Model indication method, model generation method and related apparatus

Abstract

Provided in the present application are a model indication method, a model generation method and a related apparatus, which are used for enabling a first apparatus to acquire a user-level model, or enabling a second apparatus to deploy the user-level model for the first apparatus, thereby enabling the first apparatus to execute a first task on the basis of the user-level model. The method in the embodiments of the present application comprises: a first apparatus sending feature information to a second apparatus, wherein the feature information represents a feature of a first task; and the first apparatus receiving first information from the second apparatus, wherein the first information is used for indicating a user-level model, the user-level model is determined on the basis of the feature information, and common models of various levels, related to the first task, in the second apparatus, and the user-level model is used by the first apparatus to execute the first task.

Description

Model indication method, model generation method and related device Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a model indication method, a model generation method, and related devices. Background Art

[0002] To meet the vision of intelligent and inclusive future, intelligence will further evolve at the wireless network architecture level. Artificial intelligence (AI) will be further integrated with wireless networks, enabling endogenous network intelligence. This will also include intelligent terminal devices. For example, terminal types will be diversified, and terminal device connections will be more flexible and intelligent. Terminal devices will also possess certain AI capabilities. In addition to providing traditional communication and connectivity services, the network will also provide computing and AI services to better support inclusive, real-time, and highly secure AI services.

[0003] The terminal device can perform the corresponding task through the model. Therefore, how the terminal device obtains the model used to perform the task from the network is a question worth considering.

[0004] Summary of the Invention

[0005] This application provides a model indication method, a model generation method, and related devices for enabling a first device to obtain a user-level model, or enabling a second device to deploy a user-level model for the first device, thereby enabling the first device to perform a first task based on the user-level model.

[0006] In a first aspect, the present application provides a model indication method, which can be performed by a first device. The first device can be a terminal device or a network device, or a component in the terminal device or network device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can implement all or part of the terminal device functions, or a logic module or software that can implement all or part of the network device functions. The method includes:

[0007] The first device sends characteristic information to the second device. The characteristic information characterizes the characteristics of the first task. The first device receives first information from the second device. The first information indicates a user-level model. The user-level model is determined based on the characteristic information and various levels of public models related to the first task in the second device. The user-level model is used by the first device to perform the first task. This enables the first device to obtain the user-level model, or in other words, enables the second device to deploy the user-level model to the first device. This enables the first device to perform the first task based on the user-level model.

[0008] Based on the first aspect, in one possible implementation, the method further includes: the first device determining characteristic information based on task information of the first task; or, the first device determining characteristic information based on the task information of the first task and / or auxiliary information of the first device. This implementation illustrates one possible method for generating characteristic information. The first device may determine the characteristic information based on the task information of the first task and / or the auxiliary information of the first device. This thereby indicates the characteristics of the first task to the second device.

[0009] Based on the first aspect, in one possible implementation, the auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device. In this implementation, the first device further generates feature information based on the auxiliary information to facilitate requesting an appropriate model from the second device.

[0010] Based on the first aspect, in one possible implementation, the feature information includes a feature vector; the first device determines the feature information based on the task information and auxiliary information of the first task, including: the first device inputting the task information into a feature vector generation model to obtain the feature vector output by the feature vector generation model; or the first device using the task information as an input parameter of the feature vector generation model to obtain the feature vector output by the feature vector generation model. In this implementation, the first device generates the feature vector using the feature vector generation model. This allows the task information to be represented by the feature vector, and the second device does not need to translate or interpret the task information. Instead, it determines the model required by the first device based on the feature vector, which helps simplify the processing operations of the second device.

[0011] Based on the first aspect, in one possible implementation, the feature information includes a feature vector; the first device determines the feature information based on the task information of the first task and / or the auxiliary information of the first device, including: the first device inputs the task information and / or the auxiliary information into a feature vector generation model to obtain the feature vector output by the feature vector generation model; or the first device uses the task information and / or the auxiliary information as input parameters of the feature vector generation model to obtain the feature vector output by the feature vector generation model. In this implementation, the first device generates the feature vector through the feature vector generation model. This achieves the representation of the task information and auxiliary information through the feature vector, and the second device does not need to translate or interpret the task information and auxiliary information. Instead, it determines the model required by the first device through the feature vector, which helps to simplify the processing operations of the second device.

[0012] Based on the first aspect, in a possible implementation, the feature vector is a vector composed of a bit sequence.

[0013] Based on the first aspect, in one possible implementation, before obtaining the feature vector output by the feature vector generation model, the method further includes: the first device receiving second information from the second device, where the second information is used to indicate the feature vector generation model. Thus, it can be seen that the second device can configure the feature vector generation model for the first device, thereby facilitating the first device to generate the feature vector.

[0014] The second aspect of the present application provides a model indication method, which can be performed by a second device. The second device can be a terminal device or a network device, or a component in the terminal device or network device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can implement all or part of the terminal device functions, or a logic module or software that can implement all or part of the network device functions. The method includes:

[0015] The second device receives feature information from the first device, which characterizes the features of the first task. The second device generates a user-level model based on the public models and feature information at various levels related to the first task in the second device. The user-level model is used by the first device to perform the first task. The second device sends first information to the first device, which indicates the user-level model. This enables the first device to obtain the user-level model, or in other words, enables the second device to deploy the user-level model to the first device. This enables the first device to perform the first task based on the user-level model.

[0016] Based on the second aspect, in a possible implementation, the feature information includes a feature vector; the second device generates a user-level model based on the public models of various levels related to the first task and the feature information, including: the second device correlates the feature vector with the feature vectors corresponding to the public models of various levels related to the first task in the second device to obtain the correlation coefficients of the public models of various levels related to the first task; the second device weights and fuses the public models of various levels related to the first task according to the correlation coefficients of the public models of various levels related to the first task to obtain the user-level model. It can be seen that the second device determines a suitable user-level model for the first device based on the feature vector and the feature vectors corresponding to the public models of various levels related to the first task. This implements the deployment of a suitable model for the first device and improves the performance of the first device in executing the first task.

[0017] Based on the second aspect, in one possible implementation, the feature information includes a feature vector; and the method further includes: the second device sending second information to the first device, the second information being used to indicate a feature vector generation model, and the feature vector generation model being used to generate the feature vector. This enables the first device to generate the feature vector. This facilitates the second device to quickly determine an appropriate model for the first device based on the feature vector and deploy the model for the first device.

[0018] Based on the second aspect, in a possible implementation, the feature vector is a vector composed of a bit sequence.

[0019] Based on the second aspect, in one possible implementation, the characteristic information is generated based on the task information of the first task; alternatively, the characteristic information is generated based on the task information of the first task and / or auxiliary information of the first device. This implementation illustrates two possible methods for generating characteristic information. For example, the characteristic information is generated based on the task information of the first task, thereby indicating the characteristics of the first task to the second device.

[0020] Based on the second aspect, in one possible implementation, the auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device. In this implementation, the first device further generates feature information based on the auxiliary information to facilitate requesting an appropriate model from the second device.

[0021] Based on the second aspect, in one possible implementation, the task information includes at least one of the following: a task type or target performance of the first task. This facilitates the second device deploying an appropriate model for the first device, thereby improving the effect or performance achieved by the first device when executing the first task based on the model.

[0022] Based on the first or second aspect, in one possible implementation, the task type includes any of the following: channel state information (CSI) compression feedback, beam prediction, mobility management, data encryption, transceiver link, or model training. This implementation illustrates some possible task types, which facilitates implementation of the solution.

[0023] Based on the first aspect or the second aspect, in one possible implementation, each level public model related to the first task includes at least one of the following: a first-level public model or a second-level public model related to the first task. In this implementation, the level public model related to the first task includes multiple level public models, and the multiple level public models can be generated by devices at different levels. This enables a high-level device to send the corresponding level public model to a low-level device. This enables multi-level provision of AI services, which is conducive to better task execution.

[0024] Based on the first aspect or the second aspect, in a possible implementation method, the first-level public model, the second-level public model and the user-level model are divided according to the size of the execution area corresponding to the first task. It can be seen that different levels of public models can be divided according to the size of the area for executing the first task, which is conducive to the implementation of the solution. For example, the area where the first-level public model takes effect can be the largest, the area where the second-level public model takes effect can be medium, and the area where the second-level public model takes effect belongs to the area where the first-level public model takes effect. The area where the user-level public model takes effect is the smallest, and the area where the user-level public model takes effect belongs to the area where the second-level public model takes effect. Configuring multiple levels of public models for the first task is conducive to improving the execution performance of the first task.

[0025] Based on the first aspect or the second aspect, in a possible implementation, the first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task adapted to terminal devices of different capability levels, or a sector-level public model related to the first task adapted to terminal devices of different capability levels. In this implementation, a possible implementation of the first-level public model and the second-level public model is shown. The first-level public model can be used for users at the regional level to perform the first task. The second-level public model can be used for users at the cell level or sector level to perform the first task.

[0026] Based on the first aspect or the second aspect, in a possible implementation, the first-level public model, the second-level public model and the user-level model are divided according to the master-slave relationship between the tasks that the level public models are used to perform. Another possible hierarchical method of the first-level public model, the second-level public model and the user-level model is shown, which is conducive to the implementation of the solution. That is, the multi-level public models are divided from the perspective of the relationship between tasks. For example, the task that the second-level public model is used to perform is a slave task of the task that the first-level public model is used to perform. The task that the user-level model is used to perform is a slave task of the task that the second-level public model is used to perform. Thereby, the corresponding level public models are subdivided according to the execution precision of the task. It is beneficial to configure a suitable model for the first device and improve the performance achieved by the first device in performing the first task.

[0027] Based on the second aspect, in a possible implementation, the method further includes: the third device sends third information to the second device, and the third information is used to indicate the first-level public model; accordingly, the second device receives the third information from the third device; the second device adjusts the first-level public model to obtain the second-level public model. In this implementation, the third device can be understood as a device of a higher level, and the second device can be understood as a device of a lower level. The third device sends the first-level public model (which can be understood as a high-level public model) to the second device. This facilitates the second device to generate the second-level public model (which can be understood as an intermediate public model). This enables multi-level provision of AI services. It should be noted that this implementation can be used as an independent aspect, that is, this implementation can be independent of the second aspect or any implementation shown in the second aspect.

[0028] The third aspect of the present application provides a model generation method, which can be performed by a first device, which can be a terminal device or a network device, or a component in the terminal device or network device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can implement all or part of the terminal device functions, or a logic module or software that can implement all or part of the network device functions. The method includes:

[0029] A first device receives first information from a second device, the first information indicating information about a level-common model related to a first task. The first device generates a user-level model based on the first information, and the user-level model is used by the first device to perform the first task. This enables the first device to generate the user-level model based on the first information and perform the first task based on the user-level model.

[0030] Based on the third aspect, in one possible implementation, the first information includes at least one of the following: reference information or adjustment information. The reference information indicates information related to the level common model used by the first device to generate the user-level model, and the adjustment information indicates information related to adjustments to the level common model. This implementation illustrates a possible implementation of the content included in the first information. This facilitates the first device to determine the user-level model by combining the reference information and the adjustment information. This enables the first device to obtain the user-level model and perform the first task using the user-level model.

[0031] Based on the third aspect, in a possible implementation method, the first device generates a user-level model based on the first information, including: the first device determines the level public model based on the reference information; the first device adjusts the level public model according to the adjustment information to obtain the user-level model.

[0032] Based on the third aspect, in one possible implementation, the reference information includes at least one of the following: region information, task information of the first task, first indication information, or second indication information; wherein the region information indicates the region in which the first device reuses the level common model, the first indication information indicates the level common model, and the second indication information indicates the reuse method of the level common model. In this implementation, the content included in the reference information is shown, thereby facilitating the first device to determine the level common model.

[0033] Based on the third aspect, in one possible implementation, the adjustment information includes at least one of the following: adjustment information regarding model parameters of the level-common model, adjustment information regarding the model structure of the level-common model, or adjustment information regarding the quantization accuracy of model parameters of the level-common model. This enables the first device to adjust the level-common model based on the adjustment information to obtain a user-level model. This enables the first device to obtain the user-level model and perform the first task using the user-level model.

[0034] Based on the third aspect, in a possible implementation, the method also includes: the first device generates a first feature vector, the first feature vector represents the characteristics of the first task; the first device correlates the first feature vector with the second feature vector to obtain the correlation coefficient between the first feature vector and the second feature vector, the second feature vector represents the characteristics of the level public model; the first device generates a user-level model based on the first information, including: when the correlation coefficient between the first feature vector and the second feature vector falls within the threshold range of the correlation coefficient, the first device generates a user-level model based on the first information and the third indication information, the third indication information is used to indicate the reuse method of the level public model. The first device determines the user-level model in combination with the first information. Thereby, the first device obtains the user-level model and performs the first task through the user-level model.

[0035] Based on the third aspect, in one possible implementation, the first information includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information; or the method further includes: the first device receiving second information from a second device, the second information including at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information, thereby assisting the first device in generating a user-level model.

[0036] Based on the third aspect, in one possible implementation, the first device generates the first feature vector, including: the first device generating the first feature vector based on task information of the first task; or the first device generating the first feature vector based on the task information and / or auxiliary information. The first feature vector thus represents the task characteristics of the first task. This facilitates the first device comparing the first feature vector with the second feature vector to determine whether to reuse the level common model indicated by the first information.

[0037] Based on the third aspect, in one possible implementation, the auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device. In this implementation, the first device further generates a first feature vector based on the auxiliary information to facilitate the first device's determination of whether to reuse the level-common model indicated by the first information.

[0038] Based on the third aspect, in a possible implementation, the first eigenvector is a vector composed of a bit sequence.

[0039] Based on the third aspect, in one possible implementation, the method further includes: the first device receiving a second feature vector from the second device. This facilitates the first device determining whether to reuse the level common model indicated by the first information by combining the first feature vector and the second feature vector. This enables the first device to determine the user-level model.

[0040] Based on the third aspect, in a possible implementation manner, the first information is carried in a system message or a user-specific message.

[0041] Based on the third aspect, in a possible implementation, the first information includes model parameters of the level public model and / or the model structure of the level public model, thereby facilitating the first device to obtain the level public model.

[0042] Based on the third aspect, in one possible implementation, the reuse method includes any of the following: using the level-common model as the user-level model; using the model structure of the level-common model as the model structure of the user-level model; using the model parameters of the level-common model as the model parameters of the user-level model; or using the level-common model as a teacher to train the user-level model. Multiple possible implementations of the reuse method are shown, thereby enabling the first device to determine the user-level model based on the reuse method indicated by the first information.

[0043] A fourth aspect of the present application provides a model generation method, which can be performed by a second device, which can be a terminal device or a network device, or a component in the terminal device or network device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can implement all or part of the terminal device functions, or a logic module or software that can implement all or part of the network device functions. The method includes:

[0044] The second device determines first information indicating information about a level-common model related to a first task. The second device sends the first information to the first device, and the first information is used by the first device to generate a user-level model. The user-level model is used by the first device to perform the first task. This facilitates the first device generating the user-level model based on the first information. This facilitates the first device performing the first task based on the user-level model.

[0045] Based on the fourth aspect, in one possible implementation, the first information includes at least one of the following: reference information or adjustment information, where the reference information indicates information related to the level common model used by the first device to generate the user-level model, and the adjustment information indicates information related to adjustments to the level common model. This implementation illustrates a possible implementation of the content included in the first information. This facilitates the first device determining the user-level model by combining the reference information and the adjustment information. This enables the first device to obtain the user-level model and perform the first task using the user-level model.

[0046] Based on the fourth aspect, in one possible implementation, the reference information includes at least one of the following: region information, task information of the first task, first indication information, or second indication information; wherein the region information indicates the region in which the first device reuses the level common model, the first indication information indicates the level common model, and the second indication information indicates the reuse method of the level common model. In this implementation, the content included in the reference information is shown. This facilitates the first device to determine the level common model.

