Method and communication device for acquiring training data in AI model training.

Abstract

The present application provides a method for acquiring training data in training an AI model, which can be applied to a scenario in which an AI model training network element (second network element) and a training data collection network element (first network element) are logically deployed separately. The first network element receives first information from the second network element, the first information being for determining the validity of candidate training data collected by the first network element. The first network element collects candidate training data for the AI ​​model and determines the validity of the candidate training data based on the first information. When it is determined that the candidate training data is valid, the first network element sends the valid candidate training data to the second network element but does not send invalid candidate training data. When the collected candidate training data does not include valid candidate training data, the first network element indicates to the second network element that the currently collected training data is invalid and does not send the currently collected candidate training data to the second network element. In this way, waste of air interface resources can be reduced.

Description

[Technical Field]

[0001] Embodiments of the present application relate to the field of machine learning, and more specifically, to a method and a communication apparatus for acquiring training data in AI model training. [Background Art]

[0002] The present application claims priority from Chinese Patent Application No. 202211203052.X entitled "METHOD FOR OBTAINING TRAINING DATA IN AI MODEL TRAINING AND COMMUNICATION APPARATUS" filed with the China National Intellectual Property Administration on September 29, 2022, which is incorporated herein by reference in its entirety.

[0003] When an AI model is applied to some application scenarios of air interface technology, AI model training and AI model training data collection may be deployed on different network elements. In this current situation, training or updating of an AI model requires exchange of training data (for example, reference signal measurement results and / or labels) between an AI model training network element and a training data collection network element.

[0004] In existing solutions, training data between an AI model training network element and a training data collection network element is usually transmitted periodically or exchanged continuously. However, this information exchange mode may cause waste of air interface resources. [Summary of the Invention]

[0005] The present application provides a method and a communication apparatus for acquiring training data in AI model training, so as to reduce waste of air interface resources.

[0006] According to the first aspect, a method is provided for acquiring training data in AI model training, the method which can be applied to training data acquisition network elements, such as terminal devices or access network devices. The method includes the following:

[0007] The first network element receives first information from the second network element, and the first information is used to determine the effectiveness of the candidate training data collected by the first network element, and the effectiveness determination result includes valid or invalid. The first network element collects candidate training data for the AI ​​model. The first network element sends second information to the second network element based on the candidate training data and the first information, and the second information indicates the result of the effectiveness determination.

[0008] In the technical solution of this application, a first network element is a network element that collects training data for an AI model, and a second network element is a network element that trains the AI ​​model. When the second network element needs the first network element to collect training data for the AI ​​model, the second network element sends first information to the first network element. The first information indicates to the first network element that it is collecting training data for the AI ​​model, and further, it is for determining the validity of the candidate training data collected by the first network element (also abbreviated as validity determination). The first network element collects the candidate training data for the AI ​​model and determines the validity of the collected candidate training data based on the first information. The first network element then sends second information to the second network element indicating the result of the validity determination. Based on this technical solution, after completing the collection of candidate training data once, the first network element determines the validity of the collected candidate training data. Instead of providing collected data to a second network element without screening, only valid candidate training data is provided by the first network element as training data for the second network element to use. This reduces the transmission of collected invalid candidate training data, thereby reducing the waste of air interface resources.

[0009] With respect to the first aspect, in some implementations of the first aspect, the first information is for determining constraints which are for validity determination. Optionally, the constraints are: Thresholds for quality indicators and criteria for determining quality indicators, Thresholds for the amount of training data that meets the quality indicator criteria, and criteria for determining the amount of training data, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. It may include one or more of the following.

[0010] With respect to the first aspect, in some implementations of the first aspect, the first information is Thresholds for quality metrics, quality indicator criteria, Threshold for the amount of training data that satisfies the quality indicator criteria, Criteria for determining the amount of training data that meets the quality indicator criteria, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. Show one or more of them.

[0011] Any information contained above that is not indicated by the first piece of information may be predefined in the protocol.

[0012] Optionally, the first information indicating a portion of the above information may include explicitly indicating one or more pieces of information within that portion, or implicitly indicating one or more pieces of information within that portion. Explicit indication may include the first information containing one or more pieces of information within the portion of the above information explicitly indicated by the first information. Implicit indication may include the first information containing other information corresponding to one or more pieces of information within the portion of the above information implicitly indicated by the first information. Optionally, this other information may include an index having correspondence with one or more pieces of information within the portion of the above information implicitly indicated by this other information. This other information may contain one or more pieces of information, each indicating all the portions of information within the implicitly indicated portion of the above information.

