Method and apparatus for obtaining training data sets

Abstract

The present application provides a training dataset acquisition method and apparatus. A training device may request a network to transmit a training dataset, and the request information is also sent by the network device to the training device and indicates related information of the first training dataset required by the training device. In other words, in the present application, the training device may indicate the training dataset required by the network device, and the network device may transmit the training dataset indicated by the training device to the training device, and there is no need to continuously deliver the training dataset. This method can reduce waste of air interface resources and air interface overhead and improve the utilization performance of air interface resources.

Description

[Technical Field]

[0001] This application claims priority to Chinese Patent Application No. 202211214685.0, entitled "Method and Apparatus for Acquiring Training Dataset," filed with the State Intellectual Property Office of China on September 30, 2022, and Chinese Patent Application No. 202211247927.6, entitled "Method and Apparatus for Acquiring Training Dataset," filed with the State Intellectual Property Office of China on October 12, 2022, both of which are incorporated herein by reference in their entireties.

[0002] TECHNICAL FIELD Embodiments of the present application relate to the field of communications, and more particularly to a method and apparatus for obtaining a training dataset. [Background technology]

[0003] Currently, an artificial intelligence (AI) model may be deployed on a training device (e.g., a terminal device) for training and updating. When the training device trains an AI model, the network device continuously (e.g., periodically) transmits a training dataset to the training device to support the training device in training the AI ​​model. The network device stops transmitting the training dataset to the training device only after the training device transmits model training completion instruction information to the network device. However, in the process of the training device training the AI ​​model, the training dataset transmitted by the network device to the training device is not needed by the training device, resulting in resource waste. Therefore, how the training device obtains the training dataset when training an AI model becomes a technical problem that needs to be solved. Summary of the Invention [Means for solving the problem]

[0004] The embodiments of the present application provide a training dataset acquisition method to reduce the waste of air interface resources and air interface overhead and improve the utilization performance of air interface resources.

[0005] According to a first aspect, a training dataset acquisition method is provided. The method may be performed by a training device or by a component (e.g., a chip or circuit) of the training device. This is not limited thereto. For example, the training device may be a terminal device.

[0006] The method includes the steps of: transmitting first information to a network device, the first information indicating related information of a first training data set that the training device requests the network device to transmit; The method includes receiving a first training data set from the network device, the first training data set being based on the relevant information indicated by the first information, and the first training data set being used to train an artificial intelligence (AI) model.

[0007] Based on the above technical solution, in the present application, the training device may request the network to send a training dataset, and the request information is also sent by the network device to the training device, indicating related information of the first training dataset required by the training device. In other words, in the present application, the training device may indicate the training dataset required by the network device, and the network device may send the training dataset indicated by the training device to the training device, and there is no need to continuously deliver the training dataset. This method can reduce waste of air interface resources and air interface overhead and improve the utilization performance of air interface resources.

[0008] In a possible implementation, the relevant information includes at least one of the following: information about the size of the first training dataset, configuration information of the inputs of the AI ​​model, or configuration information of the reference signals used to train the AI ​​model.

[0009] Based on the above technical solution, in the present application, the first information may indicate information about the size of the first training dataset, configuration information about the input of the AI ​​model, or configuration information about the reference signal used to train the AI ​​model. In this way, the size of the first training dataset required by the training device can be explicitly or implicitly indicated, so that the network device can distribute the first training dataset based on the instruction of the training device to improve the utilization of air interface resources.

[0010] In a possible implementation, the information about the size of the first training dataset is determined by the training device based on the size of the training dataset required to complete training of the AI ​​model.

[0011] For example, the training device may use historical information to determine the total number of training data sets required when the AI ​​model is trained from an initial state of the AI ​​model (e.g., the initial state of the AI ​​model is 0) to a converged state. For example, the training device may determine, based on historical experience, that the full codebook needs to be swept a total of 60,000 times to obtain a training data set for training the AI ​​model.

[0012] Based on the aforementioned technical solution, in the present application, the training device may determine the number of training datasets required to train the AI ​​model based on historical experience, and indicate the number to the network device, so that the network device distributes the training datasets based on the instruction, thereby reducing the waste of air interface resources.

[0013] In a possible implementation, before the step of transmitting the first information to the network device, the method further includes the steps of determining a first performance of the AI ​​model, and determining information regarding the size of the first training dataset based on the first performance of the AI ​​model and a second performance of the AI ​​model, where the first performance is a current performance of the AI ​​model and the second performance is a target performance of the AI ​​model.

[0014] For example, when monitoring an AI model, the training device may compare the current performance of the AI ​​model obtained by monitoring with model performance in the historical information corresponding to the AI ​​model in a converged state, and estimate the size of the first training data set required to achieve the expected model performance based on the current performance of the AI ​​model.

[0015] Based on the above technical solution, in the present application, the training device may estimate the number of training datasets required when the AI ​​model is trained to a convergence state based on the performance of the AI ​​model, and indicate the number to the network device, so that the network device distributes the training datasets based on the instruction, thereby reducing the waste of air interface resources.

[0016] In a possible implementation, the configuration information of the reference signal includes at least one of the following: an identifier of the reference signal, a time domain resource of the reference signal, a frequency domain resource of the reference signal, a transmission periodicity of the reference signal, or a type of transmitted reference signal.

[0017] For example, the type of the reference signal is SSB, CSI-RS, or SRS. The identifier of the reference signal may also be understood as an identifier of a reference signal group. For example, the configuration information of the reference signal includes group identifiers of N (N is an integer greater than or equal to 1) reference signal groups, where each reference signal group in the N reference signal groups has the same group identifier and includes at least one reference signal. Similarly, the time domain resources of the reference signals, the frequency domain resources of the reference signals, the transmission periodicity of the reference signals, and the type of the transmitted reference signals may also be understood as the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, the transmission periodicity of the N reference signal groups, and the type of the N transmitted reference signal groups.

[0018] Based on the aforementioned technical solution, in the present application, since the training dataset may be a measurement result of a reference signal, the training device may determine the configuration information of the reference signal based on historical information, and may use the configuration information of the reference signal to indirectly indicate the number of training datasets required.

[0019] In the present application, "configuration information for the input of the AI ​​model" may be understood as, for example, a training device determining the input information of the AI ​​model based on historical information related to AI model training. For example, the input information of the AI ​​model is a measurement result of a reference signal corresponding to a sparse beam pattern. In this case, the training device may determine that the sparse beam pattern includes a beam at a specific position in the full codebook. For example, the training device may report information such as an identifier of the sparse beam pattern, an identifier of the reference signal corresponding to the sparse beam pattern, or a measurement result of the reference signal to the first network device, so that the first network device can distribute a training data set corresponding to the beam pattern.

[0020] Based on the above technical solutions, in the present application, since the input information of the AI ​​model can be the measurement result of the reference signal, the training device may determine the configuration information of the input of the AI ​​model based on historical information, and may indirectly indicate the number of required training datasets by using the configuration information of the input of the AI ​​model.

[0021] In a possible implementation, the first information includes at least one of the following: identification information of the AI ​​model, information regarding the application scenario of the AI ​​model, usage information of the AI ​​model, or information regarding the computing capabilities of the training device.

[0022] In the present application, for example, the network device may store a first mapping relationship, where the first mapping relationship is a correspondence relationship between an identifier of each AI model and a size of a training dataset corresponding to the identifier of the AI ​​model. The training device may indicate the size of the training dataset required to train the AI ​​model by indicating the identifier of the AI ​​model to the network device.

[0023] It should be noted that the "mapping relationship" in this application may also be expressed as an "association relationship" or a "correspondence." It should be understood that the "mapping relationship" in the embodiments of this application may be stored or recorded using a function relationship, a table, a mapping relationship, etc. In the following embodiments, the "mapping relationship" may be configured by a network device or may be predefined in a protocol, etc. This is not limited thereto.

[0024] An application scenario of an AI model or a use of an AI model may be understood as the AI ​​model being used in a beam management scenario, a CSI feedback scenario, a positioning scenario, etc. The training device may indicate the size of the training dataset required for training the AI ​​model by indicating the application scenario of the AI ​​model or the use of the AI ​​model to the network device.

[0025] In the present application, the training device may further report its computing power. For example, the information regarding the computing power reported by the training device may include at least one of the following: the capabilities of the training device's processor (e.g., a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), a neural network processing unit (NPU), a field-programmable gate array (FPGA), etc.), the size of the training device's storage space, the size of the training device's memory, the battery level of the training device, etc. This is not limited to this. The training device may indicate the maximum number of training datasets that can be processed during training of the AI ​​model by reporting the training device's computing power to the network device.

[0026] In a possible implementation, the method further includes the steps of training an AI model based on the first training dataset and determining performance of the AI ​​model; transmitting second information to the network device based on the performance of the AI ​​model, wherein the second information indicates related information of the second training dataset that the training device requests the network device to transmit; and receiving the second training dataset from the network device, wherein the second training dataset is a training dataset based on the related information indicated by the second information, and the second training dataset is used to train the AI ​​model.

[0027] For example, the amount of data in the second training dataset may be less than the amount of data in the first training dataset. The training device may then further train the AI ​​model based on the second training dataset and perform repeated iterations. For example, the training device may perform model training based on the second training dataset, redetermine the performance of the AI ​​model, and determine the size of the required third training dataset based on the performance of the AI ​​model. It is assumed that the training device may perform training L times (L is an integer greater than 1) until the training device determines that the AI ​​model has converged ("the model has converged" may also be understood as the AI ​​model reaching target performance).

[0028] Based on the aforementioned technical solution, in the present application, the training device may determine the size of the training dataset required for the next training by measuring the AI ​​model training performance, so that the AI ​​model training efficiency may be improved and the utilization performance of air interface resources may be improved.

[0029] According to a second aspect, there is provided a training dataset acquisition method, which may be performed by a first network device or a component (e.g., a chip or circuit) of the first network device, but is not limited thereto.

[0030] For the same beneficial effects corresponding to the technical solution on the network side, please refer to the description of the beneficial effects on the training device side, and the details will not be described again in this specification.

[0031] The method includes the steps of receiving first information from a training device, the first information indicating relevant information of a first training dataset that the first network device is requested to transmit, and transmitting the first training dataset to the training device based on the relevant information indicated by the first information, the first training dataset being used to train an artificial intelligence AI model.

[0032] In a possible implementation, the relevant information includes at least one of the following: information about the size of the first training dataset, configuration information of the inputs of the AI ​​model, or configuration information of the reference signals used to train the AI ​​model.

[0033] In a possible implementation, the information about the size of the first training dataset is determined based on the size of the training dataset required to complete the training of the AI ​​model.

[0034] In a possible implementation, the configuration information of the reference signal includes at least one of the following: an identifier of the reference signal, a time domain resource of the reference signal, a frequency domain resource of the reference signal, a transmission periodicity of the reference signal, or a type of transmitted reference signal.

[0035] In a possible implementation, the first information includes at least one of the following: identification information of the AI ​​model, information regarding the application scenario of the AI ​​model, usage information of the AI ​​model, or information regarding the computing capabilities of the training device.

[0036] In a possible implementation, the method further includes a step of acquiring third information from the second network device, wherein the third information is relevant information for training the AI ​​model, and the first network device is a target network device to which the training device is to be handed over from the second network device, and the step of sending a first training data set to the training device based on the relevant information indicated by the first information includes a step of sending the first training data set to the training device based on the relevant information indicated by the first information and the third information.

[0037] Based on the aforementioned technical solution, in the present application, the first network device may, in combination with various information, comprehensively determine whether to support training of the AI ​​model, and the network device does not continuously deliver the training data set to the training device, so that unnecessary occupation of air interface resources may be reduced, air interface overhead may be reduced, and air interface resource utilization performance may be improved.

[0038] In a possible implementation, the third information includes at least one of the following: information regarding the size of the training dataset that the training device requests the second network device to transmit; information regarding the size of the training dataset required to complete training of the AI ​​model; identification information of the AI ​​model; or information regarding the computing capabilities of the training device.

[0039] For example, the first network device stores a first mapping relationship, where the first mapping relationship is a mapping relationship between an identifier of an AI model and a size of a training dataset corresponding to the identifier of the AI ​​model.

[0040] For example, "information regarding the size of the training dataset that the training device requests the second network device to transmit" may be understood as the training device requesting information regarding the size of the required training dataset from the second network device when the second network device is connected to the training device. For example, the training device may also determine the size of the training dataset requested from the second network device based on historical information. For example, "information regarding the size of the training dataset required to complete training of the AI ​​model" may be understood as the second network device storing the total size of the training dataset required to train the AI ​​model by the training device.

[0041] Based on the aforementioned technical solution, in the present application, the first network device may obtain information for training an AI model from the second network device, so that the first network device can jointly determine the size of training data that can be distributed to the training device based on the instruction information of the training device and the information synchronized from the second network device, so as to improve the utilization performance of air interface resources.

[0042] In a possible implementation, the third information further includes information regarding a duration for transmitting the training dataset and / or information regarding a method for transmitting the training dataset, and the method further includes determining whether the first network device and / or the training device have the capability to support training of the AI ​​model based on the third information and resource usage of the first network device.

[0043] The "information regarding how to transmit the training dataset" may be understood as, for example, the second network device periodically transmitting the training dataset to the training device. For example, the network device may identify a time period in a day when the data demand volume is the smallest, which may alternatively be understood as a time period when the minimum air interface resources are occupied or when the air interface resources are sufficient. In this case, the training dataset may be provided for the AI ​​model. For example, the network device may select a time period each day to support the update of the AI ​​model (the number of training devices and / or the number of AI models is not limited herein). In another example, the second network device transmits the training dataset to the training device at intervals. Compared with periodic transmission, this solution has some improvement in flexibility. If the network device discovers and / or determines that the current data demand volume is relatively small and the air interface resources are sufficient, the network device determines that the network device can support the update of the AI ​​model. In other words, in this implementation, the network device may decide to transmit the training dataset to the training device based on the current air interface resource occupancy status. Therefore, this implementation does not have an obvious periodic feature.

[0044] In this application, "resource usage status" may also be understood as "resource occupancy status," "occupancy of air interface resources," etc. For example, existing protocol frameworks define the maximum number of reference signals that can be configured by a network device (e.g., 64 CSI-RS). If a network device finds that all reference signals are configured for another function, the network device may determine that the current resources are occupied and cannot configure reference signal resources for a training device to support updating an AI model.

[0045] Based on the above technical solution, in the present application, the first network device may obtain information for training an AI model from the second network device, so that the first network device can jointly determine the size of training data that can be distributed to the training device based on the instruction information of the training device, the information synchronized from the second network device, and the occupancy status of the air interface resource, so as to improve the utilization of the air interface resource.

[0046] According to a third aspect, there is provided a training dataset acquisition method, which may be performed by a first network device or a component (e.g., a chip or circuit) of the first network device, but is not limited thereto.

[0047] The method includes: a first network device obtaining third information from a second network device, the third information being relevant information for training an artificial intelligence (AI) model, the first network device being a target network device to which the training device is handed over from the second network device; the first network device receiving the first information from the training device, the first information being used to request the first network device to send a training dataset; the first network device determining a first training dataset to be sent based on the third information; the first network device sending the first training dataset to the training device based on the first information, and the first training dataset being used to train the AI ​​model.

[0048] Based on the above technical solution, in the present application, the first network device may obtain information for training an AI model from the second network device, so that the first network device can jointly determine the size of training data that can be distributed to the training device based on the instruction information of the training device and the information synchronized from the second network device, so as to improve the utilization performance of air interface resources.

[0049] In a possible implementation, the third information includes at least one of the following: information regarding the size of the training dataset that the training device requests the second network device to transmit; information regarding the size of the training dataset required to complete training of the AI ​​model; identification information of the AI ​​model; or information regarding the computing capabilities of the training device.

[0050] In a possible implementation, the first network device stores a first mapping relationship, the first mapping relationship being a mapping relationship between an identifier of the AI ​​model and a size of a training dataset corresponding to the identifier of the AI ​​model.

[0051] In a possible implementation, the third information further includes information regarding a duration for transmitting the training data set and / or information regarding a method for transmitting the training data set. The method further includes, based on the third information and resource usage of the first network device, determining, by the first network device, whether the first network device and / or the training device have the capability to support training of the AI ​​model.

[0052] In a possible implementation, determining a first training data set to be transmitted by the first network device based on the third information includes determining a first training data set to be transmitted by the first network device based on the first information and the third information, wherein the first information includes indicating related information of the first training data set that the first network device is requested to transmit.

[0053] In a possible implementation, the relevant information includes at least one of the following: information about the size of the first training dataset, input information for the AI ​​model, or configuration information for the reference signal used to train the AI ​​model.

[0054] In a possible implementation, the information about the size of the first training dataset is determined based on the size of the training dataset required to complete the training of the AI ​​model.

[0055] In a possible implementation, the configuration information of the reference signal includes at least one of the following: an identifier of the reference signal, a time domain resource of the reference signal, a frequency domain resource of the reference signal, a transmission periodicity of the reference signal, or a type of transmitted reference signal.

[0056] In a possible implementation, the first information includes at least one of the following: identification information of the AI ​​model, information regarding the application scenario of the AI ​​model, usage information of the AI ​​model, or information regarding the computing capabilities of the training device.

[0057] In a possible implementation, the method further includes receiving second information from the training device, where the second information indicates related information of a second training dataset that the first network device is requested to transmit, and the second information is determined based on performance of the AI ​​model, the performance of the AI ​​model being determined by training based on the first training dataset; and determining the second training dataset to be transmitted based on the second information.

[0058] According to a fourth aspect, there is provided a communication method, which may be performed by a training device or by a component (e.g., a chip or circuit) of the training device, without being limited thereto.

[0059] The method includes: measuring N reference signal groups to obtain N measurement result groups corresponding to the N reference signal groups, where each reference signal group within the N reference signal groups includes at least one reference signal, each reference signal group having the same group identifier, and N is an integer greater than 1; receiving fourth information from the network device, where the fourth information indicates M reference signal groups among the N reference signal groups; and determining first input information for an artificial intelligence (AI) model based on the fourth information and the N measurement result groups corresponding to the N reference signal groups, where the first input information includes the M measurement result groups corresponding to the M reference signal groups. The AI ​​model is used to obtain first output information based on the first input information, where the first output information includes group identifiers for each of K reference signal groups within the N reference signal groups, where each group identifier for the K reference signal groups corresponds to the K measurement result group with the best channel quality within the N measurement result groups. Each measurement result group may include one or more measurement results.

[0060] Based on the above technical solution, in the present application, in multiple subsequent training processes, the sparse beam pattern may be the beam pattern indicated by the fourth information. During each training, the training device may perform one full codebook sweep based on the training dataset distributed by the first network device. Because the channel condition (which can also be understood as the channel environment) varies over time, the measurement results of the reference signals obtained after each full codebook sweep are imperfectly the same. Therefore, during each training, M reference signal groups within N reference signal groups correspond to different measurement results, and the training labels determined by the training device are also different. In other words, the input information and training labels of the AI ​​model change accordingly. However, these changes are essentially caused by changes in the channel condition, and the beam pattern does not change. In other words, in the solution provided in the present application, the only variable in the AI ​​model training process is the channel condition. Compared to another solution in which both the beam pattern and the channel condition change during the AI ​​model training process, the solution provided in the present application can accelerate the convergence speed of the AI ​​model, improve model training efficiency, and reduce air interface resource occupation.

[0061] In the present application, the group identifier of one group of reference signals may correspond to one beam identifier, and N group identifiers of N reference signal groups correspond to N beam identifiers.

[0062] In a possible implementation, the first output information further includes group identifiers for each of the remaining (NK) reference signal groups within the N measurement result groups, and each group identifier for the (NK) reference signal groups corresponds to one of the (NK) measurement result groups.

[0063] In the present application, classification methods and regression methods may be used during AI model training, with different training methods corresponding to different input and output information for the AI ​​model. For example, in a classification method, the input information for the AI ​​model is the measurement results of a reference signal, and the output information is the identifiers of the K beams with the best channel qualities in the full codebook predicted by the AI ​​model. As another example, in a regression method, the input information for the AI ​​model is the measurement results of a reference signal (e.g., the RSRP, RSRQ, and SINR of the reference signal), and the output information is the measurement results of all reference signals in the full codebook predicted by the AI ​​model.

[0064] In a possible implementation, the fourth information includes N fields, the N fields correspond one-to-one to the N reference signal groups, and the bit values ​​of M fields within the N fields are different from the bit values ​​of the remaining (NM) fields, and the fourth information indicating M reference signal groups among the N reference signal groups specifically includes the M fields in the fourth information indicating M reference signal groups.

[0065] Based on the above technical solution, in this application, the network device may indicate the sparse beam pattern to the training device by indicating the bit value of each field in the fourth information. In other words, the training device may obtain input information for the AI ​​model by analyzing the fourth information, so as to accelerate the convergence speed of the AI ​​model and improve the training efficiency of the AI ​​model.

[0066] In a possible implementation, the method further includes a step of receiving fifth information from the network device, wherein the fifth information indicates P reference signal groups among the N reference signal groups, and the fifth information includes N fields, the N fields having a one-to-one correspondence with the N reference signal groups, and the bit values ​​of P fields among the N fields are different from the bit values ​​of the remaining (NP) fields, and the fifth information indicating P reference signal groups among the N reference signal groups specifically includes the P fields in the fifth information indicating the P reference signal groups.

[0067] Based on the above technical solution, in this application, the network device may indicate multiple sparse beam patterns to the training device by indicating bit values ​​of each field in the fourth information and the fifth information. In other words, the training device may obtain multiple input information for the AI ​​model by analyzing the fourth information and the fifth information, so as to accelerate the convergence speed of the AI ​​model and improve the training efficiency of the AI ​​model.