[0047] Based on the fourth aspect, in one possible implementation, the adjustment information includes at least one of the following: adjustment information for model parameters of the level-common model, adjustment information for the model structure of the level-common model, or adjustment information for the quantization accuracy of the model parameters of the level-common model. This enables the first device to adjust the level-common model based on the adjustment information to obtain a user-level model. This enables the first device to obtain the user-level model and perform the first task using the user-level model.

[0048] Based on the fourth aspect, in a possible implementation, the first information also includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information, and the third indication information is used to indicate a reuse method of the level public model; or, the method further includes: the second device sends second information to the first device, and the second information includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information. In this implementation, the content that the first information also includes is shown. This facilitates the first device to determine the user-level model in combination with the first information. Alternatively, the second device sends the second information to the first device. This enables the first device to obtain the user-level model and perform the first task through the user-level model.

[0049] Based on the fourth aspect, in one possible implementation, the reuse method includes any of the following: using the level-common model as the user-level model; using the model structure of the level-common model as the model structure of the user-level model; using the model parameters of the level-common model as the model parameters of the user-level model; or using the level-common model as a teacher to train the user-level model. Multiple possible implementations of the reuse method are shown, thereby enabling the first device to determine the user-level model based on the reuse method indicated by the first information.

[0050] Based on the fourth aspect, in one possible implementation, the method further includes: the second device sending a second feature vector to the first device, where the second feature vector is used to characterize a feature of the level-common model. This facilitates the first device determining whether to reuse the level-common model indicated by the first information by combining the first feature vector and the second feature vector, thereby enabling the first device to determine the user-level model.

[0051] Based on the fourth aspect, in a possible implementation manner, the first information is carried in a system message or a user-specific message.

[0052] Based on the fourth aspect, in a possible implementation, the first information includes model parameters of the level public model and / or the model structure of the level public model, so that the first device obtains the level public model.

[0053] Based on the third or fourth aspect, in one possible implementation, each level of public models associated with the first task includes at least one of the following: a first-level public model or a second-level public model associated with the first task. In this implementation, the level of public models associated with the first task includes multiple level public models, and the multiple level public models can be generated by devices at different levels. This enables a high-level device to send the corresponding level public model to a low-level device. This enables multi-level provision of AI services, which facilitates better task execution.

[0054] Based on the third aspect or the fourth aspect, in a possible implementation method, the first-level public model, the second-level public model and the user-level model are divided according to the size of the execution area corresponding to the first task. It can be seen that different levels of public models can be divided according to the size of the area for executing the first task, which is conducive to the implementation of the solution. For example, the area where the first-level public model takes effect can be the largest, the area where the second-level public model takes effect can be medium, and the area where the second-level public model takes effect belongs to the area where the first-level public model takes effect. The area where the user-level public model takes effect is the smallest, and the area where the user-level public model takes effect belongs to the area where the second-level public model takes effect. Configuring multiple levels of public models for the first task is conducive to improving the execution performance of the first task.

[0055] Based on the third aspect or the fourth aspect, in a possible implementation, the first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task adapted to terminal devices of different capability levels, or a sector-level public model related to the first task adapted to terminal devices of different capability levels. In this implementation, a possible implementation of the first-level public model and the second-level public model is shown. The first-level public model can be used for users within a regional level to perform the first task. The second-level public model can be used for users within a cell level or a sector level to perform the first task.

[0056] Based on the third aspect or the fourth aspect, in a possible implementation, the first-level public model, the second-level public model and the user-level model are divided according to the master-slave relationship between the tasks that the level public models are used to perform. Another possible hierarchical method of the first-level public model, the second-level public model and the user-level model is shown, which is conducive to the implementation of the solution. That is, the multi-level public models are divided from the perspective of the relationship between tasks. For example, the task that the second-level public model is used to perform is a slave task of the task that the first-level public model is used to perform. The task that the user-level model is used to perform is a slave task of the task that the second-level public model is used to perform. Thereby, the corresponding level public models are subdivided according to the execution precision of the task. It is beneficial to configure a suitable model for the first device and improve the performance achieved by the first device in performing the first task.

[0057] Based on the fourth aspect, in a possible implementation, the level public model related to the first task includes a second-level public model; the method further includes: a third device sends a third information to the second device, and the third information is used to indicate the first-level public model; accordingly, the second device receives the third information from the third device; the second device adjusts the first-level public model to obtain a second-level public model. In this implementation, the third device can be understood as a device of a higher level, and the second device can be understood as a device of a lower level. The third device sends a first-level public model (which can be understood as a high-level public model) to the second device. This facilitates the second device to generate a second-level public model (which can be understood as an intermediate public model). This enables multi-level provision of AI services. It should be noted that this implementation can be an independent aspect, that is, this implementation can be independent of the fourth aspect or any of the implementations shown in the fourth aspect.

[0058] A fifth aspect of the present application provides a model indication method, which can be performed by a first device, which can be a terminal device or a network device, or a component in the terminal device or network device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can implement all or part of the terminal device functions, or a logic module or software that can implement all or part of the network device functions. The method includes:

[0059] A first device sends a first request to a second device, the first request being for a target-level public model. The first device then receives first information from the second device, the first information being for indicating the target-level public model. This allows the second device to deploy an appropriate model for the first device, enabling the first device to perform a first task based on the target-level public model.

[0060] Based on the fifth aspect, in a possible implementation manner, the first request includes an identifier of a target-level public model.

[0061] Based on the fifth aspect, in a possible implementation, the identifier of the target-level public model includes at least one of the following: a model provider identifier, a service identifier, a region identifier, or an index of the target-level public model in the region indicated by the region identifier; wherein the model provider identifier is used to indicate the provider of the target-level public model, the service identifier is used to indicate the first task, the target-level public model is used to perform the first task, and the region identifier is used to indicate the region in which the target-level public model is effective. In this implementation, a possible format for the identifier of the target-level public model is shown. Thus, the identifier of the target-level public model indicates information such as the task to which the target-level public model is applicable, the region in which it is effective, and the provider of the target-level public model. This facilitates the first device to request a suitable level public model.

[0062] Based on the fifth aspect, in one possible implementation, the first request also includes task information for the first task, and the target-level public model is used by the first device to perform the first task. The first task is a task to be performed by the first device. This facilitates the second device indicating an appropriate model to the first device. Specifically, this may include the model structure and model parameters.

[0063] Based on the fifth aspect, in one possible implementation, the method further includes: the first device receiving level public model library information from the second device, the level public model library information including the identifier of at least one level public model, the at least one level public model including a target level public model. This enables the first device to obtain the identifier of the at least one level public model and select the identifier of a level public model therefrom to request the second device to indicate the target level public model to the first device. This facilitates the first device to obtain the target level public model and perform the first task based on the target level public model.

[0064] Based on the fifth aspect, in one possible implementation, the level-level public model library information also includes a performance indicator of at least one level-level public model. This facilitates the first device to learn the performance of each level-level public model, facilitating the first device to request an appropriate level-level public model. This improves the performance of the first device in executing the first task.

[0065] A sixth aspect of the present application provides a model indication method, which can be performed by a second device, which can be a terminal device or a network device, or a component in the terminal device or network device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can implement all or part of the terminal device functions, or a logic module or software that can implement all or part of the network device functions. The method includes:

[0066] The second device receives a first request from the first device, the first request being for a target-level public model; and sends first information to the first device, the first information being for indicating the target-level public model. This enables the second device to deploy an appropriate model for the first device, enabling the first device to perform a first task based on the target-level public model.

[0067] Based on the sixth aspect, in a possible implementation manner, the first request includes an identifier of a target-level public model.

[0068] Based on the sixth aspect, in a possible implementation, the identifier of the target-level public model includes at least one of the following: a model provider identifier, a service identifier, a region identifier, or an index of the target-level public model in the region indicated by the region identifier; wherein the model provider identifier is used to indicate the provider of the target-level public model, the service identifier is used to indicate the first task, the target-level public model is used by the first device to perform the first task, and the region identifier is used to indicate the region in which the target-level public model is effective. In this implementation, a possible format for the identifier of the target-level public model is shown. Thus, the identifier of the target-level public model indicates information such as the task to which the target-level public model is applicable, the region in which it is effective, and the provider of the target-level public model. This facilitates the first device to request a suitable level public model.

[0069] Based on the sixth aspect, in one possible implementation, the first request also includes task information for a first task, and the target-level public model is used by the first device to perform the first task. The first task is a task to be performed by the first device. This facilitates the second device indicating an appropriate model to the first device. Specifically, this may include the model structure and model parameters.

[0070] Based on the sixth aspect, in one possible implementation, the method further includes: the second device sending level public model library information to the first device, the level public model library information including the identifier of at least one level public model, the at least one level public model including a target level public model. This enables the first device to obtain the identifier of the at least one level public model and select the identifier of a level public model therefrom to request the second device to indicate the target level public model to the first device. This facilitates the first device to obtain the target level public model and perform the first task based on the target level public model.

[0071] Based on the sixth aspect, in one possible implementation, the level-level public model library information further includes a performance indicator of at least one level-level public model. This facilitates the first device to learn the performance of each level-level public model, facilitating the first device to request an appropriate level-level public model. This improves the performance of the first device in executing the first task.

[0072] Based on the fifth or sixth aspect, in one possible implementation, the target-level public model is a first-level public model or a second-level public model. For example, the target-level public model is a first-level public model or a second-level public model related to the first task. Multiple level public models can be generated by devices at different levels. This enables a high-level device to send a corresponding level public model to a low-level device. This enables multi-level AI service provision, facilitating better task execution.

[0073] Based on the fifth aspect or the sixth aspect, in a possible implementation method, the first-level public model and the second-level public model are divided according to the size of the execution area corresponding to the first task. It can be seen that different levels of public models can be divided according to the size of the area for executing the first task, which is conducive to the implementation of the solution. For example, the area where the first-level public model takes effect can be the largest, the area where the second-level public model takes effect can be medium, and the area where the second-level public model takes effect belongs to the area where the first-level public model takes effect. Configuring multiple levels of public models for the first task is conducive to improving the execution performance of the first task.

[0074] Based on the fifth aspect or the sixth aspect, in a possible implementation, the first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task adapted to terminal devices of different capability levels, or a sector-level public model related to the first task adapted to terminal devices of different capability levels. In this implementation, a possible implementation of the first-level public model and the second-level public model is shown. The first-level public model can be used for users within a regional level to perform the first task. The second-level public model can be used for users within a cell level or a sector level to perform the first task.

[0075] Based on the fifth aspect or the sixth aspect, in a possible implementation, the first-level public model, the second-level public model and the user-level model are divided according to the master-slave relationship between the tasks that the level public model is used to perform. Another possible hierarchical method of the first-level public model and the second-level public model is shown, which is conducive to the implementation of the solution. That is, the multi-level public models are divided from the perspective of the relationship between the tasks. For example, the task that the second-level public model is used to perform is a slave task of the task that the first-level public model is used to perform. Thereby, the corresponding level public model is subdivided according to the execution precision of the task. It is beneficial to configure a suitable model for the first device and improve the performance achieved by the first device in performing the first task.

[0076] Based on the sixth aspect, in a possible implementation, the target level public model is a second level public model; the method further includes: a third device sends a second information to a second device, and the second information is used to indicate a first level public model; accordingly, the second device receives the second information from the third device, and the first device adjusts the first level public model to obtain the second level public model. In this implementation, the third device can be understood as a device of a higher level, and the second device can be understood as a device of a lower level. The third device sends a first level public model (which can be understood as a high-level public model) to the second device. This facilitates the second device to generate a second level public model (which can be understood as an intermediate public model). This enables multi-level provision of AI services. It should be noted that this implementation can be used as an independent aspect, that is, this implementation can be independent of the sixth aspect or any of the implementations shown in the sixth aspect.

[0077] A seventh aspect of the present application provides a first device, including:

[0078] A transceiver module is used to send characteristic information to the second device, where the characteristic information represents the characteristics of the first task; and to receive first information from the second device, where the first information is used to indicate a user-level model. The user-level model is determined based on the characteristic information and various levels of public models related to the first task in the second device, and the user-level model is used by the first device to perform the first task.

[0079] Based on the seventh aspect, in a possible implementation, the first device further includes a processing module;

[0080] The processing module is configured to determine the characteristic information according to the task information of the first task; or determine the characteristic information according to the task information of the first task and / or the auxiliary information of the first device.

[0081] Based on the seventh aspect, in a possible implementation manner, the auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device.

[0082] Based on the seventh aspect, in one possible implementation method, the feature information includes a feature vector; the processing module is specifically used to: input the task information into the feature vector generation model to obtain the feature vector output by the feature vector generation model; or, use the task information as an input parameter of the feature vector generation model to obtain the feature vector output by the feature vector generation model.

[0083] Based on the seventh aspect, in one possible implementation method, the feature information includes a feature vector; the processing module is specifically used to: input the task information and / or auxiliary information into the feature vector generation model to obtain the feature vector output by the feature vector generation model; or, use the task information and / or auxiliary information as input parameters of the feature vector generation model to obtain the feature vector output by the feature vector generation model.

[0084] Based on the seventh aspect, in a possible implementation, the feature vector is a vector composed of a bit sequence.

[0085] Based on the seventh aspect, in a possible implementation, the transceiver module is further used to: receive second information from a second device, where the second information is used to indicate a feature vector generation model.

[0086] An eighth aspect of the present application provides a second device, including:

[0087] a transceiver module, configured to receive characteristic information from the first device, where the characteristic information represents characteristics of the first task;

[0088] a processing module, configured to generate a user-level model based on public models of various levels and feature information related to the first task in the second device, wherein the user-level model is used by the first device to perform the first task;

[0089] The transceiver module is further configured to send first information to the first device, where the first information is used to indicate the user-level model.

[0090] Based on the eighth aspect, in a possible implementation method, the feature information includes a feature vector; the processing module is specifically used to: correlate the feature vector with the feature vector corresponding to the public models of each level related to the first task in the second device to obtain the correlation coefficients of the public models of each level related to the first task; weight and fuse the public models of each level related to the first task according to the correlation coefficients of the public models of each level related to the first task to obtain a user-level model.

[0091] Based on the eighth aspect, in a possible implementation, the feature information includes a feature vector; the transceiver module is further used to: send second information to the first device, the second information is used to indicate a feature vector generation model, and the feature vector generation model is used to generate the feature vector.

[0092] Based on the eighth aspect, in a possible implementation, the feature vector is a vector composed of a bit sequence.

[0093] Based on the eighth aspect, in a possible implementation manner, the feature information is generated based on the task information of the first task; or, the feature information is generated based on the task information of the first task and / or the auxiliary information of the first device.

[0094] Based on the eighth aspect, in a possible implementation manner, the auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device.

[0095] Based on the eighth aspect, in a possible implementation manner, the task information includes at least one of the following: a task type or target performance of the first task.

[0096] Based on the seventh aspect or the eighth aspect, in a possible implementation manner, the task information includes at least one of the following: a task type or target performance of the first task.

[0097] Based on the seventh aspect or the eighth aspect, in a possible implementation method, the task type includes any one of the following: CSI compression feedback, beam prediction, mobility management, data encryption, transceiver link, or model training.

[0098] Based on the seventh aspect or the eighth aspect, in a possible implementation, each level of public model related to the first task includes at least one of the following: a first-level public model related to the first task, or a second-level public model.

[0099] Based on the seventh aspect or the eighth aspect, in a possible implementation, the first-level public model, the second-level public model, and the user-level model are divided according to the size of the execution area corresponding to the first task.

[0100] Based on the seventh aspect or the eighth aspect, in a possible implementation method, the first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task and adapted to terminal devices of different capability levels, or a sector-level public model related to the first task and adapted to terminal devices of different capability levels.

[0101] Based on the seventh aspect or the eighth aspect, in a possible implementation, the first-level public model, the second-level public model, and the user-level model are divided in a master-slave relationship between tasks that the level public models are used to perform.

[0102] Based on the eighth aspect, in a possible implementation method, the transceiver module is also used to: the method also includes: receiving third information from a third device, the third information is used to indicate the first-level public model; the processing module is also used to: adjust the first-level public model to obtain a second-level public model.

