training response or a data collection response, the model training response or data collection response including a first identifier, the model training response being used to send training data for training a second model, and the data collection response being used to send training data for training a first model.

[0024] This allows the core network devices to know the model corresponding to the received training data based on the first identifier.

[0025] In one possible implementation, the core network device also sends a first identifier and a model file of the first model to the access network device. This allows the access network device to directly obtain the model.

[0026] In another possible implementation, the core network device also sends a first identifier and a model address of the first model to the access network device. This allows the access network device to download the model (or the model file of the first model) based on the model address, so that it can use the model to perform inference when needed.

[0027] In another possible implementation, the core network device stores the trained first model in a network element with storage capabilities. The core network device sends a first identifier and Analytics Data Repository Function (ADRF) information to the access network device, allowing the access network device to retrieve the model from the ADRF based on the first identifier and ADRF information. Alternatively, the core network device sends a storage identifier and ADRF information to the access network device, allowing the access network device to retrieve the model from the ADRF based on the storage identifier and ADRF information.

[0028] In one possible implementation, the first response also includes model interoperability information. Model interoperability information is the information required for cross-vendor model deployment between devices from different vendors. The model requester (i.e., the model user) informs the model trainer of this model interoperability information so that the trainer can train a model that meets the requester's requirements, thus ensuring that the model can be deployed and run correctly during use. For example, this model interoperability information could be the model file format or / and the model execution environment. The model execution environment refers to information such as the operating system or / and device hardware and software configuration. Based on this model interoperability information, it can be ensured that access network devices can use models from core network devices correctly.

[0029] Secondly, embodiments of this application provide a communication method. This method can be applied to core network side equipment, such as core network elements or modules within core network elements (e.g., circuits, chips, or chip systems), or logical nodes, logical modules, or software capable of implementing all or part of the core network element functions. Taking the application of this method to core network equipment as an example, in this method, the core network equipment sends a first request to the access network equipment, the first request being used to request the access network equipment to train a first model; or, the first request being used to request the access network equipment to report training data for training the first model. The core network equipment also receives a first response from the access network equipment, the first response including training data for training a second model or training data for training the first model, the first response also including a first identifier used to identify the first model. The core network equipment further sends a second request to an application device, the second request being used to request the application device to train a second model, the second request including training data for training the second model, the second request also including a second identifier used to identify the second model, the first model and the second model forming a dual-end model, the first model corresponding to the access network equipment, and the second model corresponding to the terminal equipment.

[0030] In this embodiment, the access network device assigns a first identifier, and the core network device assigns a second identifier. The first identifier is used to identify a first model, and the second identifier is used to identify a second model. Thus, when performing inference applications, the access network device or the terminal device can obtain the model used by the inference application by acquiring the first or second identifier.

[0031] Thirdly, embodiments of this application provide a communication method. This method can be applied to core network side equipment, such as core network elements or modules within core network elements (e.g., circuits, chips, or chip systems), or logical nodes, logical modules, or software capable of implementing all or part of the core network element functions. Taking the application of this method to core network equipment as an example, in this method, the core network equipment sends a first request to the access network equipment, the first request being used to request the access network equipment to train a first model; or, the first request being used to request the access network equipment to report training data for training the first model. The core network equipment also receives a first response from the access network equipment, the first response including training data for training a second model or training data for training the first model, the first response also including a first identifier used to identify the first model. The core network equipment further sends a second request to the application equipment, the second request being used to request the application equipment to train a second model, the second request including training data for training the second model. The core network equipment also receives a second identifier from the application equipment, which is used to identify a second model. The first model and the second model together form a dual-end model. The first model corresponds to the access network equipment, and the second model corresponds to the terminal equipment.

[0032] In this embodiment, the access network device assigns a first identifier, and the application device assigns a second identifier. The first identifier is used to identify a first model, and the second identifier is used to identify a second model. Thus, when performing inference applications, the access network device or the terminal device can obtain the model used by the inference application by acquiring the first identifier or the second identifier.

[0033] In one possible implementation, the core network device receives a response message from the application device, which indicates the training result of the second model and includes the aforementioned second identifier.

[0034] Based on either the second or third aspect above, in one possible implementation, the core network device stores association information of a first identifier, a second identifier, a first vendor identifier, and a second vendor identifier, wherein the first vendor identifier corresponds to the access network device and the second vendor identifier corresponds to the terminal device.

[0035] Based on either the second or third aspect described above, in one possible implementation, the core network device further receives a third request for an identifier of the model for performing inference. This third request includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device. The core network device also sends either the first identifier, or the second identifier, or both the first and second identifiers, where the first and second models are the models for performing inference.

[0036] When performing model inference, access network devices and terminal devices can obtain model identifiers from core network devices by sending a first vendor identifier and a second vendor identifier, thereby determining the matching dual-end model to perform the inference task. Then, the matching model is obtained through the obtained model identifiers for model inference.

[0037] Fourthly, embodiments of this application provide a communication method. This method can be applied to core network side equipment, such as core network elements or modules within core network elements (e.g., circuits, chips, or chip systems), or logical nodes, logical modules, or software capable of implementing all or part of the core network element functions. Taking the application of this method to core network equipment as an example, in this method, the core network equipment sends a first request to the access network equipment, the first request being used to request the access network equipment to train a first model; or, the first request being used to request the access network equipment to report training data for training the first model, the first request including a first identifier used to identify the first model. The core network equipment also sends a second request to the application equipment, the second request being used to request the application equipment to train a second model, the second request including training data for training the second model, the first model and the second model forming a dual-end model, the first model corresponding to the access network equipment, and the second model corresponding to the terminal equipment.

[0038] In this embodiment of the application, the core network device assigns a first identifier, which is used to identify a first model. Thus, when performing inference applications, the access network device can obtain the model used by the inference application by acquiring this first identifier.

[0039] In one possible implementation, the second request includes a second identifier used to identify the second model.

[0040] In this example, the core network device also assigns a second identifier to identify the second model. Thus, when performing inference applications, the terminal device can obtain the model used by the inference application by acquiring this second identifier.

[0041] In one possible implementation, the first identifier and the second identifier are the same.

[0042] That is, the first identifier (or the second identifier) ​​is used to associate the first model and the second model.

[0043] In another possible implementation, the core network device also receives a second identifier from the application device, which is used to identify the second model.

[0044] In this example, the application device assigns a second identifier to identify the second model. Thus, when the terminal device is performing inference, it can obtain the model used by the inference application by acquiring this second identifier.

[0045] In one possible implementation, the core network device also receives a response message from the application device, which indicates the training result of the second model and includes a second identifier.

[0046] In one possible implementation, the core network device also stores association information of a first identifier, a second identifier, a first vendor identifier, and a second vendor identifier, wherein the first vendor identifier corresponds to the access network device and the second vendor identifier corresponds to the terminal device.

[0047] In one possible implementation, the core network device further receives a third request for an identifier of the model for performing inference. This third request includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device. The core network device also sends either the first identifier, or the second identifier, or both, where the first model and the second model are the models for performing inference.

[0048] When performing model inference, access network devices and terminal devices can obtain model identifiers from core network devices by sending a first vendor identifier and a second vendor identifier, thereby determining the matching dual-end model to perform the inference task. Then, the matching model is obtained through the obtained model identifiers for model inference.

[0049] In one possible implementation, the core network device also sends a model training request or a data collection request, which includes a first identifier. The model training request is used to instruct the access network device to train a first model, and the data collection request is used to obtain training data for training the first model.

[0050] In this way, the access network device can know which model to train or send training data corresponding to the first identifier based on the first identifier.

[0051] Fifthly, embodiments of this application provide a communication method. This method can be applied to network-side devices, such as network-side access network devices, modules (e.g., circuits, chips, or chip systems) within the access network device, or logical nodes, logical modules, or software capable of implementing all or part of the functions of the access network device. Taking the application of this method to an access network device as an example, in this method, the access network device receives a first request from a core network device. This first request is used to request the access network device to train a first model or report training data for training the first model. The access network device also sends a first response to the core network device. This first response includes training data for training a second model or training data for training the first model. The first response also includes a first identifier, which identifies the first model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to a terminal device.

[0052] In this embodiment, the access network device assigns a first identifier, which is used to identify a first model and also to identify a second model. In this example, the first model and the second model are associated based on the first identifier. Thus, when the access network device or terminal device performs inference applications, it can obtain the model used by the inference application by acquiring the first identifier.

[0053] In one possible implementation, the first identifier is also used to identify the second model.

[0054] That is, the access network device assigns a first identifier, which is used to identify the first model and also to identify the second model.

[0055] In one possible implementation, the access network device also generates a first identifier in response to the first request.

[0056] In one possible implementation, the first request includes a candidate identifier; the access network device further generates a first identifier based on the candidate identifier. The first identifier is within the range of candidate identifiers.

[0057] In this example, the access network device determines the first identifier based on the candidate identifier sent by the core network device. This ensures that the model identifier determined by different access network devices is unique to the core network device. If different access network devices determine the model identifier themselves, it is possible that different access network devices set the same model identifier, which would make it impossible to distinguish the models corresponding to different access network devices during inference applications.

[0058] In one possible implementation, the access network device further sends a second request for an identifier of the model for performing inference. This second request includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device. The access network device also receives a first identifier, whereby the first identifier identifies both the first and second models as the models for performing inference.

[0059] When performing model inference, access network devices and terminal devices can obtain model identifiers from core network devices by sending a first vendor identifier and a second vendor identifier, thereby determining the matching dual-end model to perform the inference task. Then, the matching model is obtained through the obtained model identifiers for model inference.

[0060] In another possible implementation, the access network device further sends a second request for an identifier of the model performing inference. This second request includes a first vendor identifier and a second vendor identifier, which are used to determine the model performing the inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device. The access network device also receives the first identifier and the second identifier, whereby the second identifier identifies the second model.

[0061] Optionally, the access network device stores the first identifier. Alternatively, the access network device stores the association between the first identifier and the first model. In this way, the access network device can determine the first model based on the first identifier.

[0062] In one possible implementation, the access network device further receives first information, which is used by the access network device to obtain a first model, the first information including a first identifier.

[0063] In one possible implementation, the first information also includes a first model.

[0064] In another possible implementation, the first information also includes a first address, which is used to obtain the first model.

[0065] Sixthly, embodiments of this application provide a communication method. This method can be applied to a terminal device, such as a terminal device or a communication / processing module within the terminal device, or a circuit or chip in the terminal device responsible for communication functions (e.g., a modem chip, also known as a baseband chip, or a system-on-a-chip (SoC) chip containing a modem core, or a system-in-package (SIP) chip), or a circuit or chip in the terminal device responsible for processing functions (e.g., a graphics processing unit (GPU)). Taking the application of this method to a terminal device as an example, in this method, the terminal device sends first information to the core network device. This first information is used to request the identifier of a model for performing inference. The first information includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device. The terminal device also receives second information from the core network device. The second information includes a first identifier, which is used to identify a second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0066] In this example, the terminal device sends a first vendor identifier and a second vendor identifier to the core network device, enabling the core network device to determine the identifier of the model required for inference. This allows the terminal device to obtain the necessary model for inference applications and perform model inference.

[0067] In one possible implementation, the first identifier is also used to identify the first model.

[0068] In another possible implementation, the second information also includes a second identifier used to identify the first model.

[0069] In this example, the terminal device receives a first identifier and a second identifier from the core network device. Then, the terminal device can send the second identifier to the access network device so that the access network device can obtain the model it needs to use to perform model inference.

[0070] In one possible implementation, the terminal device also receives a first identifier from an application device used to train a second model.

[0071] Optionally, the terminal device may also store a first identifier. Alternatively, the terminal device may store the association between the first identifier and the second model. In this way, the terminal device can determine the second model based on the first identifier.

[0072] Seventhly, embodiments of this application provide a communication method. This method can be applied to an application device, such as an application device or a module within an application device (e.g., a circuit, chip, or chip system), or a logical node, logical module, or software capable of implementing all or part of the functions of the application device. Taking the application device as an example, in this method, the application device receives first information, which includes training data for training a second model, and the first information instructs the application device to train the second model. The application device also sends second information, which includes a first identifier used to identify the second model, wherein the first model and the second model form a dual-end model, the first model corresponding to an access network device, and the second model corresponding to a terminal device.

[0073] In this embodiment, the application device sends a first identifier to the core network device or the terminal device, the first identifier being used to identify the second model. This allows the terminal device to obtain the necessary model for inference applications and perform model inference.

[0074] In one possible implementation, the application device also generates a first identifier.

[0075] Eighthly, this application provides a communication device that has the functions of the first aspect described above. For example, the communication device includes modules, units, or means that perform the operations involved in the first aspect. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0076] In one implementation, the communication device includes: a communication module for sending a first request to an access network device, the first request being for requesting the access network device to train a first model; or, the first request being for requesting the access network device to report training data for training the first model.

[0077] The communication module is also used to receive a first response from the access network device. The first response includes training data for training a second model or training data for training a first model, and the first response also includes a first identifier for identifying the first model.

[0078] The communication module is also used to send the aforementioned first identifier to the application device, which is also used to identify the second model. The application device is used to train the second model, and the first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0079] In one possible implementation, the communication device further includes a storage module for storing association information between a first identifier, a first vendor identifier, and a second vendor identifier, wherein the first vendor identifier corresponds to an access network device and the second vendor identifier corresponds to a terminal device.

[0080] For other implementations of this example, please refer to the first part of the document, which will not be repeated here.

[0081] Ninthly, this application provides a communication device that has the functions of the second aspect described above. For example, the communication device includes modules, units, or means that perform the operations involved in the second aspect described above. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0082] In one implementation, the communication device includes: a communication module for sending a first request to an access network device, the first request being for requesting the access network device to train a first model; or, the first request being for requesting the access network device to report training data for training the first model.

[0083] The communication module is also configured to receive a first response from an access network device, the first response including training data for training a second model or training data for training a first model, and the first response also including a first identifier for identifying the first model.

[0084] The communication module is also used to send a second request to the application device. The second request is used to request the application device to train a second model. The second request includes training data for training the second model. The second request also includes a second identifier for identifying the second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0085] For other implementations of this example, please refer to the second part of the document, which will not be repeated here.