[0068] In a possible implementation, the method further includes receiving configuration information from the network device, wherein the configuration information indicates one or more of the following: time domain resources for the N reference signal groups; frequency domain resources for the N reference signal groups; transmission periodicities for the N reference signal groups; or group identifiers for the N reference signal groups.

[0069] For example, the configuration information and the fourth information and / or the fifth information may also be transmitted in the same message. This is not limited thereto.

[0070] Based on the aforementioned technical solution, in the present application, a network device may send configuration information of N reference signal groups to a training device, so that the training device can obtain a training dataset by measuring the N reference signal groups to train an AI model.

[0071] According to a fifth aspect, there is provided a communication method, which may be performed by a network device or a component (e.g., a chip or circuit) of the network device, without being limited thereto.

[0072] The method includes transmitting N reference signal groups to a training device, where each reference signal group among the N reference signal groups includes at least one reference signal, each reference signal group has the same group identifier, and N is an integer greater than 1; and transmitting fourth information to the training device, where the fourth information indicates M reference signal groups among the N reference signal groups, and the M reference signal groups are used to determine first input information, wherein the AI ​​model is used to obtain first output information based on the first input information, the first output information including respective group identifiers of K reference signal groups among the N reference signal groups, and the respective group identifiers of the K reference signal groups correspond to the K measurement result groups with best channel quality among the N measurement result groups corresponding to the N reference signal groups. Each measurement result group may include one or more measurement results.

[0073] In a possible implementation, the first output information further includes group identifiers for each of the remaining (NK) reference signal groups within the N measurement result groups, and each group identifier for the (NK) reference signal groups corresponds to one of the (NK) measurement result groups.

[0074] In a possible implementation, the fourth information includes N fields, the N fields correspond one-to-one to the N reference signal groups, and the bit values ​​of M fields within the N fields are different from the bit values ​​of the remaining (NM) fields, and the fourth information indicating M reference signal groups among the N reference signal groups specifically includes the M fields in the fourth information indicating M reference signal groups.

[0075] In a possible implementation, the method further includes a step of transmitting fifth information to the training device, wherein the fifth information indicates P reference signal groups among the N reference signal groups, the fifth information includes N fields, the N fields correspond one-to-one to the N reference signal groups, and the bit values ​​of P fields among the N fields are different from the bit values ​​of the remaining (NP) fields, and the fifth information indicating P reference signal groups among the N reference signal groups specifically includes the P fields in the fourth information indicating P reference signal groups.

[0076] In a possible implementation, the method further includes a step of transmitting configuration information to the training device, wherein the configuration information indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or group identifiers of the N reference signal groups.

[0077] According to a sixth aspect, there is provided a communication method, which may be performed by a training device or a component (e.g., a chip or circuit) of the training device, without being limited thereto.

[0078] The method includes steps of receiving a second reference signal set, the second reference signal set including N reference signal groups, each reference signal group including at least one reference signal, where N is an integer greater than 1; and receiving second beam designation information, the second beam designation information indicating beams corresponding to the first reference signal set, the beams corresponding to the first reference signal set being a subset of a plurality of beams corresponding to the second reference signal set, the beams corresponding to the first reference signal set being used to determine first input information for an AI model in the training device; The method includes a step in which first input information is based on measurement results of beams corresponding to a first reference signal set, the first reference signal set including M reference signal groups, N being an integer greater than M and M being an integer greater than or equal to 1; an AI model is used to obtain first output information based on the first input information, the first output information indicating K beams predicted to have the best channel quality among a plurality of beams corresponding to a second reference signal set, K being an integer greater than or equal to 1 and less than N; and the AI ​​model labels the K beams having the best channel quality in the measurement results of the second reference signal set.

[0079] Based on the above technical solution, in the present application, the training device may determine a sparse beam pattern using the received second beam instruction information. It can also be understood that the training device may determine a specific beam in the full codebook that constitutes the sparse beam pattern based on the second beam instruction information, and may further determine input information for the AI ​​model. In this case, the sparse beam pattern does not change, and only the channel conditions change during the training process. Therefore, this solution can accelerate the convergence of the AI ​​model and improve the training efficiency of the AI ​​model.

[0080] In a possible implementation, the first output information indicating the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set includes at least one of the following: information regarding the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set; or group identifiers for each of the K reference signal groups, where the K reference signal groups correspond to the K measurement results predicted to have the best channel quality among the N measurement results corresponding to the N reference signal groups, and where a predefined or preconfigured correspondence exists between the group identifiers for each of the K reference signal groups and the K beams; or multiple beam information corresponding to the N reference signal groups and N measurement results corresponding to the beam information; or group identifiers for each of the N reference signal groups and N measurement results for the N reference signal groups, where a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

[0081] Based on the above technical solution, in the present application, the output information of the AI ​​model may be slightly different depending on the implementation form of the algorithm of the AI ​​model. For example, the output information of the AI ​​model in the classification method is information about K (K is an integer greater than 0) beams that are predicted to have the best channel quality among multiple beams corresponding to the second reference signal set. In another example, the output information of the AI ​​model in the regression method is N measurement results corresponding to N reference signal groups.

[0082] In a possible implementation, the second beam instruction information indicating the beam corresponding to the first set of reference signals includes the second beam instruction information indicating the position of the beam corresponding to the first set of reference signals within a plurality of beams corresponding to the second set of reference signals.

[0083] Based on the above technical solution, in the present application, the second beam designation information may indicate the position of a beam corresponding to the first reference signal set among a plurality of beams corresponding to the second reference signal set to indicate a sparse beam pattern. It may also be understood that the training device may determine a beam in the full codebook corresponding to a reference signal in a reference signal group in the first reference signal set based on the second beam designation information to determine input information for the AI ​​model.

[0084] In a possible implementation, the second beam instruction information includes N fields, the N fields having a one-to-one correspondence with multiple beams corresponding to the second reference signal set, and the bit values ​​of M fields among the N fields are different from the bit values ​​of the remaining (NM) fields, and the second beam instruction information indicating a beam corresponding to the first reference signal set includes that the M fields in the second beam instruction information correspond to the first reference signal set.

[0085] Based on the above technical solution, it can be understood that in the present application, M fields among N fields may directly indicate a sparse beam pattern, or M fields among N fields may directly indicate specific beams in a full codebook that constitute a sparse beam pattern, so that the training device can determine the input information of the AI ​​model.

[0086] In a possible implementation, the method further includes a step of transmitting first configuration information to the training device, wherein the first configuration information indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, group identifiers of the N reference signal groups, or beam information of the N reference signal groups.

[0087] Based on the aforementioned technical solution, in the present application, the first configuration information may indicate how the training device needs to receive the N reference signal groups, for example, the time-frequency resources on which the training device needs to receive the N reference signal groups.

[0088] In a possible implementation, the second beam instruction information indicating beams corresponding to the M reference signal groups includes the second beam instruction information including group identifiers or beam information for the M reference signal groups, the M reference signal groups being part of N reference signal groups, and a predefined or preconfigured correspondence existing between the N reference signal groups and the N beams.

[0089] Based on the above technical solution, since there is a one-to-one correspondence between N reference signal groups and N beams, a sparse beam pattern may be indicated by indicating group identifiers or beam information of M reference signal groups.

[0090] In a possible implementation, third configuration information of the N reference signal groups is transmitted to the training device. When the second beam designation information includes group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes the group identifiers of each of the N reference signal groups and indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or beams of the N reference signal groups. The M reference signal groups being part of the N reference signal groups means that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups, or when the second beam instruction information includes beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes beam information for each of the N reference signal groups and indicates one or more of the following: group identifiers of the N reference signal groups, time domain resources, frequency domain resources of the N reference signal groups, or transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups means that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups.

[0091] Based on the above technical solution, in the present application, the third configuration information may include identifiers of N reference signal groups, and the M reference signal groups belong to the N reference signal groups. Therefore, the second beam instruction information may include identifiers of the M reference signal groups or beam information of the M reference signal groups to indicate specific beams in the full codebook that constitute a sparse beam pattern, so that the training device can determine input information for the AI ​​model.

[0092] In a possible implementation, the method further includes a step of measuring N reference signal groups to obtain N measurement results, the N measurement results corresponding to N beams, and the N measurement results including measurement results of beams corresponding to the first reference signal set.

[0093] Based on the above technical solution, in the present application, the training device may obtain N measurement result groups by measuring N reference signal groups, and determine input information for the AI ​​model based on second beam information.

[0094] According to a seventh aspect, there is provided a communication method. The method may be performed by a network device or a component (e.g., a chip or circuit) of the network device. This is not limited thereto.

[0095] The method includes the steps of transmitting a second reference signal set to a training device, the second reference signal set including N reference signal groups, each reference signal group including at least one reference signal, where N is an integer greater than 1; and transmitting second beam designation information to the training device, the second beam designation information indicating beams corresponding to the first reference signal set, the beams corresponding to the first reference signal set being a subset of a plurality of beams corresponding to the second reference signal set, and the beams corresponding to the first reference signal set being used to determine first input information for an AI model in the training device. the AI ​​model is used to obtain first output information based on the first input information, the first output information indicating K beams predicted to have best channel qualities among a plurality of beams corresponding to a second reference signal set, K being an integer greater than or equal to 1 and less than N, and labels of the AI ​​model are the K beams having best channel qualities in the measurement results of the second reference signal set.

[0096] In a possible implementation, the first output information indicating the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set includes at least one of the following: information regarding the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set; or group identifiers for each of the K reference signal groups, where the K reference signal groups correspond to the K measurement results predicted to have the best channel quality among the N measurement results corresponding to the N reference signal groups, and where a predefined or preconfigured correspondence exists between the group identifiers for each of the K reference signal groups and the K beams; or multiple beam information corresponding to the N reference signal groups and N measurement results corresponding to the beam information; or group identifiers for each of the N reference signal groups and N measurement results for the N reference signal groups, where a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

[0097] In a possible implementation, the second beam instruction information indicating the beam corresponding to the first set of reference signals includes the second beam instruction information indicating the position of the beam corresponding to the first set of reference signals within a plurality of beams corresponding to the second set of reference signals.

[0098] In a possible implementation, the second beam instruction information includes N fields, the N fields having a one-to-one correspondence with multiple beams corresponding to the second reference signal set, and the bit values ​​of M fields among the N fields are different from the bit values ​​of the remaining (NM) fields, and the second beam instruction information indicating a beam corresponding to the first reference signal set includes that the M fields in the second beam instruction information correspond to the first reference signal set.

[0099] In a possible implementation, the method further includes a step of transmitting first configuration information to the training device, wherein the first configuration information indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, group identifiers of the N reference signal groups, or beam information of the N reference signal groups.

[0100] In a possible implementation, the second beam instruction information indicating beams corresponding to the M reference signal groups includes the second beam instruction information including group identifiers or beam information for the M reference signal groups, the M reference signal groups being part of N reference signal groups, and a predefined or preconfigured correspondence existing between the N reference signal groups and the N beams.

[0101] In a possible implementation, the method further includes transmitting third configuration information of the N reference signal groups to the training device. When the second beam designation information includes group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes the group identifiers of each of the N reference signal groups and indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or beams of the N reference signal groups. The M reference signal groups being part of the N reference signal groups means that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups, or when the second beam instruction information includes beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes beam information for each of the N reference signal groups and indicates one or more of the following: group identifiers of the N reference signal groups, time domain resources, frequency domain resources of the N reference signal groups, or transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups means that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups.

[0102] According to an eighth aspect, there is provided a communication method. The method may be performed by a terminal device or a component (e.g., a chip or circuit) of the terminal device. This is not limited thereto. The terminal device may function as an inference device.

[0103] The method includes the steps of receiving a first reference signal set, the first reference signal set including M reference signal groups, each reference signal group including at least one reference signal, where M is an integer greater than or equal to 1; and receiving first beam indication information, the first beam indication information indicating beams corresponding to the first reference signal set, and the first reference signal set being used to determine first input information of an AI model, the first input information being based on measurement results of the M reference signal groups included in the first reference signal set, the beams corresponding to the first reference signal set being a subset of a plurality of beams corresponding to a second reference signal set, the second reference signal set including N reference signal groups, where N is an integer greater than or equal to M; and the AI ​​model is used to obtain first output information based on the first input information, the first output information indicating K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set, where K is an integer greater than or equal to 1 and K is less than N.

[0104] Based on the above technical solution, in this application, in the model inference phase, the network device may indicate the input information of the model to the terminal device, so that the terminal device determines the input information of the model, thereby improving the accuracy of the information output by the terminal device during model inference.

[0105] In a possible implementation, the first output information indicating the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set includes at least one of the following: information regarding the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set; or group identifiers for each of the K reference signal groups, where the K reference signal groups correspond to the K measurement results predicted to have the best channel quality among the N measurement results corresponding to the N reference signal groups, and where a predefined or preconfigured correspondence exists between the group identifiers for each of the K reference signal groups and the K beams; or multiple beam information corresponding to the N reference signal groups and N measurement results corresponding to the beam information; or group identifiers for each of the N reference signal groups and N measurement results for the N reference signal groups, where a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

[0106] Based on the above technical solution, in the present application, the inferred output information of the AI ​​model may vary slightly depending on the implementation of the algorithm of the AI ​​model. For example, the inferred output information of the AI ​​model in the classification method is information about K (K is an integer greater than 0) beams predicted to have the best channel quality among multiple beams corresponding to the second reference signal set. In another example, the inferred output information of the AI ​​model in the regression method is N measurement results corresponding to N reference signal groups.

[0107] In a possible implementation, the first beam instruction information indicating a beam corresponding to the first set of reference signals includes the first beam instruction information indicating a position of the beam corresponding to the first set of reference signals within a plurality of beams corresponding to the second set of reference signals.

[0108] Based on the above technical solution, in the present application, the first beam designation information may indicate the position of a beam corresponding to the first reference signal set among a plurality of beams corresponding to the second reference signal set to indicate a sparse beam pattern. It may also be understood that the terminal device may determine, based on the first beam designation information, a beam in the full codebook corresponding to a reference signal in a reference signal group in the first reference signal set to determine input information for the AI ​​model.

[0109] In a possible implementation, the first beam instruction information includes N fields, the N fields having a one-to-one correspondence with multiple beams corresponding to the second reference signal set, and the bit values ​​of M fields among the N fields are different from the bit values ​​of the remaining (NM) fields, and the first beam instruction information indicating a beam corresponding to the first reference signal set includes that the M fields in the first beam instruction information correspond to the first reference signal set.

[0110] Based on the above technical solution, it can be understood that in the present application, M fields out of N fields may directly indicate a sparse beam pattern, or M fields out of N fields may directly indicate specific beams in a full codebook that constitute a sparse beam pattern, so that the terminal device can determine the input information of the AI ​​model.

[0111] In a possible implementation, the method further includes receiving first configuration information, wherein the first configuration information indicates one or more of the following: time domain resources of the M reference signal groups, frequency domain resources of the M reference signal groups, transmission periodicities of the M reference signal groups, group identifiers of the M reference signal groups, or beam information of the M reference signal groups.

[0112] Based on the aforementioned technical solution, in the present application, the first configuration information may indicate how the terminal device needs to receive the M reference signal groups, for example, the time-frequency resources on which the terminal device needs to receive the M reference signal groups.

[0113] In a possible implementation, the first beam designation information indicates beams corresponding to the M reference signal groups. In a possible implementation, the first beam designation information indicating beams corresponding to the M reference signal groups includes the first beam designation information including group identifiers or beam information for the M reference signal groups, the M reference signal groups being part of N reference signal groups, and a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

[0114] Based on the above technical solution, since there is a one-to-one correspondence between N reference signal groups and N beams, a sparse beam pattern may be indicated by indicating group identifiers or beam information of M reference signal groups.

[0115] In a possible implementation, the first beam direction information is included in second configuration information of the M reference signal groups, and when the first beam direction information includes group identifiers of the M reference signal groups, the second configuration information further includes one or more of time domain resources, frequency domain resources, transmission periodicity, or beam information of the M reference signal groups, or when the first beam direction information includes beam information of the M reference signal groups, the second configuration information further includes one or more of time domain resources, frequency domain resources, transmission periodicity, or group identifiers of the M reference signal groups.

[0116] Based on the above technical solution, in the present application, the second configuration information may include first beam indication information, where the first beam indication information includes group identifiers of M reference signal groups or beam information of M reference signal groups, and indicates a specific beam in a full codebook that constitutes a sparse beam pattern, so that the terminal device can determine the input information of the AI ​​model.

[0117] In one possible implementation form, the method further includes receiving third configuration information of the N reference signal groups. When the first beam instruction information includes group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes group identifiers of each of the N reference signal groups and indicates one or more of the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, the transmission periodicity of the N reference signal groups, or the beams of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes the N group identifiers of the N reference signal groups including the M group identifiers of the M reference signal groups; or, when the first beam instruction information includes beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes beam information of each of the N reference signal groups and indicates one or more of the group identifiers of the N reference signal groups, the time domain resources, the frequency domain resources of the N reference signal groups, or the transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes the N group identifiers of the M reference signal groups.

[0118] Based on the above technical solution, in the present application, the third configuration information may include identifiers of N reference signal groups, and the M reference signal groups belong to the N reference signal groups. Therefore, the first beam indication information may include identifiers of the M reference signal groups or beam information of the M reference signal groups to indicate specific beams in the full codebook that constitute a sparse beam pattern, so that the terminal device can determine input information for the AI ​​model.

[0119] In a possible implementation, the method further includes obtaining first output information based on the first input information using the AI ​​model, and transmitting the first output information.

[0120] Based on the above technical solution, in this application, after obtaining the inference output information through AI model inference, the terminal device may further feed back the output information to the network device, so that the network device can send a corresponding reference signal to the terminal device based on the output information. The terminal device measures the reference signal again, determines the reference signal with the best measurement result, uses the beam identifier corresponding to the reference signal as the finally selected beam, and uses the beam to communicate with the network device.

[0121] According to a ninth aspect, there is provided a communication method. The method may be performed by a network device or by a component (e.g., a chip or circuit) of the network device. This is not limited thereto.

[0122] The method includes the steps of: transmitting a first reference signal set to a terminal device, the first reference signal set including M reference signal groups, each reference signal group including at least one reference signal, where M is an integer greater than or equal to 1; and transmitting first beam indication information to the terminal device, the first beam indication information indicating beams corresponding to the first reference signal set, and the first reference signal set being used to determine first input information of an AI model, the first input information being based on measurement results of the M reference signal groups included in the first reference signal set, the beams corresponding to the first reference signal set being a subset of a plurality of beams corresponding to a second reference signal set, the second reference signal set including N reference signal groups, where N is an integer greater than or equal to M; and the AI ​​model is used to obtain first output information based on the first input information, the first output information indicating K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set, where K is an integer greater than or equal to 1 and K is less than N.

[0123] In a possible implementation, the first output information indicating the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set includes at least one of the following: information regarding the K beams predicted to have the best channel quality among the multiple beams corresponding to the second reference signal set; or group identifiers for each of the K reference signal groups, where the K reference signal groups correspond to the K measurement results predicted to have the best channel quality among the N measurement results corresponding to the N reference signal groups, and where a predefined or preconfigured correspondence exists between the group identifiers for each of the K reference signal groups and the K beams; or multiple beam information corresponding to the N reference signal groups and N measurement results corresponding to the beam information; or group identifiers for each of the N reference signal groups and N measurement results for the N reference signal groups, where a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

[0124] In a possible implementation, the first beam instruction information indicating a beam corresponding to the first set of reference signals includes the first beam instruction information indicating a position of the beam corresponding to the first set of reference signals within a plurality of beams corresponding to the second set of reference signals.

[0125] In a possible implementation, the first beam instruction information includes N fields, the N fields having a one-to-one correspondence with multiple beams corresponding to the second reference signal set, and the bit values ​​of M fields among the N fields are different from the bit values ​​of the remaining (NM) fields, and the first beam instruction information indicating a beam corresponding to the first reference signal set includes that the M fields in the first beam instruction information correspond to the first reference signal set.

[0126] In a possible implementation, the method further includes a step of transmitting first configuration information to a terminal device, wherein the first configuration information indicates one or more of the following: time domain resources of the M reference signal groups, frequency domain resources of the M reference signal groups, transmission periodicities of the M reference signal groups, group identifiers of the M reference signal groups, or beam information of the M reference signal groups.

[0127] In a possible implementation, the first beam designation information indicates beams corresponding to the M reference signal groups. In a possible implementation, the first beam designation information indicating beams corresponding to the M reference signal groups includes the first beam designation information including group identifiers or beam information for the M reference signal groups, the M reference signal groups being part of N reference signal groups, and a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

[0128] In a possible implementation, the first beam direction information is included in second configuration information of the M reference signal groups, and when the first beam direction information includes group identifiers of the M reference signal groups, the second configuration information further includes one or more of time domain resources, frequency domain resources, transmission periodicity, or beam information of the M reference signal groups, or when the first beam direction information includes beam information of the M reference signal groups, the second configuration information further includes one or more of time domain resources, frequency domain resources, transmission periodicity, or group identifiers of the M reference signal groups.

[0129] In a possible implementation, the method further includes transmitting third configuration information of the N reference signal groups to the terminal device. When the first beam instruction information includes group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes group identifiers of each of the N reference signal groups and indicates one or more of the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, the transmission periodicity of the N reference signal groups, or the beams of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes the N group identifiers of the N reference signal groups including the M group identifiers of the M reference signal groups; or, when the first beam instruction information includes beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes beam information of each of the N reference signal groups and indicates one or more of the group identifiers of the N reference signal groups, the time domain resources, the frequency domain resources of the N reference signal groups, or the transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes the N group identifiers of the M reference signal groups.