[0103] A ninth aspect of the present application provides a first device, including:

[0104] a transceiver module, configured to receive first information from a second device, wherein the first information is used to indicate information of a level public model related to a first task;

[0105] The processing module is used to generate a user-level model according to the first information, and the user-level model is used for the first device to perform the first task.

[0106] Based on the ninth aspect, in a possible implementation method, the first information includes at least one of the following: reference information, or adjustment information, the reference information is used to indicate the relevant information of the level public model referenced by the first device to generate the user-level model, and the adjustment information is used to indicate the relevant information for adjusting the level public model.

[0107] Based on the ninth aspect, in a possible implementation, the processing module is specifically used to: determine the level public model according to the reference information; and adjust the level public model according to the adjustment information to obtain the user-level model.

[0108] Based on the ninth aspect, in a possible implementation method, the reference information includes at least one of the following: area information, task information of the first task, first indication information, or second indication information; wherein, the area information is used to indicate the area where the first device reuse level public model is effective, the first indication information is used to indicate the level public model, and the second indication information is used to indicate the reuse method of the level public model.

[0109] Based on the ninth aspect, in a possible implementation, the adjustment information includes at least one of the following: adjustment information of the model parameters of the level common model, adjustment information of the model structure of the level common model, or adjustment information of the quantization accuracy of the model parameters of the level common model.

[0110] Based on the ninth aspect, in a possible implementation method, the processing module is also used to: generate a first eigenvector, the first eigenvector represents the characteristics of the first task; correlate the first eigenvector with the second eigenvector to obtain the correlation coefficient between the first eigenvector and the second eigenvector, the second eigenvector represents the characteristics of the level public model; the processing module is specifically used to: when the correlation coefficient between the first eigenvector and the second eigenvector falls within the threshold range of the correlation coefficient, generate a user-level model according to the first information and the third indication information, and the third indication information is used to indicate the reuse method of the level public model.

[0111] Based on the ninth aspect, in a possible implementation manner, the first information further includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information; or,

[0112] The transceiver module is further configured to receive second information from a second device, where the second information includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information.

[0113] Based on the ninth aspect, in a possible implementation, the processing module is specifically used to: generate a first feature vector based on task information of the first task; or generate a first feature vector based on task information of the first task and / or auxiliary information of the first device.

[0114] Based on the ninth aspect, in a possible implementation, the auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device.

[0115] Based on the ninth aspect, in a possible implementation, the first eigenvector is a vector composed of a bit sequence.

[0116] Based on the ninth aspect, in a possible implementation, the transceiver module is further used to: receive a second feature vector from a second device.

[0117] Based on the ninth aspect, in a possible implementation method, the first information is carried in a system message or a user-specific message.

[0118] Based on the ninth aspect, in a possible implementation manner, the first information includes model parameters of the level common model and / or the model structure of the level common model.

[0119] Based on the ninth aspect, in one possible implementation method, the reuse method includes any one of the following: using the level public model as the user-level model; using the model structure of the level public model as the model structure of the user-level model; using the model parameters of the level public model as the model parameters of the user-level model; or using the level public model as a teacher to train the user-level model.

[0120] A tenth aspect of the present application provides a second device, including:

[0121] a processing module, configured to determine first information, the first information indicating information of a level public model related to a first task;

[0122] The transceiver module is used to send first information to the first device, the first information is used by the first device to generate a user-level model, and the user-level model is used by the first device to perform a first task.

[0123] Based on the tenth aspect, in a possible implementation method, the first information includes at least one of the following: reference information, or adjustment information, the reference information is used to indicate the relevant information of the level public model referenced by the first device to generate the user-level model, and the adjustment information is used to indicate the relevant information for adjusting the level public model.

[0124] Based on the tenth aspect, in a possible implementation method, the reference information includes at least one of the following: area information, task information of the first task, first indication information, or second indication information; wherein, the area information is used to indicate the area where the first device reuse level public model is effective, the first indication information is used to indicate the level public model, and the second indication information is used to indicate the reuse method of the level public model.

[0125] Based on the tenth aspect, in a possible implementation, the adjustment information includes at least one of the following: adjustment information of the model parameters of the level common model, adjustment information of the model structure of the level common model, or adjustment information of the quantization accuracy of the model parameters of the level common model.

[0126] Based on the tenth aspect, in a possible implementation, the first information includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information, and the third indication information is used to indicate a reuse method of the level public model.

[0127] Based on the tenth aspect, in a possible implementation method, the reuse method includes any one of the following: using the level public model as the user-level model; using the model structure of the level public model as the model structure of the user-level model; using the model parameters of the level public model as the model parameters of the user-level model; or using the level public model as a teacher to train the user-level model.

[0128] Based on the tenth aspect, in a possible implementation, the transceiver module is further used to: send a second feature vector to the first device, where the second feature vector is used to characterize features of the level public model.

[0129] Based on the tenth aspect, in a possible implementation manner, the first information is carried in a system message or a user-specific message.

[0130] Based on the tenth aspect, in a possible implementation manner, the first information includes model parameters of the level common model and / or the model structure of the level common model.

[0131] Based on the ninth aspect or the tenth aspect, in a possible implementation, each level of public model related to the first task includes at least one of the following: a first-level public model related to the first task, or a second-level public model.

[0132] Based on the ninth aspect or the tenth aspect, in a possible implementation, the first-level public model, the second-level public model, and the user-level model are divided according to the size of the execution area corresponding to the first task.

[0133] Based on the ninth aspect or the tenth aspect, in a possible implementation method, the first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task and adapted to terminal devices of different capability levels, or a sector-level public model related to the first task and adapted to terminal devices of different capability levels.

[0134] Based on the ninth aspect or the tenth aspect, in a possible implementation, the first-level public model, the second-level public model, and the user-level model are divided in a master-slave relationship between tasks that the level public models are used to perform.

[0135] Based on the tenth aspect, in a possible implementation method, the level public model related to the first task includes a second-level public model; the transceiver module is also used to: receive third information from a third device, the third information is used to indicate the first-level public model; adjust the first-level public model to obtain the second-level public model.

[0136] In an eleventh aspect, the present application provides a first device, including:

[0137] The transceiver module is configured to send a first request to the second device, the first request being used to request a target-level public model; and receive first information from the second device, the first information being used to indicate the target-level public model.

[0138] Based on the eleventh aspect, in a possible implementation manner, the first request includes an identifier of a target-level public model.

[0139] Based on the eleventh aspect, in a possible implementation method, the identifier of the target-level public model includes at least one of the following: a model provider identifier, a service identifier, an area identifier, or an index of the target-level public model in the area indicated by the area identifier; wherein, the model provider identifier is used to indicate the provider of the target-level public model, the service identifier is used to indicate the first task, the target-level public model is used to perform the first task, and the area identifier is used to indicate the area where the target-level public model is effective.

[0140] Based on the eleventh aspect, in a possible implementation, the first request further includes task information of the first task, and the target-level public model is used by the first device to perform the first task.

[0141] Based on the eleventh aspect, in a possible implementation, the transceiver module is further used to: receive level public model library information from the second device, the level public model library information includes an identifier of at least one level public model, and the at least one level public model includes a target level public model.

[0142] Based on the eleventh aspect, in a possible implementation manner, the level public model library information further includes a performance indicator of at least one level public model.

[0143] A twelfth aspect of the present application provides a second device, including:

[0144] The transceiver module is configured to receive a first request from a first device, the first request being used to request a target-level public model; and send first information to the first device, the first information being used to indicate the target-level public model.

[0145] Based on the twelfth aspect, in a possible implementation manner, the first request includes an identifier of a target-level public model.

[0146] Based on the twelfth aspect, in a possible implementation method, the identifier of the target-level public model includes at least one of the following: a model provider identifier, a service identifier, an area identifier, or an index of the target-level public model in the area indicated by the area identifier; wherein, the model provider identifier is used to indicate the provider of the target-level public model, the service identifier is used to indicate the first task, the target-level public model is used by the first device to perform the first task, and the area identifier is used to indicate the area where the target-level public model is effective.

[0147] Based on the twelfth aspect, in a possible implementation, the first request also includes task information of the first task, and the target-level public model is used by the first device to perform the first task.

[0148] Based on the twelfth aspect, in a possible implementation, the transceiver module is further used to: send level public model library information to the first device, the level public model library information includes an identifier of at least one level public model, and the at least one level public model includes a target level public model.

[0149] Based on the twelfth aspect, in a possible implementation manner, the level public model library information also includes a performance indicator of at least one level public model.

[0150] Based on the eleventh aspect or the twelfth aspect, in a possible implementation manner, the target-level public model is a first-level public model or a second-level public model.

[0151] Based on the eleventh aspect or the twelfth aspect, in a possible implementation manner, the first-level public model and the second-level public model are divided according to the size of the execution area corresponding to the first task.

[0152] Based on the eleventh aspect or the twelfth aspect, in a possible implementation method, the first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task and adapted to terminal devices of different capability levels, or a sector-level public model related to the first task and adapted to terminal devices of different capability levels.

[0153] Based on the eleventh aspect or the twelfth aspect, in a possible implementation, the first-level public model, the second-level public model, and the user-level model are divided in a master-slave relationship between tasks that the level public models are used to perform.

[0154] Based on the twelfth aspect, in one possible implementation method, the target level public model is a second level public model; the transceiver module is also used to: receive second information from a third device, the second information is used to indicate the first level public model; the processing module is also used to: adjust the first level public model to obtain a second level public model.

[0155] In a thirteenth aspect, the present application provides a first apparatus, comprising: a processor and a memory. The memory stores a computer program or computer instructions, and the processor is configured to call and execute the computer program or computer instructions stored in the memory, so that the processor implements any one of the implementations of the first, third, or fifth aspects.

[0156] Optionally, the first device further includes a transceiver, and the processor is used to control the transceiver to transmit and receive signals.

[0157] In a fourteenth aspect, the present application provides a second apparatus, comprising: a processor and a memory. The memory stores a computer program or computer instructions, and the processor is configured to call and execute the computer program or computer instructions stored in the memory, so that the processor implements any one of the implementations of the second, fourth, or sixth aspects.

[0158] Optionally, the second device further includes a transceiver, and the processor is used to control the transceiver to transmit and receive signals.

[0159] In a fifteenth aspect, the present application provides a first device, comprising a processor and an interface circuit, wherein the processor is configured to communicate with other devices via the interface circuit and execute the method described in the first, third, or fifth aspects. The processor comprises one or more processors.

[0160] In a sixteenth aspect, the present application provides a second device, comprising a processor and an interface circuit, wherein the processor is configured to communicate with other devices via the interface circuit and execute the method described in the second, fourth, or sixth aspect. The processor may comprise one or more processors.

[0161] In a seventeenth aspect, the present application provides a first device, comprising a processor, configured to be connected to a memory and configured to call a program stored in the memory to execute the method described in the first, third, or fifth aspects above. The memory may be located within or outside the first device. The processor may include one or more processors.

[0162] In an eighteenth aspect, the present application provides a second device, comprising a processor, connected to a memory, configured to invoke a program stored in the memory to execute the method described in the second, fourth, or sixth aspects. The memory may be located within or outside the second device. The processor may include one or more processors.

[0163] In one implementation, the first device of the first aspect, third aspect, fifth aspect, seventh aspect, ninth aspect, and eleventh aspect may be a chip or a chip system.

[0164] In one implementation, the second device of the second aspect, fourth aspect, sixth aspect, eighth aspect, tenth aspect, and twelfth aspect may be a chip or a chip system.

[0165] In a nineteenth aspect, the present application provides a computer program product comprising computer instructions, characterized in that when the computer program product is run on a computer, the computer is enabled to execute any one of the implementation methods of any one of the first to sixth aspects.

[0166] The twentieth aspect of the present application provides a computer-readable storage medium comprising computer instructions, which, when executed on a computer, enables the computer to execute any one of the implementation methods of any one of the first to sixth aspects.

[0167] In aspect 21 of the present application, a chip device is provided, comprising a processor for calling a computer program or computer instruction in a memory so that the processor executes any one of the implementation methods of any one of the above-mentioned aspects 1 to 6.

[0168] Optionally, the processor is coupled to the memory via an interface.

[0169] In aspect 22 of the present application, a communication system is provided, which includes a first device and a second device; the first device is used to execute the method shown in aspect 1, aspect 3 or aspect 5, and the second device is used to execute the method shown in aspect 2, aspect 4 or aspect 6.

[0170] As can be seen from the above technical solution, the first device sends feature information to the second device. This feature information characterizes the features of the first task. Then, the first device receives the first information from the second device. The first information is used to indicate a user-level model, which is determined based on the feature information and the various levels of public models related to the first task in the second device. The user-level model is used by the first device to perform the first task. This enables the first device to obtain the user-level model, or in other words, enables the second device to deploy the user-level model for the first device. This enables the first device to perform the first task based on the user-level model. BRIEF DESCRIPTION OF THE DRAWINGS

[0171] FIG1 is a schematic diagram of a communication system according to an embodiment of the present application;

[0172] FIG2A is another schematic diagram of a communication system according to an embodiment of the present application;

[0173] FIG2B is another schematic diagram of a communication system according to an embodiment of the present application;

[0174] FIG2C is another schematic diagram of a communication system according to an embodiment of the present application;

[0175] FIG3 is a schematic diagram of an embodiment of the model indication method of the present application;

[0176] FIG4 is a schematic diagram of an embodiment of a model generation method according to an embodiment of the present application;

[0177] FIG5 is a schematic diagram of another embodiment of the model indication method according to an embodiment of the present application;

[0178] FIG6 is a schematic diagram of the identification of the target-level public model according to an embodiment of the present application;

[0179] FIG7 is a schematic structural diagram of the first device according to an embodiment of the present application;

[0180] FIG8 is another structural diagram of the first device according to an embodiment of the present application;

[0181] FIG9 is a schematic structural diagram of a second device according to an embodiment of the present application;

[0182] FIG10 is another schematic structural diagram of the second device according to an embodiment of the present application;

[0183] FIG11 is a schematic structural diagram of a device according to an embodiment of the present application;

[0184] FIG12 is a schematic structural diagram of a terminal device according to an embodiment of the present application;

[0185] FIG13 is a schematic structural diagram of a network device according to an embodiment of the present application. DETAILED DESCRIPTION

[0186] The embodiments of the present application provide a model indication method, a model generation method, and related devices for enabling a first device to obtain the user-level model, or enabling a second device to deploy the user-level model for the first device, thereby enabling the first device to perform a first task based on the user-level model.

[0187] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0188] References to "one embodiment" or "some embodiments" in this application mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in 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 "including," "comprising," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0189] In the description of this application, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" in this article is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, "at least one" means one or more, and "plurality" means two or more. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can mean: a, b, c; a and b; a and c; b and c; or a, b, and c. Among them, a, b, and c can be single or multiple.

[0190] It is understood that in this application, "indication" can include direct indication, indirect indication, explicit indication, and implicit indication. When describing that a certain indication information is used to indicate A, it can be understood that the indication information carries A, directly indicates A, or indirectly indicates A.

[0191] In this application, the information indicated by the indication information is referred to as the information to be indicated. In the specific implementation process, there are many ways to indicate the information to be indicated, such as but not limited to, the information to be indicated can be directly indicated, such as the information to be indicated itself or the index of the information to be indicated, etc., or the information to be indicated can be indirectly indicated by indicating other information, wherein there is an association between the other information and the information to be indicated. It is also possible to indicate only a part of the information to be indicated, while the other parts of the information to be indicated are known or agreed in advance. For example, the indication of specific information can also be achieved with the help of the arrangement order of each information agreed in advance (such as specified in the protocol), thereby reducing the indication overhead to a certain extent.

[0192] The information to be indicated can be sent as a whole or divided into multiple sub-information and sent separately. The transmission period and / or transmission timing of these sub-information can be the same or different. The specific transmission method is not limited in this application. The transmission period and / or transmission timing of these sub-information can be predefined, for example, according to a protocol, or can be configured by the transmitting device through sending configuration information to the receiving device.