[0013] Optionally, the above correspondences may be predefined, pre-stored, or pre-configured in the protocol. In the case of pre-configuration, the correspondences between multiple indices and multiple values ​​of combinations of one or more items in the above information may be configured by radio resource control (RRC) signaling.

[0014] Optionally, the first information may be carried in control information, such as downlink control information (DCI).

[0015] Optionally, the above quality indicators may include one or more quality indicators, each of which has a corresponding threshold and corresponding decision criterion. For example, a quality indicator may include one or more quality indicators of a reference signal measurement result or a label quality indicator. The label is used as a truth value for comparison for AI model training. For example, the label may include one or more of location information, beam pattern, channel measurement results, etc.

[0016] Optionally, the threshold for quality metrics may include the above threshold for the amount of training data that satisfies the quality metrics criteria, and the quality metrics criteria may include criteria for the amount of training data that satisfies the quality metrics criteria.

[0017] In this application, the first information may be for determining constraints and can be implemented in several ways. Several examples are used below for illustrative purposes.

[0018] In the example, the first piece of information indicates thresholds for one or more quality metrics, and the criteria for one or more quality metrics are predefined in the protocol. For example, the quality metrics include the signal-to-interference plus noise ratio (SINR) of the training data and the amount of training data. The criterion for SINR is that the SINR is greater than or equal to the threshold Q. The criterion for the amount of training data is that the amount of training data is greater than or equal to the threshold N. For example, the first piece of information indicates Q and N, and both the SINR criterion and the training data amount criterion are predefined in the protocol.

[0019] In this example, the criteria for quality indicators are predefined in the protocol, thus reducing indication overhead.

[0020] In another example, the first piece of information indicates thresholds for one or more quality metrics and criteria for one or more quality metrics. For example, quality metrics include the SINR of the training data and the amount of training data. The criterion for SINR is that the SINR is greater than or equal to the threshold Q. The criterion for the amount of training data is that the amount of training data is greater than or equal to the threshold N. For example, the first piece of information indicates Q and N. In addition, the first piece of information includes an information field, which indicates the SINR criterion and the training data amount criterion. For example, if the value of the information field is 1, it indicates that "the SINR is greater than or equal to Q and the amount of training data is greater than or equal to N," and if the value of the information field is 0, it indicates that "the SINR is greater than Q and the amount of training data is greater than N."

[0021] In this example, the first piece of information indicates the threshold and criteria for quality metrics, and therefore the second network element can adaptively update the constraints based on changes in the requirements for the training data. This is applicable to scenarios where constraints change frequently, can improve the adaptability of the AI ​​model to different application scenarios, and can improve the probability of collecting training data that satisfies the requirements in different application scenarios.

[0022] In still another example, the first information indicates thresholds for some quality indicators, and thresholds for other quality indicators and determination criteria for quality indicators are predefined in a protocol. For example, the quality indicators include the SINR of training data and the amount of training data, the determination criterion for SINR is that the SINR of the training data is greater than or equal to a threshold Q, and the determination criterion for the amount of training data is that the amount of training data is greater than or equal to a threshold N. For example, the first information indicates Q, and the threshold N for the amount of training data, the determination criterion for SINR, and the determination criterion for the amount of training data may be predefined in the protocol.

[0023] In this example, thresholds for quality indicators and determination criteria for quality indicators that have a long change period in the application scenario are predefined in the protocol, which can reduce signaling overhead. Thresholds for quality indicators and determination criteria for quality indicators that change frequently are indicated by the first information, so the requirements for the required training data can be flexibly adjusted. In this example, both signaling overhead and the flexibility of constraint update can be considered.