[0086] In a tenth aspect, this application provides a communication device that has the functions of the second aspect described above. For example, the communication device includes modules, units, or means that perform the operations involved in the third aspect described above. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0087] In one implementation, the communication device includes: a communication module for sending a first request to an access network device, the first request being for requesting the access network device to train a first model; or, the first request being for requesting the access network device to report training data for training the first model.

[0088] The communication module is also configured to receive a first response from an access network device, the first response including training data for training a second model or training data for training a first model, and the first response also including a first identifier for identifying the first model.

[0089] The communication module is also used to send a second request to the application device, the second request being used to request the application device to train a second model, the second request including training data for training the second model.

[0090] The communication module is also used to receive a second identifier from the application device, which is used to identify a second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0091] For other implementations of this example, please refer to the documentation in the third section, which will not be elaborated here.

[0092] In the eleventh aspect, this application provides a communication device that has the functions of the third aspect above. For example, the communication device includes modules, units, or means that perform the operations involved in the fourth aspect above. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0093] In one implementation, the communication device includes: a communication module for sending a first request to an access network device, the first request being for requesting the access network device to train a first model; or, the first request being for requesting the access network device to report training data for training the first model, the first request including a first identifier for identifying the first model.

[0094] The communication module is also used to send a second request to the application device, the second request being used to request the application device to train a second model, the second request including training data for training the second model, the first model and the second model forming a dual-end model, the first model corresponding to the access network device, and the second model corresponding to the terminal device.

[0095] For other implementations of this example, please refer to the description in Section 4, which will not be repeated here.

[0096] In the twelfth aspect, this application provides a communication device that has the functions of the fourth aspect above. For example, the communication device includes modules, units, or means that perform the operations involved in the fifth aspect above. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0097] In one implementation, the communication device includes: a communication module for receiving a first request from a core network device, the first request being for requesting an access network device to train a first model or to report training data for training the first model.

[0098] The communication module is also used to send a first response to the core network device. The first response includes training data for training the second model or training data for training the first model. The first response also includes a first identifier for identifying the first model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0099] For other implementations of this example, please refer to the description in Section 5, which will not be repeated here.

[0100] In a thirteenth aspect, this application provides a communication device that has the functions of the fifth aspect above. For example, the communication device includes modules, units, or means that perform the operations involved in the sixth aspect above. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0101] In one implementation, the communication device includes: a communication module for sending first information to a core network device, the first information being used to request an identifier of a model for performing inference, the first information including a first vendor identifier and a second vendor identifier, the first vendor identifier and the second vendor identifier being used to determine the model for performing inference, the first vendor identifier corresponding to an access network device, and the second vendor identifier corresponding to a terminal device.

[0102] The communication module is also used to receive second information from the core network equipment. The second information includes a first identifier, which is used to identify a second model. The first model and the second model form a dual-end model. The first model corresponds to the access network equipment, and the second model corresponds to the terminal equipment.

[0103] For other implementations of this example, please refer to the documentation in Section VI, which will not be elaborated upon here.

[0104] In a fourteenth aspect, this application provides a communication device that has the functions of the sixth aspect above. For example, the communication device includes modules, units, or means that perform the operations involved in the seventh aspect above. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0105] In one implementation, the communication device includes: a communication module for receiving first information, the first information including training data for training a second model, the first information instructing an application device to train the second model.

[0106] The communication module is also used to send second information, which includes a first identifier for identifying a second model. The first model and the second model together form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0107] For other implementations of this example, please refer to the description in Section VII, which will not be repeated here.

[0108] In a fifteenth aspect, this application provides a communication device including an interface circuit and one or more processors. The one or more processors are coupled to a memory. The memory stores part or all of the necessary computer program or instructions for implementing the functions involved in any possible design or implementation of any of the first to seventh aspects described above. The one or more processors are executable to carry out the computer program or instructions, which, when executed, cause the communication device to implement the methods in any possible design or implementation of any of the first to seventh aspects described above. The interface circuit is used to implement communication functions within the communication device and / or communication functions between the communication device and other devices or components.

[0109] In one possible design, the processor is used to communicate with other devices or components through the interface circuit.

[0110] In one possible design, the communication device may also include the memory.

[0111] In a sixteenth aspect, this application provides a computer-readable storage medium storing a computer program that is executed by a processor to implement the method provided in any of the possible embodiments of the first to seventh aspects.

[0112] In a seventeenth aspect, this application provides a computer program product that, when run on a computer, causes the computer to perform a method as provided in any of the possible embodiments of the first to seventh aspects.

[0113] It is understood that the apparatus described in aspects eight through fifteen, the computer storage medium described in aspect sixteen, or the computer program product described in aspect seventeen are all used to perform the methods provided in any of aspects one through seven. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here. Attached Figure Description

[0114] The accompanying drawings used in the embodiments of this application are described below.

[0115] Figure 1 is a schematic diagram of a communication system provided in an embodiment of this application;

[0116] Figure 2 is a schematic diagram of an encoder and decoder provided in an embodiment of this application;

[0117] Figure 3a is a schematic diagram of a dual-end model training method provided in an embodiment of this application;

[0118] Figure 3b is a schematic diagram of another dual-end model training method provided in the embodiments of this application;

[0119] Figures 4-17 are schematic diagrams of the communication method provided in the embodiments of this application;

[0120] Figure 18 is a schematic diagram of the structure of a communication device provided in an embodiment of this application;

[0121] Figure 19 is a schematic diagram of the structure of another communication device provided in an embodiment of this application. Detailed Implementation

[0122] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0123] In this application, the terms "system" and "network" are used interchangeably. Unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship; for example, A / B can mean A or B. "And / or" in this application merely describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. Furthermore, in the description of this application, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be one or more. Furthermore, to facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish between network elements and similar items with essentially the same function. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that "first" and "second" are not necessarily different. In the embodiments of this application, A or / and B can be understood as one or more of A and B. In the embodiments of this application, A, B, and / or C can be understood as one or more of A, B, and C.

[0124] References to "one embodiment" or "some embodiments" in the embodiments described in this application mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0125] Furthermore, in the embodiments of this application, the words "exemplary," "for example," etc., are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" in this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the term "exemplary" is intended to present the concept in a concrete manner.

[0126] In the embodiments of this application, the terms "information," "signal," "message," "channel," and "singaling" may sometimes be used interchangeably. It should be noted that, without emphasizing their distinction, their intended meanings are consistent. Similarly, "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing their distinction, their intended meanings are consistent. Furthermore, the " / " mentioned in this application can be used to indicate an "or" relationship.

[0127] The following detailed embodiments further illustrate the objectives, technical solutions, and beneficial effects of this application. It should be understood that the following are merely specific embodiments of this application and are not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made based on the technical solutions of this application should be included within the scope of protection of this application.

[0128] In the various embodiments of this application, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions of different embodiments are consistent and can be referenced by each other. The technical features of different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0129] The technology provided in this application can be applied to various communication systems, such as fourth-generation (4G) communication systems (e.g., Long Term Evolution (LTE) systems), fifth-generation (5G) communication systems, wireless local area network (WLAN) systems, satellite communication systems, integrated systems of multiple systems, or future communication systems. Among these, 5G communication systems can also be referred to as new radio (NR) systems.

[0130] In a communication system, a network element can send signals to or receive signals from another network element. These signals can include information, signaling, or data. The term "network element" can also be replaced by an entity, network entity, device, communication equipment, communication module, node, communication node, etc. This application uses a network element as an example for description. For instance, a communication system may include at least one terminal and at least one access network device. The access network device can send downlink signals to the terminal, and / or the terminal can send uplink signals to the access network device. Furthermore, it is understood that if the communication system includes multiple terminals, these terminals can also exchange signals; that is, both the signal-sending network element and the signal-receiving network element can be a terminal.

[0131] Figure 1 is a schematic diagram of the architecture of a communication system 1000 used in an embodiment of this application. As shown in Figure 1, the communication system includes a wireless access network 100 and a core network 200. Optionally, the communication system 1000 may also include an Internet 300. The wireless access network 100 may include at least one network device (110a and 110b in Figure 1) and at least one terminal device (120a-120j in Figure 1). The terminal device is wirelessly connected to the network device, and the network device is wirelessly or wiredly connected to the core network. The core network device and the network device may be independent physical devices, or the functions of the core network device and the logical functions of the network device may be integrated on the same physical device, or a single physical device may integrate some of the functions of the core network device and some of the functions of the network device. Terminal devices and network devices can be interconnected via wired or wireless means. Figure 1 is only a schematic diagram; the communication system may also include other network devices, such as wireless relay devices and wireless backhaul devices, which are not shown in Figure 1.

[0132] Optionally, in practical applications, the wireless communication system may include multiple network devices (also known as access network devices) and multiple terminal devices simultaneously. A network device can serve one or more terminal devices simultaneously. A terminal device can also access one or more network devices simultaneously. This application embodiment does not limit the number of terminal devices and network devices included in the wireless communication system.

[0133] In this context, a network device can be an entity on the network side used to transmit or receive signals. A network device can also be an access device that allows terminal devices to wirelessly connect to the wireless communication system; for example, a network device can be a base station. Base stations can broadly encompass various names such as, or be interchangeable with, those listed below, including: radio access network (RAN) node, NodeB, evolved NodeB (eNB), next-generation base station (gNB), access network equipment in open radio access network (O-RAN), relay station, access point, transmitting and receiving point (TRP), transmitting point (TP), master eNB (MeNB), secondary eNB (SeNB), multi-standard radio (MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmission node, transceiver node, building baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), centralized unit (CU), and distributed unit (CU). Network devices include units (DU), radio units (RU), centralized unit control plane (CU-CP) nodes, centralized unit user plane (CU-UP) nodes, and positioning nodes. Base stations can be macro base stations, micro base stations, relay nodes, donor nodes, or similar entities, or combinations thereof. Network equipment can also refer to communication modules, modems, or chips installed within the aforementioned devices or apparatuses. Network equipment can also be mobile switching centers and devices that function as base stations in device-to-device (D2D), vehicle-to-everything (V2X), and machine-to-machine (M2M) communications, as well as devices that function as base stations in future communication systems. Network equipment can support networks using the same or different access technologies. The embodiments of this application do not limit the specific technologies or device forms used in the network equipment.

[0134] Network devices can be fixed or mobile. For example, base stations 110a and 110b are stationary and are responsible for wireless transmission and reception in one or more cells from terminal device 120. The helicopter or drone 120i shown in Figure 1 can be configured as a mobile base station, and one or more cells can move depending on the location of the mobile base station 120i. In other examples, the helicopter or drone (120i) can be configured as a terminal device communicating with base station 110b.

[0135] In this application, the communication device used to implement the above-mentioned network access functions can be an access network device, a network device with some access network functions, or a device capable of supporting the implementation of access network functions, such as a chip system, hardware circuit, software module, or hardware circuit plus software module. This device can be installed in the access network device or used in conjunction with the access network device. In the method of this application, the example of an access network device being used as the communication device to implement the access network device functions is described.

[0136] A terminal device can be a user-side entity used to receive or transmit signals, such as a mobile phone. Terminal devices can be used to connect people, things, and machines. Terminal devices can communicate with one or more core networks via network devices. Terminal devices include handheld devices with wireless connectivity, other processing devices connected to a wireless modem, or vehicle-mounted devices. Terminal devices can be portable, pocket-sized, handheld, computer-embedded, or vehicle-mounted mobile devices. Terminal devices can be widely used in various scenarios, such as cellular communication, D2D, V2X, point-to-point (P2P), machine-to-machine (M2M), machine-type communication (MTC), Internet of Things (IoT), virtual reality (VR), augmented reality (AR), industrial control, autonomous driving, telemedicine, smart grids, smart furniture, smart offices, smart wearables, smart transportation, smart cities, drones, robots, remote sensing, passive sensing, positioning, navigation and tracking, autonomous delivery and mobility, etc.Examples of terminal devices include: user equipment (UE) conforming to the 3rd Generation Partnership Project (3GPP) standard, fixed equipment, mobile equipment, handheld devices, wearable devices, cellular phones, smartphones, session initiated protocol (SIP) phones, laptops, personal computers, smart books, vehicles, satellites, global positioning system (GPS) devices, target tracking devices, drones, helicopters, aircraft, ships, remote control devices, smart home devices, industrial equipment, personal communication service (PCS) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), wireless network cameras, tablets, handheld computers, mobile internet devices (MIDs), wearable devices such as smartwatches, VR devices, AR devices, wireless terminals in industrial control, terminals in vehicle-to-everything (V2X) systems, wireless terminals in self-driving vehicles, wireless terminals in smart grids, wireless terminals in transportation safety, and smart city applications. Wireless terminals in various scenarios include smart gas pumps, high-speed rail terminals, and smart home terminals such as smart speakers, smart coffee machines, and smart printers. Terminal devices can be wireless devices in these scenarios or devices installed on wireless devices, such as communication modules, modems, or chips. Terminal devices can also be called terminals, user equipment (UE), mobile stations (MS), mobile terminals (MT), etc. Terminal devices can also be used in future wireless communication systems. Terminal devices can be used in dedicated network equipment or general-purpose equipment. The embodiments of this application do not limit the specific technologies or device forms used in the terminal devices.

[0137] Optionally, the terminal device can be used to act as a base station. For example, the UE can act as a scheduling entity, providing sidelink signaling between UEs in V2X, D2D, or point-to-point (P2P) scenarios. As shown in Figure 1, cellular phone 120a and car 120b communicate with each other using sidelink signaling. Cellular phone 120a communicates with smart home device 120e without relaying communication signals through base station 110b.

[0138] In this application, the communication device used to implement the functions of the terminal device can be a terminal device, a terminal device having some of the functions of the aforementioned terminal device, or a device capable of supporting the implementation of the functions of the aforementioned terminal device, such as a chip system. This device can be installed in the terminal device or used in conjunction with the terminal device. In this application, the chip system can be composed of chips or include chips and other discrete components. The technical solutions provided in this application are described using the example of a terminal device or UE as the communication device.

[0139] It should be understood that the number and type of each device in the communication system shown in Figure 1 are for illustrative purposes only, and this application is not limited thereto. In actual applications, the communication system may include more terminal devices, more access network devices, and other network elements, such as core network devices and / or network elements used to implement artificial intelligence functions.