[0130] In a possible implementation, the method further includes receiving the first output information from the terminal device.

[0131] According to a tenth aspect, there is provided a communication device configured to perform a method according to any one of possible implementation forms of the first, fourth, sixth, and eighth aspects. Specifically, the device may include a unit and / or module, such as a transceiver unit and / or a processing unit, configured to perform a method according to any one of possible implementation forms of the first, fourth, sixth, and eighth aspects.

[0132] In one implementation, the device is a training device, an inference device, or a terminal device. When the device is a communication device, the communication unit may be a transceiver or an input / output interface, and the processing unit may be at least one processor. Optionally, the transceiver may be a transceiver circuit. Optionally, the input / output interface may be an input / output circuit.

[0133] In another implementation, the apparatus is a chip, chip system, or circuit used in a training device, an inference device, or a terminal device. When the apparatus is a chip, chip system, or circuit used in a communication device, the communication unit may be an input / output interface, interface circuit, output circuit, input circuit, pin, associated circuit, etc. of the chip, chip system, or circuit, and the processing unit may be at least one processor, processing circuit, logic circuit, etc.

[0134] According to an eleventh aspect, there is provided a communication device. The device is configured to perform a method according to any one of possible implementation forms of the second, third, fifth, seventh, and ninth aspects. Specifically, the device may include a unit and / or module, such as a transceiver unit and / or a processing unit, configured to perform a method according to any one of possible implementation forms of the second, third, fifth, seventh, and ninth aspects.

[0135] In one implementation, the device is a network device or a first network device. When the device is a communication device, the communication unit may be a transceiver or an input / output interface, and the processing unit may be at least one processor. Optionally, the transceiver may be a transceiver circuit. Optionally, the input / output interface may be an input / output circuit.

[0136] In another implementation, the apparatus is a chip, a chip system, or a circuit used in the network device or the first network device. When the apparatus is a chip, a chip system, or a circuit used in the communication device, the communication unit may be an input / output interface, an interface circuit, an output circuit, an input circuit, a pin, an associated circuit, etc. of the chip, the chip system, or the circuit, and the processing unit may be at least one processor, a processing circuit, a logic circuit, etc.

[0137] According to a twelfth aspect, there is provided a communications device. The device includes at least one processor configured to execute a computer program or instructions stored in a memory to perform a method according to any one of possible implementation forms of any one of the first, fourth, sixth, and eighth aspects. Optionally, the device further includes a memory configured to store the computer program or instructions. Optionally, the device further includes a communications interface, and the processor reads the computer program or instructions stored in the memory via the communications interface.

[0138] In one implementation, the device is a training device, an inference device, or a terminal device.

[0139] In another implementation, the apparatus is a chip, chip system, or circuit used in a training device, an inference device, or a terminal device.

[0140] According to a thirteenth aspect, there is provided a communications device. The device includes at least one processor configured to execute a computer program or instructions stored in a memory to perform a method according to any one of the possible implementation forms of any one of the second, third, fifth, seventh, and ninth aspects. Optionally, the device further includes a memory configured to store the computer program or instructions. Optionally, the device further includes a communications interface, and the processor reads the computer program or instructions stored in the memory via the communications interface.

[0141] In one implementation, the apparatus is a network device or a first network device.

[0142] In another implementation, the apparatus is a chip, chip system, or circuit used in the network device or the first network device.

[0143] According to a fourteenth aspect, the present application provides a processor including an input circuit, an output circuit, and a processing circuit configured to receive a signal via the input circuit and to transmit a signal via the output circuit, such that the processor performs a method according to any one of the possible implementation forms of any one of the first to ninth aspects.

[0144] In a specific implementation process, the processor may be one or more chips, the input circuit may be an input pin, the output circuit may be an output pin, and the processing circuit may be a transistor, a gate circuit, a trigger, or any logic circuit, etc. The input signal received by the input circuit may be, for example, but not limited to, received and input by a transceiver, and the signal output by the output circuit may be, for example, but not limited to, output to a transmitter and transmitted by the transmitter, and the input circuit and the output circuit may be the same circuit, or a circuit may be used as an input circuit and an output circuit at different times. The specific implementation forms of the processor and various circuits are not limited to the embodiments of the present application.

[0145] Unless otherwise specified, or if operations such as transmitting and acquiring / receiving related to a processor do not contradict the actual functions or internal logic in the relevant description, operations may be understood as operations such as outputting, receiving, and inputting performed by a processor, or as transmitting and receiving operations performed by a radio frequency circuit and an antenna, which is not limited in this application.

[0146] According to a fifteenth aspect, there is provided a processing device including a processor and a memory, wherein the processor is configured to read instructions stored in the memory, receive signals via a transceiver, and transmit signals via a transmitter, to perform a method according to any one of the possible implementation forms of any one of the first to ninth aspects.

[0147] Optionally, there are one or more processors and one or more memories.

[0148] Optionally, the memory may be integral with the processor, or the memory and processor may be separately located.

[0149] In certain implementations, the memory may be a non-transitory memory, such as a read-only memory (ROM). The memory and the processor may be integrated into one chip or may be separately located on separate chips. The type of memory and the manner in which the memory and the processor are located are not limited in the embodiments of the present application.

[0150] It should be understood that a related data exchange process, such as transmitting instruction information, may be a process of outputting instruction information from a processor, and receiving capability information may be a process of receiving capability information input by a processor. Specifically, data output by a processor may be output to a transmitter, and input data received by a processor may be from a transceiver. The transmitter and transceiver may be collectively referred to as a transceiver.

[0151] The processing device according to the fifteenth aspect may be one or more chips. The processor in the processing device may be implemented using hardware or software. When the processor is implemented using hardware, the processor may be a logic circuit, an integrated circuit, or the like. When the processor is implemented using software, the processor may be a general-purpose processor and is implemented by reading software code stored in a memory. The memory may be integrated into the processor or may be located outside the processor and exist independently.

[0152] According to a sixteenth aspect, there is provided a computer-readable storage medium storing program code for execution by a device, the program code being used to perform a method according to any one of the possible implementations of the first to ninth aspects.

[0153] According to a seventeenth aspect, there is provided a computer program product comprising instructions, which, when executed on a computer, enable the computer to perform a method according to any one of the possible implementations of the first to ninth aspects.

[0154] According to an eighteenth aspect, there is provided a chip system, the chip system including a processor configured to call a computer program from a memory and execute the computer program, such that a device in which the chip system is installed executes a method according to an implementation form of any one of the first to ninth aspects.

[0155] According to a nineteenth aspect, there is provided a communication system including a training device and a network device, wherein the training device is configured to perform a method according to any one of possible implementation forms of the first aspect, and the network device is configured to perform a method according to any one of possible implementation forms of the second aspect.

[0156] According to a twentieth aspect, there is provided a communication system including a first network device configured to perform a method according to any one of the possible implementation forms of the third aspect.

[0157] According to a twenty-first aspect, there is provided a communication system including a training device and a network device, wherein the training device is configured to perform a method according to any one of possible implementation forms of the fourth aspect, and the network device is configured to perform a method according to any one of possible implementation forms of the fifth aspect.

[0158] According to a twenty-second aspect, there is provided a communication system including a training device and a network device, wherein the training device is configured to perform a method according to any one of possible implementation forms of the sixth aspect, and the network device is configured to perform a method according to any one of possible implementation forms of the seventh aspect.

[0159] According to a twenty-third aspect, there is provided a communication system including an inference device such as a terminal device and a network device, wherein the inference device such as the terminal device is configured to perform a method according to any one of the possible implementation forms of the eighth aspect, and the network device is configured to perform a method according to any one of the possible implementation forms of the ninth aspect. [Brief explanation of the drawings]

[0160] [Figure 1] 1 is a diagram of the structure of a communication system. [Figure 2] 1 is a diagram of a neuron structure. [Figure 3] FIG. 1 is a diagram of the hierarchical relationship of a neural network. [Figure 4] FIG. 1 is a diagram of a framework for AI model training and inference according to the present application. [Figure 5] 5 is a schematic flowchart of a training dataset acquisition method 500 according to the present application. [Figure 6] 6 is a schematic flowchart of a training dataset acquisition method 600 according to the present application. [Figure 7] 7 is a schematic flowchart of a method 700 for obtaining input information for an AI model according to the present application. [Figure 8] 8 is a schematic flow chart of a communication method 800 according to the present application. [Figure 9] 9 is a schematic flow chart of a communication method 900 according to the present application. [Figure 10] 1 is a block diagram of a communication device 100 according to the present application. [Figure 11]2 is a block diagram of a communication device 200 according to the present application. DETAILED DESCRIPTION OF THE INVENTION

[0161] The following describes the technical solutions of the embodiments in this application in conjunction with the accompanying drawings.

[0162] The technology provided in this application may be applied to various communication systems. For example, the communication system may be a fourth-generation (4G) communication system (e.g., a long-term evolution (LTE) system), a fifth-generation (5G) communication system, a worldwide interoperability for microwave access (WiMAX) or wireless local area network (WLAN) system, a satellite communication system, a future communication system such as a sixth-generation (6G) mobile communication system, or an aggregate system of multiple systems. A 5G communication system may also be referred to as a new radio (NR) system, a satellite communication system, a future communication system such as a sixth-generation (6G) mobile communication system, or an aggregate system of multiple systems.

[0163] A device in a communication system may transmit a signal to or receive a signal from another device, and the signal may include information, signaling, data, etc. A device may also be replaced with an entity, a network entity, a communication device, a communication module, a node, a communication node, etc. For example, a communication system may include at least one terminal device and at least one network device. In another example, a communication system may include one training device and at least one network device. The network device may transmit a downlink signal to the terminal device, and / or the terminal device may transmit an uplink signal to the access network device. In addition, when a communication system includes multiple terminal devices, it will be understood that the multiple terminal devices can also transmit signals to each other. In other words, both the signal transmitting network element and the signal receiving network element may be terminal devices. It will be understood that the terminal device in this application may be replaced with a first device, and the network device may be replaced with a second device, and the terminal device and the network device perform the corresponding communication method in this disclosure.

[0164] The method provided in the embodiments of the present application may be applied to a wireless communication system such as 5G, 6G, or satellite communication. FIG. 1 is a simplified diagram of a wireless communication system according to one embodiment of the present application. As shown in FIG. 1, the wireless communication system includes a radio access network 100 (an example of a network device). The radio access network 100 may be a next-generation (e.g., 6G or higher version) radio access network or a conventional (e.g., 5G, 4G, 3G, or 2G) radio access network. One or more communication devices (120a-120j, collectively referred to as 120) may be interconnected or connected to one or more network devices (110a and 110b, collectively referred to as 110) in the radio access network 100. Optionally, FIG. 1 is merely a diagram. The wireless communication system may further include other devices, for example, a core network device, a wireless relay device, and / or a wireless backhaul device not shown in FIG. 1.

[0165] Optionally, in practical application, a wireless communication system may include multiple network devices (e.g., access network devices) or multiple communication devices. One network device may provide service to one or more communication devices simultaneously. One communication device may access one or more network devices simultaneously. The number of communication devices and network devices included in a wireless communication system is not limited in this embodiment of the present application.

[0166] The network device may be an entity configured to transmit or receive signals on the network side. The network device may also be an access device for a communication device to access a wireless communication system in a wireless manner. For example, the network device may be a base station. The base station may broadly cover various names as follows, or may be replaced with the names as follows, for example, NodeB (NodeB), evolved NodeB (eNB), next generation NodeB (gNB), access network device in an open radio access network (O-RAN), relay station, access point, transmitting and receiving point (TRP), transmitting point (TP), master station MeNB, secondary station SeNB, multi-standard radio (MSR) node, home base station, network controller, access node, radio node, access point (AP), transmitting node, transceiver node, baseband unit (BBU), radio remote unit (RRU), active antenna unit (AAU), radio frequency head (RRH), central unit (CU), distributed unit (DU), radio unit (RU), central unit control plane (CU-CP) node, central unit user plane (CU-UP) node, and positioning node. The base station may be a macro base station, a micro base station, a relay node, a donor node, analog, or a combination thereof. Alternatively, the network device may be a communication module, modem, or chip located in any of the aforementioned devices or apparatus.Alternatively, the network device may be a mobile switching center, a device performing base station functions for device-to-device (D2D), vehicle-to-everything (V2X), and machine-to-machine (M2M) communications, a network-side device for a 6G network, a device performing base station functions for a future communications system, etc. The network devices may support networks using the same access technology or different access technologies. The embodiments of the present application do not limit the specific technologies and device forms used for the network devices.

[0167] Network devices may be fixed or mobile. For example, base stations 110a and 110b (examples of network devices) are stationary and responsible for radio transmission and reception in one or more cells from communication device 120. A helicopter or unmanned aerial vehicle 120i shown in FIG. 1 may be configured to function as a mobile base station, and one or more cells may move based on the location of mobile base station 120i. In another example, a helicopter or unmanned aerial vehicle 120i may be configured to function as a communication device that communicates with base station 110b.

[0168] In this application, for example, a communication device configured to perform the aforementioned network functions may be an access network device, a network device having some functions of an access network, or a device capable of supporting the implementation of the functions of an access network, such as a chip system, a hardware circuit, a software module, or a combination of a hardware circuit and a software module. The device may be installed in or used with an access network device.

[0169] The communication device may be an entity, such as a mobile phone, configured to receive or transmit signals at a user site. The communication device may be configured to connect people, objects, and machines. The communication device may communicate with one or more core networks via a network device. The communication device may include handheld devices with wireless connectivity, other processing devices connected to a wireless modem, in-vehicle devices, etc. The communication device may be portable, pocket-sized, handheld, computer-integrated, or in-vehicle mobile devices. The communication device 120 may be widely used in various scenarios, such as cellular communications, device-to-device (D2D), vehicle-to-vehicle / vehicle-to-infrastructure (V2X), peer-to-peer (P2P), machine-to-machine (M2M), machine-to-machine (MTC), Internet of Things (IoT), virtual reality (VR), augmented reality (AR), industrial control, autonomous driving, telemedicine, smart grid, smart furniture, smart office, smart wearables, smart transportation, smart city, unmanned aerial vehicles, robotics, remote sensing, passive sensing, positioning, navigation and tracking, and autonomous delivery and mobility.Some examples of communication device 120 include 3GPP user equipment (UE), fixed devices, mobile devices, handheld devices, wearable devices, mobile phones, smartphones, Session Initiation Protocol (SIP) phones, notebook computers, personal computers, smartbooks, vehicles, satellites, Global Positioning System (GPS) devices, target tracking devices, unmanned aerial vehicles, helicopters, aircraft, watercraft, remote control devices, smart home devices, industrial devices, personal communication service (PCS) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), wireless network cameras, tablet computers, palmtop computers, mobile internet devices (MIDs), wearable devices such as smart watches, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, terminals in internet of vehicular systems, wireless terminals in self driving, wireless terminals in smart grids, and transportation safety devices. The communication device 120 may be a wireless terminal in a smart safety system, a wireless terminal such as a smart fuel in a smart city, a terminal device in a high-speed train, a wireless terminal such as a smart speaker, a smart coffee machine, or a smart printer in a smart home. The communication device 120 may be a wireless device in the various scenarios mentioned above, or a device disposed within a wireless device, such as a communication module, modem, or chip within the device. The communication device may also be called a terminal, terminal device, user equipment (UE), mobile station (MS), or mobile terminal (MT).Alternatively, the communication device may be a communication device in a future wireless communication system. The communication device may be used in a dedicated network device or a general-purpose device. In the embodiments of the present application, the specific technology used by the communication device and the specific device form are not limited.

[0170] Optionally, the communication device may be configured to function as a base station. For example, the UE may function as a scheduling entity that provides sidelink signals between UEs, such as V2X, D2D, or P2P. As shown in FIG. 1, the mobile phone 120a and the car 120b communicate with each other using sidelink signals. The mobile phone 120a communicates with the smart home device 120e without relaying the communication signals through the base station 110b.

[0171] In this application, a communication device configured to implement the functions of a communication device may be a terminal device, a terminal device having some functions of a communication device, or a device capable of supporting the implementation of the functions of a communication device, such as a chip system. The device may be installed in or used with a terminal device. In this application, the chip system may include a chip, or may include a chip and other discrete components.

[0172] Optionally, a wireless communication system typically includes cells, and a base station provides cell management, providing communication services to multiple mobile stations (MSs) within the cell. The base station includes a baseband unit (BBU) and a remote radio unit (RRU). The BBU and the RRU may be located in different locations. For example, the RRU is remote and located in a high-traffic area, and the BBU is located in a central equipment room. Alternatively, the BBU and the RRU may be located in the same equipment room. Alternatively, the BBU and the RRU may be different components in the same rack. Optionally, one cell may correspond to one carrier or component carrier.

[0173] It should be understood that the number and types of devices in the communication system shown in Figure 1 are used as an example only, and the present application is not limited thereto. In actual application, the communication system may further include more terminal devices and more network devices, and may further include other network elements, for example, core network devices and / or network elements configured to perform artificial intelligence functions.

[0174] In order to facilitate understanding of the technical solutions provided in this application, the following first provides a brief description of the technical terms used in this application, and it is understood that the description is not intended to limit the present application.

[0175] 1.Artificial Intelligence (AI) technology (1) AI model An AI model is a specific implementation of an AI technology function. An AI model describes the mapping relationship between the model's inputs and outputs. The type of AI model may be a neural network, a linear regression model, a decision tree model, a support vector machine (SVM), a Bayesian network, a Q-learning model, or another machine learning (ML) model.

[0176] Based on different specific methods and / or techniques for implementing artificial intelligence, AI models may be specifically referred to as machine learning models, deep learning models, or reinforcement learning models. Machine learning is a method for implementing artificial intelligence. The purpose of this method is to design and analyze several algorithms (i.e., "models") that a computer can automatically "learn." The designed algorithms are called "machine learning models." Machine learning models are a type of algorithm that automatically analyzes data to obtain rules and then predicts unknown data according to the rules. There are various types of machine learning models. Machine learning models are classified into supervised learning models and unsupervised learning models based on whether model training depends on labels corresponding to the training data. The following discussion will focus primarily on "supervised learning models."

[0177] (2) Supervised learning model A "supervised learning model" is a model obtained after the parameters of an initial AI model are determined based on the data in a given training dataset and the labels corresponding to each data item in the training dataset. The process of determining the parameters of an initial AI model based on the data in the training dataset and the labels corresponding to the data is also called "supervised learning" (or "supervised training"). The labels of the data in the training dataset are usually manually annotated to identify the correct answer for the data in a specific task. Typical supervised learning models include support vector machines, neural network models, logistic regression models, decision trees, naive Bayes models, and Gaussian discriminant models. Supervised learning models are usually used for classification or regression. Quantitative output is called "regression" and can also be understood as the "continuous variable prediction" of the AI ​​model. Qualitative output is called "classification" and can also be understood as the "discrete variable prediction" of the AI ​​model.

[0178] (3) Deep neural network (DNN) DNNs are a specific implementation of machine learning. According to the general approximation theorem, neural networks can theoretically approximate any continuous function, and as a result, they have the ability to learn any mapping. Traditional communication systems require extensive expert knowledge to design communication modules. However, DNN-based deep learning communication systems can automatically discover implicit pattern structures from large datasets, establish mapping relationships between data, and achieve better performance than traditional modeling methods.

[0179] The concept of DNN comes from the neuron structure of the brain. As shown in Figure 2, each neuron performs a weighted sum operation on the input values ​​of the neuron and outputs the weighted sum result as a nonlinear function. Specifically, if the input of the neuron is x=[x0,...,x n ] and the weights corresponding to the inputs are d=[d0,...,d n] and the weighted summation bias is assumed to be b. The nonlinear function can take many forms. For example, if the nonlinear function is max{0,x}, the execution effect of the neuron is

number

[0180] DNNs typically have a multi-layer structure. Each layer of a DNN may contain multiple neurons. The input layer processes incoming values ​​using neurons and then forwards the processed values ​​to the intermediate hidden layer. Similarly, the hidden layer forwards the calculation results to the final output layer to generate the final output of the DNN, as shown in Figure 3. DNNs typically have two or more hidden layers, and the hidden layers typically directly affect the information extraction and function fitting capabilities of a DNN. Increasing the number of hidden layers or the width of each layer can improve the function fitting capabilities of a DNN. The weights of each neuron are the parameters of the DNN model. The model parameters are optimized during the training process to ensure that the DNN has the capabilities of data feature extraction and mapping relationship representation.

[0181] (4) AI model training and inference An AI model must be trained before it can be used to solve a specific technical problem. As shown in Figure 4, AI model training is a process in which a specific initial model is used to calculate training data, and parameters in the initial model are adjusted using a specific method based on the calculation results, so that the model gradually learns specific rules and has specific functions. After training, an AI model with stable functions can be used for inference. AI model inference is the process of using a trained AI model to calculate input data and obtain predicted inference results.

[0182] In the training phase, a training set for a deep learning model must first be constructed based on the objective. The training set includes multiple training data, each of which is labeled. The label of the training data is the correct answer for a specific question, and the label may represent the objective of training the deep learning model using the training data.

[0183] When a deep learning model is trained, training data may be input in batches to the deep learning model obtained after parameter initialization, and the deep learning model performs calculations on the training data (i.e., "inference") to obtain prediction results for the training data. The prediction results obtained by inference and the labels corresponding to the training data are used as data for calculating a loss based on a loss function. The loss function is a function used in the model training phase to calculate the difference between the model's prediction results for the training data and the labels of the training data (i.e., the loss value). The loss function may be implemented using different mathematical functions. Common expressions of the loss function include the mean squared error loss function, the logarithmic loss function, and the least squares method. Model training is an iterative process. In each iteration, different training data is inferred and a loss value is calculated. The purpose of multiple iterations is to continuously update the parameters of the deep learning model and find a parameter configuration that minimizes or gradually stabilizes the loss value of the loss function.