[0024] In yet another example, the first information indicates a threshold of a part of quality indicators and one piece of index information, and the index information is used for determining the judgment criterion of the part of quality indicators, the threshold of another quality indicator in constraints, and the judgment criterion of another quality indicator. For example, the first information indicates the SINR threshold Q and index 0, where index 0 indicates that the threshold of the amount of training data is N, the judgment criterion for SINR is that the SINR of the training data is greater than or equal to Q, and the judgment criterion for the amount of training data is that the amount of training data is at least N. Optionally, index 0 is an index value in one of a plurality of application scenarios. For example, the plurality of application scenarios include, but are not limited to, CSI prediction, uplink positioning, downlink positioning, or beam management, and index 0 is one of one or more indexes corresponding to a beam management scenario. Optionally, index 0 is an index value in a specific application scenario. For example, among a plurality of indexes corresponding to an uplink positioning scenario, index 0 is one of the plurality of indexes.

[0025] In yet another example, the first piece of information represents an index piece of information used to determine thresholds and criteria for one or more quality metrics. For example, the first piece of information represents index 0, where index 0 indicates that the threshold for the SINR of the training data is Q, the threshold for the amount of training data is N, the criterion for SINR is that the SINR of the training data is greater than or equal to Q, and the criterion for the amount of training data is that the amount of training data is at least N. Optionally, index 0 is an index value in one of several application scenarios. For example, the application scenarios may include, but are not limited to, CSI prediction, uplink positioning, downlink positioning, or beam management, and index 0 is one of one or more indices corresponding to the beam management scenario. Optionally, index 0 is an index value in a specific application scenario. For example, in a group of indices corresponding to an uplink positioning scenario, index 0 is one of several indices.

[0026] In the last two examples, the correspondence between index information and quality metric thresholds and / or quality metric criteria is predefined in the protocol and is used only as an example. Alternatively, in other implementations, this correspondence may include, but is not limited to, pre-storing or pre-configuring.

[0027] The above describes an example in which the first piece of information is used to determine a constraint. This application is not limited to the above example.

[0028] Optionally, with respect to the first embodiment, in some implementations of the first embodiment, the quality metrics indicated by the first information include quality metrics for the AI ​​model's labels.

[0029] Optionally, the training data for the AI ​​model may further include labels. In the example, in an uplink or downlink positioning application scenario, the labels are location information. Optionally, the quality metrics in the constraints may further include quality metrics for the labels. For example, the quality metrics for the labels may include a distance threshold between locations of different samples. Optionally, the quality metrics indicated by the first information may further include quality metrics for the labels of the AI ​​model.

[0030] With respect to the first aspect, in some implementations of the first aspect, the second information includes the first training data, the second information indicates that the candidate training data collected by the first network element is valid, and the first training data is valid data in the candidate training data.

[0031] In this implementation, after the first network element determines the validity of the collected candidate training data based on the first information, if it determines that the current collection is valid, the first network element sends the second information to the second network element. The second information may be valid candidate training data (i.e., the first training data), and invalid candidate training data is not sent, thus reducing the waste of air interface resources.

[0032] In this application, valid candidate training data is provided to a second network element for training or updating an AI model; that is, valid candidate training data actually becomes training data. Invalid candidate training data is candidate training data that does not satisfy the constraints.

[0033] In addition, the first network element does not send invalid candidate training data to the second network element, so the second network element does not receive invalid or unsuitable training data, thereby avoiding contamination of the entire training dataset. Furthermore, adverse effects on AI model training by the second network element are also avoided, such as inaccurate AI performance gain assessment, AI model overfitting, weak generalization ability, and poor scenario adaptability caused by AI model training performed using invalid candidate training data.

[0034] With respect to the first embodiment, in some implementations of the first embodiment, the second information indicates that the candidate training data collected by the first network element is invalid.

[0035] In this implementation, if the first network element determines, based on the first information, that the current collection of candidate training data is invalid, then the first network element sends the second information to the second network element. The second information only indicates that the currently collected candidate training data is invalid and does not provide the collected candidate training data to the second network element, thus reducing the waste of air interface resources. In addition, the second network element does not receive invalid or unsuitable candidate training data, thus avoiding contamination of the entire training dataset. Furthermore, any adverse effects on AI model training by the second network element are also avoided, such as inaccurate AI performance gain assessment, AI model overfitting, weak generalization ability, and poor scenario adaptability caused by AI model training performed using invalid candidate training data.

[0036] With respect to the first aspect, in some implementations of the first aspect, the first information is for determining constraints for determining the validity of candidate training data collected by the first network element.

[0037] In this implementation, when a second network element indicates to the first network element, based on the first information, that it should collect training data for an AI model, the first information is further used by the first network element to determine the constraints that the training data to be collected must satisfy. Thus, the first network element provides the first network element with a basis for performing screening (i.e., validity determination) after collecting the candidate training data to determine whether the collected candidate training data is valid.