[0140] It is understandable that all or part of the functions implemented by one or more of the terminal devices, access network devices, core network devices, or network elements used to implement artificial intelligence functions can be virtualized, that is, implemented through one or more of dedicated or general-purpose processors and corresponding software modules. Among these, the terminal devices and access network devices involve air interface transmission, and the transmit and receive functions of this interface can be implemented in hardware. Core network devices, such as operation administration and maintenance (OAM) network elements, can also be virtualized. Optionally, one or more of the functions of the virtualized terminal devices, access network devices, core network devices, or network elements used to implement artificial intelligence functions can be implemented by cloud devices, such as cloud devices in over-the-top (OTT) systems.

[0141] In this application, the phrase "sending information to... (e.g., a terminal)" or the related illustrations in the accompanying drawings can be understood as the destination of the information being the terminal. This can include sending information directly or indirectly to the terminal. Similarly, the phrase "receiving information from... (e.g., a terminal)" or "receiving information from... (e.g., a terminal)" or the related illustrations in the accompanying drawings can be understood as the source of the information being the terminal. This can include receiving information directly or indirectly from the terminal. Information may undergo necessary processing between the source and destination, such as format changes, but the destination can understand the valid information from the source. Similar expressions in this application can be interpreted similarly, and will not be elaborated further here.

[0142] The following describes some of the network elements and their functions involved in the embodiments of this application.

[0143] 1. Network Data Analytics Function (NWDAF)

[0144] NWDAF possesses data collection, training, analysis, and inference capabilities, and can be used to collect relevant data from network elements, third-party service servers, terminal devices, or network management systems (such as OAM) within the 3GPP network. NWDAF performs analysis and training based on this data and provides the analysis results to network elements, third-party service servers, terminal devices, or network management systems. These analysis results can assist the network in selecting service quality parameters, performing traffic routing, or selecting background data transmission strategies, among other things.

[0145] 2. Management Data Analytics Function (MDAF)

[0146] MDAF provides management data analysis services for one or more network elements, network slice instances, or network slice subnet instances. MDAF can collect events, status, and performance data related to the management of networks and services, and generate data analyses for managing networks and services, such as service experience analysis, network slice throughput analysis, and end-to-end latency analysis.

[0147] 3. Application Function (AF)

[0148] AF represents the interaction between the application and other control elements in the 3GPP network. This interaction includes providing service quality of service (QoS) policy requests, routing policy requests, or providing third-party services to the network side.

[0149] 4. Network Exposure Function (NEF)

[0150] The Network Provider Framework (NEF) exposes 3GPP network capabilities and events to external entities, and receives relevant external information. The NEF is primarily responsible for supporting secure interaction between 3GPP networks and third-party applications regarding network capabilities and events. For example, the NEF can securely expose network capabilities and events to third parties to enhance or improve application service quality. Alternatively, the NEF can ensure the secure acquisition of relevant data from third parties by the 3GPP network to enhance intelligent network decision-making. The NEF also supports recovering structured data from or storing structured data in a unified database.

[0151] 5. Unified Data Management (UDM)

[0152] UDM is primarily used for managing user subscription information, access authorization information, and registration / mobility context management. UDM is also used for generating authentication credentials, user identification processing (such as storing and managing permanent user identities), and access control.

[0153] 6. Unified Data Repository (UDR) function

[0154] UDR primarily provides storage capabilities for contract data, policy data, and capability-related data.

[0155] 7. Analysis and data storage function ADRF

[0156] ADRF provides the ability to store, retrieve, update, and delete AI data or models.

[0157] 8. Network Management System (NMS)

[0158] NMS is a management system responsible for the operation, management, and maintenance of the network. For example, NMS is a function / network element in OAM.

[0159] 9. Element Management System (EMS)

[0160] EMS manages one or more network elements of a specific category. For example, EMS manages access network devices. A network element management system can manage all characteristics of each network element individually.

[0161] To facilitate understanding of the solutions in the embodiments of this application, the terms that may be involved in the embodiments of this application are explained below.

[0162] (1) AI Model

[0163] An AI model is an algorithm or computer program that enables AI functionality. It represents the mapping relationship between the model's input and output. Types of AI models include neural networks, linear regression models, decision tree models, support vector machines (SVMs), Bayesian networks, Q-learning models, and other machine learning (ML) models.

[0164] It is understood that communication systems may include network elements with artificial intelligence capabilities. In one possible design, AI functions (such as AI modules or AI entities) can be configured within existing network elements in the communication system to implement AI-related operations (such as training and / or inference of AI models). For example, these existing network elements could be access network equipment (such as gNB), terminal equipment, core network equipment, or operation and maintenance management network elements. Alternatively, in another possible design, an independent network element can be introduced into the communication system to perform AI-related operations (such as training AI models). This independent network element can be called an AI network element or an AI node, and this disclosure does not limit the terminology. The AI ​​network element can be directly connected to the access network equipment in the communication system, or it can be indirectly connected to the access network equipment through a third-party network element. This third-party network element can be a core network element such as an authentication management function (AMF) network element or a user plane function (UPF) network element, an OAM (Operational Information Management) system, a cloud server, or other network elements; this application does not limit the specific network elements involved.

[0165] (2) Two-ended model

[0166] A two-sided model, also known as a bilateral model, collaborative model, or dual model, can consist of two sub-models (which are, for example, parts of a complete model). These two sub-models must be matched and jointly trained and inferred. Alternatively, a two-sided model can be composed of two combined models, again requiring matching and joint training and inference. The two sub-models or the two models themselves can be deployed on different nodes.

[0167] For example, the encoder for compressing CSI and the decoder for recovering compressed CSI involved in the embodiments of this application are a two-end model. The encoder for compressing CSI is used in conjunction with the decoder for compressing CSI, and it can be understood that the encoder and decoder are matched AI models.

[0168] (3) Training data

[0169] Training data (or training dataset) is used to train an AI model. Training data can include the input to the AI ​​model, or it can include both the input and the corresponding target output. A training dataset includes one or more training data sets, which can include training sample data input to the AI ​​model, or the target output of the AI ​​model. The target output can also be referred to as a label, sample label, or labeled sample.

[0170] In the field of communications, training data (or training datasets) can include simulation data collected through simulation platforms, experimental data collected from experimental scenarios, or measured data collected in actual communication networks. Because the geographical environment and channel conditions where the data is generated vary—for example, indoor / outdoor conditions, movement speed, frequency bands, or antenna configurations—the collected data can be categorized during acquisition. For instance, data with the same channel propagation environment and antenna configuration can be grouped together.

[0171] For example, for a two-end model, training data can be used to train the model on one side first, and then new training data can be generated based on the training data corresponding to that side model, which can be used to train the model on the other side; or the training data can be used to train both-end models simultaneously.

[0172] (4) Model file

[0173] A model file is a file corresponding to a runnable model. By deploying the model file to the corresponding hardware or software devices, the model corresponding to that model file can be run.

[0174] The model file includes model parameters, which may include one or more of the following: model structure parameters (e.g., number of layers, and / or weights), model input parameters (e.g., input dimensions, number of input ports), or model output parameters (e.g., output dimensions, number of output ports). The input dimension refers to the size of an input data set; for example, if the input data is a sequence, the corresponding input dimension indicates the length of the sequence. The number of input ports refers to the quantity of input data. Similarly, the output dimension refers to the size of an output data set; for example, if the output data is a sequence, the corresponding output dimension indicates the length of the sequence. The number of output ports refers to the quantity of output data.

[0175] (5) Channel Information

[0176] In communication systems, network devices need to determine the configuration of downlink data channels, modulation and coding strategies, and precoding for scheduling terminal devices based on channel information. Channel information can also be referred to as Channel State Information (CSI) or Channel Environment Information. Channel information is information that reflects channel characteristics or channel quality. In this application, the meaning of CSI is not limited to channel impulse response (CIR), power delay profile (PDP), channel quality indication (CQI), precoding matrix indicator (PMI), rank indicator (RI), or CSI-RS resource indicator (CRI). It can also be one or more of the following: channel response information (such as the channel response matrix), weight information corresponding to the channel response, reference signal receiving power (RSRP), or signal to interference plus noise ratio (SINR).

[0177] CSI measurement refers to the process by which the receiver measures channel information based on a reference signal transmitted by the transmitter; that is, it estimates the channel information using channel estimation methods. For example, the reference signal may include one or more of the following: channel state information reference signal (CSI-RS), synchronizing signal / physical broadcast channel block (SSB), sounding reference signal (SRS), or demodulation reference signal (DMRS). CSI-RS, SSB, and DMRS can be used to measure downlink CSI. SRS and DMRS can be used to measure uplink CSI.

[0178] In this process, the terminal device needs to feed back the aforementioned CSI to the network device. CSI feedback consumes significant air interface resources. To reduce CSI feedback overhead, AI technology is introduced into the wireless communication network, resulting in an AI model-based CSI feedback method. The terminal device uses the AI ​​model to compress the CSI and feeds back the compressed CSI to the network device, which then uses the AI ​​model to reconstruct the compressed CSI.

[0179] Taking Figure 2 as an example, the encoder in Figure 2 can perform CSI compression, and the decoder can perform CSI reconstruction (or decompression). For instance, the encoder can be deployed in the terminal device, and the decoder can be deployed in the network device. The terminal device can generate CSI feedback information z from the raw CSI information V through the encoder. The terminal device reports a CSI report, which may include the CSI feedback information z. The network device can reconstruct the CSI feedback information z through the decoder, thus obtaining the CSI recovery information V'.

[0180] (6) Training method of dual-end model

[0181] Referring to Figures 3a and 3b, two training methods for the dual-end model are shown. As shown in Figure 3a, the core network device trains the access network device-side model, and the application device trains the terminal device-side model. For example, the core network device collects training data and trains the access network device-side model. Then, the core network device sends the trained model (Model 1 shown in Figure 3a) to the access network device. The core network device also sends the training data to the application device, which trains the terminal device-side model. The application device then sends the trained model (Model 2 shown in Figure 3a) to the terminal device.

[0182] For example, Figure 3b shows another training method for the dual-end model. As shown in Figure 3b, the access network device trains the access network device-side model, and the application device trains the terminal device-side model. For instance, the access network device collects data to train the access network device-side model. The access network device also sends the training dataset to the application device through the core network device. The application device further trains the terminal device-side model and sends the trained model (Model 2 shown in Figure 3b) to the terminal device. It should be noted that while Figure 3b shows the access network device training the access network device-side model, alternatively, the EMS can train the access network device-side model, and then the EMS sends the trained model to the access network device.

[0183] The above describes the network architecture that can be implemented in the embodiments of this application. The method of the embodiments of this application will be described in detail below.

[0184] Example 1

[0185] Referring to Figure 4, which is a schematic diagram of a communication method provided in an embodiment of this application, this embodiment uses the method shown in Figure 3a for model training, that is, the core network device trains the access network device-side model. As shown in Figure 4, the method may include steps 401-410, as follows:

[0186] 401. The core network device sends Request 1 to the access network device. Request 1 is used to obtain the training data of the model on the access network device side. Accordingly, the access network device receives Request 1.

[0187] For example, core network equipment may be NWDAF, MDAF, or NMS.

[0188] For example, the access network device can be a gNB or EMS, etc.

[0189] Optionally, request 1 in step 401 may include one or more of the following information: candidate identifier, model type identifier, training method indication information, training data indication, training data target, and training data filtering information.

[0190] The following section introduces various types of information and their functions:

[0191] 1. Candidate Identifiers, used to assist in generating model identifiers. For example, candidate identifiers can provide a range for generating model identifiers. For instance, candidate identifiers can be a range of model IDs or a list of model IDs. For example, the access network device can determine a model identity (model ID) based on the candidate identifier, such as the first identifier. Optionally, the model identifier can also be called an association identity. Alternatively, the first identifier can also be an association identity. Association identities are used to associate sub-models of a two-end model or different parts of a two-end model. For example, if the range of model identifiers represents the range of model ID values ​​[0001, 1000], the access network device can generate, for example, 0123 as the model ID, where [0001, 1000] represents the set of values ​​from 0001 to 1000. For example, the list of model identifiers represents a set of possible values ​​for a model ID. For instance, if the list of model identifiers represents a set of possible values ​​for a model ID as {0123,0234,0512,0999}, then the access network device can select 0512 as the model ID from this set.

[0192] 2. Model Type ID: This identifier indicates the model to be trained. For example, it can indicate a model used for service experience analysis or / and a model used for network element load analysis. For instance, the model type ID can represent the purpose of the model. For example, it can indicate a model used for CSI compression. Alternatively, it can indicate the model deployment type. For example, it can indicate a dual-end model deployed on both the network side and the terminal device side. Or, it can identify both the function and deployment type. For example, it can indicate a dual-end model for CSI compression.

[0193] 3. Training method indication: This information indicates the training method of the model. For example, it may indicate the network element for training a specific model. For instance, it might indicate that the core network device trains the access network device-side model, or that the access network device trains the access network device-side model.

[0194] 4. Training data indication: This indicates the content of the training data to be collected. For example, the training data indication could be the target CSI or (and) CSI feedback information. The target CSI refers to the labels used in model training. The CSI feedback information refers to the model's training data. Optionally, the content of the training data can also be represented using event IDs. Event IDs are used to identify a type of event or a type of data. For example, different values ​​of the event ID correspond to different CSI information (info), which includes the aforementioned target CSI or (and) CSI feedback information.

[0195] 5. Target of training data: This parameter indicates the target training data, specifying which terminal devices' data should be used as training data. This parameter can be a terminal device identifier, indicating the acquisition of data from a single terminal device. Alternatively, it can be a list of terminal device identifiers or a group identifier, indicating the acquisition of data from a group of terminal devices. Or, it can be any terminal device, indicating the acquisition of data from any terminal device.

[0196] 6. Training data filter information indicates the range of training data to be acquired. For example, training data filter information includes region information (e.g., a list of tracking areas (TAs) / cells), indicating which regions' data to acquire. For instance, data from terminal devices within the regions represented by the tracking area list or cell list can be acquired as training data. Another example is slice information (e.g., single network slice selection assistance information (S-NSSAI), indicating which slice's data to acquire.