[0184] (5) Training dataset and inference data A "training dataset" is used to train an AI model. The training dataset may include the input of the AI ​​model, or may include the input and target output of the AI ​​model. The training dataset includes one or more training data. The training data may be training samples input to the AI ​​model or the target output of the AI ​​model. The target output may also be called a "label" or a "label sample." Training datasets are an important part of machine learning. Essentially, model training involves learning some features from the training data so that the output of the AI ​​model is as close as possible to the target output, e.g., minimizing the difference between the output of the AI ​​model and the target output. The composition and selection of the training dataset can determine to some extent the performance of the trained AI model. Model performance may be measured, for example, by a "loss value" or "inference accuracy."

[0185] A loss function may also be defined in the training process of an AI model (e.g., a neural network). The loss function describes the gap or difference between the output value of the AI ​​model and the target output value. The specific form of the loss function is not limited in this application. The training process of the AI ​​model is a process in which the model parameters of the AI ​​model are adjusted so that the value of the loss function is less than a threshold or so that the value of the loss function meets a target requirement. For example, the AI ​​model is a neural network, and adjusting the model parameters of the neural network includes adjusting at least one of the number and width of layers of the neural network, the weights of neurons, or parameters of activation functions of neurons.

[0186] "Inference data" may be used as input to a trained AI model for AI model inference. During model inference, the inference data is input to the AI ​​model to obtain a corresponding output, i.e., an inference result.

[0187] 2. Beam Management (1) Beam In this application, a "beam" may also be understood as a "spatial filtering parameter," "spatial filter," or "spatial parameters." A beam for transmitting a signal may generally be called a transmission beam (Tx beam), or may be called a spatial domain transmit filter or a spatial domain transmit parameter. A beam for receiving a signal may be called a reception beam (Rx beam), or may be called a spatial domain receive filter or a spatial domain receive parameter.

[0188] In new radio (NR) protocols, a beam may be, for example, a spatial filtering parameter (e.g., a spatial receive filtering parameter or a spatial transmit filtering parameter). However, it should be understood that this application does not exclude the possibility of defining other terms that have the same or similar meaning in future protocols.

[0189] (2) Beam sweeping "Beam sweeping" refers to the transmission of beams in predefined directions with a fixed periodicity to cover a specific spatial region at a specific periodicity or time period. For example, during initial access, a UE needs to synchronize with the system and receive minimal system information. Therefore, a bearer synchronization signal and a physical broadcast channel (PBCH) block (SSB) are used to perform sweeping and transmission at a fixed periodicity. A channel state information-reference signal (CSI-RS) can also use the beam sweeping technique. However, if all predefined beam directions need to be covered, the overhead of the CSI-RS would be excessively high. In this case, the CSI-RS is transmitted only in a specific subset of predefined beam directions based on the location of the served terminal device.

[0190] (3) Beam measurement "Beam measurement" is a process in which a network device or terminal device measures the quality and characteristics of a received beamformed signal. In the beam management process, the terminal device or network device may obtain information such as the reference signal received power (RSRP), reference signal receiving quality (RSRQ), and signal to interference plus noise ratio (SINR) of the reference signal using SSB and CSI-RS to identify the optimal beam.

[0191] (4) Beam determination A network device or a terminal device selects a transmission beam or a reception beam to be used by the network device or the terminal device. A downlink beam may be determined by the terminal device. For example, the determination criterion for a downlink beam is that the maximum received signal strength of the beam is greater than a certain threshold. In the uplink direction, the terminal device transmits a sounding reference signal (SRS) in the direction of the network device, and the network device measures the SRS to determine the optimal uplink beam.

[0192] Currently, an AI model may be deployed to a training device (e.g., a terminal device) for training and updating. To support the training device when the training device trains the AI ​​model, the network device continuously (e.g., periodically) transmits a training dataset to the training device when the training device trains the AI ​​model, and the network device stops transmitting the training dataset to the training device only after the training device transmits model training completion instruction information to the network device. However, in the process of the training device training the AI ​​model, the training dataset transmitted by the network device to the training device is not required by the training device, resulting in resource waste. Therefore, how the training device obtains the training dataset when training the AI ​​model becomes a technical problem that needs to be solved.

[0193] In consideration of this, the present application provides a model training method. A training device may request a network to transmit a training dataset, and the request information is also sent by the network device to the training device, indicating related information of the training dataset required by the training device. In other words, in the present application, the training device may indicate the training dataset required by the network device, and the network device may transmit the training dataset indicated by the training device to the training device, and the training dataset does not need to be delivered continuously. This method can reduce waste of air interface resources and air interface overhead, and compared with other methods, this method improves the utilization performance of air interface resources.

[0194] It should be noted that the "training device" in this application may be understood as, for example, a terminal device. In other words, the terminal device can communicate with a network device, and the terminal device also has the ability to support model training. In another example, the "training device" may be understood as a device specifically used for model training. For example, the device may only provide model training functionality. When the device determines that model training is complete, the device may transmit the trained model to the required terminal device.

[0195] In this application, "model training" is described using "AI model training" as an example, and it is assumed that the AI ​​model is deployed to a training device.

[0196] Figure 5 is a schematic flowchart of a training dataset acquisition method 500 according to the present application. Next, each step shown in Figure 5 will be described. Please note that the steps shown with dashed lines in Figure 5 are optional and will not be described in detail in the following description. The method includes the following steps:

[0197] Optionally, step 501: the training device decides to monitor the performance of the AI ​​model.

[0198] For example, the training device is handed over from the second cell to the first cell, with the first network device serving the first cell and the second network device serving the second cell, in which case the training device determines to monitor the performance of the AI ​​model, or the first network device indicates the training device monitoring the performance of the AI ​​model.

[0199] In another example, the training device discovers that the input information of the AI ​​model changes. For example, the training device discovers that the sparse beam pattern does not belong to the beam patterns in the full codebook. In this case, the training device decides to monitor the performance of the AI ​​model.

[0200] Optionally, step 502: the first network device transmits a training dataset to a training device.

[0201] After entering the model monitoring phase, the first network device may configure reference signal resources for the training device and transmit N reference signal groups (i.e., training data sets) to the training device. For example, in a beam management scenario, the training device may obtain input information and labels for the AI ​​model through full codebook sweeping (which may also be understood as full beam sweeping).

[0202] Specifically, in a possible implementation, the first network device may transmit N reference signal groups to the training device, and the training device may measure the N reference signal groups to obtain N corresponding measurement result groups. For example, each of the N measurement result groups includes the RSRP of the reference signal. The training device may then determine any one of the N measurement result groups as input information for the AI ​​model, or the training device may use the N measurement result groups as input information for the AI ​​model. In another possible implementation, the first network device may present measurement results corresponding to M reference signal groups in the N reference signal groups to the training device as input information for the AI ​​model (see the description of method 700 for details). Based on the configuration of the AI ​​model, for example, the training device may use some maximum RSRP values ​​among the RSRP values ​​of all reference signals in the N measurement result groups as a label for the AI ​​model during training. The training device may then use the input information to obtain output information (i.e., the inference result of the AI ​​model) of the AI ​​model and compare the output information with the label. For example, the model prediction performance may be measured using training loss or training accuracy. For example, a threshold may be set. If the training loss (or accuracy) is greater than or equal to the threshold, it indicates that the AI ​​model meets the requirements of the new cell or the requirements of the new input information. In other words, the AI ​​model may continue to be used. If the training loss (or accuracy) is less than the threshold, the training device may determine that the AI ​​model is not suitable for the requirements of the new cell or the requirements of the new input information. In this case, the training device determines that the AI ​​model needs to be updated. In this embodiment, it is assumed that the training device determines that the AI ​​model needs to be updated, and then steps 503 to 506 need to be performed.

[0203] Step 503: The training device transmits first information to the first network device, where the first information indicates related information of the first training data set that the training device requests the first network device to transmit.

[0204] In response, the first network device receives first information from the training device.

[0205] In the present application, the related information of the first training dataset may include, for example, at least one of the following: information about the size of the first training dataset, configuration information of the inputs of the AI ​​model, or configuration information of the reference signals used to train the AI ​​model.

[0206] In the present application, for example, the training device may determine the "information about the size of the first training data set" in the following manner.

[0207] In a possible implementation, information about the size of the first training dataset may be determined by the training device based on the size of the training dataset required to complete training of the AI ​​model. For example, information about the size of the first training dataset may be determined by the training device based on the size of the training dataset required to complete training of the AI ​​model. For example, the training device may use historical information to determine the total number of training datasets required when the AI ​​model is trained from an initial state of the AI ​​model (e.g., the initial state of the AI ​​model is 0) to a converged state. For example, the training device may determine, based on historical experience, that the full codebook needs to be swept a total of 60,000 times to obtain a training dataset for training the AI ​​model.

[0208] In another possible implementation, before step 501, the method further includes: the training device determining a first performance of the AI ​​model; and the training device determining information regarding the size of the first training dataset based on the first performance of the AI ​​model and a second performance of the AI ​​model, where the first performance is the current performance of the AI ​​model and the second performance is a target performance of the AI ​​model. For example, when monitoring the AI ​​model, the training device may compare the current performance of the AI ​​model obtained by the monitoring with model performance in the historical information corresponding to the AI ​​model in a converged state, and estimate the size of the first training dataset required to achieve the expected model performance based on the current performance of the AI ​​model.

[0209] It should be noted that in this application, the performance of an AI model may be determined by measuring information such as "training loss" and "training accuracy." For example, the "training loss" may be compared to a threshold, and the comparison result may be used to measure the performance of the AI ​​model during training.

[0210] In the present application, the training device may indicate the size of the training dataset required by the first network device, so that the first network device distributes the training dataset based on the instruction of the training device, thereby preventing the first network device from continuously transmitting data to the training device during the model training process of the training device, thereby reducing the waste of air interface resources and improving the utilization performance of air interface resources.

[0211] In the present application, "reference signal configuration information" may include, for example, at least one of a reference signal identifier, a reference signal time domain resource, a reference signal frequency domain resource, a reference signal transmission periodicity, or a transmitted reference signal type. For example, the reference signal type is SSB, CSI-RS, or SRS. The reference signal identifier may also be understood as a reference signal group identifier. For example, the reference signal configuration information includes group identifiers of N reference signal groups (N is an integer equal to or greater than 1), where each reference signal group in the N reference signal groups has the same group identifier and includes at least one reference signal. Similarly, the reference signal time domain resource, the reference signal frequency domain resource, the reference signal transmission periodicity, and the transmitted reference signal type may also be understood as the N reference signal group time domain resources, the N reference signal group frequency domain resources, the N reference signal group transmission periodicity, and the N reference signal group type.

[0212] The "resource" in this application may be a frequency domain resource, a time domain resource, a resource block (RB), a physical resource block (PRB), etc. This is not limited in this application.

[0213] In the present application, the training device may determine configuration information for the reference signal using historical information related to AI model training and indicate the configuration information to the first network device so that the network device can configure the reference signal for the training device. The training device obtains input information and labels for the AI ​​model by measuring the configured reference signal to continue training the AI ​​model. Furthermore, in this embodiment, in the process of multiple iterations of training the AI ​​model, the first network device may also indicate input information for the AI ​​model to accelerate the convergence of the AI ​​model. For a specific implementation, see method 700 below. In other words, method 500 may also be combined with method 700.

[0214] In the present application, "configuration information for the input of the AI ​​model" may be understood as, for example, a training device determining the input information of the AI ​​model based on historical information related to AI model training. For example, the input information of the AI ​​model is a measurement result of a reference signal corresponding to a sparse beam pattern. In this case, the training device may determine that the sparse beam pattern includes a beam at a specific position in the full codebook. For example, the training device may report information such as an identifier of the sparse beam pattern, an identifier of the reference signal corresponding to the sparse beam pattern, or a measurement result of the reference signal to the first network device, so that the first network device can distribute a training data set corresponding to the beam pattern.

[0215] In the present application, the phrase "the first information indicates related information of the first training dataset that the training device requests the first network device to transmit" may be specifically implemented in the following implementations. In a possible implementation, the first information may explicitly indicate related information of the first training dataset that the training device requests the first network device to transmit. For example, the first information may include information about the size of the first training dataset, configuration information of the input of the AI ​​model, and configuration information of the reference signal used to train the AI ​​model. Specifically, it is assumed that the training device needs to sweep the full codebook 60,000 times to determine the size of the first training dataset. In this case, the first information may include information indicating a specific value of the number of sweeps. In another possible implementation, the first information may implicitly indicate related information of the first training dataset that the training device requests the first network device to transmit. For example, the first information may indicate the size of the first training dataset by including an index, and the first network device may query the index to determine a specific value corresponding to the index. Other related information that is of the first training data set and is indicated by the first information may be understood in the same way and is not shown one by one.

[0216] Optionally, in the present application, the first information further includes at least one of the following: identification information of the AI ​​model, information regarding the application scenario of the AI ​​model, usage information of the AI ​​model, and information regarding the computing capability of the training device.

[0217] For example, the network device may store a first mapping relationship. The first mapping relationship is a correspondence relationship between the identifier of each AI model and the size of the training dataset corresponding to the identifier of the AI ​​model. For example, the first mapping relationship may be in the form of a table. As shown in Table 1, AI model #1 corresponds to training dataset #A, AI model #2 corresponds to training dataset #B, and AI model #3 corresponds to training dataset #C. The training device may indicate the size of the required training dataset by transmitting the identifier of the AI ​​model.

[0218] In another example, an application scenario of an AI model or a use of an AI model may be understood as the AI ​​model being used in a beam management scenario, a CSI feedback scenario, a positioning scenario, etc. The training device may indicate to the first network device the size of the training dataset required to train the AI ​​model by indicating the application scenario or the use of the AI ​​model.

[0219] In another example, the training device may report information about its computing power to the first network device. In the present application, the information about the computing power reported by the training device includes at least one of the following: the capabilities of the training device's processor (e.g., a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), a neural network processing unit (NPU), a field-programmable gate array (FPGA), etc.), the size of the training device's storage space, the size of the training device's memory, the battery level of the training device, etc. This is not limited to this. By reporting the training device's computing power to the network device, the training device may indicate the maximum number of training datasets that can be processed during training of the AI ​​model. For example, the training device may sweep the full codebook up to 40,000 times. In this case, the size of the training dataset sent to the training device by the first network device does not exceed the computing power of the training device.

[0220] [Table 1]

[0221] Step 504: The first network device transmits a first training data set to the training device based on the relevant information indicated by the first information.

[0222] In response, the training device receives a first training data set from the first network device.

[0223] In the present application, the first training dataset is a training dataset based on the related information of the first training dataset indicated by the first information, and the first training dataset is used to train the AI ​​model. In other words, the first training dataset is determined based on the related information of the first training data indicated by the first information.

[0224] Optionally, in the present application, the first network device may further obtain third information from the second network device. The third information is relevant information for training the AI ​​model on the second network device. For example, the third information includes at least one of the following: information about the size of the training dataset that the training device requests the second network device to transmit, information about the size of the training dataset required to complete training of the AI ​​model, identification information of the AI ​​model, or information about the computing capability of the training device. Optionally, the third information further includes information about the duration for transmitting the training dataset by the second network device and / or information about the method for transmitting the training dataset by the second network device. Specifically, the first network device may transmit the first training dataset to the training device based on the relevant information indicated by the first information and the third information. For a specific implementation form, please refer to the following description of method 600.

[0225] Optionally, step 505: the training device trains the AI ​​model based on the first training dataset.

[0226] For example, after receiving the first training data set, the training device may determine input information and labels. In a possible implementation, the first training data set is N reference signal groups, and the terminal device may measure the N reference signal groups to obtain N corresponding measurement result groups. For example, each of the N measurement result groups includes the RSRP of a reference signal, and the training device may then determine any one of the N measurement result groups as input information for the AI ​​model. In another possible implementation, the network device may display measurement results corresponding to M reference signal groups among the N reference signal groups to the terminal device as input information for the AI ​​model (see the description of method 700 for details). Based on the configuration of the AI ​​model, for example, the training device may use the RSRP measurement results of all reference signals in the N measurement result groups as labels for the AI ​​model. The training device may then use the input information to obtain output information for the AI ​​model and compare the output information with the labels to obtain a training loss for the AI ​​model. The above process may be understood as training one of the AI ​​models. The training device may measure the performance of the AI ​​model based on the training loss and training accuracy of the current model training and may determine a training data set required for the next model training. For example, the size of the training dataset may be determined by the training device based on the model performance evaluation results. For example, if the training device determines that the model performance has significantly improved after the first training of the AI ​​model, the amount of data in the training dataset may be reduced. Specifically, the following steps are further included:

[0227] The training device may train an AI model based on a first training dataset and determine the performance of the AI ​​model. The training device may transmit second information to the network device based on the performance of the AI ​​model, where the second information indicates related information of a second training dataset that the training device requests the network device to transmit. The training device receives the second training dataset from the network device, where the second training dataset is based on the related information indicated by the second information, and the second training dataset is used to train the AI ​​model. For example, the amount of data in the second training dataset may be less than the amount of data in the first training dataset. Thereafter, the training device may still train the AI ​​based on the second training dataset and repeatedly perform iterations. For example, the training device performs model training based on the second training dataset, re-determines the performance of the AI ​​model, and determines the size of a required third training dataset based on the performance of the AI ​​model. It is assumed that the training device may perform training L (L is an integer greater than 1) times until the training device determines that the AI ​​model has converged ("the model has converged" may also be understood as meaning that the AI ​​model has reached target performance).

[0228] Optionally, step 506: the training device sends a model training completion indication to the first network device.

[0229] After completing training of the AI ​​model, the training device may enter a model inference phase.

[0230] Based on the above technical solution, in this application, the training device may indicate the relevant information of the required training data set to the network, so that the network device may send the training data set to the training device based on the instruction, and there is no need to continuously deliver the training data set. This method can reduce the waste of air interface resources and air interface overhead, and improve the utilization performance of air interface resources.

[0231] Method 500 mainly describes that the training device determines relevant information of the required training dataset so that the network device can deliver the training dataset based on the request of the training device, thereby reducing the waste of air interface resources and improving the utilization performance of air interface resources.The following method 600 mainly describes that when the training device performs cell handover, the new network device after the handover can obtain relevant information for training the AI ​​model from the previous network device before the handover, and based on the information, determine the training dataset that needs to be sent to the training device.

[0232] 6 is a schematic flowchart of a training dataset acquisition method 600 according to the present application. In the method 600, it is considered that when the training device performs a cell handover, the training device determines that it needs to perform model monitoring. It is assumed that the training device determines that a model update needs to be performed on the AI ​​model through model monitoring. The method 600 includes the following steps:

[0233] Optionally, in steps 601 and 602, the training device determines to monitor the performance of the AI ​​model. In this embodiment, it is still assumed that the training device determines that the AI ​​model needs to be updated, and steps 603 to 608 need to be performed. Specifically, for the implementation of steps 601 and 602, please refer to steps 501 and 502 of method 500. The details will not be described again.

[0234] Step 603: The training device sends first information to the first network device, where the first information is used to request the first network device to send a training dataset.

[0235] For example, if the training device determines, by monitoring the performance of the AI ​​model, that the AI ​​model needs to be updated, the training device may send first information to the first network device.

[0236] Step 604: The first network device obtains third information from the second network device, and the third information is relevant information for training the AI ​​model.

[0237] In one possible implementation, the first network device may send request information to the second network device, where the request information is used to request obtaining information for training the AI ​​model, and the second network device may synchronize relevant information for training the AI ​​model to the first network device based on the request information. In another possible implementation, the second network device may actively provide relevant information for training the AI ​​model to the first network device.

[0238] For example, the third information may include at least one of the following: information regarding the size of the training dataset that the training device requests the second network device to transmit; information regarding the size of the training dataset required to complete training of the AI ​​model; identification information of the AI ​​model; or information regarding the computing capabilities of the training device.

[0239] "Information regarding the size of the training dataset that the training device requests the second network device to transmit" may be understood as the training device requesting information regarding the size of the required training dataset from the second network device when the second network device connects to the training device. For example, the training device may also determine the size of the training dataset requested from the second network device based on historical information. "Information regarding the size of the training dataset required to complete the training of the AI ​​model" may be understood as the second network device storing the total size of the training dataset required to train the AI ​​model by the training device. For example, the network devices (e.g., the first network device and the second network device) may together maintain a first mapping relationship. The first mapping relationship is a mapping relationship between the identifiers of the AI ​​models and the sizes of the training datasets corresponding to the identifiers of the AI ​​models. For example, the first mapping relationship may be in the form of a table, as shown in Table 1. In other words, the amount of training data required to complete one AI model training in the current cell for each AI model is collected based on historical experience. In other words, the information in Table 1 may be synchronized between the network devices. For example, Table 1 may be stored in both the first network device and the second network device. If the identifier of the AI ​​model is not present in Table 1, the first network device may determine the size of the training dataset during current model training based on the size of the training dataset delivered to the training device by the second network device. Note that some AI models may not be adapted to the current cell environment at all. The first network device may indicate that the AI ​​model cannot be trained and advise the training device to replace the AI ​​model. Alternatively, the second network device may synchronize information about the computing capabilities of the training device to the first network device, so that the first network device determines the size of the training dataset to be sent to the training device.

[0240] It should be noted that the "mapping relationship" in this application may also be expressed as an "association relationship" or a "correspondence." It should be understood that the "mapping relationship" in the embodiments of this application may be stored or recorded using a function relationship, a table, a mapping relationship, etc. In the following embodiments, the "mapping relationship" may be configured by a network device or may be predefined in a protocol, etc. This is not limited thereto.