[0193] It can be understood that "sending" and "receiving" in this application indicate the direction of signal transmission. For example, "sending information to XX" can be understood as the destination of the information is XX, which can include direct sending through the air interface, and also include indirect sending through the air interface by other units or modules. "Receiving information from YY" can be understood as the source of the information is YY, which can include direct receiving from YY through the air interface, and also include indirect receiving from YY through the air interface from other units or modules. "Sending" can also be understood as the "output" of the chip interface, and "receiving" can also be understood as the "input" of the chip interface.

[0194] In other words, sending and receiving can be performed between devices, for example, between a network device and a terminal device, or can be performed within a device, for example, sending or receiving between components, modules, chips, software modules or hardware modules within the device through a bus, wiring or interface.

[0195] It is understandable that information may be processed between the source and destination of information transmission, such as coding, modulation, etc., but the destination can understand the valid information from the source. Similar expressions in this application can be understood similarly and will not be repeated.

[0196] The technical solution of the present application can be applied to cellular communication systems related to the 3rd Generation Partnership Project (3GPP). For example, the fourth generation (4G) communication system, the fifth generation (5G) communication system, and the communication system after the fifth generation communication system. For example, the sixth generation communication system. For example, the fourth generation communication system may include a long term evolution (LTE) communication system. The fifth generation communication system may include a new radio (NR) communication system. The technical solution of the present application can also be applied to wireless fidelity (WiFi) systems, communication systems that support the integration of multiple wireless technologies, device-to-device (D2D) systems, vehicle to everything (V2X) communication systems, or satellite communication systems, etc. The wireless communication systems involved in this application also include but are not limited to: narrowband Internet of Things (NB-IoT) system, global system for mobile communications (GSM), enhanced data rate for GSM evolution (EDGE), wideband code division multiple access (WCDMA), code division multiple access 2000 (CDMA2000), or time division-synchronization code division multiple access (TD-SCDMA).

[0197] The communication system to which the technical solution of the present application is applicable includes a first device and a second device. Optionally, the communication system further includes a third device. Optionally, the first device, the second device, and the third device can be devices of three capability levels or three different layers.

[0198] Optionally, the first device is a terminal device, or a chip, chip system, or processor within the terminal device, or a logic module or software that implements some or all of the terminal device functions. The second device is an access network device, or a chip, chip system, or processor within the access network device, or a logic module or software that implements some or all of the access network device functions. In this implementation, optionally, the third device is a core network device or other AI node, or a chip, chip system, or processor within the core network device or other AI node, or a logic module or software that implements some or all of the core network device functions or other AI node functions.

[0199] Optionally, the first device is an access network device, or a chip, chip system, or processor within the access network device, or a logic module or software that implements some or all of the functions of the access network device. The second device is a core network device, or a chip, chip system, or processor within the core network device, or a logic module or software that implements some or all of the functions of the core network device. In this implementation, optionally, the third device is a core network device, or a chip, chip system, or processor within the core network device, or a logic module or software that implements some or all of the functions of the core network device.

[0200] Optionally, the first device is a weak terminal device, or a chip, chip system, or processor in a weak terminal device, or a logic module or software that implements some or all of the functions of a weak terminal device. The second device is a strong terminal device, or a chip, chip system, or processor in a strong terminal device, or a logic module or software that implements some or all of the functions of a strong terminal device. In this implementation, optionally, the third device is a network device or other AI node, or a chip, chip system, or processor in a network device or other AI node, or a logic module or software that implements some or all of the functions of a network device or other AI node. The network device can be an access network device or a core network device.

[0201] FIG1 is a schematic diagram of a communication system according to an embodiment of the present application. As shown in FIG1 , the first device is a terminal device 101, and the second device is a base station 103. The technical solution of the present application can be implemented between the terminal device 101 and the base station 103. Alternatively, the first device is the terminal device 101, and the second device is the terminal device 102. The technical solution of the present application can be implemented between the terminal device 101 and the terminal device 102. As shown in FIG1 , the communication system includes an independent AI network element 104, or AI modules can exist in the base station and the terminal device, respectively, to provide AI services.

[0202] Optionally, the communication system further includes a core network device 105. The third apparatus may be the core network device 105 shown in FIG1 .

[0203] The following introduces the terminal equipment and network equipment involved in this application.

[0204] Terminal equipment, also known as user equipment (UE), mobile station (MS), mobile terminal (MT), or customer premise equipment (CPE), is a device that includes wireless communication capabilities (providing voice and data connectivity to users). Examples include handheld devices, in-vehicle devices, and machine-type communication (MTC) terminals with wireless connectivity. Currently, terminal equipment can include: mobile phones, tablets, laptops, PDAs, mobile internet devices (MIDs), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals used in industrial control, self-driving vehicles, remote medical surgery, smart grids, transportation safety, smart cities, and smart homes. For example, wireless terminals in unmanned driving can be drones, helicopters, or airplanes. For example, wireless terminals in the Internet of Vehicles can be onboard equipment, complete vehicle equipment, onboard modules, vehicles, or ships. Wireless terminals in industrial control can be cameras, robots, or robotic arms. Wireless terminals in smart homes can be televisions, air conditioners, vacuum cleaners, speakers, or set-top boxes.

[0205] It should be noted that the terminal device can be a device or apparatus with a chip, or a device or apparatus with integrated circuits, or a chip, module or control unit in the device or apparatus shown above, and this application does not limit this specifically.

[0206] A network device is a device deployed in a radio access network (RAN) to provide wireless communication capabilities for terminal devices. A network device can also be called an access network (RAN) entity, access node, network node, or communication device.

[0207] Specifically, the network device may be an access network device of a cellular system related to the 3rd Generation Partnership Project (3GPP). For example, a fourth-generation (4G) mobile communication system or a 5G mobile communication system. The network device may also be an access network device in an open access network (O-RAN or ORAN) or a cloud radio access network (CRAN). Alternatively, the network device may be an access network device in a communication system obtained by integrating two or more of the above communication systems.

[0208] The network equipment includes, but is not limited to, evolved Node B (eNB), radio network controller (RNC), Node B (NB), base station controller (BSC), base transceiver station (BTS), home base station (e.g., home evolved NodeB, or home Node B, HNB), baseband unit (BBU), access point (AP) in wireless fidelity (WIFI) system, macro base station, micro base station, wireless relay node, donor node, wireless controller in CRAN scenario, wireless backhaul node, transmission point (TP) or transmission and receiving point (TRP). The network equipment can also be an access network equipment in a 5G mobile communication system. For example, a next generation NodeB (gNB) in a new radio (NR) system, a TRP, a TP, or one or a group of antenna panels (including multiple antenna panels) of a base station in a 5G mobile communication system. Alternatively, the network device may also be a network node constituting a gNB or a transmission point. For example, a centralized unit (CU), a distributed unit (DU), a CU-control plane (CP), a CU-user plane (UP), or a radio unit (RU). The CU and DU may be configured separately or included in the same network element. For example, a BBU. The RU may be included in a radio frequency device or radio frequency unit. For example, in a remote radio unit (RRU), an active antenna unit (AAU), or a remote radio head (RRH). Alternatively, the network device may also be a server, a wearable device, a vehicle, or an on-board device. For example, in V2X technology, the network device may be a road side unit (RSU).

[0209] It should be noted that in different systems, CU (or CU-CP and CU-UP), DU or RU may have different names, but those skilled in the art can understand their meanings. For example, in an open radio access network (ORAN) system, CU may also be called an open centralized unit (O-CU) or an open CU, DU may also be called an open distributed unit (O-DU), a centralized unit control plane (CU-CP) may also be called an open centralized unit control plane (O-CU-CP) or an open CU-CP, a centralized unit user plane (CU-UP) may also be called an open centralized unit user plane (O-CU-UP) or an open CU-UP, and RU may also be called an open radio unit (O-RU). This application does not limit this. Any of the CU, CU-CP, CU-UP, DU and RU in this application may be implemented by a software module, a hardware module, or a combination of a software module and a hardware module.

[0210] Optionally, for network elements in the ORAN system, each network element may implement the protocol layer functions shown in Table 1 below.

[0211] Table 1

[0212] It should be noted that the network device can be one or more network elements shown in Table 1 above, and this application does not make any specific limitations.

[0213] It should be noted that Table 1 above is merely an example. In actual applications, the protocol layer functions supported by each network element are not limited. For example, each network element may support more protocol layer functions, or the protocol layer functions supported by each network element may be specifically configured based on actual conditions. This application does not impose any specific limitations.

[0214] The following describes the architecture of the CU and DU of an access network device. The access network device includes at least one CU and at least one DU. Optionally, the access network device also includes at least one RU.

[0215] The following is an introduction using the example of an access network device including a CU and a DU. The CU has some functions of the core network, and the CU may include a CU-CP and a CU-UP. The CU and the DU may be configured according to the protocol layer functions of the wireless network they implement. For example, the CU is configured to implement the functions of the packet data convergence protocol (PDCP) layer and the protocol layers above it (for example, the RRC layer and / or the SDAP layer). The DU is configured to implement the functions of the protocol layers below the PDCP layer (for example, the RLC layer, the MAC layer, and / or the physical (PHY) layer). For another example, the CU is configured to implement the functions of the protocol layers above the PDCP layer (such as the RRC layer and / or the SDAP layer), and the DU is configured to implement the functions of the PDCP layer and the protocol layers below it (for example, the RLC layer, the MAC layer, and / or the PHY layer, etc.).

[0216] When a CU includes a CU-CP and a CU-UP, the CU-CP is used to implement the control plane functions of the CU, and the CU-UP is used to implement the user plane functions of the CU. For example, when the CU is configured to implement the functions of the PDCP layer, RRC layer, and SDAP layer, the CU-CP is used to implement the RRC layer functions and the control plane functions of the PDCP layer, and the CU-UP is used to implement the SDAP layer functions and the user plane functions of the PDCP layer.

[0217] The CU-CP can interact with the network elements in the core network for implementing control plane functions. The network elements in the core network for implementing control plane functions may be access and mobility function network elements, such as the access and mobility management function (AMF) in the 5G mobile communication system. The access and mobility function network element is responsible for mobility management in the mobile network, such as location update of the terminal device, registration network of the terminal device, switching of the terminal device, etc. The CU-UP can interact with the network elements in the core network for implementing user plane functions. The network elements in the core network for implementing user plane functions, such as the user plane function (UPF) in the 5G mobile communication system, are responsible for forwarding and receiving data in the terminal device.

[0218] The above configuration of CU and DU is only an example, and the functions of CU and DU can also be configured as needed. For example, the CU or DU can be configured to have the functions of more protocol layers, or the CU or DU can be configured to have partial processing functions of the protocol layer. For example, some functions of the RLC layer and the functions of the protocol layers above the RLC layer are set in the CU, and the remaining functions of the RLC layer and the functions of the protocol layers below the RLC layer are set in the DU. For another example, the functions of the CU or DU can be divided according to the service type or other system requirements. For example, according to the delay, the functions whose processing time needs to meet the smaller delay requirement are set in the DU, and the functions that do not need to meet the delay requirement are set in the CU.

[0219] The DU and RU can work together to implement the functions of the PHY layer. A DU can be connected to one or more RUs. The functions of the DU and RU can be configured in various ways according to the design. For example, the DU is configured to implement the baseband function, and the RU is configured to implement the mid-RF function. For another example, the DU is configured to implement the high-layer functions in the PHY layer, and the RU is configured to implement the low-layer functions in the PHY layer or to implement the low-layer functions and the RF functions. The high-layer functions in the physical layer may include a part of the functions of the physical layer, which is closer to the MAC layer, and the low-layer functions in the physical layer may include another part of the functions of the physical layer, which is closer to the mid-RF side.

[0220] Optionally, the network device may also be a core network device. The core network device is responsible for access control, registration management, service management, mobility management, etc. for terminal devices accessing the network. For example, the core network device is an AMF.

[0221] It should be noted that the network device can be the device or apparatus shown above, or it can be a component (for example, a chip), module, or unit in the device or apparatus shown above, and this application does not limit it specifically.

[0222] The communication system provided in the present application can introduce AI network elements to implement some or all AI-related operations. The AI ​​network element can also be called an AI node, AI device, AI entity, AI module, AI model, or AI unit, etc. The AI ​​network element can be a network element built into the communication system. For example, the AI ​​network element can be an AI module built into: an access network device, a core network device, a cloud server, or a network management (operation, administration and maintenance, OAM) to implement AI-related functions. The OAM can be a network management device as a core network device and / or a network management device as an access network device. Alternatively, the AI ​​network element can also be a network element independently set up in the communication system. Optionally, the terminal device or the chip built into the terminal device can also include an AI entity to implement AI-related functions. The following describes another communication system provided in an embodiment of the present application in conjunction with Figure 2A.

[0223] Figure 2A is another schematic diagram of a communication system according to an embodiment of the present application. As shown in Figure 2A, network elements in the communication system are connected through interfaces (for example, NG interfaces, or Xn interfaces) or air interfaces. One or more AI modules are provided in these network element nodes (for example, core network equipment, access network nodes (RAN nodes), terminals or one or more devices in OAM) (for the sake of clarity, only one is shown in Figure 2A). The access network node can be a separate RAN node, or it can include multiple RAN nodes, for example, including CU and DU. One or more AI modules can also be provided in the CU and / or DU. Optionally, the CU can also be split into CU-CP and CU-UP. One or more AI models are provided in the CU-CP and / or CU-UP.

[0224] The AI ​​module is used to implement the corresponding AI function. The AI ​​modules deployed in different network elements can be the same or different. The model of the AI ​​module can implement different functions according to different parameter configurations. The model of the AI ​​module can be configured based on one or more of the following parameters: structural parameters (such as the number of neural network layers, the width of the neural network, the connection relationship between layers, the weight of the neuron, the activation function of the neuron, or at least one of the bias in the activation function), input parameters (such as the type of input parameters and / or the dimension of the input parameters), or output parameters (such as the type of output parameters and / or the dimension of the output parameters). Among them, the bias in the activation function can also be called the bias of the neural network.

[0225] An AI module can have one or more models. A model can infer an output, which includes one or more parameters. The learning, training, or inference processes of different models can be deployed on different nodes or devices, or on the same node or device.

[0226] It should be noted that the access network node shown in Figure 2A can be used as a whole, or the access network node shown in Figure 2A can adopt a CU and DU separated architecture. Figure 2B shows a schematic diagram of a communication system in which the access network node is used as a whole and the access network node is connected to a terminal device. Figure 2C shows a schematic diagram of a communication system in which the access network node adopts a CU and DU separated architecture.

[0227] Currently, terminal devices can perform corresponding tasks through models. Therefore, how the terminal device obtains the model used to perform the task from the network is a question worth considering. For details, please refer to the relevant introduction of the embodiments shown in Figures 3 to 5 below.

[0228] The technical solution of this application is introduced below in conjunction with specific embodiments.

[0229] FIG3 is a schematic diagram of an embodiment of the model indication method of the present application. Referring to FIG3 , the method includes:

[0230] 301. A first device sends characteristic information to a second device. Correspondingly, the second device receives the characteristic information from the first device.

[0231] The feature information represents the features of the first task.

[0232] Optionally, the embodiment shown in FIG3 further includes step 301 a , and step 301 a may be performed before step 301 .

[0233] 301a. The first device determines feature information according to task information of the first task and / or auxiliary information of the first device.

[0234] The task information of the first task includes at least one of the following: a task type or target performance of the first task.

[0235] Optionally, the task type of the first task includes any of the following: CSI compression feedback, beam prediction, mobility management, data encryption, transceiver link, or model training. The task types shown here are only some examples and are not limited in this application. The target performance refers to the execution effect or execution performance required to be achieved by executing the first task through the model. For example, if the first task is CSI compression, the target performance can be the recovery accuracy of the channel feedback.

[0236] The auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device.

[0237] Optionally, the statistical parameter includes at least one of the following: service, quality of service (QoS), quality of experience (QoE), or channel characteristics between the first device and the second device. The environmental parameter includes the temperature or humidity of the environment in which the first device is located.

[0238] Optionally, the characteristic information is represented by a characteristic vector, or the characteristic information is represented by other forms. For example, the characteristic information is represented by a characteristic matrix, which is not specifically limited in this application. The following is an introduction using a characteristic vector as an example.