[0197] In one possible implementation, before step 401, the following steps are included: configuring the core network equipment with information such as a first vendor and a second vendor, as well as the address of the access network equipment and the address of the application device (e.g., server) corresponding to the terminal equipment. The first vendor corresponds to the access network equipment, and the second vendor corresponds to the terminal equipment. For example, the operator, the first vendor (e.g., A), and the second vendor (e.g., B) negotiate offline to determine that a dual-end model for training between the first and second vendors is needed. The operator can configure the information of the first and second vendors, as well as the access network equipment address and the terminal equipment server address, to the core network equipment (e.g., configure it on core network elements) through the network management system. It should be noted that the first vendor is a network equipment vendor, such as the manufacturer of the access network equipment or the owner of the access network equipment (e.g., the operator). It is understood that the first vendor can also be a core network equipment vendor. The second vendor can be, for example, the manufacturer of the terminal equipment or a chip manufacturer.

[0198] Optionally, after configuring the above information, the core network device can trigger dual-end model training, that is, trigger the core network device to send request 1 to the access network device.

[0199] 402. The access network device sends a response message 1 to the core network device.

[0200] For example, response message 1 is used to provide feedback on training data; for instance, response message 1 includes the training data. For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in response message 1. Accordingly, the core network device receives response message 1. Optionally, the core network device receives the first identifier.

[0201] In this example, the access network device assigns a first identifier, or in other words, the access network device generates a first identifier. This first identifier is used to associate a first model and a second model. Alternatively, the first identifier associates the first model and the second model. The first model corresponds to the access network device. For example, the first model is used by the access network device, or the first model is deployed on the access network device. This first model can also be called an access network device-side model. The first identifier is also used to identify the first model. That is, the first identifier corresponds to the first model.

[0202] Optionally, in response to request 1, the access network device generates a first identifier. For example, the access network device generates a first identifier after receiving request 1.

[0203] Optionally, the access network device generates a first identifier based on the candidate identifier received in step 401.

[0204] Optionally, response message 1 may also include model interoperability information. This model interoperability information is required for cross-vendor deployment of models between devices from different vendors. The model requester (i.e., the model user) informs the model trainer of this interoperability information so that the trainer can train a model that meets the requester's requirements, ensuring the model can be deployed and run correctly during use. For example, this interoperability information may include the model file format and / or the model execution environment. The model execution environment refers to information such as the operating system and / or device hardware and software configuration.

[0205] For example, the current training method involves the core network device training the first model and sending it to the access network device, meaning the model is ultimately used by the access network device. To ensure that the access network device can use the model correctly, the access network device instructs the core network device on model interoperability information. Based on this model interoperability information, it can be guaranteed that the access network device can use the model from the core network device correctly.

[0206] 403. The core network equipment trains the first model based on the training data.

[0207] 404a. The core network equipment sends a notification message 1 to the access network equipment.

[0208] For example, notification message 1 is used to send a first model that has been trained to an access network device. For example, notification message 1 includes a first identifier. Accordingly, the access network device receives notification message 1.

[0209] Optionally, the notification message 1 may also include one or more of the following information: model file, model address, model accuracy information, and model size information.

[0210] In one possible implementation, the core network device can directly send the model file of the first model to the access network device. This allows the access network device to directly obtain the model. Optionally, the access network device can store the model file locally and use it to perform inference when needed.

[0211] In another possible implementation, the core network device can send the model address of the first model to the access network device. The access network device can then download the model (or the model file of the first model) based on this model address, so that it can use the model to perform inference when needed. For example, the access network device can download the model immediately upon receiving the notification message and save it locally, or it can download the model only when model inference is required.

[0212] In another possible implementation, after training the first model, the core network device can also store the first model in a network element with storage capabilities, such as an ADRF (Advanced Transaction Registry). In this case, the core network device can send a first identifier and ADRF information to the access network device, allowing the access network device to retrieve the model from the ADRF based on the first identifier and ADRF information. Alternatively, the core network device can send a storage transaction ID and ADRF information to the access network device, allowing the access network device to retrieve the model from the ADRF based on the storage transaction ID and ADRF information. This storage identifier corresponds to the stored model information, and the ADRF can identify the previously stored model based on this identifier. The ADRF information may include an ADRF identifier and / or the ADRF address. For example, the access network device sends a request message to the ADRF based on the ADRF identifier or ADRF address, and this request message contains the storage identifier. The ADRF determines the stored model based on the storage identifier and sends the model to the access network device.

[0213] Core network devices can also provide access network devices with other relevant model information, such as model accuracy and model size. The access network devices then select a suitable model from multiple trained models for deployment based on model accuracy and / or model size.

[0214] Optionally, the access network device stores the first identifier. Alternatively, the access network device stores the association between the first identifier and the first model. In this way, the access network device can determine the first model based on the first identifier.

[0215] 404b. The core network equipment sends a notification message 2 to the application equipment.

[0216] For example, notification message 2 is used to instruct the application device to train the terminal device-side model (or second model). For example, notification message 2 includes training dataset information. For example, the core network device sends a first identifier to the application device. Accordingly, the application device receives notification message 2. Optionally, the application device receives the first identifier. Optionally, the aforementioned training dataset information and the first identifier can be carried in the same message.

[0217] As shown in Figure 3a, the core network device needs to send the training dataset to the application device so that the application device can train the terminal device-side model. Therefore, the core network device needs to determine which application device is associated with the training dataset, or in other words, which application device receives and uses the training dataset to train the terminal device-side model. Optionally, the core network device can determine which application device to send the notification message 2 to based on the configuration information in step 401. For example, the application device could be an application server (AF) or an application server (AS). When the application device is an AS, it can be understood that the AS is the server corresponding to the terminal device. For example, if the terminal device is from vendor A, then the AS refers to vendor A's server.

[0218] For example, the first identifier is also used to identify the second model. That is, the first identifier corresponds to the second model. The second model corresponds to the terminal device. For example, the second model is used by the terminal device. Optionally, the second model is deployed on the terminal device.

[0219] In other words, the first identifier is used to identify the first model and the second model.

[0220] Optionally, in the case of an untrusted AF (or an AF not deployed by the operator), the core network equipment can send the notification message to the AF through a network open function (e.g., NEF).

[0221] The training dataset information includes the training data used to train the first model, as well as intermediate data obtained during the training of the first model (such as training rounds and / or the model input / output data corresponding to each round of training). It is understood that this training dataset information can also be the same as the training data used to train the first model. This training dataset is used for training the second model. For example, the model deployed on the terminal device side (i.e., the second model) is the encoder of the dual-end model, and the model deployed on the access network device side (i.e., the first model) is the decoder. The output of the first model can be used as the input of the second model. Therefore, after the core network device trains the first model, it can send the corresponding model input / output information as the training dataset to the application device, which then uses this training dataset to complete the training of the second model.

[0222] Optionally, the notification message 2 may also include the following information: model interoperability information. This model interoperability information is used to align the input and / or output formats of the first and second models for model inference. For example, the first and second models may jointly perform model inference. For instance, the first model may need to use the output of the second model as input to perform model inference, and the first and second models may need to align their input and / or output formats, such as ensuring that the output dimensions of the second model match the input dimensions of the first model. Optionally, the model interoperability information sent by the core network device to the application device may be based on the model interoperability information sent by the access network device to the core network device in step 402.

[0223] 404c: Core network equipment stores the correspondence between the first identifier, the first vendor, and the second vendor.

[0224] Optionally, the correspondence can also be called associated information.

[0225] For an introduction to the first and second manufacturers, please refer to the description in step 401 above, which will not be repeated here.

[0226] For example, the core network device stores the association information of the first identifier, the first vendor identifier, and the second vendor identifier. For instance, the core network device stores the association information of <<first identifier, first vendor identifier, second vendor identifier>>.

[0227] For example, the core network device may store the association information locally. Alternatively, the core network device may store the association information in a storage function network element (e.g., UDM / UDR). For instance, the core network device may send the association information to a storage function network element, which will then store the association information.

[0228] 405. The application device trains the second model based on the received training dataset information.

[0229] 406. The application device sends a notification message to the terminal device.

[0230] For example, notification message 3 is used to send the trained second model to the terminal device. For example, the application device sends a first identifier to the terminal device. Optionally, the trained second model and the first identifier can be carried in the same message. For example, notification message 3 includes the first identifier. Accordingly, the terminal device receives notification message 3.

[0231] Optionally, the terminal device stores the first identifier. Alternatively, the terminal device stores the association between the first identifier and the second model. In this way, the terminal device can determine the second model based on the first identifier.

[0232] Optionally, the notification message 3 may also include one or more of the following information: model file, model address, model accuracy information, and model size information.

[0233] The application device can directly send model files to the terminal device, or it can send a model address to the terminal device, which the terminal device can then use to download the model. The terminal device can download the model immediately upon receiving a notification message and save it locally, or it can download the model only when it needs to perform model inference. The application device can also provide the terminal device with other relevant information about the model, such as model accuracy information and model size information. For a detailed description of this part, please refer to the description in step 404a above; it will not be repeated here.

[0234] 407. The access network device sends Request 2 to the core network device to request the identifier of the model to be used for model inference. Request 2 contains the first vendor identifier and the second vendor identifier.

[0235] Specifically, when model inference needs to be performed, the access network device determines the model to be used. For example, when a terminal device connects to the access network device and needs to send CSI information, the terminal device or the access network device negotiates and determines to use a CSI dual-end model to perform CSI information compression and decompression. Then, the access network device sends the aforementioned request 2 to the core network device.

[0236] 408. The core network device sends a first identifier to the access network device. Accordingly, the access network device receives the first identifier.

[0237] For example, the core network device determines the first identifier based on the stored correspondence between the first identifier and the first and second vendors. For instance, the core network device searches the stored association between the first vendor identifier, the second vendor identifier, and the first identifier to obtain the first identifier.

[0238] Optionally, the core network device queries the storage function network element (e.g., UDM / UDR) for the first identifier based on the first vendor identifier and the second vendor identifier. For example, the core network device sends the first vendor identifier and the second vendor identifier to the storage function network element (e.g., UDM / UDR), and the storage function network element determines the first identifier based on the association between the first vendor identifier, the second vendor identifier, and the first identifier. Then, the storage function network element sends the queried first identifier to the core network device.

[0239] 409. The terminal device sends Request 3 to the core network device to request the identifier of the model to be used for model inference. Request 3 contains the first vendor identifier and the second vendor identifier.

[0240] For example, the terminal device can send a request via AF / NEF or core network elements such as AMF to request the identifier of the model needed for inference. Optionally, the terminal device sends a request to AF / NEF / AMF to request the identifier of the model needed for model inference; this request includes a first vendor identifier and a second vendor identifier. Then, AF / NEF / AMF sends a request to the core network device to request the identifier of the model needed for model inference; this request also includes the first vendor identifier and the second vendor identifier. Subsequently, the core network device sends the first identifier to AF / NEF / AMF. AF / NEF / AMF then sends this first identifier to the terminal device.

[0241] For example, when a terminal device connects to an access network device and needs to send CSI information, the terminal device or the access network device negotiates and determines to use a CSI end-to-end model to perform CSI information compression and decompression. Then, the terminal device sends request 3 to the core network device.

[0242] 410. The core network equipment sends a first identifier to the terminal equipment. Accordingly, the terminal equipment receives the first identifier.

[0243] For example, the core network device determines the first identifier based on the stored correspondence between the first identifier and the first and second vendors. For example, the core network can determine the first identifier by referring to step 408.

[0244] Optionally, terminal devices and access network devices may also request the identifier of the model to be used for inference from core network devices in the following manner.

[0245] In one implementation, in step 408, after receiving the first identifier from the core network device, the access network device sends the first identifier to the terminal device. In this implementation, the terminal device does not need to request the identifier of the model required for inference from the core network device. That is, steps 409-410 are unnecessary.

[0246] In another implementation, the terminal device requests an identifier of the model needed for inference from the core network device. For example, in step 407, the terminal device sends request 2 to the core network device to request an identifier of the model needed for model inference, and request 2 includes a first vendor identifier and a second vendor identifier.

[0247] Correspondingly, in step 408, the terminal device receives a first identifier from the core network device. Then, the terminal device sends the first identifier to the access network device. In this implementation, the access network device does not need to request the identifier of the model required for inference from the core network device.

[0248] In this embodiment, the access network device assigns a dual-end model identifier (e.g., a first identifier), and the identifier of the model on the access network device side is the same as the identifier of the model on the terminal device side. This model identifier allows the dual-end models on the access network device and terminal device sides to be associated, achieving model association between the dual-end models. When performing model inference, the access network device and the terminal device can obtain the model identifier from the core network device by sending a first vendor identifier and a second vendor identifier, thereby determining the matching dual-end model to perform the inference task. Then, the matching model is obtained through the obtained model identifier for model inference.

[0249] Example 2

[0250] Figure 5 shows a schematic diagram of another communication method provided in an embodiment of this application. Unlike Embodiment 1, this embodiment uses the method shown in Figure 3b for model training, i.e., the access network device trains the access network device-side model. As shown in Figure 5, the method may include steps 501-510, as follows:

[0251] 501. The core network device sends Request 1 to the access network device, Request 1 instructing the core network device to request the access network device to perform training of the first model. Accordingly, the access network device receives Request 1.

[0252] Optionally, the request 1 may include one or more of the following information: candidate identifier, model type identifier, training method indication information, training data indication, training data target, and training data filtering information.

[0253] For details on this part, please refer to the description of step 401 in the embodiment shown in Figure 4, which will not be repeated here.

[0254] 502. The access network device sends response message 1 to the core network device, wherein response message 1 includes training dataset information.

[0255] For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in response message 1. Accordingly, the core network device receives response message 1. Optionally, the core network device receives the first identifier.

[0256] In this example, the access network device assigns a first identifier. This first identifier is used to associate a first model with a second model. Furthermore, this first identifier is also used to identify the first model.

[0257] The access network device collects training data to train the first model. The aforementioned training dataset information is used for training the second model.

[0258] 503. The core network equipment sends a notification message 1 to the application equipment.

[0259] For example, notification message 1 is used to instruct the application device to train a second model. For example, notification message 1 includes training dataset information. For example, the core network device sends a first identifier to the application device. Accordingly, the application device receives notification message 1. Optionally, the application device receives the first identifier. Optionally, the aforementioned training dataset information and the first identifier can be carried in the same message. For example, the first identifier can be carried in notification message 1.

[0260] In this example, the first model and the second model are associated through a first identifier. This first identifier is also used to identify the second model.

[0261] 504. Core network equipment stores the correspondence between the first identifier, the first manufacturer, and the second manufacturer.

[0262] For example, core network equipment stores the correspondence between the first identifier, the first vendor identifier, and the second vendor identifier.