[0241] Optionally, the third information further includes information regarding a duration for which the second network device transmits the training dataset and / or information regarding a method for transmitting the training dataset by the second network device. "Information regarding a duration for which the second network device transmits the training dataset" may also be understood as the time for training the AI ​​model when the training device is connected to the second network device. In other words, the time required to train the AI ​​model until convergence is completed.

[0242] The "information regarding how the second network device transmits the training dataset" may be understood as, for example, the second network device periodically transmitting the training dataset to the training device. For example, the network device may identify a time period in a day when the data demand volume is the smallest, which may alternatively be understood as a time period when the minimum air interface resources are occupied or when the air interface resources are sufficient. In this case, the training dataset may be provided to the AI ​​model. For example, the network device may select a time period each day to support the update of the AI ​​model (the number of training devices and / or the number of AI models is not limited herein). In another example, the second network device transmits the training dataset to the training device at intervals. Compared with periodic transmission, this solution has some improvement in flexibility. If the network device discovers and / or determines that the current data demand volume is relatively small and the air interface resources are sufficient, the network device determines that the network device can support the update of the AI ​​model. In other words, in this implementation, the network device may determine to transmit the training dataset to the training device based on the current air interface resource occupancy status. Therefore, this implementation does not have any obvious periodic characteristics.

[0243] The first network device may determine, based on the information, whether the first network device has the capability to support training of the AI ​​model, and the first network may also determine, based on information (e.g., information regarding the computing capabilities of the training device), whether the training device has the capability to support training of the AI ​​model. For example, for some AI models, the first network device may determine, based on historical information, that the training device is unable to train the AI ​​model to a converged state. In this case, it may be understood that the training device does not support training of the AI ​​model. In another example, the first network device may determine, based on information synchronized by the second network device, that there are insufficient air interface resources for delivering a training dataset to the training device. In this case, it may be understood that the first network device does not support training of the AI ​​model.

[0244] It should be noted that the present application places no restriction on the order between step 603 and step 604. For example, step 603 and step 604 may alternatively be performed simultaneously.

[0245] Step 605: The first network device determines a first training data set to be transmitted based on the third information.

[0246] As described in step 604, the first network device obtains relevant information for training the AI ​​model by the training device from the second network device and determines the size of the first training dataset to be distributed by the training device. Furthermore, the first network device may simultaneously provide services to multiple training devices or terminal devices. For example, the first network device may need to distribute a training dataset to another training device, and the first network device may also need to transmit control information to multiple terminal devices. In this case, the air interface resources of the first network device are very insufficient. Therefore, the first network device needs to comprehensively determine the size of the training dataset to be distributed to the training device based on the current usage status of the air interface resources. For example, the first network device determines, based on the third information, that the training dataset to be transmitted to the training device is Training Dataset #A. However, because the first network device does not have sufficient time-frequency resources to transmit Training Dataset #A, the first network device may decide to transmit only a portion of the training dataset.

[0247] In this application, "resource usage status" may also be understood as "resource occupancy status," "occupancy of air interface resources," etc. For example, existing protocol frameworks define the maximum number of reference signals that can be configured by a network device (e.g., 64 CSI-RS). If a network device finds that all reference signals are configured for another function, the network device may determine that the current resources are occupied and cannot configure reference signal resources for a training device to support updating an AI model.

[0248] Optionally, the first information indicates information related to the first training dataset that the training device requests the first network device to transmit. For example, the first information may include at least one of the following: information on the size of the requested training dataset, configuration information of the input of the AI ​​model, configuration information of the reference signal used to train the AI ​​model, and identification information of the AI ​​model. Optionally, the first information further includes at least one of the following: identification information of the AI ​​model, information on the application scenario of the AI ​​model, usage information of the AI ​​model, or information on the computing capability of the training device. In other words, this embodiment may be further combined with method 500. Specifically, the training device may indicate the size of the requested first training dataset to the network device. In this case, the first network device may comprehensively determine the size of the training dataset to be transmitted to the training device based on the third information, the first information, and the usage status of the air interface resource.

[0249] Step 606: The first network device transmits a first training data set to the training device.

[0250] In response, the training device receives a first training data set from the first network device.

[0251] In this application, the first training data set is used to train an AI model.

[0252] Step 607: The training device trains the AI ​​model based on the first training dataset.

[0253] Specifically, for the process of the training device training the AI ​​model based on the first training data in this embodiment, please refer to step 505 of method 500. The details will not be described again.

[0254] Optionally, in step 608, the training device sends a model training completion indication to the network device.

[0255] After completing training of the AI ​​model, the training device may enter a model inference phase.

[0256] In this embodiment, the network device may combine various information to comprehensively determine whether to support training of the AI ​​model, and the network device does not continuously deliver the training dataset to the training device, so that unnecessary occupation of air interface resources may be reduced, air interface overhead may be reduced, and air interface resource utilization performance may be improved.

[0257] In method 500 and method 600, the training data set acquisition method is described from the perspective of the training device and the network device, respectively.

[0258] The present application further considers the problem of how to specifically train an AI model during model training. In a possible solution, a network device transmits a training dataset to a training device, and the training device obtains input information and labels based on the training dataset transmitted by the network device. In this solution, during model training, all training datasets are input to the AI ​​model for training. This model training method requires a long training time to converge the model. The following method 700 further considers how specifically the AI ​​model can be quickly converged during training. Figure 7 is a schematic flowchart of a method 700 for obtaining input information for an AI model according to the present application. The method 700 includes the following steps:

[0259] Optionally, in steps 701 and 702, the training device determines to monitor the performance of the AI ​​model. In this embodiment, it is still assumed that the training device determines that the AI ​​model needs to be updated, and steps 703 to 710 need to be performed. Specifically, for the implementation of steps 701 and 702, please refer to steps 501 and 502 of method 500. The details will not be described again.

[0260] Optionally, step 703: the training device sends first information to the first network device, where the first information is used to request the first network device to send a training dataset.

[0261] In response, the first network device receives first information from the training device.

[0262] Optionally, the first information indicates information related to the first training dataset that the training device requests the first network device to transmit. For example, the first information may include at least one of the following: information on the size of the requested training dataset, configuration information of the input of the AI ​​model, configuration information of the reference signal used to train the AI ​​model, and identification information of the AI ​​model. Optionally, the first information further includes at least one of the following: identification information of the AI ​​model, information on the application scenario of the AI ​​model, usage information of the AI ​​model, or information on the computing capability of the training device. In other words, this embodiment may be further combined with method 500. Specifically, the training device may indicate the size of the requested training dataset to the network device.

[0263] Optionally, step 704: The first network device determines the size of a first training data set to be sent to the training device.

[0264] For example, after receiving the first information from the training device, the first network device may obtain third information from the second network device, where the third information is relevant information for training the AI ​​model. For example, the third information may include at least one of the following: information about the size of the training dataset that the training device requests the second network device to transmit, information about the size of the training dataset required to complete training of the AI ​​model, identification information of the AI ​​model, or information about the computing capability of the training device. Optionally, the third information further includes information about the duration for transmitting the training dataset by the second network device and / or information about the method for transmitting the training dataset by the second network device. For example, the first network device may determine the first training dataset to be transmitted to the training device based on the third information. For a specific implementation, see the description of method 600. In another example, the first network device may determine the first training dataset to be transmitted to the training device based on the relevant information indicated by the first information and the third information. In another example, the first network device may determine the first training data set to be sent to the training device based on the related information indicated by the first information, the third information, and the resource usage of the first network device. In other words, this embodiment may be combined with method 600. Specifically, the network device may comprehensively determine the size of the first training data set sent to the training device.

[0265] Optionally, step 705: The first network device transmits N (N is an integer greater than 1) reference signal groups to the training device.

[0266] In response, the training device receives N reference signal groups from the first network device.

[0267] In this embodiment, "the first network device transmits N reference signal groups to the training device" may also be understood as the first network device transmitting a first training data set to the training device. For example, in a beam management scenario, the first network device may configure reference signal resources for the training device and transmit the reference signal resources (e.g., N reference signal groups) to the training device. Each reference signal group in the N reference signal groups includes at least one reference signal, and each reference signal group has the same group identifier. The group identifier corresponding to each reference signal group may be understood as a beam identification number, e.g., a beam identifier, or the group identifier corresponding to each reference signal group may be understood as a resource identifier for each reference signal group. In other words, the first network device may instruct the training device to perform full codebook beam sweeping. That is, the first network device instructs the training device to measure the N reference signal groups.

[0268] Before step 705, the method may further include the first network device transmitting configuration information to the training device, where the configuration information may indicate one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or group identifiers of the N reference signal groups. For example, in one possible implementation, the configuration information may include information on the sizes of the time domain resources of the N reference signal groups, the sizes of the frequency domain resources of the N reference signal groups, the transmission periodicities of the N reference signal groups, etc. In this case, the configuration information may be understood to be explicitly indicated. In another possible implementation, the configuration information may carry indices of the time domain resources of the N reference signal groups, indices of the frequency domain resources of the N reference signal groups, indices of the transmission periodicities of the N reference signal groups, etc. In this case, it may also be understood that the configuration information is implicitly indicated.

[0269] Step 706: The training device measures the N reference signal groups and obtains N measurement result groups corresponding to the N reference signal groups.

[0270] For example, each of the N measurement result groups may include at least one measurement quantity. Specifically, for each reference signal in each reference signal group, one or more of the RSRP, RSRQ, SINR, etc. of the reference signal may be measured. In other words, each measurement result group may include one or more of the RSRP, RSRQ, and SINR measurement results measured for each reference signal in the reference signal group. Each measurement result group corresponds to the same group identifier. As shown in Table 2, reference signal group #A includes reference signal #A1 and reference signal #A2. The training device may measure reference signal #A1 and reference signal #A2 separately. For example, if the RSRP and SINR of reference signal #A1 and the RSRP and SINR of reference signal #A2 are measured, the measurement result of reference signal #A1 includes the RSRP and SINR measurement results of reference signal #A1, and the measurement result of reference signal #A2 includes the RSRP and SINR measurement results of reference signal #A2.

[0271] [Table 2]

[0272] In this embodiment, the training device may determine the training labels based on the full codebook beam sweeping results. Specifically, if the AI ​​model is trained using a regression method, the training device may determine, for example, the RSRP measurement results of the N reference signal groups as the training labels. Alternatively, if the AI ​​model is trained using a classification method, the training device may determine, for example, the beam identifiers corresponding to the K measurement result groups with the best channel qualities among the N reference signal groups as the training labels.

[0273] Step 707: The first network device sends fourth information to the training device, where the fourth information indicates M reference signal groups among the N reference signal groups.

[0274] For example, in a beam management scenario, the fourth information may indicate a sparse beam pattern. It may also be understood that in this embodiment, the first network device may indicate input information of the AI ​​model to the training device. For example, the first network device may indicate to the training device that the input information of the AI ​​model is specifically a beam pattern including a specific beam in the full codebook. Assuming that there are 64 beams in the full codebook, the fourth information may indicate that the input information is a pattern including a specific beam among the 64 beams. Specifically, the following two implementation forms may be possible.

[0275] Implementation form 1 In a possible implementation, the beam identifiers in the sparse beam pattern ("beam identifiers" may also be understood as "reference signal group group identifiers") match the identifiers of the beams in the full codebook. For example, the full codebook has 64 beams, and the beam identifiers are beam #1 to beam #64, respectively. However, the sparse beam pattern transmitted by the first network device includes 16 beams, beam #16 to beam #32. For example, in implementation 1, there is a one-to-one correspondence between the beam identifiers and the reference signal group identifiers. The network device may indicate the sparse beam pattern by indicating the beam identifiers (an example of beam information) and / or the reference signal group identifiers to the training device.

[0276] Implementation form 2 In another possible implementation, it is assumed that the beam identifiers in the sparse beam pattern do not match the identifiers of the beams in the full codebook. It may be understood that the beam identifiers in the sparse beam pattern do not correspond to the identifiers of the beams in the full codebook. For example, the full codebook has 64 beams, and the beam identifiers are beam #1 to beam #64. However, the sparse beam pattern transmitted by the first network device includes 16 beams, beam #1 to beam #16. However, beam #1 to beam #16 in the sparse beam pattern cannot be analyzed by the training device as beam #1 to beam #16 in the full codebook. For example, the sparse beam pattern may include 16 beams, beam #1 to beam #16, which actually correspond to beam #1, beam #4, beam #8, beam #12, beam #16, beam #20, beam #24, beam #28, beam #32, beam #36, beam #40, beam #44, beam #48, beam #52, beam #56, and beam #60 in the full codebook. In this case, the sparse beam pattern may be indicated using the following solution in this scenario. For example, the fourth information may include N fields, where the N fields correspond one-to-one with the N reference signal groups, and the bit values ​​of M fields in the N fields are different from the bit values ​​of the remaining (NM) fields. The first network device may indicate the M reference signal groups using the M fields. For example, the bit values ​​of the M fields are all "1" and the bit values ​​of the remaining (NM) fields are all "0". In this case, the training device may determine that the input information is a measurement result corresponding to a specific reference signal group. Thus, a network device may specifically indicate that a sparse beam pattern includes a particular beam in the full codebook by using a field.

[0277] In a possible implementation, the first network device may further transmit fifth information to the training device, where the fifth information indicates P reference signal groups among the N reference signal groups. It may also be understood that the first network device may provide the training device with measurement results of reference signals corresponding to another beam pattern as input information for the AI ​​model. In other words, in the present application, the first network device may provide the training device with measurement results of reference signals corresponding to multiple beam patterns as input information for the AI ​​model to perform model training, so that the AI ​​model obtained by training may converge for all beam patterns.

[0278] Optionally, there may be no order between step 705 and step 707. For example, the configuration information and the fourth information may be sent in the same message. This is not a limitation.

[0279] Step 708: The training device determines first input information of the AI ​​model based on the fourth information and the N measurement result groups corresponding to the N reference signal groups.

[0280] In step 705, the training device measures the N reference signal groups and obtains N measurement result groups corresponding to the N reference signal groups. Therefore, the training device may determine, based on the fourth information, a particular group of measurement results within the N measurement result groups that can be used as input information for the AI ​​model.

[0281] It is assumed that the training device measures 64 reference signal groups, and the first network device indicates, as input to the AI ​​model, measurement results corresponding to the second, fourth, eighth, and sixteenth reference signal groups. In this case, the training device may use the measurement results of the reference signals of the second, fourth, and eighth groups as input information for the AI ​​model. For example, the first input information may be RSRP measurement results of the M reference signal groups.

[0282] Step 709: The training device obtains first output information based on the first input information.

[0283] In this embodiment, the "first output information" may also be understood as an inference output result of the AI ​​model. When the AI ​​model is trained using a regression method, the first input information may be, for example, RSRP measurement results for M reference signal groups. In this case, the first output information may include N group identifiers corresponding to the RSRP measurement results for N reference signal groups. When the AI ​​model is trained using a classification method, the first input information may be, for example, measurement results corresponding to M reference signal groups. In this case, the first output information may be K group identifiers corresponding to K reference signal groups having optimal channel quality measurement results in the N reference signal groups, and may be understood as, for example, K beam identifiers.

[0284] The training device may compare the first output information with the label to obtain a training loss for the AI ​​model. A classification method for model training is used as an example. The first output information is K group identifiers corresponding to K reference signal groups with optimal channel quality measurement results among N reference signal groups inferred by the training device. The training labels determined by the training device are assumed to be K group identifiers corresponding to K reference signal groups with optimal channel quality measurement results among the N reference signal groups during full codebook sweeping. In this case, the training device may compare the output results with the training labels, determine the performance of the AI ​​model, and adjust the model parameters. The above process may be understood as training one of the AI ​​models. The training device may measure the performance of the AI ​​model based on the training loss and training accuracy of the current model training and determine the training dataset required for the next model training. For example, the size of the training dataset may be determined by the training device based on the model performance evaluation results. For example, if the training device determines that the model performance has significantly improved after the first training of the AI ​​model, the amount of data in the training dataset may be reduced. Specifically, the following steps may be further included:

[0285] The training device may train an AI model based on the first training dataset and determine the performance of the AI ​​model. The training device may transmit second information to the network device based on the performance of the AI ​​model, where the second information indicates related information of the second training dataset that the training device requests the network device to transmit. The training device receives the second training dataset from the network device, where the second training dataset is based on the related information indicated by the second information, and the second training dataset is used to train the AI ​​model. For example, the amount of data in the second training dataset may be less than the amount of data in the first training dataset. Thereafter, the training device may continue to train the AI ​​based on the second training dataset and repeatedly perform iterations. It is assumed that the training device performs training Q (Q is an integer greater than 1) times until the training device determines that the AI ​​model has converged ("the model has converged" may be understood as the AI ​​model reaching a target performance).

[0286] It should be noted that in this embodiment, when the training device subsequently performs model training based on the requested training data set, the sparse beam pattern of the AI ​​model may be fixed, or in the subsequent Q training processes, the sparse beam pattern may still be the beam pattern indicated by the fourth information in step 707. It should be understood that during each training, the training device performs one full codebook sweep based on the training data set distributed by the first network device. Because the channel conditions (which may also be understood as the channel environment) change over time, the measurement results of the reference signals obtained after each full codebook sweep are imperfectly the same. Therefore, during each training, the M reference signal groups among the N reference signal groups correspond to different measurement results, and the training labels determined by the training device are also different. In other words, the input information and training labels of the AI ​​model change correspondingly. However, these changes are essentially caused by changes in the channel conditions, and the beam pattern does not change. In other words, in the solution provided in this embodiment, the only variable in the AI ​​model training process is the channel conditions.

[0287] In another solution, during AI model training, the input information for the AI ​​model is the measurement results of all reference signals obtained after full codebook sweeping. In other words, the beam pattern and channel conditions change during each training, causing the model convergence performance to deteriorate during AI model training. However, according to the method 700 provided in the present application, during model training, the network device may indicate which measurement results among the measurement results obtained by full codebook sweeping are the input information for the AI ​​model, i.e., only changes in channel conditions. Compared with the previous solution, this solution can accelerate the convergence speed of the AI ​​model, improve model training efficiency, and reduce the occupation of air interface resources.

[0288] Optionally, step 710: the training device sends a model training completion indication to the first network device.

[0289] After completing training of the AI ​​model, the training device may enter a model inference phase. For example, the first network device may then transmit a sparse beam pattern and a corresponding reference signal to the training device, and the training device may obtain input information for the AI ​​model by measuring the reference signal. The AI ​​model uses a classification method, and the input of the AI ​​model is assumed to be the measurement results of the reference signal. In this case, the AI ​​model may output K beam identifiers through inference. The K beam identifiers are beams corresponding to the K measurement results with the best channel qualities among the measurement results of the reference signals in the full codebook estimated by the training device. The training device may feed back the K beam identifiers to the first network device. The first network device may again transmit K reference signal groups corresponding to the K beams to the training device. The training device again measures the K reference signal groups, determines one reference signal group with the best measurement result, uses the beam identifier corresponding to the reference signal group as the finally selected beam, and communicates with the first network device using the beam.

[0290] Based on the above technical solution, in this application, the network device may provide the input information of the AI ​​model to the training device, so that only the channel conditions change during the AI ​​model training process, which can accelerate the convergence speed of the AI ​​model, improve the model training efficiency, and also reduce the occupation of air interface resources.

[0291] Method 700 provides a method for obtaining input information for an AI model. The following method 800 provides a communication method. This method describes in more detail a method for obtaining input information for an AI model in a training phase. The method may be implemented independently of or applied in combination with the above-mentioned methods for obtaining input information. As shown in FIG. 8, method 800 includes the following steps:

[0292] Step 801: A network device transmits a second set of reference signals to a training device.

[0293] In response, the training device receives a second set of reference signals from the network device.

[0294] In this embodiment, the second reference signal set includes N reference signal groups, each of which includes at least one reference signal. The "second reference signal set" may also be understood as a reference signal set corresponding to a full codebook beam. For example, each beam in the full codebook beam may correspond to one group of reference signals in the second reference signal set.

[0295] In a possible implementation, before step 801, the method may further include the network device transmitting first configuration information to the training device, where the first configuration information may indicate one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, group identifiers of the N reference signal groups, or beams of the N reference signal groups.

[0296] Each reference signal group has a group identifier.

[0297] It should be noted that in the present application, when each reference signal group in the N reference signal groups includes only one reference signal, the group identifier of the reference signal group is the identifier of the reference signal, i.e., the group identifier may be replaced with the identifier of the reference signal.

[0298] In another possible implementation, before step 801, the method may further include the network device transmitting third configuration information to the training device, wherein the third configuration information includes group identifiers for each of the N reference signal groups, and the third configuration information may indicate one or more of the following: time domain resources for the N reference signal groups, frequency domain resources for the N reference signal groups, transmission periodicities for the N reference signal groups, or beams for the N reference signal groups.

[0299] In yet another possible implementation, before step 801, the method may further include the network device transmitting first configuration information to the training device, where the first configuration information includes beam information for each of the N reference signal groups, and the third configuration information may indicate one or more of the following: time domain resources for the N reference signal groups, frequency domain resources for the N reference signal groups, transmission periodicities for the N reference signal groups, or group identifiers for the N reference signal groups.

[0300] Step 802: The network device sends second beam indication information to the training device, where the second beam indication information indicates a beam corresponding to the first reference signal set.

[0301] In response, the training device receives second beam instruction information from the network device.

[0302] In this embodiment, the beams corresponding to the first reference signal set are a subset of the multiple beams corresponding to the second reference signal set, and the first reference signal set includes M reference signal groups, where N is an integer greater than M and M is an integer greater than or equal to 1.

[0303] The network device may transmit the second beam instruction information to the training device, which may be understood as the network device indicating the sparse beam pattern to the training device. In other words, the second beam instruction information may indicate to the training device that the sparse beam pattern that the training device needs to sweep is a pattern including a specific beam in the full codebook.