[0239] Optionally, the first device inputs the task information and / or auxiliary information into the feature vector generation model to obtain the feature vector output by the feature vector generation model; or, the first device uses the task information and / or auxiliary information as input parameters of the feature vector generation model to obtain the feature vector output by the feature vector generation model.

[0240] Optionally, the feature vector can be a vector composed of a bit sequence. It is understandable that the first device represents the task information and / or auxiliary information of the first task through a feature vector, that is, the first device translates the task information and / or auxiliary information into a feature vector. The first device can specifically generate the feature vector through the feature vector generation model. This makes it easier for the second device to obtain the feature vector to determine the characteristics of the first task. There is no need for the second device to interpret the task information and / or auxiliary information to obtain the characteristics of the first task. This facilitates the second device to quickly combine the feature vector to deploy a suitable user-level model for the first device.

[0241] Optionally, the embodiment shown in Figure 3 further includes step 301b. Step 301b may be performed before step 301a.

[0242] 301b. The second device sends second information to the first device. The second information is used to indicate the feature vector generation model. Correspondingly, the first device receives the second information from the second device.

[0243] 302. The second device generates a user-level model based on public models of various levels and feature information related to the first task in the second device.

[0244] The user-level model is used by the first device to perform the first task.

[0245] Optionally, the various levels of public models associated with the first task include one or more levels of public models. Different levels of public models within the one or more levels of public models may be generated by devices of different capability levels or hierarchies. For example, core network devices may be the highest-level devices, and access network devices may be the intermediate-level devices. For example, the various levels of public models associated with the first task include at least one of the following: a first-level public model or a second-level public model associated with the first task. The first-level public model may be generated by a core network device. The second-level public model may be generated by an access network device. The first-level public model may also be referred to as a primary public model, and the second-level public model may also be referred to as an advanced public model. The following description uses the first-level public model and the second-level public model as examples. It should be noted that the various levels of public models associated with the first task may also include multiple levels of public models, which are not specifically limited in this application. For example, the various levels of public models associated with the first task may also include a third-level public model. The first-level public model may be referred to as a primary public model, the second-level public model may be referred to as an intermediate public model, and the third-level public model may be referred to as an advanced public model.

[0246] In a possible implementation, the one or more levels of public models are divided according to the size of the execution area corresponding to the first task.

[0247] Optionally, the first-level public model is a global public model related to the first task. The global public model may be a public model used by users within a region to perform the first task. The region may include one or more cells, or one or more sectors. The second-level public model is a cell-level public model or a sector-level public model related to the first task.

[0248] For example, the global public model is applicable to users in the Hangzhou area performing the first task, and the global public model has relatively good generalization performance. The second-level public model is applicable to users in the cell covered by the second device in the Hangzhou area performing the first task. There are multiple cells in the Hangzhou area, and each cell has different communication, geography, users, and task characteristics of the first task. The second-level public model can be generated based on the characteristics of the cell covered by the second device. The second-level public model is more applicable to users in the cell covered by the second device performing the first task.

[0249] Optionally, the first-level public model is a global public model related to the first task. The second-level public model is a cell-level public model related to the first task adapted to terminal devices of different capability levels, or a sector-level public model related to the first task adapted to terminal devices of different capability levels.

[0250] For example, the global public model is applicable to users in the Hangzhou area performing the first task, and the global public model has relatively good generalization performance. The second-level public model can be a cell-level public model or a sector-level public model. The cell-level public model can be further divided into cell-level public models applicable to terminal devices of different capability levels. The sector-level public model can be further divided into sector-level public models applicable to terminal devices of different capability levels.

[0251] In this implementation, the public models at each level associated with the first task can be understood as public models for executing the first task within regions of varying sizes. The regions where public models at different levels are effective have a containment or inclusion relationship. For example, the region where a first-level public model is effective contains the region where a second-level public model is effective.

[0252] In another possible implementation, the one or more level public models are divided according to the master-slave relationship between the tasks that the level public models are used to perform. The tasks that different level public models are used to perform have a master-slave relationship. The slave task can be a subdivision of the master task.

[0253] For example, the first-level public model is an image recognition public model, the second-level public model is an animal image recognition public model, and the user-level model is a bird image recognition public model. In other words, the first-level public model is used to perform image recognition tasks, the second-level public model is used to perform animal image recognition tasks, and the user-level model is used to perform bird image recognition tasks.

[0254] For another example, the first-level public model is a public model for regional positioning, the second-level public model is a public model for intra-regional positioning, and the user-level model is a public model for intra-regional indoor positioning. In other words, the first-level public model is used for regional positioning tasks, the second-level public model is used for intra-regional positioning tasks, and the user-level model is used for intra-regional indoor positioning tasks.

[0255] For another example, the first-level public model is a public model for identifying moving objects, the second-level public model is a public model for identifying vehicles within moving objects, and the user-level public model is a public model for identifying bicycles within moving objects. In other words, the first-level public model is used to perform the task of identifying moving objects, the second-level public model is used to perform the task of identifying vehicles within moving objects, and the user-level model is used to perform the task of identifying bicycles within moving objects.

[0256] It can be seen from this that the task used to execute by the first-level public model can be understood as the main task. The task used to execute by the second-level public model can be understood as the first subordinate task. The task used to execute by the second-level public model and the task used to execute by the third-level public model are both subordinate tasks of the task used to execute by the first-level public model. The task used to execute by the third-level public model is a subordinate task of the task used to execute by the second-level public model.

[0257] In this implementation, the public models at various levels related to the first task can be understood as tasks executed by the public models at various levels having a master-slave relationship. For example, the task executed by the first-level public model is the master task of the task executed by the second-level public model.

[0258] Optionally, the feature information includes a feature vector. The above step 302 specifically includes steps 302a to 302b.

[0259] 302a. The second device correlates the feature vector with feature vectors corresponding to public models of various levels related to the first task in the second device to obtain correlation coefficients of public models of various levels related to the first task.

[0260] The first device performs correlation processing on the feature vector and the feature vectors corresponding to the public models of each level related to the first task in the second device to obtain correlation coefficients of the public models of each level related to the first task.

[0261] For example, the feature vector is represented as X, the feature vector corresponding to the level public model 1 related to the first task is represented as Y, and the feature vector corresponding to the level public model 2 related to the first task is represented as Z. Then the correlation coefficient between the feature vector X and the feature vector Y is expressed as The correlation coefficient between the eigenvector X and the eigenvector Z is expressed as In the above formula, the function E(a) represents the expected value of a, and D(a) represents the variance of a.

[0262] 302b. The second device performs weighted and fusion processing on the public models of each level related to the first task according to the correlation coefficients of the public models of each level related to the first task to obtain a user-level model.

[0263] Specifically, the second device uses the correlation coefficients of the public models at each level related to the first task as the weights of the public models at each level. Then, the second device performs weighted and fused processing on the public models at each level related to the first task according to the correlation coefficients of the public models at each level related to the first task to obtain a user-level model. For example, the public models at each level related to the first task include level public model 1 and level public model 2. The correlation coefficient of level public model 1 is ρ XY , the correlation coefficient of the level public model 2 is ρ XZ The second device will XY As the weighting coefficient of the level public model 1, ρ XZ As the weight coefficient of the level public model 2. Then, the second device combines the weight coefficients of the public models of each level with the weighted fusion of the public models of each level related to the first task to obtain the user-level public model.

[0264] 303. The second device sends first information to the first device. The first information is used to indicate the user-level model. Correspondingly, the first device receives the first information from the second device.

[0265] Specifically, the first information includes the model structure and / or model parameters of the user-level model.

[0266] Optionally, the embodiment shown in FIG3 further includes steps 301c to 301d. Steps 301c to 301d may be performed before step 302.

[0267] 301c. The third device sends third information to the second device. The third information is used to indicate the first-level public model. Correspondingly, the second device receives the third information from the third device.

[0268] The first-level public model is a first-level public model related to the first task. For example, the first-level public model can be called a global public model.

[0269] Specifically, the second device receives task request information sent from one or more first devices. Each task request information includes at least one of the following: the type of the first task, the local parameters of the first device (for example, the communication capability, computing capability, etc. of the first device), or the geographic location of the first device. The second device integrates the task request information sent by the one or more first devices to obtain cell-level task request information. The cell-level task request information corresponds to the first task. The second device sends the cell-level task request information to the third device. Then, the third device completes the training of the first-level public model based on the cell-level task request information. For example, the third device can train the first-level public model in conjunction with the second device or other AI nodes. The third device sends the third information to the second device, and the third information is used to indicate the first-level public model. Similarly, for other tasks, the third device can obtain the first-level public models related to multiple tasks respectively, and indicate the first-level public models related to the multiple tasks respectively to the second device. For example, as shown in Table 2, the third device can obtain the global public models related to tasks 1 to N respectively. The third device can send the global public models related to tasks 1 to N respectively as shown in Table 2 to the second device.

[0270] Table 2

[0271] 301d. The second device adjusts the first-level public model to obtain a second-level public model.

[0272] Specifically, the second device may adjust the model parameters and / or model structure of the first-level public model to obtain the second-level public model.

[0273] Optionally, the second device may receive first-level public models associated with multiple tasks from a third device. The second device may adjust the first-level public model associated with each task to obtain a second-level public model corresponding to each task. For example, the second-level public model is a cell-level public model or a sector-level public model. For example, as shown in Table 3 or Table 4:

[0274] Table 3

[0275] Table 4

[0276] For example, as shown in Table 2, the second device is an access network device, and the third device is a core network device. After the core network device receives the cell-level task request information from the access network device, the core network device determines that there are a large number of users in the Hangzhou area requesting the user-level model of Task 1. The core network device can train the global public model 1 for Task 1, and the global public model 1 has relatively good generalization performance. The access network device is an access network device in the Hangzhou area. The access network device can combine the characteristics of the cell covered by the access network device (for example, the communication conditions, geographical location, number of users, etc. of the cell), the global public model related to Task 1, and the data set of the cell covered by the access network device to train the cell-level public model 1 or the sector-level public model 1. For example, the access network device adjusts the scale of the global public model related to Task 1. For example, the model parameter of the global public model is 100M (megabytes), and the access network device simplifies the model parameter to 10M.

[0277] It should be noted that there is no fixed execution order between steps 301c to 301d and step 301. Step 301 may be executed first, and then steps 301c to 301d; or steps 301c to 301d may be executed first, and then step 301; or steps 301c to 301d and step 301 may be executed simultaneously depending on the circumstances. Optionally, if the embodiment shown in FIG3 further includes steps 301a to 301b, there is no fixed execution order between steps 301a to 301b and step 301. Step 301 may be executed first, and then steps 301a to 301b; or steps 301a to 301b may be executed first, and then step 301; or steps 301 and steps 301a to 301b may be executed simultaneously depending on the circumstances, and this application does not make any specific limitations.

[0278] It should be noted that steps 301c to 301d can be performed independently or in combination with other steps of the embodiment shown in FIG3 , and this application does not limit this. In other words, steps 301c to 301d can be used as independent embodiments and are not dependent on other steps of the embodiment shown in FIG3 .

[0279] In an embodiment of the present application, a first device sends feature information to a second device. The feature information characterizes the features of a first task. Then, the first device receives first information from the second device. The first information is used to indicate a user-level model, which is determined based on the feature information and various levels of public models related to the first task in the second device. The user-level model is used by the first device to perform the first task. This enables the first device to obtain the user-level model, or in other words, enables the second device to deploy the user-level model for the first device. This enables the first device to perform the first task based on the user-level model.

[0280] FIG4 is a schematic diagram of an embodiment of a model generation method according to an embodiment of the present application. Referring to FIG4 , the method includes:

[0281] 401. A second device sends first information to a first device. Correspondingly, the first device receives the first information from the second device.

[0282] The first information is used to indicate information of a level public model related to the first task.

[0283] It is understood that the one or more levels of public models associated with the first task may be generated by devices of different capability levels or hierarchies. For example, core network devices may be the highest-level devices, and access network devices may be the intermediate-level devices. For example, the various levels of public models associated with the first task may include at least one of the following: a first-level public model or a second-level public model associated with the first task. The first-level public model may be generated by the core network device. The second-level public model may be generated by the access network device. The first-level public model may also be referred to as a primary public model, and the second-level public model as a high-level public model. The following description uses the first-level public model and the second-level public model as examples. It should be noted that the various levels of public models associated with the first task may also include multiple levels of public models, which are not specifically limited in this application. For example, the various levels of public models associated with the first task may also include a third-level public model. The first-level public model may be referred to as a primary public model, the second-level public model may be referred to as an intermediate public model, and the third-level public model may be referred to as a high-level public model. For details on the division of the one or more levels of public models, please refer to the relevant description in step 302 of the embodiment shown in FIG. 3 , and will not be repeated here. Therefore, the level public model associated with the first task indicated by the first information may be the first-level public model or the second-level public model. For example, if the second device is an access network device, in step 401, the level public model associated with the first task is the second-level public model. For example, a cell-level public model or a sector-level public model.

[0284] The following describes two possible implementations of the first information. This application is still applicable to other implementations, and this application does not limit them specifically.

[0285] 1. The first information includes at least one of the following: reference information or adjustment information. The reference information is used to indicate relevant information of the level public model referenced by the first device when generating the user-level model. The adjustment information is used to indicate relevant information for adjusting the level public model.

[0286] Optionally, the reference information includes at least one of the following: area information, task information of the first task, first indication information, or second indication information.

[0287] The area information is used to indicate the area in which the first device reuses the public model of the level. The first indication information is used to indicate the public model of the level, and the second indication information is used to indicate the reuse mode of the public model of the level.

[0288] The reuse method includes any of the following: using the public model at this level as a user-level model; using the model structure of the public model at this level as the model structure of the user-level model; using the model parameters of the public model at this level as the model parameters of the user-level model; or using the public model at this level as a teacher to train a user-level model. The user-level model is used by the first device to perform the first task.

[0289] Optionally, the adjustment information includes at least one of the following: adjustment information of the model parameters of the level common model, adjustment information of the model structure of the level common model, or adjustment information of the quantization accuracy of the model parameters of the level common model. For example, the adjustment information of the model parameters of the level common model includes scaling the model parameters of the level common model, and / or adding bias to the model parameters of the level common model. The adjustment information of the model structure of the level common model includes adjusting the number of neural network layers, the number of neurons, and / or the activation function adopted by the level common model. The adjustment information of the quantization accuracy of the model parameters of the level common model includes the quantization accuracy of the model parameters of the level common model, and / or the quantization accuracy of the model parameters transferred during the training process, etc.

[0290] Optionally, the first information is carried in RRC signaling, media access control control element (MAC CE) or downlink control information (DCI), which is not specifically limited in this application.

[0291] The following describes a possible format of the first information. The following format does not limit the format of the first information. The first information may also be in other formats, which is not specifically limited in this application.

[0292] As can be seen from the first information above, AreaInfo is used to describe the area where the first device reuses the public model of this level. MissionInfo is used to describe the task information of the first task. For example, the task type of the first task. For some possible implementation methods of the task type, please refer to the relevant introduction of the embodiment shown in Figure 3 above. ModelIndex is used to indicate the level public model related to the first task, or a part of the level public model related to the first task. ReferenceType is used to indicate the reuse method of the public model of this level. For example, when the value of ReferenceType is type0, it means that the first device directly uses the public model of this level, that is, the public model of this level is used as the user-level model. When the value of ReferenceType is type1, it means that the first device can use the model structure of the public model of this level as the model structure of the user-level model. When the value of ReferenceType is type2, it means that the first device can use the public model of this level as a teacher to train the user-level model, that is, the user-level model is trained based on the public model of the level.

[0293] In this implementation, the first device is preconfigured with one or more task-related level public models. Alternatively, the second device sends the one or more task-related level public models to the first device via a system message or a user-specific message. For example, the first device is a terminal device and the second device is an access network device. The second device further sends the first information to the first device.

[0294] 2. The first information includes model parameters and / or model structure of the level public model related to the first task.