[0263] For a description of steps 504-510, please refer to the description of steps 404c-410 in the embodiment shown in Figure 4 above, which will not be repeated here.

[0264] In this embodiment, the access network device assigns dual-end model identifiers, and the identifiers of the access network device-side model and the terminal device-side model are the same.

[0265] Example 3

[0266] Figure 6 is a schematic diagram of another communication method provided in an embodiment of this application. This embodiment uses the method shown in Figure 3a for model training, i.e., the core network device trains the access network device-side model. The difference from Embodiment 1 is that in this embodiment, the core network device allocates the model identifier to the terminal device side. As shown in Figure 6, the method may include steps 601-610, as follows:

[0267] 601. The core network device sends Request 1 to the access network device, Request 1 being used to obtain training data for the first model. Accordingly, the access network device receives Request 1.

[0268] For a description of this step, please refer to the description of step 401 in the embodiment shown in Figure 4, which will not be repeated here.

[0269] 602. The access network device sends a response message 1 to the core network device.

[0270] For example, the access network device sends a first identifier to the core network device. Optionally, the access network device stores the first identifier. Alternatively, the access network device stores the association between the first identifier and the first model. In this way, the access network device can determine the first model based on the first identifier.

[0271] For a description of this step, please refer to the description of step 402 in the embodiment shown in Figure 4, which will not be repeated here.

[0272] 603. The core network equipment trains the first model based on the training data.

[0273] 604a. The core network equipment sends a notification message 1 to the access network equipment.

[0274] For a description of this step, please refer to the description of step 404a in the embodiment shown in Figure 4, which will not be repeated here.

[0275] 604b. The core network equipment sends a notification message 2 to the application equipment.

[0276] For example, notification message 2 is used to instruct the application device to train a second model. For example, notification message 2 includes training dataset information. For example, the core network device sends a second identifier to the application device. Accordingly, the application device receives notification message 2. Optionally, the application device receives the second identifier. Optionally, the aforementioned training dataset information and the second identifier can be carried in the same message. For example, the second identifier can be carried in notification message 2.

[0277] In this example, the core network device assigns a second identifier, or in other words, the core network device generates a second identifier. This second identifier is used to associate the first model with the second model. That is, both the first and second identifiers are used to associate the first model with the second model. The second identifier is also used to identify the second model. For example, the second identifier and the first identifier are different. Of course, the second identifier and the first identifier can also be the same; this solution does not impose any restrictions on this.

[0278] Optionally, the core network device generates a second identifier in any step 604b and prior to it.

[0279] 604c. Core network equipment stores the first identifier, the second identifier, and the association information of the first vendor and the second vendor.

[0280] For example, the core network device stores the association information of the first identifier, the second identifier, the first vendor identifier, and the second vendor identifier. For instance, the core network device stores the correspondence between <<first identifier, first vendor identifier>> and <second identifier, second vendor identifier>>.

[0281] 605. The application device trains the second model based on the received training dataset information.

[0282] 606. The application device sends a notification message to the terminal device.

[0283] For example, notification message 3 is used to send the trained second model to the terminal device. For example, the application device sends a second identifier to the terminal device. Optionally, the trained second model and the second identifier can be carried in the same message. For example, notification message 3 includes the second identifier. Accordingly, the terminal device receives notification message 3.

[0284] Optionally, the terminal device stores a second identifier. Alternatively, the terminal device stores the association between the second identifier and the second model. In this way, the terminal device can determine the second model based on the second identifier.

[0285] For example, notification message 3 carries a second identifier and the model file of the trained second model. Alternatively, notification message 3 carries a second identifier and the model address of the trained second model.

[0286] 607. The access network device sends Request 2 to the core network device to request the identifier of the model to be used for model inference. Request 2 includes a first vendor identifier and a second vendor identifier.

[0287] For a description of this step, please refer to the description of step 407 in the embodiment shown in Figure 4, which will not be repeated here.

[0288] 608. The core network device sends a first identifier to the access network device. Accordingly, the access network device receives the first identifier.

[0289] For a description of this step, please refer to the description of step 408 in the embodiment shown in Figure 4, which will not be repeated here.

[0290] 609. The terminal device sends Request 3 to the core network device to request the identifier of the model to be used for model inference. Request 3 includes the first vendor identifier and the second vendor identifier.

[0291] 610. The core network equipment sends a second identifier to the terminal equipment. Accordingly, the terminal equipment receives the second identifier.

[0292] Optionally, terminal devices and access network devices may also request the identifier of the model to be used for inference from core network devices in the following manner.

[0293] In one implementation, after the access network device sends request 2 to the core network device in step 607, the core network device sends a first identifier and a second identifier to the access network device. Then, the access network device sends the second identifier to the terminal device. In this implementation, the terminal device does not need to request the identifier of the model required for inference from the core network device. That is, steps 609-610 are unnecessary.

[0294] In another implementation, the terminal device requests an identifier of the model needed for inference from the core network device. For example, in step 607, the terminal device sends request 2 to the core network device to request an identifier of the model needed for model inference, and request 2 includes a first vendor identifier and a second vendor identifier.

[0295] Correspondingly, in step 608, the terminal device receives a first identifier and a second identifier from the core network device. Then, the terminal device sends the first identifier to the access network device. In this implementation, the access network device does not need to request the identifier of the model required for inference from the core network device.

[0296] In this way, both the terminal device and the access network device know which model to use for model inference. After determining the model to be used, the terminal device and the access network device use the determined dual-end model to perform model inference.

[0297] In this embodiment, the access network device assigns the identifier of the access network device-side model, and the core network device assigns the identifier of the terminal device-side model.

[0298] Example 4

[0299] Figure 7 is a schematic diagram of another communication method provided in an embodiment of this application. Unlike Embodiment 3, this embodiment uses the method shown in Figure 3b for model training, i.e., the access network device trains the access network device-side model. As shown in Figure 7, the method may include steps 701-710, as follows:

[0300] 701. The core network device sends Request 1 to the access network device, which instructs the access network device to perform model training on its side. Accordingly, the access network device receives Request 1.

[0301] For details on this part, please refer to the description of step 501 in the embodiment shown in Figure 5, which will not be repeated here.

[0302] 702. The access network device sends response message 1 to the core network device, wherein response message 1 includes a first identifier and training dataset information. Accordingly, the core network device receives response message 1.

[0303] In this example, the access network device assigns a first identifier.

[0304] For details on this part, please refer to the description of step 502 in the embodiment shown in Figure 5, which will not be repeated here.

[0305] 703. Core network equipment sends notification message 1 to application equipment.

[0306] For example, notification message 1 is used to instruct the application device to train a second model. For example, notification message 1 includes training dataset information. For example, the core network device sends a second identifier to the application device. Accordingly, the application device receives notification message 1. Optionally, the application device receives the second identifier. Optionally, the aforementioned training dataset information and the second identifier can be carried in the same message. For example, the second identifier can be carried in notification message 1.

[0307] In this example, the core network device assigns a second identifier.

[0308] 704. Core network equipment stores the correspondence between the first identifier, the second identifier, the first vendor, and the second vendor.

[0309] For a description of steps 704-710, please refer to the description of steps 604c-610 in the embodiment shown in Figure 6 above, which will not be repeated here.

[0310] In this embodiment, the access network device assigns the identifier of the access network device-side model, and the core network device assigns the identifier of the terminal device-side model.

[0311] Example 5

[0312] Figure 8 shows a schematic diagram of another communication method provided in an embodiment of this application. This embodiment uses the method shown in Figure 3a for model training, that is, the core network device trains the access network device-side model. The difference from embodiment 3 is that in this embodiment, the application device assigns the identifier of the terminal device-side model. As shown in Figure 8, the method may include steps 801-810, as follows:

[0313] For a description of steps 801-804a, please refer to the description of steps 401-404a in Embodiment 1 shown in Figure 4, which will not be repeated here.

[0314] 804b. The core network equipment sends a notification message 2 to the application equipment.

[0315] For example, notification message 2 is used to instruct the application device to train a second model. For example, notification message 2 includes training dataset information. Accordingly, the application device receives notification message 2.

[0316] 805. The application device trains the second model based on the training dataset information.

[0317] 806a. The application device sends a second identifier to the core network device.

[0318] In this example, the application device assigns a second identifier, or in other words, the application device generates a second identifier. This second identifier is used to associate the first model with the second model. For example, the second identifier is different from the first identifier. Of course, the second identifier and the first identifier can also be the same; this solution does not impose any restrictions on this.

[0319] For an introduction to the generation of the second identifier by the application device, please refer to the description of the candidate identifier in steps 401 and 402 of the embodiment shown in Figure 4, which will not be repeated here.

[0320] 806b. The application device sends a notification message to the terminal device.

[0321] For example, notification message 3 is used to send the trained second model to the terminal device. Optionally, notification message 3 includes a second identifier. Accordingly, the terminal device receives notification message 3.

[0322] 806c, core network equipment stores the association information of the first identifier, the second identifier, the first vendor, and the second vendor.

[0323] For example, core network equipment stores the correspondence between a first identifier, a first vendor identifier, a second identifier, and a second vendor identifier. For instance, the correspondence between the first identifier, the first vendor identifier, the second identifier, and the second vendor identifier can be stored as <<First Identifier, First Vendor Identifier>, <Second Identifier, Second Vendor Identifier>>.

[0324] For a description of steps 807-810, please refer to the description of steps 607-610 in the embodiment shown in Figure 6, which will not be repeated here.

[0325] In this embodiment, the access network device assigns the identifier of the access network device-side model, and the application device assigns the identifier of the terminal device-side model.

[0326] Example 6

[0327] Figure 9 shows a schematic diagram of another communication method provided in an embodiment of this application. Unlike embodiment 5, this embodiment uses the method shown in Figure 3b for model training, i.e., the access network device trains the access network device-side model. As shown in Figure 9, the method may include steps 901-909, as follows:

[0328] 901. The core network device sends Request 1 to the access network device, which instructs the core network device to request the access network device to perform model training on the access network device side. Accordingly, the access network device receives Request 1.

[0329] 902. The access network device sends a model training response message 1 to the core network device, wherein the response message 1 includes training dataset information.

[0330] For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in the model training response message. Accordingly, the core network device receives the response message 1. Optionally, the core network device receives the first identifier.

[0331] In this example, the access network device assigns a first identifier.

[0332] For a description of steps 903-909, please refer to the description of steps 804b-810 in the embodiment shown in Figure 8, which will not be repeated here.

[0333] In this embodiment, the access network device assigns the identifier of the access network device-side model, and the application device assigns the identifier of the terminal device-side model.

[0334] Example 7

[0335] Figure 10 is a schematic diagram of another communication method provided in an embodiment of this application. This embodiment uses the method shown in Figure 3a for model training, that is, the core network device trains the access network device-side model. The difference from Embodiment 1 is that in this embodiment, the core network device allocates the dual-end model identifiers. As shown in Figure 10, the method may include steps 1001-1009, as follows:

[0336] 1001. The core network device sends Request 1 to the access network device. Request 1 is used to obtain training data for the first model, wherein Request 1 includes a first identifier. Accordingly, the access network device receives Request 1.

[0337] In this example, the core network device assigns a first identifier, or in other words, the core network device generates a first identifier. This first identifier is used to associate the first model with the second model. For a description of this step, please refer to the description of generating the first identifier based on the candidate identifier in steps 401 and 402 of the embodiment shown in Figure 4, which will not be repeated here.

[0338] 1002. The access network device sends a response message 1 to the core network device.

[0339] For example, response message 1 is used to provide feedback on training data. For instance, response message 1 includes the training data. For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in response message 1.

[0340] For a description of steps 1003-1004a, please refer to the description of steps 403 and 404a in the embodiment shown in Figure 4, which will not be repeated here.

[0341] 1004b. The core network equipment sends a notification message 2 to the application equipment.

[0342] For example, notification message 2 is used to instruct the application device to train a second model. For example, notification message 2 includes training dataset information. For example, the core network device sends a first identifier to the application device. Accordingly, the application device receives notification message 2. Optionally, the application device receives the first identifier. Optionally, the aforementioned training dataset information and the first identifier can be carried in the same message. For example, the first identifier can be carried in notification message 2.

[0343] In this example, the core network device associates the first model with the second model using the same identifier.

[0344] 1004c. Core network equipment stores the correspondence between the first identifier, the first vendor, and the second vendor.

[0345] For a description of steps 1005-1010, please refer to the description of steps 405-410 in the embodiment shown in Figure 4, which will not be repeated here.

[0346] In this example, dual-end model identifiers are assigned by the core network equipment, and the identifiers of the access network equipment-side model and the terminal equipment-side model are the same.

[0347] Example 8

[0348] Figure 11 is a schematic diagram of another communication method provided in an embodiment of this application. Unlike embodiment 7, this embodiment uses the method shown in Figure 3b for model training, i.e., the access network device trains the access network device-side model. As shown in Figure 11, the method may include steps 1101-1110, as follows:

[0349] 1101. The core network device sends Request 1 to the access network device, Request 1 instructing the core network device to request the access network device to perform model training on the access network device side, wherein Request 1 includes a first identifier. Accordingly, the access network device receives Request 1.

[0350] In this example, the core network device assigns a first identifier, or in other words, the core network device generates a first identifier. This first identifier is used to associate the first model with the second model.

[0351] 1102. The access network device sends a response message 1 to the core network device, wherein the response message 1 includes training dataset information.

[0352] For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in response message 1. Accordingly, the core network device receives response message 1. Optionally, the core network device receives the first identifier.

[0353] For a description of steps 1103a-1109, please refer to the description of steps 1004b-1010 in the embodiment shown in Figure 10, which will not be repeated here.

[0354] In this example, dual-end model identifiers are assigned by the core network equipment, and the identifiers of the access network equipment-side model and the terminal equipment-side model are the same.

[0355] Example 9

[0356] Figure 12 shows a schematic diagram of another communication method provided in an embodiment of this application. This embodiment uses the method shown in Figure 3a for model training, that is, the core network device trains the access network device side model. The difference from Embodiment 7 is that the dual-end model identifiers assigned by the core network device are different in this embodiment. As shown in Figure 12, the method may include steps 1201-1210, as follows:

[0357] 1201. The core network device sends Request 1 to the access network device. Request 1 is used to obtain training data for the first model, wherein Request 1 includes a first identifier. Accordingly, the access network device receives Request 1.