[0304] In this embodiment, the second beam designation information indicating the beam corresponding to the first reference signal set may be, for example, the second beam designation information indicating the position of the beam corresponding to the first reference signal set among multiple beams corresponding to the second reference signal set. For example, there are 64 beams in the full codebook, and the beam identifiers are beam #1 to beam #64, respectively. However, the sparse beam pattern transmitted by the network device includes 16 beams. In this case, the second beam information may indicate which beams in the full codebook beams are among the 16 beams (i.e., which beams in the full codebook correspond to the first reference signal set). Specifically, the following two implementation forms are possible.

[0305] Mounting form A The second beam information includes N fields, where the N fields correspond one-to-one to the multiple beams corresponding to the second reference signal set, and the bit values ​​of M fields in the N fields are different from the bit values ​​of the remaining (NM) fields. The network device may indicate the first reference signal set using the M fields. For example, the bit values ​​of the M fields are all "1" and the bit values ​​of the remaining (NM) fields are all "0." For example, there are 64 beams in the full codebook, and the beam identifiers are beam #1 to beam #64, respectively. However, the sparse beam pattern transmitted by the network device includes 16 beams, beam #1 to beam #16. Field #1 to field #16 in the second beam indication information indicate beam #1 to beam #16, respectively, in the full codebook. For example, field #1 to field #16 in the second beam indication information indicate reference signal group #1 to reference signal group #16, respectively, in the first reference signal set. Implementation A may also be understood to allow the network device to directly indicate beam positions.

[0306] Mounting form B The second beam instruction information includes group identifiers or beam information (e.g., beam identifiers) of the M reference signal groups, where the M reference signal groups are part of N reference signal groups, and a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams. For example, the N reference signal groups correspond one-to-one to the N beams. For example, the group identifiers of the N reference signal groups correspond one-to-one to the N beam identifiers. For example, there are 64 beams in the full codebook, and the beam identifiers are beam #1 to beam #64, respectively. However, the sparse beam pattern transmitted by the network device includes 16 beams, beam #16 to beam #32. For example, in implementation B, there is a one-to-one correspondence between the beam identifiers and the reference signal group identifiers. The network device may indicate the sparse beam pattern by indicating the beam identifiers (an example of beam information) and / or the reference signal group identifiers to the training device. Compared to implementation A, in implementation B, the beam position is indirectly indicated using the group identifiers or beam information of the reference signals. Since the beam information corresponds to the group identifier of the reference signal, the relationship between the group identifier / beam information of the M reference signal groups and the group identifier / beam information of the N reference signal groups is fixed.

[0307] Step 803: The training device determines a first input information for the AI ​​model.

[0308] For example, after receiving the second reference signal set, the training device may measure N reference signal groups in the second reference signal set and obtain measurement results for the N reference signal groups. In this case, the training device may determine measurement results corresponding to beams indicated by the second beam indication information based on the second beam indication information and the measurement results for the N reference signal groups in the second reference signal set, and use the measurement results corresponding to the beams indicated by the second beam indication information as first input information for the AI ​​model. That is, the measurement results for beams corresponding to a sparse beam pattern are used as the first input information.

[0309] Corresponding to two implementations of step 802, after determining the sparse beam pattern, the training device may determine that input information for the AI ​​model is measurement results corresponding to reference signals at specific beam positions in the full codebook. For example, in implementation A, the training device may use measurement results for reference signal groups #16 to #32 corresponding to beams #16 to #32 in the full codebook as input information for the AI ​​model. As another example, in implementation B, the training device may use measurement results for reference signal group #1, reference signal group #4, reference signal group #8, reference signal group #12, reference signal group #16, reference signal group #20, reference signal group #24, reference signal group #28, reference signal group #32, reference signal group #36, reference signal group #40, reference signal group #44, reference signal group #48, reference signal group #52, reference signal group #56, and reference signal group #60 as input information for the AI ​​model.

[0310] Optionally, step 804: The training device performs model training based on the determined first input information and the AI ​​model to obtain first output information.

[0311] In the present application, the first output information indicates K beams that are predicted to have the best channel quality among a plurality of beams corresponding to the second reference signal set, where K is an integer greater than or equal to 1 and K is less than N.

[0312] For example, as described above, the classification method and the regression method may be used in implementations based on different algorithms for training an AI model. For example, when the AI ​​model is trained using the classification method, the first output information may include information on K beams predicted to have the best channel quality among a plurality of beams corresponding to the second reference signal set. Alternatively, the first output information may include group identifiers or beam information for each of the K reference signal groups, where the K reference signal groups correspond to the K measurement results predicted to have the best channel quality among the N measurement results corresponding to the N reference signal groups, and a predefined or preconfigured correspondence exists between the group identifiers for each of the K reference signal groups and the K beams, i.e., the K beam information. For example, when the AI ​​model is trained using a regression method, the first output information may include multiple beam information corresponding to the N reference signal groups and N predicted measurement results corresponding to the beam information. Alternatively, the first output information may include group identifiers for each of the N reference signal groups and N predicted measurement results for the N reference signal groups, where a predefined or preconfigured correspondence exists between the N reference signal groups and the N beam information.

[0313] In this embodiment, the "first output information" may also be understood as a training output result of the AI ​​model. When the AI ​​model is trained using a regression method, the first input information may be, for example, RSRP measurement results of M reference signal groups. In this case, the first output information may include predicted RSRP measurement results of reference signals corresponding to N beams and N group identifiers corresponding to the RSRP measurement results. When the AI ​​model is trained using a classification method, the first input information may be, for example, measurement results of M reference signal groups corresponding to M beams. In this case, the first output information may be beam identifiers of K beams predicted to have optimal channel quality measurement results for the N beams or K group identifiers corresponding to the K beams.

[0314] The training device may compare the first output information with the labels to obtain a training loss for the AI ​​model. A classification method for model training is used as an example. The first output information is K group identifiers corresponding to K beams with optimal channel quality measurement results among the N beams inferred by the training device. The training labels determined by the training device are assumed to be K group identifiers corresponding to K reference signal groups with optimal channel quality measurement results among the N reference signal groups corresponding to the N beams during full codebook sweeping. In this case, the training device may compare the output results with the training labels, determine the performance of the AI ​​model, and adjust the model parameters. The above process may be understood as training one of the AI ​​models. The training device may measure the performance of the AI ​​model based on the training loss and training accuracy of the model training, and repeatedly perform iterations until the model converges.

[0315] Based on the above technical solution, the network device may provide the input information of the AI ​​model to the training device, so that for the same sparse beam pattern, the channel conditions change during the AI ​​model training process, which can accelerate the convergence speed of the AI ​​model, improve the model training efficiency, and also reduce the occupation of air interface resources.

[0316] It should be appreciated that the above steps may be repeated for multiple sparse beam patterns.

[0317] The above-described methods 700 and 800 mainly describe specific implementation solutions for obtaining input information during the model training phase. The following method 900 in FIG. 9 mainly describes that the method can also be used during the model inference phase. In method 900, it is assumed that an AI model has been trained. For example, the model may be trained using method 700 or method 800. In another example, the AI ​​model may be trained using an existing solution. For example, multiple trained AI models may be pre-configured directly on an inference device, such as a terminal device. For example, each AI model is trained using one or more sparse beam patterns to complete the AI ​​model. Thus, the model inference phase may be initiated directly. It should be understood that the method described in FIG. 9 may be applied independently of or in combination with the method in FIG. 8. In conjunction with the present application, each related identical term is represented by an X, and for ease of distinction, it refers to configuration information during the training process and configuration information during the inference process, respectively. For example, configuration information refers to configuration information during the training process or configuration information during the inference process.

[0318] Step 901: A network device transmits a first set of reference signals to a terminal device.

[0319] In response, the terminal device receives a first set of reference signals from the network device.

[0320] In this embodiment, the first reference signal set includes M reference signal groups, each of which includes at least one reference signal, where M is an integer greater than or equal to 1. The "first reference signal set" may be understood as a reference signal set corresponding to a sparse beam pattern. For example, the sparse beam pattern is a subset of a full-codebook beam, or the reference signal set corresponding to the sparse beam pattern belongs to a subset of a second reference signal set corresponding to a full-codebook beam (i.e., the first reference signal set is a subset of the second reference signal set), and the second reference signal set includes N reference signal groups, each of which includes at least one reference signal.

[0321] In a possible implementation, before step 901, the method may further include the network device transmitting first configuration information to the terminal device, where the first configuration information may indicate one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, group identifiers of the N reference signal groups, or beam information of the N reference signal groups.

[0322] In another possible implementation, before step 901, the method may further include the network device transmitting second configuration information to the terminal device, the second configuration information including first beam information, the first beam information including group identifiers of the M reference signal groups, and the second configuration information may further include one or more of time domain resources of the M reference signal groups, frequency domain resources of the M reference signal groups, transmission periodicities of the M reference signal groups, or beam information.

[0323] In a possible implementation, before step 901, the method may further include the network device transmitting second configuration information to the terminal device, where the second configuration information includes first beam information, where the first beam information includes beam information for the M reference signal groups, and the second configuration information may further include one or more of time domain resources for the M reference signal groups, frequency domain resources for the M reference signal groups, transmission periodicities for the M reference signal groups, or group identifiers.

[0324] In yet another possible implementation, before step 901, the method may further include: the network device transmitting third configuration information to the terminal device, where the third configuration information includes group identifiers for each of the N reference signal groups, and the third configuration information may indicate one or more of the following: time domain resources for the N reference signal groups, frequency domain resources for the N reference signal groups, transmission periodicities for the N reference signal groups, or beams for the N reference signal groups. The N group identifiers for the N reference signal groups include M group identifiers for the M reference signal groups.

[0325] In yet another possible implementation, before step 901, the method may further include: the network device transmitting third configuration information to the terminal device, where the third configuration information includes beam information for each of the N reference signal groups, and the third configuration information may indicate one or more of the following: time domain resources for the N reference signal groups, frequency domain resources for the N reference signal groups, transmission periodicities for the N reference signal groups, or group identifiers for the N reference signal groups. The N beam information for the N reference signal groups includes M beam information for the M reference signal groups.

[0326] Step 902: The network device sends first beam indication information to the terminal device, where the first beam indication information indicates a beam corresponding to the first reference signal set.

[0327] In response, the terminal device receives first beam instruction information from the network device.

[0328] In this embodiment, the beams corresponding to the first set of reference signals are a subset of the plurality of beams corresponding to the second set of reference signals.

[0329] The network device may transmit the first beam indication information to the terminal device, which may be understood as the network device indicating a sparse beam pattern to the terminal device. In other words, the first beam indication information may indicate to the terminal device that the sparse beam pattern that the terminal device needs to sweep is a pattern including a specific beam in the full codebook.

[0330] In this embodiment, the second beam designation information indicating the beam corresponding to the first reference signal set may be, for example, the second beam designation information indicating the position of the beam corresponding to the first reference signal set among multiple beams corresponding to the second reference signal set. For example, there are 64 beams in the full codebook, and the beam identifiers are beam #1 to beam #64, respectively. However, the sparse beam pattern transmitted by the first network device includes 16 beams. In this case, the second beam information may indicate which beams in the full codebook beams are among the 16 beams (i.e., which beams in the full codebook correspond to the first reference signal set). Specifically, the following two implementation forms are possible.

[0331] Mounting form A The first beam information includes N fields, where the N fields correspond one-to-one to multiple beams corresponding to the second reference signal set. The bit values ​​of M fields in the N fields are different from the bit values ​​of the remaining (NM) fields. The network device may indicate the first reference signal set using the M fields. For example, the bit values ​​of the M fields are all "1" and the bit values ​​of the remaining (NM) fields are all "0." For example, there are 64 beams in the full codebook, and the beam identifiers are beam #1 to beam #64, respectively. However, the sparse beam pattern transmitted by the network device includes 16 beams, beam #1 to beam #16. Field #1 to field #16 in the second beam indication information indicate beam #1 to beam #16, respectively, in the full codebook. For example, field #1 to field #16 in the second beam indication information indicate reference signal group #1 to reference signal group #16, respectively, in the first reference signal set. Implementation A may also be understood to allow the network device to directly indicate beam positions.

[0332] Mounting form B The first beam indication information includes group identifiers or beam information (e.g., beam identifiers) of the M reference signal groups, where the M reference signal groups are part of N reference signal groups, and a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams. For example, the N reference signal groups correspond one-to-one to the N beams. For example, the group identifiers of the N reference signal groups correspond one-to-one to the N beam identifiers. For example, there are 64 beams in the full codebook, and the beam identifiers are beam #1 to beam #64, respectively. However, the sparse beam pattern transmitted by the first network device includes 16 beams, beam #16 to beam #32. For example, in implementation B, there is a one-to-one correspondence between the beam identifiers and the reference signal group identifiers. The network device may indicate the sparse beam pattern by indicating the beam identifiers (an example of beam information) and / or the reference signal group identifiers to the terminal device. Compared to implementation A, in implementation B, the beam position is indirectly indicated using the group identifiers or the beam information of the reference signals. Since the beam information corresponds to the group identifier of the reference signal, the relationship between the group identifier / beam information of the M reference signal groups and the group identifier / beam information of the N reference signal groups may be fixed.

[0333] Optionally, step 903: the terminal device determines an AI model based on the received first beam instruction information, i.e., the sparse beam pattern.

[0334] It is assumed that multiple AI models are preconfigured on the terminal device, in which case the terminal device may determine an AI model based on a sparse beam pattern. For example, the terminal device may identify a received sparse beam pattern and determine an AI model corresponding to the sparse beam pattern from multiple locally preconfigured AI models by identifying the beam pattern. It may also be understood that the terminal device may determine an AI model having a best match from multiple AI models based on the sparse beam pattern.

[0335] In this application, the terminal device identifying a sparse beam pattern may be understood as the terminal device needing to determine that the received beam pattern includes a beam at a specific position within the full codebook beam. In other words, the terminal device needs to establish a relationship between the received sparse beam pattern and the beam in the full codebook, i.e., to determine that the beam in the sparse beam pattern is a specific beam in the full codebook. Based on the configurations of the network device and the terminal device, the terminal device may determine the sparse beam pattern based on Implementation A and / or Implementation B.

[0336] Step 904: The terminal device determines first input information of the AI ​​model.

[0337] For example, after receiving the first reference signal set, the terminal device may measure M reference signal groups in the first reference signal set to obtain measurement results of the M reference signal groups, and may determine first input information of the AI ​​model based on the measurement results of the M reference signal groups.

[0338] Corresponding to two implementations of step 902, after determining the sparse beam pattern, the terminal device may determine that input information for the AI ​​model is measurement results corresponding to reference signals at specific beam positions in the full codebook. For example, in implementation A, the terminal device may use measurement results for reference signal groups #16 to #32 corresponding to beams #16 to #32 in the full codebook as input information for the AI ​​model. As another example, in implementation B, the terminal device may use measurement results for reference signal group #1, reference signal group #4, reference signal group #8, reference signal group #12, reference signal group #16, reference signal group #20, reference signal group #24, reference signal group #28, reference signal group #32, reference signal group #36, reference signal group #40, reference signal group #44, reference signal group #48, reference signal group #52, reference signal group #56, and reference signal group #60 as input information for the AI ​​model.

[0339] Optionally, step 905: the terminal device performs model inference based on the determined first input information and the AI ​​model to obtain first output information.

[0340] In the present application, the first output information indicates K beams that are predicted to have the best channel quality among a plurality of beams corresponding to the second reference signal set, where K is an integer greater than or equal to 1 and K is less than N.

[0341] For this step, please refer to the description of the first output information in the inference process included in the training process in step 804 of Figure 8. The details will not be described again in this specification. The difference between this step and step 804 is that in this step, the first output information and label do not need to be obtained by executing a loss, but the first output information is a usable prediction result.

[0342] Optionally, step 906: The terminal device sends the first output information to the network device.

[0343] The network device may then again transmit a reference signal corresponding to the first output information to the terminal device, and the terminal device again measures the reference signal, determines the reference signal having the best measurement result, uses the beam identifier corresponding to the reference signal as the finally selected beam, and communicates with the network device using the beam.

[0344] It should be understood that in this application, the optimal measurement result may include the maximum RSRP value or the maximum SINR value, or may be another evaluation criterion, and is not limited in this application.

[0345] For example, assume that the AI ​​model uses a classification method. In this case, the AI ​​model may output K beam identifiers through inference. The K beam identifiers are beams corresponding to K measurement results with the best channel quality among the measurement results of reference signals in the full codebook estimated by the terminal device. The terminal device may feed back the K beam identifiers to the first network device. The first network device may again transmit K reference signal groups corresponding to the K beams to the terminal device. The terminal device again measures the K reference signal groups, determines one reference signal group with the best measurement result, uses the beam identifier corresponding to the reference signal group as the finally selected beam, and communicates with the first network device using the beam.

[0346] Based on the above technical solution, in this application, the terminal device may identify a sparse beam pattern and further determine the input information of the AI ​​model, so that the model inference result is more accurate.

[0347] It should be understood that the examples of methods 500 to 900 in the embodiments of the present application are intended only to help those skilled in the art understand the embodiments of the present application, and are not intended to limit the embodiments of the present application to the specific scenarios in the examples. It is apparent that those skilled in the art can make various equivalent modifications or variations to the examples of methods 500 to 900, and such modifications or variations also fall within the scope of the embodiments of the present application.

[0348] It will be further understood that some optional features in the embodiments of the present application may be independent of other features in some scenarios and may be combined with other features in some scenarios, without limitation.

[0349] It should be further understood that the embodiments described in this application may be independent solutions or may be combined based on internal logic. All of these solutions fall within the scope of protection of this application. Furthermore, the interpretation or explanation of terms in the embodiments may be mutually referenced or interpreted in the embodiments. This is not limited.

[0350] It is to be understood that "predefine" in this application may be understood as "define," "predefine," "store," "prestore," "prenegotiate," "preconfigure," "build," or "pre-baked."

[0351] It will be understood that in this application, both "when" and "when" mean that the device performs processing corresponding to the objective case, but do not constitute any limitation on time, the device does not necessarily have to have a decision operation during implementation, and do not imply any other limitation.

[0352] It will be understood that "and / or" in this specification describes an associative relationship between related objects and represents three possible relationships. For example, A and / or B can represent the following cases: when only A is present, when both A and B are present, and when only B is present, and A and B can be singular or plural. In the text descriptions of this application, the symbol " / " generally indicates an "or" relationship between related objects. In formulas of this application, the symbol " / " indicates a "divide by" relationship between related objects.

[0353] The above mainly describes the solutions provided in the embodiments of the present application from the perspective of interactions between nodes. It should be understood that to implement the aforementioned functions, nodes such as training devices and network devices include corresponding hardware structures and / or software modules for performing the functions. In combination with the examples described in the embodiments disclosed in the present application, those skilled in the art should be able to recognize that the units and algorithm steps can be implemented by hardware or a combination of hardware and computer software in the present application. Whether the functions are performed by hardware or by hardware driven by computer software depends on the specific application and design constraints of the technical solution. Those skilled in the art can implement the functions described in each specific application using various methods, but such implementation forms should not be considered beyond the scope of the present application.

[0354] It should be understood that, to implement the functions of the foregoing embodiments, the network device and the training device include corresponding hardware structures and / or software modules for performing the functions. Those skilled in the art should easily recognize that the units and method steps in the examples described with reference to the embodiments disclosed in this application can be implemented by hardware or a combination of hardware and computer software. Whether the functions are performed by hardware or by hardware driven by computer software depends on the specific application scenario and design constraints of the technical solution.

[0355] 10 and 11 are diagrams of possible communication device structures according to embodiments of the present application. The communication device may be configured to perform the functions of the training device or the network device in the above-described method embodiments, and thus may also perform the beneficial effects of the above-described method embodiments. In embodiments of the present application, the communication device may be one of the terminal devices 120a to 120j (an example of a training device or an example of an inference device) shown in FIG. 1, the network device 110a or 110b shown in FIG. 1, or a module (e.g., a chip) used in the terminal device or the network device.

[0356] 10, the communication device 100 includes a processing unit 120 and a transceiver unit 110. The communication device 100 is configured to perform the functions of a training device or a network device in the method embodiments shown in FIGS.

[0357] When the communication device 100 is configured to perform the functions of the training device in the method embodiment shown in FIG. 5, the transceiver unit 110 is configured to transmit first information, the first information indicating related information of a first training dataset that the device requests to be transmitted, the transceiver unit 110 is further configured to receive the first training dataset, the first training dataset being a training dataset based on the related information indicated by the first information, and the first training dataset being used to train an artificial intelligence AI model.

[0358] In a possible implementation, the processing unit 120 is configured to train an AI model based on the first training data set and determine performance of the AI ​​model. The processing unit 120 is further configured to control the transceiver unit 110 to transmit second information based on the performance of the AI ​​model, where the second information indicates related information of the second training data set that the device requests to be transmitted. The transceiver unit 110 is configured to receive the second training data set, where the second training data set is based on the related information indicated by the second information, and the second training data set is used to train the AI ​​model.

[0359] When the communication device 100 is configured to perform the functions of the first network device in the method embodiment shown in FIG. 5, the transceiver unit 110 is configured to receive first information, where the first information indicates relevant information of a first training data set that the device requests to transmit, and the processing unit 120 is configured to control the transceiver unit 110 to transmit the first training data set based on the relevant information indicated by the first information, where the first training data set is used to train an artificial intelligence AI model.

[0360] In a possible implementation, the transceiver unit 110 is configured to obtain third information, where the third information is relevant information for training the AI ​​model, and the processing unit 120 is configured to control the transceiver unit 110 to transmit the first training data set based on the relevant information indicated by the first information includes the processing unit 120 being configured to control the transceiver unit 110 to transmit the first training data set based on the relevant information indicated by the first information and the third information.