[0295] Optionally, the first information is carried in a system message or a user-specific message.

[0296] In one possible implementation, the first information further includes at least one of the following: a threshold range for the correlation coefficient, an index of a level-specific public model, or third indication information, where the third indication information is used to indicate a reuse method for the level-specific public model. For details on the reuse method, please refer to the aforementioned description and will not be repeated here. For details on the role of the threshold range for the correlation coefficient, please refer to the following description.

[0297] In another possible implementation, the second device sends the second information to the first device. Accordingly, the first device receives the second information from the second device. The second information includes at least one of the following: a threshold range of the correlation coefficient, an index of the level public model, or a third indication information. It should be noted that, in the two methods shown above, one is that the threshold range of the correlation coefficient, the index of the level public model, and the third indication information are all carried in the first information. The other is that the threshold range of the correlation coefficient, the index of the level public model, and the third indication information are carried in the second information. In actual applications, there may also be other implementation methods, which are not limited in this application. For example, the second device separately sends the threshold range of the correlation coefficient, the index of the level public model, and each information in the third indication information, or the second device sends the threshold range of the correlation coefficient and the index of the level public model together through one message, and then sends the third indication information separately, which is not limited in this application.

[0298] The following describes a possible format of the second information. This application is still applicable to other formats and does not limit them.

[0299] As can be seen from the second information above, Range is used to indicate the threshold range of the correlation coefficient, in which the maximum value is value 1 and the minimum value is value 2. ModelIndex is used to indicate the index of the level public model. MultiplexType is used to indicate the reuse method of the level public model. When the value of MultiplexType is type0, it means that the first device directly uses the level public model, that is, the level public model is used as the user-level model. When the value of MultiplexType is type1, it means that the first device can use the model structure of the level public model as the model structure of the user-level model. When the value of MultiplexType is type2, it means that the first device can use the level public model as a teacher to train the user-level model, that is, the user-level model is obtained by training with the level public model as the benchmark. The format of the first information can also be similar to the format of the second information, which is not limited in this application.

[0300] In this implementation, the second device may send the one or more task-related level public models to the first device via a system message or a user-specific message. For example, the first device is a terminal device, and the second device is an access network device. The access network device sends the one or more task-related second-level public models to the terminal device. For example, the one or more task-related second-level public models may be the one or more task-related cell-level public models or sector-level public models.

[0301] 402. The first device generates a user-level model according to the first information.

[0302] The user-level model is used by the first device to perform the first task.

[0303] Step 402 is described below in conjunction with two possible implementations of the first information.

[0304] 1. The first information includes reference information and adjustment information. The following describes a possible implementation of step 402. Optionally, the above step 402 specifically includes step 402a and step 402b.

[0305] 402a. The first device determines a level public model according to reference information.

[0306] Specifically, the reference information includes at least one of the following: region information, task information of the first task, first indication information, or second indication information. The first device determines the public model of the level based on the first indication information. The first device determines the region in which the public model of the level is effective based on the region information. The first device determines the reuse mode of the public model of the level based on the second indication information.

[0307] 402b. The first device adjusts the level public model according to the adjustment information to obtain a user-level model.

[0308] For example, the first device may adjust the model parameters and / or model structure of the level public model to obtain a user-level model.

[0309] 2. The first information includes model parameters and / or model structure of the level public model related to the first task.

[0310] In a possible implementation manner, the first information includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information.

[0311] In another possible implementation, the second device sends the second information to the first device. Correspondingly, the first device receives the second information from the second device. The second information includes at least one of the following: a threshold range of a correlation coefficient, an index of a level public model, or third indication information.

[0312] Optionally, the embodiment shown in FIG4 further includes step 401a. Step 401a may be performed before step 401.

[0313] 401a. The first device generates a first feature vector.

[0314] The first eigenvector represents the characteristics of the first task.

[0315] Optionally, the first device generates a first feature vector based on task information of the first task and / or auxiliary information of the first device. The task information of the first task includes at least one of the following: a task type or target performance of the first task. For details about the task type and target performance of the first task, please refer to the relevant description of step 301a in the embodiment shown in Figure 3 above, which will not be repeated here. The auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device.

[0316] Optionally, the statistical parameter includes at least one of the following: service, QoS, QoE of the first device, or channel characteristics between the first device and the second device, etc. The environmental parameter includes the temperature or humidity of the environment where the first device is located.

[0317] Optionally, the first feature vector is a vector composed of a bit sequence. It is understandable that the first device represents the task information and / or auxiliary information of the first task through the first feature vector, that is, the first device translates the task information and / or auxiliary information into the form of a first feature vector. Optionally, the first device uses the task information and / or auxiliary information of the first task as input parameters of a feature vector generation model to obtain a first feature vector output by the feature vector generation model. The feature vector generation model can be pre-configured in the first device or sent by the second device to the first device, and this application does not limit this.

[0318] Optionally, the above step 402 specifically includes: when the correlation coefficient between the first eigenvector and the second eigenvector falls within a threshold range of the correlation coefficient, the first device generates a user-level model according to the first information and the third indication information.

[0319] The second eigenvector represents the characteristics of the public model at this level.

[0320] Specifically, the first device correlates the first eigenvector with the second eigenvector, that is, the first device performs a correlation operation on the first eigenvector and the second eigenvector to obtain a correlation coefficient between the first eigenvector and the second eigenvector. If the correlation coefficient falls within the threshold range of the correlation coefficient, the first device determines the level public model based on the first information, and generates a user-level model based on the level public model and the third indication information. For example, the third indication information instructs the first device to directly use the level public model, then the first device uses the level public model as the user-level public model. For another example, the third indication information instructs the first device to reuse the model structure of the level public model, then the first device uses the model structure of the level public model as the model structure of the user-level model, and then trains the user-level model to obtain the model parameters of the user-level model.

[0321] As can be seen, in this embodiment, the first device can determine the user-level model through the first information, which is beneficial for the first device to quickly obtain the user-level model and also reduces the air interface transmission overhead of the second device deploying the model for the first device.

[0322] Optionally, the embodiment shown in FIG4 further includes step 401b. Step 401b may be performed before step 402.

[0323] 401b: The second device sends the second feature vector to the first device. Correspondingly, the first device receives the second feature vector from the second device.

[0324] Optionally, the second device generates feature vectors corresponding to more task-related level public models, and sends the feature vectors corresponding to the more task-related level public models to the first device.

[0325] Optionally, the second device may send a system message or a user-specific message to the first device. The system message or user-specific message includes the level common models associated with the multiple tasks. Specifically, the message may include model parameters, model structures, and / or model indexes of the level common models associated with the multiple tasks.

[0326] It should be noted that there is no fixed order for executing steps 401a to 401b and step 401. Step 401 may be executed first, followed by steps 401a to 401b; or steps 401a to 401b may be executed first, followed by step 401; or steps 401a to 401b and step 401 may be executed simultaneously, depending on the circumstances. This application does not impose any specific restrictions.

[0327] It should be noted that there is no fixed order for executing step 401a and step 401b. Step 401a may be executed first, then step 401b; or step 401b may be executed first, then step 401a; or, depending on the circumstances, step 401a and step 401b may be executed simultaneously. This application does not impose any specific restrictions.

[0328] Optionally, the embodiment shown in FIG4 further includes steps 401c to 401d. Steps 401c to 401d may be performed before step 402.

[0329] 401c. The third device sends third information to the second device. The third information is used to indicate the first-level public model. Correspondingly, the second device receives the third information from the third device.

[0330] 401d. The second device adjusts the first-level public model to obtain a second-level public model.

[0331] Steps 401c to 401d are similar to steps 301c to 301d in the embodiment shown in FIG. 3 . For details, please refer to the relevant introduction of steps 301c to 301d in the embodiment shown in FIG. 3 , which will not be repeated here.

[0332] It should be noted that steps 401c to 401d can be performed independently or in combination with other steps of the embodiment shown in FIG4 , and this application does not limit this. In other words, steps 401c to 401d can be used as independent embodiments and are not dependent on other steps of the embodiment shown in FIG4 .

[0333] It should be noted that there is no fixed order for executing steps 401c to 401d and step 401. Step 401 may be executed first, followed by steps 401c to 401d; or steps 401c to 401d may be executed first, followed by step 401; or, depending on the circumstances, steps 401 and 401c to 401d may be executed simultaneously, and this application does not impose any specific restrictions.

[0334] It should be noted that if the embodiment shown in Figure 4 also includes steps 401a to 401b, there is no fixed execution order between steps 401a to 401b and steps 401c to 401d. Steps 401a to 401b can be executed first, and then steps 401c to 401d; or, steps 401c to 401d can be executed first, and then steps 401a to 401b; or, steps 401a to 401b and steps 401c to 401d can be executed simultaneously depending on the situation, and this application does not limit this specifically.

[0335] In an embodiment of the present application, a first device receives first information from a second device. The first information indicates information about a level public model related to a first task. The first device generates a user-level model based on the first information, and the user-level model is used by the first device to perform the first task. This enables the first device to generate the user-level model based on the first information and perform the first task based on the user-level model.

[0336] FIG5 is a schematic diagram of another embodiment of the model indication method of the present application. Referring to FIG5 , the method includes:

[0337] 501. A first device sends a first request to a second device. The first request is used to request a target-level public model. Correspondingly, the second device receives the first request from the first device.

[0338] Optionally, the first request includes an identifier of a target-level public model.

[0339] Optionally, the identifier of the target-level public model includes at least one of the following: a model provider identifier, a service identifier, a region identifier, or an index of the target-level public model in the region indicated by the region identifier.

[0340] The model provider identifier is used to indicate the provider of the target level public model. For example, the public model at this level may be provided by Huawei or Qualcomm. For example, the length of the model provider identifier can be 1 bit, 2 bits, or 3 bits, which is not limited in this application. For example, a value of 01 for the model provider identifier indicates that the public model at this level is provided by Huawei.

[0341] The service identifier is used to indicate the first task. Optionally, the length of the service identifier can be 1 bit, 2 bits, or 3 bits, which is not limited in this application. For example, a service identifier of 001 indicates that the first task is CSI compression feedback; a service identifier of 010 indicates that the first task is beam prediction; and a service identifier of 011 indicates that the first task is a transceiver link. It should be noted that, optionally, the length of the service identifier can be determined by the number of task types of the first task.

[0342] The area identifier is used to indicate the area where the target level public model is effective. Optionally, the area identifier includes at least one of the following: an identifier of a first-level area, an identifier of a second-level area, and an identifier of a third-level area. A first-level area may be an area covered by several base stations, a second-level area may be an area covered by one base station, and a third-level area may be an area covered by one sector of a base station. Here, the third-level area is taken as an example for introduction. In actual applications, the area identifier may also include an area identifier of a lower level, which is not specifically limited in this application. Optionally, the number of bits occupied by each level of area identifier may be determined in combination with the number of divisions of each level of area. Please refer to Figure 6 for a possible format of the identifier of the target level public model. As shown in Figure 6, the model provider identifier (Provider ID) indicates the provider of the target level public model. The service identifier (Service ID) indicates the first task. The identifier of the first level area (1 st Area ID) is used to indicate the first-level area. The second-level area identifier (2 nd Area ID) is used to indicate the secondary area. The identification of the third level area (3 rd Area ID) is used to indicate the third-level area.

[0343] It should be noted that when the values ​​of a level-one region's identifier are all 0s or all 1s, the level-one common model is globally unique within the region above it. For example, if the values ​​of the level-two region's identifier and the level-three region's identifier are all 0s or all 1s, respectively, the level-one common model is valid within the level-one region, and there is no further subdivision of the level-two or level-three regions.

[0344] Optionally, the first request also includes task information for the first task. The first task is a task to be executed by the first device. The target-level public model is used to execute the first task. Optionally, the task information for the first task includes at least one of the following: a task type or target performance of the first task. For details about task types and target performance, please refer to the relevant description of the embodiment shown in Figure 3 above.

[0345] It should be noted that the level-by-level public models associated with the first task may include one or more level-by-level public models. One or more level-by-level public models may be generated by devices of different capability levels or different hierarchies. For example, core network devices may be the highest-level devices, and access network devices may be the intermediate-level devices. For example, the various level-by-level public models associated with the first task may include at least one of the following: a first-level public model or a second-level public model associated with the first task. The first-level public model may be generated by core network devices. The second-level public model may be generated by access network devices. The first-level public model may also be referred to as a primary public model, and the second-level public model may also be referred to as an advanced public model. The following description uses the first-level public model and the second-level public model as examples. It should be noted that the various level-by-level public models associated with the first task may also include multiple levels of public models, which are not specifically limited in this application. For example, the various level-by-level public models associated with the first task may also include a third-level public model. The first-level public model may be referred to as a primary public model, the second-level public model may be referred to as an intermediate public model, and the third-level public model may be referred to as an advanced public model.

[0346] Therefore, the target level public model can be a first level public model or a second level public model. For example, the first device is a terminal device and the second device is an access network device, then the target level public model can be a second level public model. For example, a cell level public model or a sector level public model. For another example, the first device is an access network device and the second device is a core network device, then the target level public model can be a first level public model. Regarding the division method of the one or more levels of public models, please refer to the relevant introduction in step 302 of the embodiment shown in Figure 3 above, which will not be repeated here.

[0347] Optionally, the embodiment shown in FIG5 further includes step 501a. Step 501a may be performed before step 501.

[0348] 501a: The second device sends the level public model library information to the first device. Correspondingly, the first device receives the level public model library information from the second device.

[0349] The first-level public model library information includes an identifier of at least one first-level public model. The at least one first-level public model includes the target-level public model. For example, the first device is a terminal device, and the second device is an access network device. The first-level public model library information includes at least one second-level public model. The at least one second-level public model is a second-level public model corresponding to one or more tasks.

[0350] Optionally, the level public model library information further includes performance indicators of at least one level public model, such as applicable scenarios, reasoning accuracy achieved in executing tasks, and / or model scale of each level public model in the at least one level public model.

[0351] Optionally, the second device may send the level public model library information to the first device via a broadcast message, a multicast message, or a unicast message.

[0352] Optionally, the embodiment shown in FIG5 further includes steps 501b to 501c. Steps 501b to 501c may be performed before step 502.

[0353] 501b. The third device sends second information to the second device. The second information is used to indicate the first-level public model. Correspondingly, the second device receives the second information from the third device.

[0354] 501c. The second device adjusts the first-level public model to obtain a second-level public model.

[0355] Steps 501b to 501c are similar to steps 301c to 301d in the embodiment shown in FIG. 3 . For details, please refer to the relevant introduction of steps 301c to 301d in the embodiment shown in FIG. 3 .

[0356] It should be noted that steps 501b to 501c can be performed independently or in combination with other steps of the embodiment shown in FIG5 , and this application does not limit this. In other words, steps 501b to 501c can be used as independent embodiments and are not dependent on other steps of the embodiment shown in FIG5 .

[0357] It should be noted that there is no fixed order for executing steps 501b to 501c relative to step 501. Step 501 may be executed first, followed by steps 501b to 501c; or, step 501 may be executed first, followed by steps 501b to 501c; or, depending on the circumstances, steps 501 and 501b to 501c may be executed simultaneously, with no specific limitation in this application. If the embodiment shown in FIG. 5 also includes step 501a, steps 501b to 501c may be executed before step 501a.

[0358] 502. The second device sends first information to the first device. The first information is used to indicate the target-level public model. Correspondingly, the first device receives the first information from the second device.

[0359] Specifically, the first information includes the model structure and / or model parameters of the target-level public model.

[0360] In an embodiment of the present application, a first device sends a first request to a second device. The first request includes an identifier of a target-level public model and is used to request the target-level public model. The first device then receives first information from the second device, the first information being used to indicate the target-level public model. This enables the second device to deploy an appropriate model for the first device, enabling the first device to perform a first task based on the target-level public model.

[0361] The following is a schematic diagram of the structure of the first device in the embodiment of the present application. Please refer to Figure 7. The first device can be used to execute the process executed by the first device in the embodiment shown in Figure 4. For details, please refer to the relevant introduction in the above method embodiment.