[0358] In this example, the core network device assigns the first identifier.

[0359] 1202. The access network device sends a response message 1 to the core network device.

[0360] For example, response message 1 is used to provide feedback on training data. For instance, response message 1 includes the training data. For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in response message 1.

[0361] For a description of steps 1203-1204a, please refer to the description of steps 403 and 404a in the embodiment shown in Figure 4, which will not be repeated here.

[0362] 1204b. The core network equipment sends a notification message 2 to the application equipment.

[0363] For example, notification message 2 is used to instruct the application device to train a second model. For example, notification message 2 includes training dataset information. For example, the core network device sends a second identifier to the application device. Accordingly, the application device receives notification message 2. Optionally, the application device receives the second identifier. Optionally, the aforementioned training dataset information and the second identifier can be carried in the same message. For example, the second identifier can be carried in notification message 2.

[0364] In this example, the core network devices associate the first model with the second model using different identifiers.

[0365] 1204c: The core network equipment stores the correspondence between the first identifier, the second identifier, the first vendor, and the second vendor.

[0366] 1205. The application device trains the second model based on the received training data.

[0367] 1206. The application device sends a notification message to the terminal device.

[0368] For example, notification message 3 is used to send the trained second model to the terminal device. Optionally, notification message 3 includes a second identifier. Accordingly, the terminal device receives notification message 3.

[0369] For a description of steps 1207-1210, please refer to the description of steps 1007-1010 in the embodiment shown in Figure 10, which will not be repeated here.

[0370] In this example, dual-end model identifiers are assigned by the core network equipment, and the identifiers of the model on the access network equipment side and the model on the terminal equipment side are different.

[0371] Example 10

[0372] Figure 13 is a schematic diagram of another communication method provided in an embodiment of this application. Unlike Embodiment 9, this embodiment uses the method shown in Figure 3b for model training, i.e., the access network device trains the access network device-side model. As shown in Figure 13, the method may include steps 1301-1310, as follows:

[0373] 1301. The core network device sends Request 1 to the access network device, Request 1 instructing the core network device to request the access network device to perform model training on the access network device side, wherein Request 1 includes a first identifier. Accordingly, the access network device receives Request 1.

[0374] In this example, the core network device assigns the first identifier.

[0375] 1302. The access network device sends response message 1 to the core network device, wherein response message 1 includes training dataset information.

[0376] For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in response message 1. Accordingly, the core network device receives response message 1. Optionally, the core network device receives the first identifier.

[0377] 1303. The core network equipment sends a notification message 1 to the application equipment.

[0378] For example, notification message 1 is used to instruct the application device to train a second model. For example, notification message 1 includes training dataset information. For example, the core network device sends a second identifier to the application device. Accordingly, the application device receives notification message 1. Optionally, the application device receives the second identifier. Optionally, the aforementioned training dataset information and the second identifier can be carried in the same message.

[0379] In this example, the core network devices associate the first model with the second model using different identifiers.

[0380] 1304. Core network equipment stores the correspondence between the first identifier, the second identifier, the first vendor, and the second vendor.

[0381] For a description of steps 1305-1310, please refer to the description of steps 1205-1210 in the embodiment shown in Figure 12, which will not be repeated here.

[0382] In this example, dual-end model identifiers are assigned by the core network equipment, and the identifiers of the model on the access network equipment side and the model on the terminal equipment side are different.

[0383] Example 11

[0384] Figure 14 shows a schematic diagram of another communication method provided in an embodiment of this application. This embodiment uses the method shown in Figure 3a for model training, that is, the core network device trains the access network device-side model. The difference from Embodiment 9 is that in this embodiment, the core network device assigns the identifier of the access network device-side model, and the application device assigns the identifier of the terminal device-side model. As shown in Figure 14, the method may include steps 1401-1410, as follows:

[0385] For a description of steps 1401-1404a, please refer to the description of steps 1001-1004a in the embodiment shown in Figure 10, which will not be repeated here.

[0386] 1404b. The core network equipment sends a notification message 2 to the application equipment.

[0387] For example, notification message 2 is used to instruct the application device to train a second model. For example, notification message 2 includes training dataset information. Accordingly, the application device receives notification message 2.

[0388] 1405. The application device trains the second model based on the received training dataset information.

[0389] 1406a. The application device sends the second identifier to the core network device.

[0390] In this example, the second identifier is assigned by the application device. For instance, this second identifier is different from the first identifier. Of course, the second identifier and the first identifier can also be the same; this solution does not impose any restrictions on this.

[0391] 1406b. The application device sends a notification message to the terminal device.

[0392] For example, notification message 3 is used to send the trained second model to the terminal device. Optionally, notification message 3 includes a second identifier. Accordingly, the terminal device receives notification message 3.

[0393] 1406c. Core network equipment stores the correspondence between the first identifier, the second identifier, the first vendor, and the second vendor.

[0394] For example, core network equipment stores the correspondence between a first identifier, a first vendor identifier, a second identifier, and a second vendor identifier. For instance, the correspondence between the first identifier, the first vendor identifier, the second identifier, and the second vendor identifier can be stored as <<First Identifier, First Vendor Identifier>, <Second Identifier, Second Vendor Identifier>>.

[0395] For a description of steps 1407-1410, please refer to the description of steps 607-610 in the embodiment shown in Figure 6, which will not be repeated here.

[0396] In this example, the core network equipment assigns the identifier of the access network equipment-side model, and the application equipment assigns the identifier of the terminal equipment-side model.

[0397] Example 12

[0398] Figure 15 shows a schematic diagram of another communication method provided in an embodiment of this application. Unlike embodiment 11, this embodiment uses the method shown in Figure 3b for model training, i.e., the access network device trains the access network device-side model. As shown in Figure 15, the method may include steps 1501-1509, as follows:

[0399] 1501. The core network device sends Request 1 to the access network device, Request 1 instructing the core network device to request the access network device to perform model training on the access network device side, wherein Request 1 includes a first identifier. Accordingly, the access network device receives Request 1.

[0400] In this example, the core network device assigns the first identifier.

[0401] 1502. The access network device sends response message 1 to the core network device, wherein response message 1 includes training dataset information.

[0402] For example, the access network device sends a first identifier to the core network device. Optionally, the first identifier is included in the model training response message. Accordingly, the core network device receives the response message 1. Optionally, the core network device receives the first identifier.

[0403] 1503. The core network equipment sends a notification message 1 to the application equipment.

[0404] For example, notification message 1 is used to instruct the application device to train a second model. For example, notification message 1 includes training dataset information. Accordingly, the application device receives notification message 1.

[0405] 1504. The application device trains the second model based on the received training dataset information.

[0406] 1505a. The application device sends a second identifier to the core network device.

[0407] In this example, the application device assigns a second identifier. For instance, this second identifier is different from the first identifier. Of course, the second identifier and the first identifier can also be the same; this solution does not impose any restrictions on this.

[0408] 1505b. The application device sends a notification message 2 to the terminal device.

[0409] For example, notification message 2 is used to send the trained second model to the terminal device. Optionally, notification message 2 includes a second identifier. Accordingly, the terminal device receives notification message 2.

[0410] 1505c: The core network equipment stores the correspondence between the first identifier, the second identifier, the first vendor, and the second vendor.

[0411] For a description of steps 1506-1509, please refer to the description of steps 1407-1410 in the embodiment shown in Figure 14, which will not be repeated here.

[0412] In this example, the core network equipment assigns the identifier of the access network equipment-side model, and the application equipment assigns the identifier of the terminal equipment-side model.

[0413] In conjunction with embodiments 1-12 above, this application also provides a communication method. Referring to FIG16, a flowchart illustrating a communication method provided in this application is shown. Optionally, this method can be applied to the aforementioned communication system, such as the communication system shown in FIG1. ​​The communication method shown in FIG16 may include steps 1601-1605. Steps 1601-1605 are as follows:

[0414] 1601. Core network equipment obtains the first identifier.

[0415] The first identifier is used to associate the first model with the second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0416] The first model corresponds to the access network equipment and can be understood as being used by the access network equipment. Optionally, the first model is deployed on the access network equipment. Correspondingly, the second model corresponds to the terminal equipment and can be understood as being used by the terminal equipment. Optionally, the second model is deployed on the terminal equipment.

[0417] In one implementation, the core network device receives a first identifier from the access network device. This first identifier is generated by the access network device. The first identifier is used to identify a first model; that is, the first identifier corresponds to the first model. Optionally, before step 1601, the core network device sends candidate identifiers to the access network device. In this way, the access network device determines the first identifier from the candidate identifiers.

[0418] The implementation of this step can be referred to in step 402 of the embodiment shown in Figure 4, or in step 502 of the embodiment shown in Figure 5.

[0419] 1603. The core network equipment stores the first identifier, the first vendor, and the association information of the second vendor.

[0420] In one implementation, the first identifier is also used to identify the second model. That is, the first identifier corresponds to the second model. The core network device sends the first identifier to the application device.

[0421] The implementation of this step is described in step 404b of the embodiment shown in Figure 4, or in step 503 of the embodiment shown in Figure 5.

[0422] For example, the core network device stores association information of a first identifier, a first vendor identifier, and a second vendor identifier. For instance, the core network device stores association information of <<first identifier, first vendor identifier, second vendor identifier>>.

[0423] Optionally, the association information of the first identifier, the first vendor, and the second vendor can also be stored by a network element with storage capabilities (e.g., UDM / UDR).

[0424] The implementation of this step is described in step 404c in the embodiment shown in Figure 4, or in step 504 in the embodiment shown in Figure 5.

[0425] 1604. The core network equipment receives the first information.

[0426] The first information is used to identify the model for which inference is requested to be performed. This first information includes a first vendor identifier and a second vendor identifier, which are used to determine the model for which inference is performed.

[0427] This initial information can come from access network equipment or from terminal equipment, etc.

[0428] The implementation of this step is described in steps 407 and 409 in the embodiment shown in Figure 4, or in steps 507 and 509 in the embodiment shown in Figure 5.

[0429] 1605. The core network equipment sends the first identifier.

[0430] The core network equipment determines the first identifier based on stored association information. For example, the core network equipment searches for the association between the first vendor identifier, the second vendor identifier, and the first identifier stored in the core network equipment, thereby obtaining the first identifier.

[0431] Optionally, the core network equipment queries the storage function network element (e.g., UDM / UDR, etc.) for the first identifier based on the first vendor identifier and the second vendor identifier.

[0432] The core network equipment sends the first identifier to the access network equipment, or the core network equipment sends the first identifier to the terminal equipment.

[0433] The implementation of this step is described in steps 408 and 410 in the embodiment shown in Figure 4, or in steps 508 and 510 in the embodiment shown in Figure 5.

[0434] The above example illustrates how an access network device generates a first identifier, which is used to identify a first model and a second model.

[0435] Alternatively, in one implementation, the method further includes step 1602: the core network device obtains a second identifier, which is used to identify the second model. That is, in this implementation, the first identifier is used to identify the first model, and the second identifier is used to identify the second model.

[0436] The following section describes how the core network device obtains the second identifier.

[0437] (1) The second identifier is generated by the core network equipment. Both the first and second identifiers are used to associate the first model with the second model. The first identifier is used to identify the first model, and the second identifier is used to identify the second model.

[0438] In this implementation, in step 1603, the core network device stores the association information of the first identifier, the second identifier, the first vendor, and the second vendor. For example, the core network device stores the association information of <<first identifier, first vendor identifier>, <second identifier, second vendor identifier>>.

[0439] The implementation of this step is described in step 604c in the embodiment shown in Figure 6, or in step 704 in the embodiment shown in Figure 7.

[0440] In this implementation, in step 1604, the core network device receives second information, which is used to request the identifier of the model for performing inference. The second information includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference.

[0441] The implementation of this step is described in steps 607 and 609 in the embodiment shown in Figure 6, or in steps 707 and 709 in the embodiment shown in Figure 7.

[0442] In this implementation, in step 1605, the core network device sends a first identifier to the access network device. Alternatively, the core network device sends a second identifier to the terminal device. Alternatively, the core network device sends both a first identifier and a second identifier to the access network device. Alternatively, the core network device sends both a first identifier and a second identifier to the terminal device.

[0443] The implementation of this step is described in steps 608 and 610 in the embodiment shown in Figure 6, or in steps 708 and 710 in the embodiment shown in Figure 7.

[0444] (2) The second identifier is generated by the application device. Both the first and second identifiers are used to associate the first model with the second model. The first identifier is used to identify the first model, and the second identifier is used to identify the second model.

[0445] For example, the core network equipment receives a second identifier from the application equipment.

[0446] In this implementation, in step 1603, the core network device stores the association information of the first identifier, the second identifier, the first vendor, and the second vendor.

[0447] The implementation of this step is described in step 806c of the embodiment shown in Figure 8, or in step 905c of the embodiment shown in Figure 9.

[0448] In this implementation, in step 1604, the core network device receives second information, which is used to request the identifier of the model for performing inference. The second information includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference.

[0449] The implementation of this step is described in steps 807 and 809 in the embodiment shown in Figure 8, or in steps 906 and 908 in the embodiment shown in Figure 9.

[0450] In this implementation, in step 1605, the core network device sends a first identifier to the access network device. Alternatively, the core network device sends a second identifier to the terminal device. Alternatively, the core network device sends both a first identifier and a second identifier to the access network device. Alternatively, the core network device sends both a first identifier and a second identifier to the terminal device.

[0451] The implementation of this step is described in steps 808 and 810 in the embodiment shown in Figure 8, or in steps 907 and 909 in the embodiment shown in Figure 9.

[0452] The above example illustrates how the access network device generates the first identifier and the core network device obtains the second identifier.

[0453] Alternatively, in one implementation, the core network device generates a first identifier, which is used to identify the first model and the second model.

[0454] For example, in step 1601, the core network device generates a first identifier. The core network device sends the first identifier to the access network device. The core network device then sends the first identifier to the application device.

[0455] The implementation of this step is described in steps 1001 and 1004b in the embodiment shown in Figure 10, or in steps 1101 and 1103a in the embodiment shown in Figure 11.

[0456] The above example illustrates how a core network device generates a first identifier, which is used to identify a first model and a second model.

[0457] Alternatively, in one implementation, the core network device generates a first identifier and a second identifier, which are used to associate the first model with the second model. The first identifier is used to identify the first model, and the second identifier is used to identify the second model.