[0361] In a possible implementation, the transceiver unit 110 is configured to receive second information, the second information indicating relevant information of a second training dataset requested to be transmitted, the second information being determined based on performance of an AI model, the performance of the AI ​​model being determined by training based on the first training dataset, and the processing unit 120 is configured to control the transceiver unit 110 to transmit the second training dataset based on the relevant information indicated by the second information, and the second training dataset is used to train the AI ​​model.

[0362] When the communication device 100 is configured to perform the functions of the first network device in the method embodiment shown in FIG. 6, the transceiver unit 110 is configured to acquire third information, where the third information is relevant information for training an artificial intelligence (AI) model; the transceiver unit 110 is configured to receive first information, where the first information is used to request transmission of a training dataset; the processing unit 120 is configured to determine the first training dataset to be transmitted based on the third information; the processing unit 120 is configured to control the transceiver unit 110 to transmit the first training dataset based on the first information; and the first training dataset is used to train the AI ​​model.

[0363] In a possible implementation, the third information further includes information regarding a duration for transmitting the training dataset and / or information regarding a method for transmitting the training dataset, and the processing unit 120 is configured to determine, based on the third information and the resource usage of the apparatus, whether the apparatus and / or the training device has the capability to support training of the AI ​​model.

[0364] In a possible implementation, configuring the processing unit 120 to determine the first training data set to be transmitted based on the third information includes configuring the processing unit 120 to determine the first training data set to be transmitted based on the first information and the third information, wherein the first information indicates related information of the first training data set that the first network device is requested to transmit.

[0365] In a possible implementation, the transceiver unit 110 is configured to receive second information, the second information indicating related information of a second training dataset that the device is requested to transmit, the second information being determined based on performance of an AI model, the performance of the AI ​​model being determined by training based on the first training dataset, and the processing unit 120 is configured to determine the second training dataset to be transmitted based on the second information.

[0366] When communications device 100 is configured to perform the function of the training device in the method embodiment shown in FIG. 7 , processing unit 120 is configured to measure N reference signal groups and obtain N measurement result groups corresponding to the N reference signal groups, where each reference signal group within the N reference signal groups includes at least one reference signal, each reference signal group has the same group identifier, and N is an integer greater than 1. Transceiver unit 110 is configured to receive fourth information, where the fourth information indicates M reference signal groups within the N reference signal groups. Processing unit 120 is configured to determine first input information for an artificial intelligence (AI) model based on the fourth information and the N measurement result groups corresponding to the N reference signal groups, where the first input information includes the M measurement result groups corresponding to the M reference signal groups. The AI ​​model is used to obtain first output information based on the first input information, where the first output information includes group identifiers for each of K reference signal groups within the N reference signal groups, where each group identifier for the K reference signal groups corresponds to the K measurement result group with the best channel quality within the N measurement result groups.

[0367] In a possible implementation, the transceiver unit 110 is configured to receive configuration information, where the configuration information indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or group identifiers of the N reference signal groups.

[0368] 7 , the transceiver unit 110 is configured to transmit N reference signal groups to the training device, where each reference signal group in the N reference signal groups includes at least one reference signal, each reference signal group has the same group identifier, and N is an integer greater than 1. The transceiver unit 110 is configured to transmit fourth information to the training device, where the fourth information indicates M reference signal groups in the N reference signal groups, where the M reference signal groups are used to determine first input information. The AI ​​model is used to obtain first output information based on the first input information, where the first output information includes respective group identifiers of K reference signal groups in the N reference signal groups, where the respective group identifiers of the K reference signal groups correspond to the K measurement result groups with best channel qualities among the N measurement result groups corresponding to the N reference signal groups.

[0369] In a possible implementation, the transceiver unit 110 is configured to transmit configuration information, where the configuration information indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or group identifiers of the N reference signal groups.

[0370] 8 , the transceiver unit 110 is configured to receive a second reference signal set, the second reference signal set including N reference signal groups, each reference signal group in the N reference signal groups including at least one reference signal, where N is an integer greater than 1. The transceiver unit 110 is further configured to receive second beam instruction information, the second beam instruction information indicating beams corresponding to the first reference signal set, the beams corresponding to the first reference signal set being a subset of multiple beams corresponding to the second reference signal set, the beams corresponding to the first reference signal set being used to determine first input information for an AI model in the training device, the first input information being based on measurement results of the beams corresponding to the first reference signal set, the first reference signal set including M reference signal groups, where N is an integer greater than M, where M is an integer greater than or equal to 1.

[0371] In a possible implementation, the transceiver unit 110 is further configured to transmit the first configuration information.

[0372] In a possible implementation, the transceiver unit 110 is further configured to transmit third configuration information.

[0373] In a possible implementation, the processing unit 120 is further configured to measure N reference signal groups to obtain N measurement results.

[0374] 8 , the transceiver unit 110 is configured to receive a second reference signal set, the second reference signal set including N reference signal groups, each reference signal group in the N reference signal groups including at least one reference signal, where N is an integer greater than 1. The transceiver unit 110 is further configured to receive second beam instruction information, the second beam instruction information indicating beams corresponding to the first reference signal set, the beams corresponding to the first reference signal set being a subset of multiple beams corresponding to the second reference signal set, the beams corresponding to the first reference signal set being used to determine first input information for an AI model in the training device, the first input information being based on measurement results of the beams corresponding to the first reference signal set, the first reference signal set including M reference signal groups, where N is an integer greater than M, where M is an integer greater than or equal to 1.

[0375] In a possible implementation, the transceiver unit 110 is configured to transmit first configuration information.

[0376] In a possible implementation, the transceiver unit 110 is configured to transmit third configuration information.

[0377] 9 , the transceiver unit 110 is configured to receive a first reference signal set, the first reference signal set including M reference signal groups, each reference signal group in the M reference signal groups including at least one reference signal, where M is an integer greater than or equal to 1. The transceiver unit 110 is further configured to receive first beam indication information, the first beam indication information indicating a beam corresponding to the first reference signal set, the first reference signal set being used to determine first input information for the AI ​​model, the first input information being based on measurement results of the M reference signal groups included in the first reference signal set, the beam corresponding to the first reference signal set being a subset of multiple beams corresponding to a second reference signal set, the second reference signal set including N reference signal groups, where N is an integer greater than or equal to M.

[0378] In a possible implementation, the transceiver unit 110 is configured to receive first configuration information.

[0379] In a possible implementation, the transceiver unit 110 is configured to receive second configuration information.

[0380] In a possible implementation, the transceiver unit 110 is configured to receive third configuration information.

[0381] In a possible implementation, the processing unit 120 is configured to obtain first output information based on the first input information, and the transceiver unit 110 is configured to transmit the first output information.

[0382] 9 , the transceiver unit 110 is configured to transmit a first reference signal set, the first reference signal set including M reference signal groups, each reference signal group in the M reference signal groups including at least one reference signal, where M is an integer greater than or equal to 1. The transceiver unit 110 is further configured to transmit first beam indication information, the first beam indication information indicating a beam corresponding to the first reference signal set, the first reference signal set being used to determine first input information for the AI ​​model, the first input information being based on measurement results of the M reference signal groups included in the first reference signal set, the beam corresponding to the first reference signal set being a subset of multiple beams corresponding to a second reference signal set, the second reference signal set including N reference signal groups, where N is an integer greater than or equal to M.

[0383] In a possible implementation, the transceiver unit 110 is configured to transmit first configuration information.

[0384] In a possible implementation, the transceiver unit 110 is configured to transmit second configuration information.

[0385] In a possible implementation, the transceiver unit 110 is configured to transmit third configuration information.

[0386] In a possible implementation, the transceiver unit 110 is configured to receive first output information.

[0387] For a more detailed description of the processing unit 120 and the transceiver unit 110, please directly refer to the relevant descriptions of the method embodiments shown in Figures 5 to 9. The details will not be described again here.

[0388] 11 , the communication device 200 includes a processor 210 and an interface circuit 220. The processor 210 is coupled to the interface circuit 220. It will be understood that the interface circuit 220 may be a transceiver or an input / output interface. Optionally, the communication device 200 may further include a memory 230 configured to store instructions to be executed by the processor 210, to store input data required by the processor 210 to execute the instructions, or to store data generated after the processor 210 executes the instructions.

[0389] When the communication device 200 is configured to perform the method shown in FIG. 5, the processor 210 is configured to perform the functions of the processing unit 120 and the interface circuit 220 is configured to perform the functions of the transceiver unit 110.

[0390] When the communication device 200 is configured to perform the method shown in FIG. 6, the processor 210 is configured to perform the functions of the processing unit 120 and the interface circuit 220 is configured to perform the functions of the transceiver unit 110.

[0391] When the communication device 200 is configured to perform the method shown in FIG. 7, the processor 210 is configured to perform the functions of the processing unit 120 and the interface circuit 220 is configured to perform the functions of the transceiver unit 110.

[0392] When the communication device 200 is configured to perform the method shown in FIG. 8, the processor 210 is configured to perform the functions of the processing unit 120 and the interface circuit 220 is configured to perform the functions of the transceiver unit 110.

[0393] When the communication device 200 is configured to perform the method shown in FIG. 9, the processor 210 is configured to perform the functions of the processing unit 120 and the interface circuit 220 is configured to perform the functions of the transceiver unit 110.

[0394] It should be understood that the processor shown in FIG. 11 may include at least one processor, and the interface circuit may also include multiple interface circuits.

[0395] For the description of the relevant contents and beneficial effects of any one of the above-provided devices, please refer to the corresponding method embodiments provided above, and the details will not be described again in this specification.

[0396] If the communication device is a chip used in a training device (or terminal device), the chip in the training device (or terminal device) performs the functions of the training device (or terminal device) in the above-mentioned method embodiments. The chip in the training device (or terminal device) receives information from another module (e.g., a radio frequency module or an antenna) in the training device (or terminal device), and the information is transmitted to the training device (or terminal device) by a network device, or the chip in the training device (or terminal device) transmits information to another module (e.g., a radio frequency module or an antenna) in the training device (or terminal device), and the information is transmitted to the network device by the training device (or terminal device).

[0397] It will be understood that the processor in this embodiment of the present application may be a Central Processing Unit (CPU), or may be another general-purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0398] Based on the methods provided in the embodiments of the present application, the present application further provides a computer program product, which stores computer program code, which, when executed on a computer, enables the computer to execute the methods executed by the training device (or terminal device) or network device in the embodiments of Methods 500 to 900.

[0399] Based on the methods provided in the embodiments of the present application, the present application further provides a computer-readable medium, which stores program code, which, when executed on a computer, enables the computer to execute the methods 500 to 900 executed by a training device (or a terminal device) or a network device.

[0400] Based on the method provided in the embodiment of the present application, the present application further provides a communication system, which includes a training device and a first network device, wherein the training device is configured to perform steps corresponding to the training device of the method 500, and the first network device is configured to perform steps corresponding to the first network device of the method 500.

[0401] Based on the method provided in the embodiment of the present application, the present application further provides a communication system, which includes a training device and a first network device, wherein the training device is configured to perform steps corresponding to the training device of the method 600, and the network device is configured to perform steps corresponding to the first network device of the method 600.

[0402] Based on the method provided in the embodiment of the present application, the present application further provides a communication system, which includes a training device and a first network device, wherein the training device is configured to perform steps corresponding to the training device of the method 700, and the network device is configured to perform steps corresponding to the first network device of the method 700.

[0403] Based on the method provided in the embodiment of the present application, the present application further provides a communication system, which includes a training device and a network device, wherein the training device is configured to perform steps corresponding to the training device of the method 800, and the network device is configured to perform steps corresponding to the network device of the method 800.

[0404] Based on the method provided in the embodiment of the present application, the present application further provides a communication system, which includes an inference device such as a terminal device and a network device, wherein the terminal device is configured to perform steps corresponding to the terminal device of the method 900, and the network device is configured to perform steps corresponding to the network device of the method 900.

[0405] The method steps in the embodiments of the present application may be implemented in hardware or software instructions that can be executed by a processor. The software instructions may include corresponding software modules. The software modules may be stored in random access memory, flash memory, read-only memory, programmable read-only memory, erasable programmable read-only memory, electrically erasable programmable read-only memory, registers, hard disk, removable hard disk, CD-ROM, or any other form of storage medium well known in the art. For example, the storage medium is coupled to the processor such that the processor can read information from and write information to the storage medium. The storage medium may alternatively be a component of the processor. The processor and the storage medium may be located in an ASIC. Further, the ASIC may be located in a base station or a terminal. Indeed, the processor and the storage medium may reside as discrete components in the base station or a terminal.

[0406] All or part of the above-described embodiments may be implemented using software, hardware, firmware, or any combination thereof. When software is used to implement the above-described embodiments, all or part of the above-described embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer programs or instructions are loaded and executed on a computer, all or part of the procedures or functions of the embodiments of the present application are performed. The computer may be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user terminal, or another programmable device. The computer program or instructions may be stored on a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer program or instructions may be transmitted from a website, computer, server, or data center to another website, computer, server, or data center via wired or wireless communication. The computer-readable storage medium may be any available medium or data storage device that can be accessed by a computer, such as a server or data center that integrates one or more available media. The available media may be magnetic media such as floppy disks, hard disks, or magnetic tapes, or optical media such as digital video disks, or semiconductor media such as solid-state drives. The computer-readable storage media may be volatile or nonvolatile storage media, or may include both types of media: volatile and nonvolatile storage media.

[0407] In various embodiments of the present application, unless otherwise specified or there is no logical contradiction, the terms and / or descriptions in different embodiments are consistent and can be cross-referenced, and the technical features in different embodiments can be combined based on their internal logical relationships to form new embodiments.

[0408] It should be understood that various numbers in the embodiments of the present application are used only for distinction to facilitate description, and are not used to limit the scope of the embodiments of the present application. The sequence numbers of the above processes do not imply an execution order, and the execution order of the processes should be determined based on the functions and internal logic of the processes. [Explanation of symbols]

[0409] 100 Wireless access network, communication equipment 110 Network device, transceiver unit 110a Network devices, base stations 110b Network devices, base stations 120 Communication device, processing unit 120a Mobile phones, terminal devices 120b Car, Terminal Device 120c terminal device 120d Terminal Device 120e Smart Home Devices, Terminal Devices 120f terminal device 120g terminal device 120h Terminal Device 120i Unmanned Aerial Vehicles, Mobile Base Stations, Terminal Devices 120j terminal device 200 Communication Equipment 210 processors 220 Interface Circuit 230 memory 500 How to obtain the training dataset 600 How to obtain training dataset 700 methods 800 Communication Methods 900 Communication Methods

Claims

1. 1. A training dataset acquisition method, the method being performed by a training device or a chip or circuit configured to be disposed within a training device, the method comprising: transmitting first information to a network device, the first information indicating related information of a first training data set that the training device requests the network device to transmit; receiving the first training data set from the network device, the first training data set being based on the related information indicated by the first information, and the first training data set being used to train an artificial intelligence (AI) model; A method comprising:

2. 2. The method of claim 1, wherein the related information includes at least one of the following: information about the size of the first training data set, configuration information of inputs to the AI ​​model, or configuration information of reference signals used to train the AI ​​model.

3. The information regarding the size of the first training data set is:

3. The method of claim 2, wherein the training device determines the size of the training dataset required to complete training of the AI ​​model.

4. Before the step of transmitting first information to a network device, the method further comprises: determining a first performance of the AI ​​model; determining the information regarding the size of the first training dataset based on the first performance of the AI ​​model and a second performance of the AI ​​model, wherein the first performance is a current performance of the AI ​​model and the second performance is a target performance of the AI ​​model; 3. The method of claim 2, further comprising:

5. 5. The method of claim 2, wherein the configuration information of the reference signal includes at least one of the following: an identifier of the reference signal, a time domain resource of the reference signal, a frequency domain resource of the reference signal, a transmission periodicity of the reference signal, or a type of the transmitted reference signal.

6. 6. The method of claim 1, wherein the first information includes at least one of the following: identification information of the AI ​​model, information about an application scenario of the AI ​​model, usage information of the AI ​​model, or information about the computing capabilities of the training device.

7. training the AI ​​model based on the first training dataset to determine performance of the AI ​​model; transmitting second information to the network device based on the performance of the AI ​​model, the second information indicating related information of a second training data set that the training device requests the network device to transmit; receiving the second training data set from the network device, the second training data set being a training data set based on the related information indicated by the second information, and the second training data set being used to train the AI ​​model; 7. The method of claim 1, further comprising:

8. 1. A training dataset acquisition method, the method being performed by a first network device or a chip or circuit configured to be located in the first network device, the method comprising: receiving first information from a training device, the first information indicating related information of a first training data set that the first network device is requested to transmit; transmitting the first training data set to the training device based on the related information indicated by the first information, wherein the first training data set is used to train the artificial intelligence (AI) model; A method comprising:

9. 9. The method of claim 8, wherein the related information includes at least one of the following: information about the size of the first training data set, configuration information of inputs to the AI ​​model, or configuration information of reference signals used to train the AI ​​model.

10. 10. The method of claim 9, wherein the information about the size of the first training data set is determined based on a size of a training data set required to complete training of the AI ​​model.

11. 11. The method of claim 9 or 10, wherein the configuration information of the reference signal includes at least one of the following: an identifier of the reference signal, a time domain resource of the reference signal, a frequency domain resource of the reference signal, a transmission periodicity of the reference signal, or a type of the transmitted reference signal.

12. 12. The method of claim 8, wherein the first information includes at least one of the following: identification information of the AI ​​model, information about an application scenario of the AI ​​model, usage information of the AI ​​model, or information about the computing capabilities of the training device.

13. The method comprises: acquiring third information from a second network device, the third information being relevant information for training the AI ​​model, and the first network device being a target network device to which the training device is handed over from the second network device; transmitting the first training data set to the training device based on the related information indicated by the first information; transmitting the first training data set to the training device based on the related information indicated by the first information and the third information.

13. The method according to any one of claims 8 to 12.

14. The third information is as follows: at least one of information about the size of a training dataset that the training device requests the second network device to transmit, information about the size of the training dataset required to complete training of the AI ​​model, the identification information of the AI ​​model, or the information about the computing capabilities of the training device; The method of claim 13.

15. 15. The method of claim 13 or 14, wherein the first network device stores a first mapping relationship, the first mapping relationship being a mapping relationship between an identifier of an AI model and a size of a training dataset corresponding to the identifier of the AI ​​model.

16. The third information further includes information regarding a duration for transmitting the training data set and / or information regarding a method for transmitting the training data set, and the method further comprises: determining whether the first network device and / or the training device has the capability to support training of the AI ​​model based on the third information and resource usage of the first network device; 16. The method according to any one of claims 13 to 15.

17. receiving second information from the training device, the second information indicating related information of a second training data set that the first network device is requested to transmit, the second information being determined based on performance of the AI ​​model, the performance of the AI ​​model being determined by training based on the first training data set; transmitting the second training data set to the training device based on the related information indicated by the second information, wherein the second training data set is used to train the AI ​​model; 17. The method of any one of claims 8 to 16, further comprising:

18. 1. A training dataset acquisition method, the method being performed by a first network device or a chip or circuit configured to be located on the first network device, the method comprising: Obtaining third information from a second network device, the third information being relevant information for training an artificial intelligence (AI) model, and the first network device being a target network device to which a training device is handed over from the second network device; receiving first information from the training device, the first information being used to request the first network device to transmit a training data set; determining a first training data set to be transmitted based on the third information; transmitting, by the first network device, the first training data set to the training device based on the first information, wherein the first training data set is used to train the AI ​​model; A method comprising:

19. 19. The method of claim 18, wherein the third information includes at least one of the following: information regarding a size of a training dataset requested to be transmitted from the second network device, information regarding a size of a training dataset required to complete training of the AI ​​model, identification information of the AI ​​model, or information regarding the computing capabilities of the training device.

20. 20. The method of claim 18 or 19, wherein the first network device stores a first mapping relationship, the first mapping relationship being a mapping relationship between an identifier of the AI ​​model and a size of a training dataset corresponding to the identifier of the AI ​​model.

21. 21. The method of claim 18, wherein the third information further includes information regarding a duration for transmitting the training dataset and / or information regarding a method for transmitting the training dataset, and the method further includes a step of determining, by the first network device, whether the first network device and / or the training device has the capability to support training of the AI ​​model based on the third information and resource usage of the first network device.

22. determining a first training data set to be transmitted based on the third information; determining the first training data set to be transmitted based on the first information and the third information, wherein the first information indicates related information of the first training data set that the first network device is requested to transmit; 22. The method of any one of claims 18 to 21.

23. 23. The method of claim 22, wherein the related information includes at least one of the following: information about the size of the first training data set, input information for the AI ​​model, or configuration information for a reference signal used to train the AI ​​model.

24. 24. The method of claim 23, wherein the information about the size of the first training data set is determined based on the size of the training data set required to complete training of the AI ​​model.

25. 24. The method of claim 23, wherein the configuration information of the reference signal includes at least one of the following: an identifier of the reference signal, a time domain resource of the reference signal, a frequency domain resource of the reference signal, a transmission periodicity of the reference signal, or a type of the transmitted reference signal.

26. 26. The method of any one of claims 22 to 25, wherein the first information includes at least one of the following: identification information of the AI ​​model, information about an application scenario of the AI ​​model, usage information of the AI ​​model, or information about the computing capabilities of the training device.