[0362] The first device 700 includes a transceiver module 701 and a processing module 702 .

[0363] The processing module 702 is used to process data. The transceiver module 701 can implement corresponding communication functions. The transceiver module 701 can also be called a communication interface or a communication module.

[0364] Optionally, the first device 700 may further include a storage module, which may be used to store instructions and / or data. The processing module 702 may read the instructions and / or data in the storage module so that the first device implements the aforementioned method embodiment.

[0365] The first device 700 can be configured to perform the actions performed by the first device in the above method embodiments. The first device 700 can be a first device or a component configurable in the first device. The processing module 702 is configured to perform the processing-related operations on the first device side in the above method embodiments. The transceiver module 701 is configured to perform the reception-related operations on the first device side in the above method embodiments.

[0366] Optionally, the transceiver module 701 may include a sending module and a receiving module. The sending module is used to perform the sending operation in the above method embodiment. The receiving module is used to perform the receiving operation in the above method embodiment.

[0367] It should be noted that the first device 700 may include a sending module but not a receiving module. Alternatively, the first device 700 may include a receiving module but not a sending module. The specific details may depend on whether the above solution executed by the first device 700 includes a sending action and a receiving action.

[0368] Optionally, the first device 700 is configured to execute the actions executed by the first device in the embodiment shown in Figure 4. For details, please refer to the relevant introduction in the embodiment shown in Figure 4, which will not be elaborated here.

[0369] For example, the first device 700 is configured to execute the following solution:

[0370] The transceiver module 701 is configured to receive first information from a second device, where the first information is used to indicate information of a level public model related to a first task;

[0371] The processing module 702 is used to generate a user-level model according to the first information, and the user-level model is used by the first device 700 to perform the first task.

[0372] It should be understood that the specific process of each module executing the above corresponding process has been described in detail in the above method embodiment, and for the sake of brevity, it will not be repeated here.

[0373] The processing module 702 in the above embodiment can be implemented by at least one processor or processor-related circuits. The transceiver module 701 can be implemented by a transceiver or transceiver-related circuits. The transceiver module 701 can also be called a communication module or communication interface. The storage module can be implemented by at least one memory.

[0374] FIG8 is another structural diagram of the first device according to an embodiment of the present application. Referring to FIG8 , the first device can be used to execute the process executed by the first device in the embodiments shown in FIG4 and FIG5 , and for details, please refer to the relevant description in the aforementioned method embodiment.

[0375] The first device 800 includes a transceiver module 801. Optionally, the first device 800 further includes a processing module 802.

[0376] The processing module 802 is used to perform data processing. The transceiver module 801 can implement corresponding communication functions. The transceiver module 801 can also be called a communication interface or a communication module.

[0377] Optionally, the first device 800 may further include a storage module, which may be used to store instructions and / or data. The processing module 802 may read the instructions and / or data in the storage module so that the first device implements the aforementioned method embodiment.

[0378] The first device 800 can be used to perform the actions performed by the first device in the above method embodiments. The first device 800 can be a first device or a component configurable in the first device. The processing module 802 is used to perform the processing-related operations on the first device side in the above method embodiments. The transceiver module 801 is used to perform the reception-related operations on the first device side in the above method embodiments.

[0379] Optionally, the transceiver module 801 may include a sending module and a receiving module. The sending module is used to perform the sending operation in the above method embodiment. The receiving module is used to perform the receiving operation in the above method embodiment.

[0380] It should be noted that the first device 800 may include a sending module but not a receiving module. Alternatively, the first device 800 may include a receiving module but not a sending module. The specific implementation depends on whether the above solution executed by the first device 800 includes a sending action and a receiving action.

[0381] Optionally, the first device 800 is configured to execute the actions executed by the first device in the embodiments shown in Figures 4 and 5. For details, please refer to the relevant introductions in the embodiments shown in Figures 4 and 5, which will not be elaborated here.

[0382] For example, the first device 800 is configured to execute the following solution:

[0383] The transceiver module 801 is used for the second device to send characteristic information, which represents the characteristics of the first task; and receive first information from the second device, which is used to indicate a user-level model. The user-level model is determined based on the characteristic information and various levels of public models related to the first task in the second device. The user-level model is used for the first device 800 to perform the first task.

[0384] For another example, the first device 800 is configured to execute the following solution:

[0385] The transceiver module 801 is configured to send a first request to a second device, where the first request is used to request a target-level public model; and receive first information from the second device, where the first information is used to indicate a target-level public model.

[0386] It should be understood that the specific process of each module executing the above corresponding process has been described in detail in the above method embodiment, and for the sake of brevity, it will not be repeated here.

[0387] The processing module 802 in the above embodiment can be implemented by at least one processor or processor-related circuits. The transceiver module 801 can be implemented by a transceiver or transceiver-related circuits. The transceiver module 801 can also be referred to as a communication module or communication interface. The storage module can be implemented by at least one memory.

[0388] FIG9 is a schematic diagram of the structure of the second device according to an embodiment of the present application. Referring to FIG9 , the second device 900 can be used to execute the process executed by the second device in the embodiments shown in FIG3 and FIG4 . For details, please refer to the relevant description of the above method embodiments.

[0389] The second device 900 includes a transceiver module 901 and a processing module 902 .

[0390] The processing module 902 is used to perform data processing. The transceiver module 901 can implement corresponding communication functions. The transceiver module 901 can also be called a communication interface or a communication module.

[0391] Optionally, the second device 900 may further include a storage module, which may be used to store instructions and / or data. The processing module 902 may read the instructions and / or data in the storage module so that the second device implements the aforementioned method embodiment.

[0392] The second device 900 can be used to perform the actions performed by the second device in the above method embodiments. The second device 900 can be a second device or a component that can be configured in the second device. The processing module 902 is used to perform the processing-related operations on the second device side in the above method embodiments. The transceiver module 901 is used to perform the reception-related operations on the second device side in the above method embodiments.

[0393] Optionally, the transceiver module 901 may include a sending module and a receiving module. The sending module is used to perform the sending operation in the above method embodiment. The receiving module is used to perform the receiving operation in the above method embodiment.

[0394] It should be noted that the second device 900 may include a sending module but not a receiving module. Alternatively, the second device 900 may include a receiving module but not a sending module. The specific implementation depends on whether the above solution executed by the second device 900 includes a sending action and a receiving action.

[0395] Optionally, the second device 900 is used to perform the actions performed by the second device in the embodiments shown in Figures 3 and 4. For details, please refer to the relevant introductions in the embodiments shown in Figures 3 and 4, which will not be expanded in detail here.

[0396] For example, the second device 900 is configured to execute the following solution:

[0397] The transceiver module 901 is configured to receive characteristic information from the first device, where the characteristic information represents characteristics of the first task;

[0398] Processing module 902, configured to generate a user-level model based on public models and feature information of various levels related to the first task in the second device, wherein the user-level model is used by the first device to perform the first task;

[0399] The transceiver module 901 is further configured to send first information to the first device, where the first information is used to indicate the user-level model.

[0400] For another example, the second device 900 is configured to execute the following solution:

[0401] A processing module 902 is configured to determine first information, where the first information indicates information of a level public model related to a first task;

[0402] The transceiver module 901 is used to send first information to the first device, where the first information is used by the first device to generate a user-level model, and the user-level model is used by the first device to perform a first task.

[0403] FIG10 is another schematic diagram of the structure of the second device according to an embodiment of the present application. Referring to FIG10 , the second device 1000 can be used to execute the process executed by the second device in the embodiment shown in FIG5 . For details, please refer to the relevant description of the above method embodiment.

[0404] The second device 1000 includes a transceiver module 1001. Optionally, the second device 1000 further includes a processing module 1002.

[0405] The processing module 1002 is used to process data. The transceiver module 1001 can implement corresponding communication functions. The transceiver module 1001 can also be called a communication interface or a communication module.

[0406] Optionally, the second device 1000 may further include a storage module, which may be used to store instructions and / or data. The processing module 1002 may read the instructions and / or data in the storage module so that the second device implements the aforementioned method embodiment.

[0407] The second device 1000 can be used to perform the actions performed by the second device in the above method embodiments. The second device 1000 can be a second device or a component configurable in the second device. The processing module 1002 is used to perform the processing-related operations on the second device side in the above method embodiments. The transceiver module 1001 is used to perform the reception-related operations on the second device side in the above method embodiments.

[0408] Optionally, the transceiver module 1001 may include a sending module and a receiving module. The sending module is used to perform the sending operation in the above method embodiment. The receiving module is used to perform the receiving operation in the above method embodiment.

[0409] It should be noted that the second device 1000 may include a sending module but not a receiving module. Alternatively, the second device 1000 may include a receiving module but not a sending module. The specific implementation depends on whether the above solution executed by the second device 1000 includes a sending action and a receiving action.

[0410] Optionally, the second device 1000 is used to perform the actions performed by the second device in the embodiment shown in Figure 5. For details, please refer to the relevant introduction of the embodiment shown in Figure 5, which will not be expanded in detail here.

[0411] For example, the second device 1000 is configured to execute the following solution:

[0412] The transceiver module 1001 is configured to receive a first request from a first device, the first request being used to request a target-level public model; and send first information to the first device, the first information being used to indicate the target-level public model.

[0413] The present application also provides an apparatus 1100. Referring to Figure 11 , apparatus 1100 includes a processor 1110 coupled to a memory 1120. Memory 1120 is configured to store computer programs, instructions, and / or data. Processor 1110 is configured to execute the computer programs, instructions, and / or data stored in memory 1120, thereby implementing the method described in the method embodiments described above. Apparatus 1100 is configured to implement the operations performed by the first apparatus or the second apparatus described in the method embodiments described above.

[0414] Optionally, the device 1100 includes one or more processors 1110.

[0415] Optionally, as shown in FIG11 , the apparatus 1100 may further include a memory 1120 .

[0416] Optionally, the memory 1120 included in the device 1100 may be one or more.

[0417] Optionally, the memory 1120 may be integrated with the processor 1110 or provided separately.

[0418] Optionally, as shown in Figure 11 , the apparatus 1100 may further include a transceiver 1130, which is configured to receive and / or transmit signals. For example, the processor 1110 is configured to control the transceiver 1130 to receive and / or transmit signals.

[0419] The present application further provides an apparatus 1200, which may be a terminal device, a processor in the terminal device, or a chip. The apparatus 1200 may be used to execute the operations performed by the first apparatus or the second apparatus in the above method embodiment.

[0420] When apparatus 1200 is a terminal device, FIG12 shows a simplified schematic diagram of the terminal device structure. As shown in FIG12, the terminal device includes a processor, a memory, and a transceiver. The memory can store computer program code, and the transceiver includes a transmitter 1231, a receiver 1232, a radio frequency circuit (not shown), an antenna 1233, and input / output devices (not shown).

[0421] The processor is mainly used to process communication protocols and communication data; control terminal devices, execute software programs and process software program data, etc.

[0422] Memory is mainly used to store software programs and data.

[0423] Radio frequency circuits are mainly used for conversion between baseband signals and radio frequency signals and for processing radio frequency signals.

[0424] Antennas are mainly used to send and receive radio frequency signals in the form of electromagnetic waves.

[0425] The input and output device may include a touch screen, a display screen, or a keyboard. The input and output device is mainly used to receive data input by the user and output data to the user. It should be noted that some types of terminal devices may not have input and output devices.

[0426] When data needs to be sent, the processor performs baseband processing on the data to be sent and outputs the baseband signal to the RF circuit. The RF circuit then performs RF processing on the baseband signal and transmits the RF signal to the outside in the form of electromagnetic waves through the antenna. When data is sent to the terminal device, the RF circuit receives the RF signal through the antenna. The RF circuit converts the RF signal into a baseband signal and outputs the baseband signal to the processor. The processor converts the baseband signal into data and processes the data. For ease of explanation, Figure 12 only shows one memory, processor, and transceiver. In actual terminal device products, there may be one or more processors and one or more memories. The memory may also be referred to as a storage medium or storage device, etc. The memory may be set independently of the processor or integrated with the processor, and this embodiment of the application does not limit this.

[0427] In the embodiment of the present application, the antenna and radio frequency circuit with transceiver functions can be regarded as the transceiver module of the terminal device, and the processor with processing function can be regarded as the processing module of the terminal device.

[0428] As shown in FIG12 , the terminal device includes a processor 1210, a memory 1220, and a transceiver 1230. The processor 1210 may also be referred to as a processing unit, a processing board, a processing module, or a processing device. The transceiver 1230 may also be referred to as a transceiver unit, a transceiver, or a transceiver device.

[0429] Optionally, the device used to implement the receiving function in transceiver 1230 is considered a receiving module, and the device used to implement the transmitting function in transceiver 1230 is considered a transmitting module. That is, transceiver 1230 includes a receiver and a transmitter. A transceiver may also be sometimes referred to as a transceiver, a transceiver module, or a transceiver circuit. A receiver may also be sometimes referred to as a receiver, a receiving module, or a receiving circuit. A transmitter may also be sometimes referred to as a transmitter, a transmitting module, or a transmitting circuit.

[0430] The processor 1210 is used to execute the processing actions of the first device or the second device in the embodiments shown in Figures 3 to 5. The transceiver 1230 is used to execute the transceiver actions of the first device or the second device in the embodiments shown in Figures 3 to 5.

[0431] It should be understood that FIG12 is merely an example and not a limitation, and the terminal device including the transceiver module and the processing module may not rely on the structures shown in FIG7 , FIG8 , or FIG12 .

[0432] When the device 1200 is a chip, the chip includes a processor, a memory, and a transceiver. The transceiver may be an input / output circuit or a communication interface. The processor may be a processing module, a microprocessor, or an integrated circuit integrated on the chip. The sending operation of the first or second device in the above method embodiments may be understood as an output of the chip, and the receiving operation of the first or second device in the above method embodiments may be understood as an input of the chip.

[0433] The present application further provides an apparatus 1300, which can be a network device or a chip. The apparatus 1300 can be used to perform the operations performed by the first apparatus or the second apparatus in the embodiments shown in FIG. 3 to FIG. 5 .

[0434] When the apparatus 1300 is a network device, for example, a base station, FIG13 shows a simplified schematic diagram of a base station structure. The base station includes a portion 1310, a portion 1320, and a portion 1330.

[0435] Part 1310 is mainly used for baseband processing, base station control, etc.; Part 1310 is usually the control center of the base station, which can usually be called a processor, used to control the base station to execute the processing operations on the first device or second device side in the above method embodiment.

[0436] Part 1320 is mainly used to store computer program code and data.

[0437] Section 1330 is primarily used for receiving and transmitting RF signals and converting RF signals to baseband signals. Section 1330 can generally be referred to as a transceiver module, transceiver, transceiver circuit, or transceiver. The transceiver module in section 1330, which can also be referred to as a transceiver or transceiver, includes an antenna 1333 and a RF circuit (not shown in the figure), where the RF circuit is primarily used for RF processing. Optionally, the device used to implement the receiving function in section 1330 can be considered a receiver, and the device used to implement the transmitting function can be considered a transmitter. That is, section 1330 includes receiver 1332 and transmitter 1331. A receiver can also be referred to as a receiving module, receiver, or receiving circuit, and a transmitter can be referred to as a transmitting module, transmitter, or transmitting circuit.

[0438] Sections 1310 and 1320 may include one or more boards, each of which may include one or more processors and one or more memories. The processor is used to read and execute programs in the memory to implement baseband processing functions and control the base station. If multiple boards are present, the boards may be interconnected to enhance processing capabilities. As an optional implementation, multiple boards may share one or more processors, multiple boards may share one or more memories, or multiple boards may simultaneously share one or more processors.

[0439] For example, in one implementation, the transceiver module in section 1330 is used to execute the transceiver-related processes performed by the first device or the second device in the embodiments shown in Figures 3 to 5. The processor in section 1310 is used to execute the processing-related processes performed by the first device or the second device in the embodiments shown in Figures 3 to 5.

[0440] It should be understood that FIG13 is merely an example and not a limitation, and the network device including the processor, memory, and transceiver may not rely on the structures shown in FIG9 , FIG10 , or FIG13 .