[0458] For example, in step 1601, the core network device generates a first identifier and a second identifier. The core network device sends the first identifier to the access network device. The core network device sends the second identifier to the application device. Optionally, the first identifier and the second identifier are different.

[0459] The implementation of this step is described in steps 1201 and 1204b in the embodiment shown in Figure 12, or in steps 1301 and 1303 in the embodiment shown in Figure 13.

[0460] In this implementation, in step 1603, the core network device stores the association information of the first identifier, the second identifier, the first vendor, and the second vendor.

[0461] The implementation of this step is described in step 1204c in the embodiment shown in Figure 12, or in step 1304 in the embodiment shown in Figure 13.

[0462] In this implementation, in step 1604, the core network device receives second information, which is used to request the identifier of the model for performing inference. The second information includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference.

[0463] The implementation of this step is described in steps 1207 and 1209 in the embodiment shown in Figure 12, or in steps 1307 and 1309 in the embodiment shown in Figure 13.

[0464] In this implementation, in step 1605, the core network device sends a first identifier to the access network device. Alternatively, the core network device sends a second identifier to the terminal device. Alternatively, the core network device sends both a first identifier and a second identifier to the access network device. Alternatively, the core network device sends both a first identifier and a second identifier to the terminal device.

[0465] The implementation of this step is described in steps 1208 and 1210 in the embodiment shown in Figure 12, or in steps 1308 and 1310 in the embodiment shown in Figure 13.

[0466] The above example illustrates how a core network device generates a first identifier and a second identifier, with the first identifier used to identify a first model and the second identifier used to identify a second model.

[0467] Alternatively, in one implementation, the core network device generates a first identifier, and the application device generates a second identifier. The first and second identifiers are used to associate the first model with the second model; the first identifier identifies the first model, and the second identifier identifies the second model.

[0468] For example, in step 1601, the core network device generates a first identifier. The core network device then sends the first identifier to the access network device.

[0469] The implementation of this step is described in step 1401 in the embodiment shown in Figure 14, or in step 1501 in the embodiment shown in Figure 15.

[0470] In step 1602, the application device generates a second identifier. The application device then sends the second identifier to the core network device.

[0471] The implementation of this step is described in step 1406a of the embodiment shown in Figure 14, or in step 1505a of the embodiment shown in Figure 15.

[0472] In this implementation, in step 1603, the core network device stores the association information of the first identifier, the second identifier, the first vendor, and the second vendor.

[0473] The implementation of this step is described in step 1406c in the embodiment shown in Figure 14, or in step 1505c in the embodiment shown in Figure 15.

[0474] In this implementation, in step 1604, the core network device receives second information, which is used to request the identifier of the model for performing inference. The second information includes a first vendor identifier and a second vendor identifier, which are used to determine the model for performing inference.

[0475] The implementation of this step is described in steps 1407 and 1409 in the embodiment shown in Figure 14, or in steps 1506 and 1508 in the embodiment shown in Figure 15.

[0476] In this implementation, in step 1605, the core network device sends a first identifier to the access network device. Alternatively, the core network device sends a second identifier to the terminal device. Alternatively, the core network device sends both a first identifier and a second identifier to the access network device. Alternatively, the core network device sends both a first identifier and a second identifier to the terminal device.

[0477] The implementation of this step is described in steps 1408 and 1410 in the embodiment shown in Figure 14, or in steps 1507 and 1509 in the embodiment shown in Figure 15.

[0478] In this embodiment, model association between two-end models is achieved by associating a first model with a second model through a first identifier, or by associating a first model with a second identifier. When performing model inference, the access network device and the terminal device obtain the model's identifier from the core network device, and then obtain the matching model for model inference.

[0479] In conjunction with embodiments 1-12 above, this application also provides a communication method. Referring to FIG17, a flowchart illustrating another communication method provided by this application is shown. Optionally, this method can be applied to the aforementioned communication system, such as the communication system shown in FIG1. ​​The communication method shown in FIG17 may include steps 1701-1709. Steps 1701-1709 are as follows:

[0480] 1701. The core network device sends a first request to the access network device, the first request being used to request the access network device to train a first model or report training data used to train the first model. Accordingly, the access network device receives the first request.

[0481] The first request is used to request the access network device to train the first model, that is, the access network device trains the model on the access network device side. The implementation of this step is described in step 501 of the embodiment shown in Figure 5.

[0482] The first request is used to request the access network device to report training data for training the first model, that is, the core network device trains the model on the access network device side. The implementation of this step is described in step 401 of the embodiment shown in Figure 4.

[0483] 1702. The access network device sends a first response to the core network device. This first response includes training data for training the second model or the aforementioned training data for training the first model. The first response also includes a first identifier for identifying the first model. Accordingly, the core network device receives the first response.

[0484] For example, when the first request is used to request the access network device to train the first model, the corresponding first response includes training data for training the second model. That is, after completing the training of the first model, the access network device sends the training data for training the second model to the core network device. The implementation of this step is described in step 502 of the embodiment shown in FIG5.

[0485] When the first request is used to request the access network device to report training data for training the first model, the corresponding first response includes the training data for training the first model. This allows the core network device to train the first model based on the training data. The implementation of this step is described in step 402 of the embodiment shown in Figure 4.

[0486] Optionally, the training data and the first identifier used to train the second model can be carried in the same message, or they can be carried in different messages.

[0487] Accordingly, the training data and the first identifier used to train the first model can be carried in the same message or in different messages.

[0488] In one possible implementation, before step 1702 is executed, the core network device further sends a candidate identifier to the access network device, which is used to determine the aforementioned first identifier. The first identifier is within the range of candidate identifiers.

[0489] 1703. The core network device sends the aforementioned first identifier to the application device, which is also used to identify the aforementioned second model. The application device is used to train the second model, and the first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device. Accordingly, the application device receives the first identifier.

[0490] In one possible implementation, the core network device sends a second request to the application device, which requests the application device to train a second model. This second request includes training data for training the second model and also includes a first identifier.

[0491] The implementation of this step is described in step 404b of the embodiment shown in Figure 4, or in step 503 of the embodiment shown in Figure 5.

[0492] 1704. The core network equipment stores the association information of the first identifier, the first vendor identifier, and the second vendor identifier. The first vendor identifier corresponds to the access network equipment, and the second vendor identifier corresponds to the terminal equipment.

[0493] The implementation of this step is described in step 404c in the embodiment shown in Figure 4, or in step 504 in the embodiment shown in Figure 5.

[0494] 1705. The application device sends the first identifier to the terminal device. The application device also sends the second model to the terminal device.

[0495] The implementation of this step is described in step 406 of the embodiment shown in Figure 4, or in step 506 of the embodiment shown in Figure 5.

[0496] 1706. The core network device receives a third request from the access network device, the third request being used to request an identifier for the model used to perform inference. The third request includes a first vendor identifier and a second vendor identifier, the first vendor identifier and the second vendor identifier being used to determine the model used to perform inference.

[0497] The implementation of this step is described in step 407 of the embodiment shown in Figure 4, or in step 507 of the embodiment shown in Figure 5.

[0498] 1707. The core network device sends a first identifier to the access network device, wherein the first model and the second model identified by the first identifier are the models for the above-mentioned inference.

[0499] For example, the core network device determines the first model and the second model identified by the first identifier as the models for performing inference based on the first vendor identifier and the second vendor identifier.

[0500] The implementation of this step is described in step 408 of the embodiment shown in Figure 4, or in step 508 of the embodiment shown in Figure 5.

[0501] 1708. The core network device receives a fourth request from the terminal device, the fourth request being for requesting an identifier for the model used to perform inference. The third request includes a first vendor identifier and a second vendor identifier, the first vendor identifier and the second vendor identifier being used to determine the model used to perform inference.

[0502] The implementation of this step is described in step 409 of the embodiment shown in Figure 4, or in step 509 of the embodiment shown in Figure 5.

[0503] 1709. The core network equipment sends a first identifier to the terminal equipment, wherein the first model and the second model identified by the first identifier are the models for performing the above-mentioned inference.

[0504] The implementation of this step is described in step 410 of the embodiment shown in Figure 4, or in step 510 of the embodiment shown in Figure 5.

[0505] In this embodiment, the first model and the second model are associated through a first identifier assigned by the access network device, thereby realizing the model association between the two-end models.

[0506] It should be understood that this application describes the steps in the above order for ease of description, and is not intended to limit the execution to the above order. The embodiments of this application do not limit the order of execution, execution time, or number of executions of one or more steps in the above embodiments.

[0507] It should be noted that, in the various embodiments of this application, unless otherwise specified or in case of logical conflict, the terms and / or descriptions between the various embodiments are consistent and can be referenced by each other. The technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0508] The methods of the embodiments of this application have been described in detail above, and the apparatus of the embodiments of this application is provided below. It is understood that the division of multiple units or modules in the various apparatus embodiments of this application is only a logical division based on function and is not intended to limit the specific structure of the apparatus. In specific implementations, some functional modules may be subdivided into more smaller functional modules, and some functional modules may be combined into a single functional module. However, regardless of whether these functional modules are subdivided or combined, the general flow executed by the apparatus is the same. For example, some apparatuses include a receiving unit and a transmitting unit. In some designs, the transmitting unit and the receiving unit can also be integrated into a communication unit, which can implement the functions implemented by the receiving unit and the transmitting unit. Typically, each unit corresponds to its own program code (or program instructions). When the program code corresponding to each unit runs on the processor, it causes the unit to be controlled by the processing unit to execute the corresponding flow and thus achieve the corresponding function.

[0509] This application also provides an apparatus for implementing any of the above methods. For example, a communication apparatus is provided that includes modules (or means) for implementing the steps performed by the core network device, access network device, application device, or terminal device in any of the above methods.

[0510] For example, referring to FIG18, which is a schematic diagram of a communication device provided in an embodiment of this application, the communication device is used to implement the aforementioned communication method, such as the modules (or means) of the steps performed by the core network equipment in the communication method shown in FIG4-FIG17.

[0511] As shown in Figure 18, the device may include a communication module 1801, as detailed below:

[0512] The communication module 1801 is used to send a first request to the access network device, the first request being used to request the access network device to train a first model; or, the first request being used to request the access network device to report training data for training the first model.

[0513] The communication module 1801 is also configured to receive a first response from an access network device. The first response includes training data for training a second model or training data for training a first model, and the first response also includes a first identifier for identifying the first model.

[0514] The communication module 1801 is also used to send the aforementioned first identifier to the application device, which is also used to identify the second model. The application device is used to train the second model, and the first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0515] For a description of each of the above modules, please refer to the description of the embodiments shown in Figures 4-17 above, and will not be repeated here.

[0516] For example, as shown in FIG18, the device may include a communication module 1801, which is used to send a first request to the access network device, the first request being used to request the access network device to train a first model; or, the first request being used to request the access network device to report training data for training the first model.

[0517] The communication module 1801 is also configured to receive a first response from an access network device, the first response including training data for training a second model or training data for training a first model, and the first response also including a first identifier for identifying the first model.

[0518] The communication module 1801 is also used to send a second request to the application device. The second request is used to request the application device to train a second model. The second request includes training data for training the second model. The second request also includes a second identifier for identifying the second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0519] The above modules can be described in the description of the embodiments shown in Figures 4-17 above, and will not be repeated here.

[0520] For example, as shown in FIG18, the device may include a communication module 1801, which is used to send a first request to the access network device, the first request being used to request the access network device to train a first model; or, the first request being used to request the access network device to report training data for training the first model.

[0521] The communication module 1801 is also configured to receive a first response from an access network device, the first response including training data for training a second model or training data for training a first model, and the first response also including a first identifier for identifying the first model.

[0522] The communication module 1801 is also used to send a second request to the application device, the second request being used to request the application device to train a second model, the second request including training data for training the second model.

[0523] The communication module 1801 is also used to receive a second identifier from the application device, the second identifier being used to identify a second model, the first model and the second model forming a dual-end model, the first model corresponding to the access network device, and the second model corresponding to the terminal device.

[0524] For a description of each of the above modules, please refer to the description of the embodiments shown in Figures 4-17 above, and will not be repeated here.

[0525] For example, as shown in FIG18, the device may include a communication module 1801, which is used to send a first request to the access network device, the first request being used to request the access network device to train a first model; or, the first request being used to request the access network device to report training data for training the first model, the first request including a first identifier, the first identifier being used to identify the first model.

[0526] The communication module 1801 is also used to send a second request to the application device. The second request is used to request the application device to train a second model. The second request includes training data for training the second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0527] For a description of each of the above modules, please refer to the description of the embodiments shown in Figures 4-17 above, and will not be repeated here.

[0528] For example, as shown in FIG18, the device may include a communication module 1801, which is used to receive a first request from a core network device, the first request being used to request an access network device to train a first model or to report training data for training the first model.

[0529] The communication module 1801 is also used to send a first response to the core network device. The first response includes training data for training the second model or training data for training the first model. The first response also includes a first identifier for identifying the first model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0530] For a description of each of the above modules, please refer to the description of the embodiments shown in Figures 4-17 above, and will not be repeated here.

[0531] For example, as shown in FIG18, the device may include a communication module 1801, which is used to send first information to the core network device. The first information is used to request the identifier of the model to be executed for inference. The first information includes a first vendor identifier and a second vendor identifier. The first vendor identifier and the second vendor identifier are used to determine the model to be executed for inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device.

[0532] The communication module 1801 is also used to receive second information from the core network device. The second information includes a first identifier, which is used to identify a second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

[0533] For a description of each of the above modules, please refer to the description of the embodiments shown in Figures 4-17 above, and will not be repeated here.

[0534] For example, as shown in FIG18, the device may include a communication module 1801, which is used to receive first information, the first information including training data for training a second model, and the first information instructing the application device to train the second model.

[0535] The communication module 1801 is also used to send second information, the second information including a first identifier, the first identifier being used to identify a second model, wherein the first model and the second model form a dual-end model, the first model corresponding to the access network device, and the second model corresponding to the terminal device.

[0536] For a description of each of the above modules, please refer to the description of the embodiments shown in Figures 4-17 above, and will not be repeated here.