27. receiving second information from the training device, the second information comprising: indicating related information of a second training data set that the first network device is requested to transmit, the second information being determined based on performance of the AI ​​model, the performance of the AI ​​model being determined by training based on the first training data set; determining the second training data set to be transmitted based on the second information; 27. The method of any one of claims 18 to 26, further comprising:

28. 1. A communication method, the method being performed by a training device or a chip or circuit configured to be disposed within a training device, the method comprising: measuring N reference signal groups to obtain N measurement result groups corresponding to the N reference signal groups, wherein each reference signal group of the N reference signal groups includes at least one reference signal, each reference signal group has the same group identifier, and N is an integer greater than 1; receiving fourth information from the network device, the fourth information indicating M reference signal groups among the N reference signal groups; determining first input information of an artificial intelligence (AI) model based on the fourth information and the N measurement result groups corresponding to the N reference signal groups, wherein the first input information includes M measurement result groups corresponding to the M reference signal groups; Including, the AI ​​model is used to obtain first output information based on the first input information, the first output information including group identifiers of K reference signal groups within the N reference signal groups, the group identifiers of the K reference signal groups corresponding to K measurement result groups having best channel qualities within the N measurement result groups, and each measurement result group including one or more measurement results; method.

29. 29. The method of claim 28, wherein the first output information further includes group identifiers for each of the remaining (N−K) reference signal groups in the N measurement result groups, and the group identifiers for each of the (N−K) reference signal groups correspond to the (N−K) measurement result groups.

30. the fourth information includes N fields, the N fields correspond one-to-one to the N reference signal groups, and bit values ​​of M fields among the N fields are different from bit values ​​of the remaining (N-M) fields; The fourth information indicating M reference signal groups among the N reference signal groups is the M fields in the fourth information indicating the M reference signal groups; 30. The method of claim 28 or 29.

31. receiving fifth information from the network device, the fifth information indicating P reference signal groups among the N reference signal groups, the fifth information including N fields, the N fields corresponding one-to-one to the N reference signal groups, and bit values ​​of P fields among the N fields being different from bit values ​​of the remaining (N-P) fields; The fifth information indicating P reference signal groups among the N reference signal groups is the P fields in the fifth information indicating the P reference signal groups; 31. The method of any one of claims 28 to 30.

32. receiving configuration information from the network device, the configuration information indicating one or more of the following: time domain resources for the N reference signal groups; frequency domain resources for the N reference signal groups; transmission periodicities for the N reference signal groups; or group identifiers for the N reference signal groups.

32. The method of any one of claims 28 to 31.

33. 1. A communication method, the method being performed by a network device or a chip or circuit configured to be disposed in a network device, the method comprising: transmitting N reference signal groups to a training device, each reference signal group of the N reference signal groups including at least one reference signal, each reference signal group having the same group identifier, and N being an integer greater than 1; transmitting fourth information to the training device, the fourth information indicating M reference signal groups within the N reference signal groups, the M reference signal groups being used to determine first input information; Including, an AI model is used to obtain first output information based on the first input information, the first output information including group identifiers of K reference signal groups within the N reference signal groups, the group identifiers of the K reference signal groups corresponding to K measurement result groups having best channel qualities within N measurement result groups corresponding to the N reference signal groups, and each measurement result group including one or more measurement results; method.

34. the first output information further includes group identifiers of the remaining (N-K) reference signal groups within the N measurement result groups, and the group identifiers of the (N-K) reference signal groups correspond to the (N-K) measurement result groups.

34. The method of claim 33.

35. the fourth information includes N fields, the N fields correspond one-to-one to the N reference signal groups, and bit values ​​of M fields among the N fields are different from bit values ​​of the remaining (N-M) fields; the fourth information indicates M reference signal groups among the N reference signal groups; the M fields in the fourth information indicating the M reference signal groups; 35. The method of claim 33 or 34.

36. transmitting fifth information to the training device, the fifth information indicating P reference signal groups among the N reference signal groups, the fifth information including N fields, the N fields corresponding one-to-one to the N reference signal groups, and bit values ​​of P fields among the N fields being different from bit values ​​of the remaining (N-P) fields; The fifth information indicating P reference signal groups among the N reference signal groups is the P fields in the fifth information indicating the P reference signal groups; 36. The method of any one of claims 33 to 35.

37. transmitting configuration information to the training device, the configuration information indicating one or more of the following: time domain resources of the N reference signal groups; frequency domain resources of the N reference signal groups; transmission periodicities of the N reference signal groups; or group identifiers of the N reference signal groups.

37. The method of any one of claims 33 to 36.

38. 1. A communication method, the method being performed by a training device or a chip or circuit configured to be disposed within a training device, the method comprising: receiving a second set of reference signals, the second set of reference signals including N reference signal groups, each reference signal group in the N reference signal groups including at least one reference signal, where N is an integer greater than 1; receiving second beam instruction information, the second beam instruction information indicating beams corresponding to a first reference signal set, the beams corresponding to the first reference signal set being a subset of a plurality of beams corresponding to the second reference signal set, the beams corresponding to the first reference signal set being used to determine first input information for an AI model in the training device, the first input information being based on measurement results of the beams corresponding to the first reference signal set, the first reference signal set including M reference signal groups, N being an integer greater than M and M being an integer equal to or greater than 1; The AI ​​model is used to obtain first output information based on the first input information, the first output information indicating K beams predicted to have best channel qualities among the plurality of beams corresponding to the second reference signal set, K is an integer greater than or equal to 1 and less than N, and labels of the AI ​​model are the K beams having best channel qualities in measurement results of the second reference signal set. method.

39. The first output information indicating K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set is determined by the following: information regarding the K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set; or a respective group identifier for K reference signal groups, the K reference signal groups corresponding to the K measurement results predicted to have the best channel qualities among the N measurement results corresponding to the N reference signal groups, and a predefined or preconfigured correspondence exists between the respective group identifiers of the K reference signal groups and the K beams; or A plurality of beam information corresponding to the N reference signal groups, and N measurement results corresponding to the beam information; or a group identifier for each of the N reference signal groups and N measurement results for the N reference signal groups, wherein a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams; 39. The method of claim 38, comprising at least one of:

40. The second beam instruction information indicating a beam corresponding to a first reference signal set is the second beam designation information indicating a position of the beam corresponding to the first set of reference signals within the plurality of beams corresponding to the second set of reference signals; 40. The method of claim 38 or 39.

41. the second beam instruction information includes N fields, the N fields having a one-to-one correspondence with the plurality of beams corresponding to the second reference signal set, and bit values ​​of M fields among the N fields are different from bit values ​​of the remaining (N-M) fields; The second beam instruction information indicating a beam corresponding to a first reference signal set is the M fields in the second beam instruction information correspond to a first reference signal set; 41. The method of any one of claims 38 to 40.

42. transmitting first configuration information to the training device, the first configuration information indicating one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, the group identifiers of the N reference signal groups, or the beam information of the N reference signal groups; 42. The method of any one of claims 38 to 41, further comprising:

43. 43. The method of any one of claims 38 to 42, wherein the second beam instruction information indicates a beam corresponding to the M reference signal groups.

44. The second beam indication information indicates beams corresponding to the M reference signal groups, the second beam instruction information includes group identifiers or beam information of the M reference signal groups, the M reference signal groups being part of the N reference signal groups, and the predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

44. The method of claim 43.

45. The method comprises: transmitting third configuration information of the N reference signal groups to the training device; when the second beam direction information includes the group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective group identifiers of the N reference signal groups and indicates one or more of the following: the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, the transmission periodicity of the N reference signal groups, or beams of the N reference signal groups; The M reference signal groups being part of the N reference signal groups may include N group identifiers of the N reference signal groups including M group identifiers of the M reference signal groups, or When the second beam indication information includes the beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective beam information of the N reference signal groups and indicates one or more of the following: the group identifiers of the N reference signal groups, the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, or the transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups.

45. The method of claim 44.

46. 46. ​​The method of any one of claims 38 to 45, further comprising the step of measuring the N reference signal groups to obtain the N measurement results, the N measurement results corresponding to the N beams, and the N measurement results including the measurement results of the beam corresponding to the first reference signal set.

47. 1. A communication method, the method being performed by a network device or a chip or circuit configured to be disposed in a network device, the method comprising: transmitting a second reference signal set to a training device, the second reference signal set including N reference signal groups, each reference signal group in the N reference signal groups including at least one reference signal, where N is an integer greater than 1; transmitting second beam instruction information to the training device, the second beam instruction information indicating beams corresponding to a first reference signal set, the beams corresponding to the first reference signal set being a subset of a plurality of beams corresponding to the second reference signal set, the beams corresponding to the first reference signal set being used to determine first input information for an AI model in the training device, the first input information being based on measurement results of the beams corresponding to the first reference signal set, the first reference signal set including M reference signal groups, N being an integer greater than M and M being an integer equal to or greater than 1; The AI ​​model is used to obtain first output information based on the first input information, the first output information indicating K beams predicted to have best channel qualities among the plurality of beams corresponding to the second reference signal set, K is an integer greater than or equal to 1 and less than N, and labels of the AI ​​model are the K beams having best channel qualities in measurement results of the second reference signal set. method.

48. The first output information indicating K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set is determined by the following: information regarding the K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set; or a respective group identifier for K reference signal groups, the K reference signal groups corresponding to the K measurement results predicted to have the best channel qualities among the N measurement results corresponding to the N reference signal groups, and a predefined or preconfigured correspondence exists between the respective group identifiers of the K reference signal groups and the K beams; or A plurality of beam information corresponding to the N reference signal groups, and N measurement results corresponding to the beam information; or a group identifier for each of the N reference signal groups and N measurement results for the N reference signal groups, wherein a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams; 48. The method of claim 47, comprising at least one of:

49. 49. The method of claim 47 or 48, wherein the second beam instruction information indicating a beam corresponding to a first set of reference signals comprises the second beam instruction information indicating a position of the beam corresponding to the first set of reference signals within the plurality of beams corresponding to the second set of reference signals.

50. the second beam instruction information includes N fields, the N fields having a one-to-one correspondence with the plurality of beams corresponding to the second reference signal set, and bit values ​​of M fields among the N fields are different from bit values ​​of the remaining (N-M) fields; The second beam instruction information indicating a beam corresponding to a first reference signal set is the M fields in the second beam direction information correspond to the first reference signal set; 50. The method of any one of claims 47 to 49.

51. transmitting first configuration information to the training device, the first configuration information indicating one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, the group identifiers of the N reference signal groups, or the beam information of the N reference signal groups; 51. The method of any one of claims 47 to 50, further comprising:

52. 52. The method of any one of claims 47 to 51, wherein the second beam instruction information indicates a beam corresponding to the M reference signal groups.

53. The second beam indication information indicates beams corresponding to the M reference signal groups, the second beam instruction information includes group identifiers or beam information of the M reference signal groups, the M reference signal groups being part of the N reference signal groups, and the predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

53. The method of claim 52.

54. The method comprises: transmitting third configuration information of the N reference signal groups to the training device; When the second beam direction information includes the group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective group identifiers of the N reference signal groups, and the second beam direction information indicates one or more of the following: the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, the transmission periodicity of the N reference signal groups, or beams of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups; or when the second beam indication information includes the beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective beam information of the N reference signal groups, the second beam indication information indicates one or more of the following: the group identifiers of the N reference signal groups, the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, or the transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups means that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups; 54. The method of claim 53, comprising:

55. 1. A communication method, the method being performed by a terminal device or a chip or circuit configured to be located in a terminal device, the method comprising: receiving a first set of reference signals, the first set of reference signals including M reference signal groups, each reference signal group in the M reference signal groups including at least one reference signal, where M is an integer greater than or equal to 1; receiving first beam instruction information, the first beam instruction information indicating a beam corresponding to the first reference signal set, the first reference signal set being used to determine first input information of the AI ​​model, the first input information being based on measurement results of the M reference signal groups included in the first reference signal set, the beam corresponding to the first reference signal set being a subset of a plurality of beams corresponding to a second reference signal set, the second reference signal set including N reference signal groups, N being an integer greater than or equal to M; The AI ​​model is used to obtain first output information based on the first input information, and the first output information indicates K beams that are predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set, where K is an integer greater than or equal to 1 and less than N. method.

56. The first output information indicating K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set may be determined by the following: information regarding the K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set; or a respective group identifier for K reference signal groups, the K reference signal groups corresponding to K measurement results predicted to have the best channel qualities among the N measurement results corresponding to the N reference signal groups, and a predefined or preconfigured correspondence exists between the respective group identifiers of the K reference signal groups and the K beams; or A plurality of beam information corresponding to the N reference signal groups, and N measurement results corresponding to the beam information; or a group identifier for each of the N reference signal groups and N measurement results for the N reference signal groups, wherein a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams; 56. The method of claim 55, comprising at least one of:

57. 57. The method of claim 55 or 56, wherein the first beam designation information indicating a beam corresponding to the first reference signal set comprises the first beam designation information indicating a position of the beam corresponding to the first reference signal set within the plurality of beams corresponding to the second reference signal set.

58. the first beam designation information includes N fields, the N fields having a one-to-one correspondence with the plurality of beams corresponding to the second reference signal set, and bit values ​​of M fields among the N fields are different from bit values ​​of the remaining (N-M) fields; The first beam designation information indicating a beam corresponding to the first reference signal set is the M fields in the first beam direction information correspond to the first reference signal set.

58. The method of any one of claims 55 to 57.

59. receiving first configuration information, wherein the first configuration information indicates one or more of the following: time domain resources of the M reference signal groups; frequency domain resources of the M reference signal groups; transmission periodicities of the M reference signal groups; group identifiers of the M reference signal groups; or beam information of the M reference signal groups; 59. The method of any one of claims 55 to 58, further comprising:

60. 60. The method of any one of claims 55 to 59, wherein the first beam instruction information indicates a beam corresponding to the M reference signal groups.

61. The first beam indication information indicates beams corresponding to the M reference signal groups, the first beam instruction information includes the group identifiers or the beam information of the M reference signal groups, the M reference signal groups are part of the N reference signal groups, and the predefined or preconfigured correspondence exists between the N reference signal groups and the N beams.

61. The method of claim 60.

62. the first beam direction information is included in second configuration information of the M groups of the reference signals; When the first beam direction information includes the group identifiers of the M reference signal groups, the second configuration information further includes one or more of the time domain resources, the frequency domain resources, the transmission periodicity, or the beam information of the M reference signal groups; or When the first beam direction information includes the beam information of the M reference signal groups, the second configuration information further includes one or more of the time domain resources, the frequency domain resources, the transmission periodicity, or the group identifier of the M reference signal groups.

62. The method of claim 61.

63. The method comprises: receiving third configuration information of the N reference signal groups; When the first beam indication information includes the group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective group identifiers of the N reference signal groups and indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or beams of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups; or when the first beam instruction information includes the beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective beam information of the N reference signal groups; indicates one or more of the following: the group identifiers of the N reference signal groups; the time domain resources of the N reference signal groups; the frequency domain resources of the N reference signal groups; or the transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes the N group identifiers of the N reference signal groups including M group identifiers of the M reference signal groups.

63. The method of claim 61 or 62.

64. using the AI ​​model based on the first input information; obtaining the first output information; transmitting the first output information; 64. The method of any one of claims 55 to 63, further comprising:

65. 1. A communication method, the method being performed by a network device or a chip or circuit configured to be disposed in a network device, the method comprising: transmitting a first reference signal set to a terminal device, the first reference signal set including M reference signal groups, each reference signal group in the M reference signal groups including at least one reference signal, where M is an integer greater than or equal to 1; a step of transmitting first beam indication information to the terminal device, the first beam indication information indicating a beam corresponding to the first reference signal set, the first reference signal set being used to determine first input information of the AI ​​model, the first input information being based on measurement results of the M reference signal groups included in the first reference signal set, the beam corresponding to the first reference signal set being a subset of a plurality of beams corresponding to a second reference signal set, the second reference signal set including N reference signal groups, N being an integer greater than or equal to M; Including, The AI ​​model is used to obtain first output information based on the first input information, and the first output information indicates K beams that are predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set, where K is an integer greater than or equal to 1 and less than N. method.

66. The first output information indicating K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set is information regarding the K beams predicted to have the best channel quality among the plurality of beams corresponding to the second reference signal set; or a respective group identifier for K reference signal groups, the K reference signal groups corresponding to the K measurement results predicted to have the best channel qualities among the N measurement results corresponding to the N reference signal groups, and a predefined or preconfigured correspondence exists between the respective group identifiers of the K reference signal groups and the K beams; or A plurality of beam information corresponding to the N reference signal groups, and N measurement results corresponding to the beam information; or a group identifier for each of the N reference signal groups and N measurement results for the N reference signal groups, wherein a predefined or preconfigured correspondence exists between the N reference signal groups and the N beams; 66. The method of claim 65, comprising at least one of:

67. wherein the first beam designation information indicates a beam corresponding to the first reference signal set, the first beam designation information indicating a position of the beam corresponding to the first reference signal set within the plurality of beams corresponding to the second reference signal set.

67. The method of claim 65 or 66.

68. the first beam designation information includes N fields, the N fields having a one-to-one correspondence with the plurality of beams corresponding to the second reference signal set, and bit values ​​of M fields among the N fields are different from bit values ​​of the remaining (N-M) fields; The first beam designation information indicating a beam corresponding to the first reference signal set is the M fields in the first beam direction information correspond to the first reference signal set; 68. The method of any one of claims 65 to 67, comprising:

69. transmitting first configuration information to the terminal device, wherein the first configuration information indicates one or more of the following: time domain resources of the M reference signal groups, frequency domain resources of the M reference signal groups, transmission periodicities of the M reference signal groups, group identifiers of the M reference signal groups, or beam information of the M reference signal groups; 69. The method of any one of claims 65 to 68, comprising:

70. 70. The method of any one of claims 65 to 69, wherein the first beam instruction information indicates a beam corresponding to the M reference signal groups.

71. The first beam indication information indicates beams corresponding to the M reference signal groups, the first beam instruction information includes the group identifiers or the beam information of the M reference signal groups, the M reference signal groups are part of the N reference signal groups, and there is the predefined or preconfigured correspondence between the N reference signal groups and the N beams; 71. The method of claim 70, comprising:

72. the first beam direction information is included in second configuration information of the M reference signal groups; When the first beam direction information includes the group identifiers of the M reference signal groups, the second configuration information further includes one or more of the time domain resources, the frequency domain resources, the transmission periodicity, or the beam information of the M reference signal groups; or When the first beam direction information includes the beam information of the M reference signal groups, the second configuration information further includes one or more of the time domain resources, the frequency domain resources, the transmission periodicity, or the group identifier of the M reference signal groups.

72. The method of claim 71.

73. The method comprises: further comprising transmitting third configuration information of the N reference signal groups to the terminal device; When the first beam indication information includes the group identifiers of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective group identifiers of the N reference signal groups and indicates one or more of the following: time domain resources of the N reference signal groups, frequency domain resources of the N reference signal groups, transmission periodicities of the N reference signal groups, or beams of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups; or When the first beam indication information includes the beam information of the M reference signal groups, the third configuration information of the N reference signal groups includes the respective beam information of the N reference signal groups and indicates one or more of the following: the group identifiers of the N reference signal groups, the time domain resources of the N reference signal groups, the frequency domain resources of the N reference signal groups, or the transmission periodicity of the N reference signal groups, and the M reference signal groups being part of the N reference signal groups includes that the N group identifiers of the N reference signal groups include M group identifiers of the M reference signal groups.

73. The method of claim 71 or 72.

74. 74. The method of any one of claims 65 to 73, further comprising receiving the first output information from the terminal device.

75. A communications device configured to perform a method according to any one of claims 1 to 7, or configured to perform a method according to any one of claims 28 to 32, or configured to perform a method according to any one of claims 38 to 46.

76. A communications device configured to perform a method according to any one of claims 8 to 17 or configured to perform a method according to any one of claims 18 to 27.

77. A communications device configured to perform a method according to any one of claims 33 to 37 or configured to perform a method according to any one of claims 47 to 54.

78. A communications device configured to perform a method according to any one of claims 55 to 64 or configured to perform a method according to any one of claims 65 to 74.

79. a processor coupled to a memory, the processor configured to invoke computer program instructions stored in the memory to perform the method of any one of claims 1 to 7, or to implement the method of any one of claims 28 to 32, or to implement the method of any one of claims 38 to 46; Communication equipment.

80. a processor coupled to a memory, the processor configured to invoke computer program instructions stored in the memory to execute the method of any one of claims 8 to 17 or to implement the method of any one of claims 18 to 27; Communication equipment.

81. a processor coupled to a memory, the processor configured to invoke computer program instructions stored in the memory to execute the method of any one of claims 33 to 37 or to implement the method of any one of claims 47 to 54; Communication equipment.

82. a processor coupled to a memory, the processor configured to invoke computer program instructions stored in the memory to execute the method of any one of claims 55 to 64 or to implement the method of any one of claims 65 to 74; Communication equipment.

83. 1. A communication system comprising: A communication device configured to perform the method according to any one of claims 1 to 7; and a communication device configured to implement the method according to any one of claims 8 to 17; or A communication device configured to perform the method according to any one of claims 1 to 7; and a communication device configured to perform the method according to any one of claims 18 to 27; or A communications device configured to implement a method according to any one of claims 28 to 32; and a communications device configured to implement a method according to any one of claims 33 to 37; or A communications device configured to implement a method according to any one of claims 38 to 46; and a communications device configured to implement a method according to any one of claims 47 to 54; or A communications device configured to implement a method according to any one of claims 55 to 64; and a communications device configured to implement a method according to any one of claims 65 to 74. a communication system including:

84. 75. A computer-readable storage medium having instructions stored thereon that, when executed on a computer, enable the computer to perform the method of any one of claims 1 to 74.

85. 75. A computer program product comprising instructions that, when executed on a computer, enable the computer to carry out the method of any one of claims 1 to 74.

Images