[0441] When device 1300 is a chip, the chip includes a transceiver, a memory, and a processor. The transceiver may be an input / output circuit or a communication interface; the processor may be a processor, microprocessor, or integrated circuit integrated on the chip. In the above method embodiments, the sending operation of the first or second device can be understood as an output of the chip, and the receiving operation of the first or second device in the above method embodiments can be understood as an input of the chip.

[0442] An embodiment of the present application further provides a computer-readable storage medium storing computer instructions for implementing the method executed by the first device or the second device in the above method embodiment.

[0443] For example, when the computer program is executed by a computer, the computer can implement the method performed by the first device or the second device in the above method embodiment.

[0444] An embodiment of the present application further provides a computer program product comprising instructions, which, when executed by a computer, enables the computer to implement the method performed by the first device or the second device in the above method embodiment.

[0445] The present application also provides a communication system including the first device described in the above embodiment and the second device described in the above embodiment. The first device is configured to perform some or all of the operations performed by the first device described in the above method embodiment, and the second device is configured to perform some or all of the operations performed by the second device described in the above method embodiment.

[0446] An embodiment of the present application further provides a chip device, including a processor, configured to call a computer program or computer instruction stored in the memory so that the processor executes the method provided in the embodiments shown in FIG. 3 to FIG. 5 .

[0447] In one possible implementation, the input of the chip device corresponds to the receiving operation in any one of the embodiments shown in FIG. 3 to FIG. 5 , and the output of the chip device corresponds to the sending operation in any one of the embodiments shown in FIG. 3 to FIG. 5 .

[0448] Optionally, the processor is coupled to the memory via an interface.

[0449] Optionally, the chip device further includes a memory, in which computer programs or computer instructions are stored.

[0450] The processor mentioned in any of the above may be a general-purpose central processing unit, a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the method provided in any of the embodiments shown in Figures 3 to 5. The memory mentioned in any of the above may be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM), etc.

[0451] Those skilled in the art can clearly understand that, for the sake of convenience and brevity of description, the explanation of the relevant contents and beneficial effects of any of the above-mentioned devices can refer to the corresponding method embodiments provided above, and will not be repeated here.

[0452] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0453] The units described as separate components may or may not be physically separate, and 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 these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0454] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0455] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the part that essentially contributes to the technical solution of the present application or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

[0456] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A model indication method, characterized in that: The method comprises: The first device sends characteristic information to the second device, where the characteristic information represents a characteristic of the first task; The first device receives first information from the second device, where the first information is used to indicate a user-level model, where the user-level model is determined based on the feature information and various levels of public models in the second device related to the first task, and where the user-level model is used by the first device to perform the first task.

2. The method according to claim 1, characterized in that: The method further comprises: The first device determines the feature information according to task information of the first task; or, the first device determines the feature information according to the task information of the first task and auxiliary information of the first device.

3. The method according to claim 2, characterized in that The feature information includes a feature vector; The first device determines characteristic information according to task information of the first task, including: The first device inputs the task information into a feature vector generation model to obtain the feature vector output by the feature vector generation model; The first device determines the characteristic information according to the task information of the first task and the auxiliary information of the first device, including: The first device inputs the task information and the auxiliary information into the feature vector generation model to obtain the feature vector output by the feature vector generation model.

4. The method according to claim 3, characterized in that Before obtaining the feature vector output by the feature vector generation model, the method further includes: The first device receives second information from the second device, where the second information is used to indicate the feature vector generation model.

5. A model indication method, characterized in that: The method comprises: The second device receives characteristic information from the first device, wherein the characteristic information represents a characteristic of the first task; The second device generates a user-level model according to the public models of various levels related to the first task in the second device and the feature information, and the user-level model is used for the first device to perform the first task; The second device sends first information to the first device, where the first information is used to indicate the user-level model.

6. The method according to claim 5, characterized in that The feature information includes a feature vector; The second device generates the user-level model according to the public models of various levels related to the first task and the feature information, including: The second device correlates the feature vector with feature vectors corresponding to public models of various levels related to the first task in the second device to obtain correlation coefficients of public models of various levels related to the first task; The second device performs weighted and fusion processing on the public models of various levels related to the first task according to the correlation coefficients of the public models of various levels related to the first task to obtain the user-level model.

7. The method according to claim 5 or 6, characterized in that: The feature information includes a feature vector; the method further includes: The second device sends second information to the first device, where the second information is used to indicate a feature vector generation model, and the feature vector generation model is used to generate the feature vector.

8. The method according to claim 3, 4, 6 or 7, characterized in that: The feature vector is a vector composed of bit sequences.

9. The method according to any one of claims 5 to 8, characterized in that The feature information is generated based on task information of the first task; or, the feature information is generated based on task information of the first task and auxiliary information of the first device.

10. The method according to claim 2 or 9, characterized in that: The auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device.

11. The method according to any one of claims 2 to 4, 9 and 10, characterized in that: The task information includes at least one of the following: a task type or target performance of the first task.

12. The method according to claim 11, characterized in that The task type includes any of the following: channel state information CSI compression, beam prediction, mobility management, data encryption, transceiver link, or model training.

13. The method according to any one of claims 1 to 12, characterized in that The public models at each level related to the first task include at least one of the following: a first-level public model or a second-level public model related to the first task.

14. The method according to claim 13, characterized in that The first-level public model, the second-level public model and the user-level model are divided according to the size of the execution area corresponding to the first task.

15. The method according to claim 14, characterized in that The first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task and adapted to terminal devices of different capability levels, or a sector-level public model related to the first task and adapted to terminal devices of different capability levels.

16. The method according to claim 13, characterized in that The first-level public model, the second-level public model and the user-level model are divided according to the master-slave relationship between the tasks that the models are used to perform.

17. The method according to any one of claims 5 to 16, characterized in that The method further comprises: The third device sends third information to the second device, where the third information is used to indicate the first-level public model; correspondingly, the second device receives the third information from the third device; The second device adjusts the first-level public model to obtain the second-level public model.

18. A model generation method, characterized in that: The method comprises: The first device receives first information from the second device, wherein the first information is used to indicate information of a level public model related to the first task; The first device generates a user-level model according to the first information, and the user-level model is used by the first device to perform the first task.

19. The method according to claim 18, characterized in that The first information includes at least one of the following: reference information or adjustment information, wherein the reference information is used to indicate relevant information of a level public model referenced by the first device to generate the user-level model, and the adjustment information is used to indicate relevant information for adjusting the level public model.

20. The method according to claim 19, characterized in that The reference information includes at least one of the following: area information, task information of the first task, first indication information, or second indication information; wherein the area information is used to indicate the area in which the first device reuses the level common model, the first indication information is used to indicate the level common model, and the second indication information is used to indicate the reuse method of the level common model.

21. The method according to claim 19 or 20, characterized in that The adjustment information includes at least one of the following: adjustment information of a model parameter of the level common model, adjustment information of a model structure of the level common model, or adjustment information of a quantization accuracy of a model parameter of the level common model.

22. The method according to claim 21, characterized in that The method further comprises: The first device generates a first feature vector, wherein the first feature vector represents a feature of the first task; The first device correlates the first feature vector with a second feature vector to obtain a correlation coefficient between the first feature vector and the second feature vector, wherein the second feature vector represents a feature of the level common model; The first device generates the user-level model according to the first information, including: When the correlation coefficient between the first feature vector and the second feature vector falls within a threshold range of the correlation coefficient, the first device generates the user-level model according to the first information and third indication information, and the third indication information is used to indicate a reuse method of the level public model.

23. The method according to claim 22, characterized in that The first information further includes at least one of the following: a threshold range of the correlation coefficient, or the third indication information; or, Before the first device generates the user-level model according to the first information, the method further includes: The first device receives second information from the second device, where the second information includes at least one of the following: a threshold range of the correlation coefficient, or the third indication information.

24. The method according to claim 22 or 23, characterized in that The first device generates a first feature vector, including: The first device generates the first feature vector according to task information of the first task; or, the first device generates the first feature vector according to the task information and auxiliary information.

25. The method according to claim 24, characterized in that The auxiliary information includes at least one of the following: capability information, location information, statistical parameters, or environmental parameters of the first device.

26. The method according to any one of claims 22 to 25, characterized in that The method further comprises: The first device receives the second feature vector from the second device.

27. The method according to claim 26, characterized in that The first feature vector is a vector composed of a bit sequence.

28. A model generation method, characterized in that: The method comprises: The second device determines first information indicating information of a level public model related to the first task; The second device sends the first information to the first device, and the first information is used by the first device to generate a user-level model, and the user-level model is used to perform the first task.

29. The method according to claim 28, characterized in that The first information includes at least one of the following: reference information or adjustment information, wherein the reference information is used to indicate relevant information of a level public model referenced by the first device to generate the user-level model, and the adjustment information is used to indicate relevant information for adjusting the level public model.

30. The method according to claim 29, characterized in that The reference information includes at least one of the following: area information, task information of the first task, first indication information, or second indication information; wherein the area information is used to indicate the area in which the first device reuses the level common model, the first indication information is used to indicate the level common model, and the second indication information is used to indicate the reuse method of the level common model.

31. The method according to claim 29 or 30, characterized in that The adjustment information includes at least one of the following: adjustment information of a model parameter of the level common model, adjustment information of a model structure of the level common model, or adjustment information of a quantization accuracy of a model parameter of the level common model.

32. The method according to claim 28, characterized in that The first information includes at least one of the following: a threshold range of a correlation coefficient, or third indication information, wherein the third indication information is used to indicate a reuse mode of the level public model; or The method further comprises: The second device sends second information to the first device, where the second information includes at least one of the following: a threshold range of the correlation coefficient, or the third indication information.

33. The method according to claim 28 or 32, characterized in that The method further comprises: The second device sends a second feature vector to the first device, where the second feature vector is used to characterize the features of the level public model.

34. The method according to any one of claims 22 to 25, 32 and 33, characterized in that The first information is carried in a system message or a user-specific message.

35. The method according to any one of claims 22 to 25, 32, 33, 34, characterized in that The first information includes model parameters of the level common model and / or model structure of the level common model.

36. The method according to any one of claims 20 to 27, 30 to 35, characterized in that The reuse method includes any one of the following: using the level public model as the user-level model; using the model structure of the level public model as the model structure of the user-level model; using the model parameters of the level public model as the model parameters of the user-level model; or using the level public model as a teacher to train the user-level model.

37. The method according to any one of claims 18 to 36, characterized in that The level public model associated with the first task includes a first level public model or a second level public model associated with the first task.

38. The method according to claim 37, characterized in that The first-level public model, the second-level public model and the user-level model are divided according to the size of the execution area corresponding to the first task.

39. The method according to claim 38, characterized in that The first-level public model is a global public model related to the first task; the second-level public model is a cell-level public model related to the first task, a sector-level public model related to the first task, a cell-level public model related to the first task and adapted to terminal devices of different capability levels, or a sector-level public model related to the first task and adapted to terminal devices of different capability levels.

40. The method according to claim 37, characterized in that The first-level public model, the second-level public model and the user-level model are divided according to the master-slave relationship between the tasks that the models are used to perform.

41. The method according to any one of claims 37 to 40, characterized in that The level public model related to the first task includes the second level public model; the method further includes: The third device sends third information to the second device, where the third information is used to indicate the first-level public model; accordingly, The second device receives the third information from the third device; The second device adjusts the first-level public model to obtain the second-level public model.

42. A model indication method, characterized in that: The method comprises: The first device sends a first request to the second device, wherein the first request is used to request a target-level public model; The first device receives first information from the second device, where the first information is used to indicate the target-level public model.

43. The method according to claim 42, characterized in that The method further comprises: The first device receives level public model library information from the second device, where the level public model library information includes an identifier of at least one level public model, and the at least one level public model includes the target level public model.

44. A model indication method, characterized in that: The method comprises: The second device receives a first request from the first device, wherein the first request is used to request the target-level public model; The second device sends first information to the first device, where the first information is used to indicate the target-level public model.

45. The method according to claim 44, characterized in that The method further comprises: The second device sends level public model library information to the first device, where the level public model library information includes an identifier of at least one level public model, and the at least one level public model includes the target level public model.

46. ​​The method according to claim 43 or 45, characterized in that The level public model library information also includes a performance indicator of the at least one level public model.

47. The method according to any one of claims 42 to 46, characterized in that The first request includes an identification of the target-level public model.

48. The method according to claim 47, characterized in that The identifier of the target-level public model includes at least one of the following: a model provider identifier, a service identifier, an area identifier, or an index of the target-level public model in the area indicated by the area identifier; wherein the model provider identifier is used to indicate the provider of the target-level public model, the service identifier is used to indicate a first task, the target-level public model is used to perform the first task, and the area identifier is used to indicate the area where the target-level public model is effective.

49. The method according to any one of claims 42 to 48, characterized in that The first request also includes task information of a first task, and the target-level public model is used by the first device to perform the first task.

50. The method according to any one of claims 42 to 49, characterized in that The target level public model is a first level public model or a second level public model.

51. The method according to claim 50, characterized in that The first-level public model and the second-level public model are divided according to the size of the execution area corresponding to the first task.

52. The method according to claim 51, characterized in that The first-level public model is a global public model, and the second-level public model is a cell-level public model, a sector-level public model, a cell-level public model adapted to terminal devices of different capability levels, or a sector-level public model adapted to terminal devices of different capability levels.

53. The method according to claim 50, characterized in that The first-level public model and the second-level public model are divided according to the master-slave relationship between the tasks that the models are used to perform.

54. The method according to any one of claims 42 to 53, characterized in that The target level public model is a second level public model, and the method further includes: The third device sends second information to the second device, where the second information is used to indicate the first-level public model; correspondingly, the second device receives the second information from the third device; The second device adjusts the first-level public model to obtain the second-level public model.

55. A first device, characterized in that: The first device includes a transceiver module; The transceiver module is used to perform the transceiver operation of the method as described in any one of claims 1 to 4, 8 to 16; or, the transceiver module is used to perform the transceiver operation of the method as described in any one of claims 42, 43, 46 to 53.

56. The first device according to claim 55, characterized in that The first device also includes a processing module; When the transceiver module is used to perform the transceiver operation of the method as claimed in any one of claims 1 to 4, 8 to 16, the processing module is used to perform the processing operation of the method as claimed in any one of claims 1 to 4, 8 to 16; or, When the transceiver module is used to perform the transceiver operation of the method as described in any one of claims 42, 43, 46 to 53, the processing module is used to perform the processing operation of the method as described in any one of claims 42, 43, 46 to 53.

57. A first device, characterized in that: The first device includes a transceiver module and a processing module; the transceiver module is used to perform the transceiver operation of the method as described in any one of claims 18 to 27 and 34 to 41, and the processing module is used to perform the processing operation of the method as described in any one of claims 18 to 27 and 34 to 41.

58. A second device, characterized in that: The second device includes a transceiver module and a processing module; The transceiver module is used to perform the transceiver operation of the method according to any one of claims 5 to 17, and the processing module is used to perform the processing operation of the method according to any one of claims 5 to 17; or, The transceiver module is used to perform the transceiver operation of the method as described in any one of claims 28 to 41, and the processing module is used to perform the processing operation of the method as described in any one of claims 28 to 41.

59. A second device, characterized in that: The second device includes a transceiver module; the transceiver module is used to perform the transceiver operation of the method as described in any one of claims 44 to 54.

60. The second device according to claim 58, characterized in that The second device comprises a processing module, which is used to perform the processing operation of the method as claimed in any one of claims 44 to 54.

61. A device, characterized in that The apparatus comprises a processor configured to execute a computer program or computer instructions in a memory to perform a method as claimed in any one of claims 1 to 54.

62. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a device, the device executes the method according to any one of claims 1 to 54.

63. A communication system, characterized in that: The communication system comprises a first device and a second device; The first device is used to perform the method as described in any one of claims 1 to 4, 8 to 16, 18 to 27, 34 to 41, 42, 43, and 46 to 53; the second device is used to perform the method as described in any one of claims 5 to 17, 28 to 41, and 44 to 54.

Images