[0537] It should be understood that the division of modules in the above devices is only a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, modules in a communication device can be implemented by a processor calling software; for example, a communication device includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of each module in the device. The processor can be, for example, a general-purpose processor, such as a central processing unit (CPU) or a microprocessor, and the memory can be internal or external to the device. Alternatively, the modules in the device can be implemented as hardware circuits. The functionality of some or all units can be achieved through the design of these hardware circuits, which can be understood as one or more processors. For example, in one implementation, the hardware circuit is an application-specific integrated circuit (ASIC), and the functionality of some or all of the above units is achieved through the design of the logical relationships between the components within the circuit. In another implementation, the hardware circuit can be implemented using a programmable logic device (PLD), such as a field-programmable gate array (FPGA), which can include a large number of logic gates. The connection relationships between the logic gates are configured through configuration files, thereby achieving the functionality of some or all of the above units. All modules of the above device can be implemented entirely through processor-called software, entirely through hardware circuits, or partially through processor-called software with the remaining parts implemented through hardware circuits.

[0538] Referring to FIG19, a hardware structure diagram of another communication device provided in an embodiment of this application is shown. The communication device 1900 shown in FIG19 includes one or more processors 1901 (a processor is illustrated in the figure).

[0539] Processor 1901 is a circuit with signal processing capabilities. In one implementation, processor 1901 can be a circuit with instruction read and execute capabilities, such as a central processing unit (CPU), microprocessor, graphics processing unit (GPU) (which can be understood as a type of microprocessor), or digital signal processor (DSP). In another implementation, processor 1901 can implement certain functions through the logical relationships of hardware circuits. These logical relationships of hardware circuits are fixed or reconfigurable. For example, processor 1901 can be a hardware circuit implemented as an ASIC or a programmable logic device (PLD), such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the processor loading instructions to implement the functions of some or all of the above modules. Furthermore, it can also be a hardware circuit designed for artificial intelligence, which can be understood as a type of ASIC, such as a neural network processing unit (NPU), tensor processing unit (TPU), or deep learning processing unit (DPU). The processor 1901 is used to execute related programs to implement the functions required by the units in the communication device of the present application embodiment, or to execute the communication method of the method embodiment of the present application.

[0540] Optionally, the communication device 1900 may also include a memory (e.g., memory 1903, memory 1904, memory 1905) (shown as dashed lines in the figure). This memory is used to store instructions executed by the processor 1901, or to store input data required by the processor 1901 to execute instructions, or to store data generated after the processor 1901 executes instructions.

[0541] Optionally, the memory may be located within the one or more processors (e.g., memory 1903), or outside the one or more processors (e.g., memory 1904, memory 1905), or may include a storage portion located within the one or more processors and a storage portion located outside the one or more processors.

[0542] In this embodiment, the memory (e.g., memory 1903, memory 1904, memory 1905) may include, but is not limited to, cache, read-only memory (ROM), random access memory (RAM), synchronous dynamic random access memory (SDRAM), hard disk drive (HDD) or solid-state drive (SSD), erasable programmable read-only memory (EPROM), or compact disc read-only memory (CD-ROM), etc. Memory is any other medium capable of carrying or storing desired program code having an instruction or data structure form and accessible by a computer, but is not limited thereto. The memory in this embodiment may also be a circuit or any other device capable of implementing storage functions for storing computer programs or instructions, and / or data.

[0543] Optionally, the communication device 1900 may also include a communication interface 1902 (shown as a dashed line in the figure). The processor 1901 and the communication interface 1902 are coupled to each other. The communication interface 1902 may be a transceiver or interface circuit, bus, module, or other type of communication interface.

[0544] The memory can store programs. When the program stored in the memory is executed by the processor 1901, the processor 1901 and the communication interface 1902 are used to execute the various steps of the communication method of the embodiments of this application.

[0545] As can be seen, each module in the above device can be one or more processors (or processing circuits) configured to implement the above methods, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms or a portion of the processing circuits in these processors.

[0546] Furthermore, the modules in the above devices can be integrated in whole or in part, or they can be implemented independently. In one implementation, these modules are integrated together as a system-on-a-chip (SOC). The SOC may include at least one processor for implementing any of the above methods or for implementing the functions of the modules of the device. The at least one processor may be of different types, such as CPU and FPGA, CPU and artificial intelligence processor, CPU and GPU, etc.

[0547] It should be noted that although the device 1900 shown in FIG. 19 only illustrates the memory, processor, and communication interface, those skilled in the art should understand that in specific implementations, device 1900 may also include other devices necessary for normal operation. Furthermore, depending on specific needs, those skilled in the art should understand that device 1900 may also include hardware devices for implementing other additional functions. Moreover, those skilled in the art should understand that device 1900 may only include the devices necessary for implementing the embodiments of this application, and not necessarily all the devices shown in FIG. 19.

[0548] This application also provides a computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform one or more steps of any of the above methods.

[0549] This application also provides a computer program product containing instructions. When the computer program product is run on a computer or processor, it causes the computer or processor to perform one or more steps of any of the methods described above.

[0550] It is understood that in this application, "instruction" can include direct instruction, indirect instruction, explicit instruction, and implicit instruction. When describing a certain instruction information to indicate A, it can be understood that the instruction information carries A, directly indicates A, or indirectly indicates A. In this application, the information indicated by the instruction information is called the information to be instructed. In specific implementation, there are many ways to indicate the information to be instructed, such as, but not limited to, directly indicating the information to be instructed, such as the information to be instructed itself or its index, or indirectly indicating the information to be instructed by indicating other information, wherein there is an association between the other information and the information to be instructed. It is also possible to indicate only a part of the information to be instructed, while the other parts of the information to be instructed are known or agreed upon in advance. For example, the instruction of specific information can also be achieved by using the arrangement order of various information in advance (e.g., as specified by a protocol), thereby reducing the instruction overhead to a certain extent. The information to be instructed can be sent as a whole or divided into multiple sub-information to be sent separately, and the sending period and / or sending time of these sub-information can be the same or different. This application does not limit the specific sending method. The sending period and / or timing of these sub-information messages can be predefined, for example, according to a protocol, or configured by the transmitting device by sending configuration information to the receiving device.

[0551] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the division of units is merely a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. The coupling, direct coupling, or communication connection shown or discussed between each other may be indirect coupling or communication connection through some interfaces, apparatuses, or units, and may be electrical, mechanical, or other forms.

[0552] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0553] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media can be read-only memory (ROM), random access memory (RAM), or magnetic media, such as floppy disks, hard disks, magnetic tapes, magnetic disks, or optical media, such as digital versatile discs (DVDs), or semiconductor media, such as solid-state disks (SSDs).

[0554] The above description is merely a specific implementation of the embodiments of this application, but the protection scope of the embodiments of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the embodiments of this application should be covered within the protection scope of the embodiments of this application. Therefore, the protection scope of the embodiments of this application should be determined by the protection scope of the claims.

Claims

1. A communication method, characterized in that, The method is applied to core network equipment, and the method includes: Send a first request to the access network device, the first request being used to request the access network device to train a first model or report training data for training the first model; The system receives a first response from the access network device. The first response includes training data for training a second model or the training data for training a first model. The first response also includes a first identifier for identifying the first model. The first identifier is sent to the application device, and the first identifier is also used to identify the second model. The application device is used to train the second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

2. The method according to claim 1, characterized in that, Sending the first identifier to the application device includes: A second request is sent to the application device, the second request being used to request the application device to train the second model, the second request including training data for training the second model, and the second request also including the first identifier.

3. The method according to claim 1 or 2, characterized in that, The method further includes: The system stores association information between the first identifier, the first vendor identifier, and the second vendor identifier, where the first vendor identifier corresponds to the access network device and the second vendor identifier corresponds to the terminal device.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: A candidate identifier is sent to the access network device, the candidate identifier being used to determine the first identifier, the first identifier being within the range of the candidate identifiers.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: A third request is received, the third request being used to request the identifier of the model for performing inference, the third request including a first vendor identifier and a second vendor identifier, the first vendor identifier and the second vendor identifier being used to determine the model for performing inference, the first vendor identifier corresponding to the access network device, and the second vendor identifier corresponding to the terminal device; Send the first identifier, wherein the first model and the second model identified by the first identifier are the models for performing inference.

6. The method according to claim 5, characterized in that, The method further includes: Based on the first vendor identifier and the second vendor identifier, the first model and the second model identified by the first identifier are determined to be the models for performing inference.

7. A communication method, characterized in that, The method is applied to core network equipment, and the method includes: Send a first request to the access network device, the first request being used to request the access network device to train a first model or report training data for training the first model; The system receives a first response from the access network device. The first response includes training data for training a second model or the training data for training a first model. The first response also includes a first identifier for identifying the first model. A second request is sent to the application device. The second request is used to request the application device to train the second model. The second request includes the training data used to train the second model. The second request also includes a second identifier used to identify the second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

8. The method according to claim 7, characterized in that, The method further includes: The system stores the association information of the first identifier, the second identifier, the first vendor identifier, and the second vendor identifier. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device.

9. The method according to claim 7 or 8, characterized in that, The method further includes: A third request is received, the third request being used to request the identifier of the model for performing inference, the third request including a first vendor identifier and a second vendor identifier, the first vendor identifier and the second vendor identifier being used to determine the model for performing inference, the first vendor identifier corresponding to the access network device, and the second vendor identifier corresponding to the terminal device; Send the first identifier and the second identifier, where the first model and the second model are the models for performing the inference.

10. The method according to claim 9, characterized in that, The method further includes: Based on the first vendor identifier and the second vendor identifier, the first model identified by the first identifier and the second model identified by the second identifier are determined to be the models for performing inference.

11. A communication method, characterized in that, The method is applied to core network equipment, and the method includes: Send a first request to the access network device, the first request being used to request the access network device to train a first model or report training data for training the first model, the first request including a first identifier, the first identifier being used to identify the first model; A second request is sent to the application device. The second request is used to request the application device to train a second model. The second request includes training data for training the second model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

12. The method according to claim 11, characterized in that, The second request includes a second identifier, which is used to identify the second model.

13. The method according to claim 12, characterized in that, The first identifier and the second identifier are the same.

14. A communication method, characterized in that, The method is applied to an access network device, and the method includes: Receive a first request from the core network device, the first request being used to request the access network device to train a first model or report training data for training the first model; A first response is sent to the core network device. The first response includes training data for training the second model or the training data for training the first model. The first response also includes a first identifier for identifying the first model. The first model and the second model form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

15. The method according to claim 14, characterized in that, The method further includes: In response to the first request, the first identifier is generated.

16. The method according to claim 15, characterized in that, The first request includes a candidate identifier; the step of generating the first identifier in response to the first request includes: The first identifier is generated based on the candidate identifiers, and the first identifier is within the range of the candidate identifiers.

17. The method according to claim 14, characterized in that, The method further includes: Receive the first identifier from the core network device.

18. The method according to claim 17, characterized in that, The first request includes the first identifier.

19. The method according to any one of claims 14 to 18, characterized in that, The first identifier is also used to identify the second model.

20. The method according to claim 19, characterized in that, The method further includes: Send a second request, the second request being used to request the identifier of the model for performing inference, the second request including a first vendor identifier and a second vendor identifier, the first vendor identifier and the second vendor identifier being used to determine the model for performing inference, the first vendor identifier corresponding to the access network device, and the second vendor identifier corresponding to the terminal device; The first identifier is received, and the first model identified by the first identifier and the second model identified by the first identifier are the models for performing inference.

21. The method according to any one of claims 14 to 18, characterized in that, The method further includes: Send a second request, the second request being used to request the identifier of the model for performing inference, the second request including a first vendor identifier and a second vendor identifier, the first vendor identifier and the second vendor identifier being used to determine the model for performing inference, the first vendor identifier corresponding to the access network device, and the second vendor identifier corresponding to the terminal device; Receive the first identifier and the second identifier, the second identifier being used to identify the second model, and the first model and the second model being the model for performing inference.

22. A communication method, characterized in that, The method is applied to a terminal device, and the method includes: Send first information to the core network device. The first information is used to request the identifier of the model for performing inference. The first information includes a first vendor identifier and a second vendor identifier. The first vendor identifier and the second vendor identifier are used to determine the model for performing inference. The first vendor identifier corresponds to the access network device, and the second vendor identifier corresponds to the terminal device. The system receives second information from a core network device. The second information includes a first identifier, which is used to identify a second model. The first model and the second model together form a dual-end model. The first model corresponds to the access network device, and the second model corresponds to the terminal device.

23. The method according to claim 22, characterized in that, The first identifier is also used to identify the first model.

24. The method according to claim 22, characterized in that, The second information also includes a second identifier, which is used to identify the first model.

25. The method according to any one of claims 22 to 24, characterized in that, The method further includes: The first identifier is received from an application device used to train the second model.

26. A communication device, characterized in that, It includes modules or units for implementing the method as described in any one of claims 1-6, or modules or units for implementing the method as described in any one of claims 7-10, or modules or units for implementing the method as described in any one of claims 11-13, or modules or units for implementing the method as described in any one of claims 14-21, or modules or units for implementing the method as described in any one of claims 22-25.

27. A communication device, characterized in that, The device includes a processor and a memory, the processor being configured to execute a computer program or computer-executable instructions stored in the memory, and / or via logic circuitry, cause the device to perform the method as claimed in any one of claims 1-6, or the method as claimed in any one of claims 7-10, or the method as claimed in any one of claims 11-13, or the method as claimed in any one of claims 14-21, or the method as claimed in any one of claims 22-25.

28. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, causes the method described in any one of claims 1-6 to be implemented; or causes the method described in any one of claims 7-10 to be implemented; or causes the method described in any one of claims 11-13 to be implemented; or causes the method described in any one of claims 14-21 to be implemented; or causes the method described in any one of claims 22-25 to be implemented.

29. A computer program product comprising instructions that, when executed on a processor, causes the method of any one of claims 1-6 to be implemented; or causes the method of any one of claims 7-10 to be implemented; or causes the method of any one of claims 11-13 to be implemented; or causes the method of any one of claims 14-21 to be implemented; or causes the method of any one of claims 22-25 to be implemented.

30. A chip, characterized in that, The chip includes at least one processor and an interface, the processor being configured to execute computer instructions or programs that, when executed, cause the chip to perform the method as described in any one of claims 1-6, or the method as described in any one of claims 7-10, or the method as described in any one of claims 11-13, or the method as described in any one of claims 14-21, or the method as described in any one of claims 22-25.

Images