[0038] With respect to the first embodiment, some implementations of the first embodiment further include the following:

[0039] If the first network element determines that the candidate training data includes the first training data that satisfies the constraints, the first network element determines that the candidate training data is valid, or If the first network element determines that the candidate training data does not contain any first training data that satisfies the constraints, the first network element determines that the candidate training data is invalid.

[0040] In this implementation, the set of candidate training data that is present in the collected candidate training data and satisfies the constraints is called the first training data. If there is no candidate training data that satisfies the constraints, it indicates that the current collection is invalid.

[0041] With respect to the first embodiment, in some implementations of the first embodiment, when the candidate training data collected by the first network element is invalid, the method further includes the following:

[0042] The first network element receives third information from the second network element, and the third information indicates to the first network element that it is recollecting candidate training data for the AI ​​model.

[0043] In this implementation, a re-collection is performed after one collection becomes invalid, and validity testing is performed on both the previously collected invalid candidate training data and the re-collected candidate training data together, potentially improving the probability of obtaining candidate training data that meets the requirements (i.e., obtaining training data). In addition, the air interface transmission configuration of the reference signal may be updated during re-collection, increasing the likelihood of obtaining high-quality candidate training data and thus increasing the probability of collecting eligible training data.

[0044] With respect to the first embodiment, some implementations of the first embodiment further include the following:

[0045] The first network element determines the air interface transmission configuration information, the air interface transmission configuration information corresponds to the updated air interface transmission configuration, and the air interface transmission configuration information collects candidate training data for the AI ​​model based on the updated air interface transmission configuration.

[0046] Information regarding the updated air interface transmission configuration is available at: Transmission power of the reference signal, The number of antenna ports used for the reference signal, Reference signal bandwidth, The frequency domain density of the reference signal, or Reference signal duration This includes one or more of the following updates.

[0047] In this implementation, when training data for the AI ​​model needs to be recollected, the air interface transmission configuration of the reference signal involved in training data collection may be updated, thereby improving or guaranteeing the quality of the reference signal. In this way, valid candidate training data can be collected, thereby guaranteeing AI model training, for example, initial training / or training during the update process. In addition, updating the air interface transmission configuration of the reference signal helps in the collection of valid candidate training data, so the efficiency of AI model training can be further improved.

[0048] In addition, by learning the status of AI model training data collection, it is possible to learn whether there is a large amount of valid candidate training data for AI model training / updating. To ensure reliable AI model-based air interface performance, it is necessary to maintain the AI ​​model in a timely manner, switch to non-AI mode, or adaptively update the configuration for collecting training data.

[0049] With respect to the first embodiment, in some implementations of the first embodiment, the third piece of information further indicates the maximum number of times k to determine validity, where k is a positive integer.

[0050] In this implementation, the third piece of information indicates the maximum number of times k to determine validity; that is, the second network element indicates the maximum number of times k to determine validity to the first network element only when it determines that recollection is necessary. This avoids the signaling waste caused by restricting the recollection process when the collection results are unknown. For example, the first network element may obtain valid candidate training data through a single collection. In this case, recollection does not need to be performed. In this case, the second network element does not need to indicate the recollected relevant information to the first network element, reducing signaling overhead.

[0051] With respect to the first aspect, in some implementations of the first aspect, the first information further indicates the maximum number of times k to determine validity, where k is a positive integer.

[0052] In this implementation, the first piece of information indicates the maximum number of times k to determine effectiveness, i.e., when training data collection begins, the second network element indicates the maximum number of times to determine effectiveness, thereby allowing the first network element to quickly enter the re-collection process after one collection failure. This reduces the interaction time between the first and second network elements and improves the efficiency of training data collection.

[0053] In the two implementations described above, the second network element indicates to the first network element the maximum number of times k to determine effectiveness, thereby allowing the first network element to quickly collect candidate training data the next time if the initially collected candidate training data is invalid, and to repeatedly collect candidate training data if the maximum number of times k to determine effectiveness cannot be exceeded. This reduces the signaling overhead of recollection indication and improves the efficiency of AI model training / updates.

[0054] With respect to the first embodiment, some implementations of the first embodiment further include the following:

[0055] The first network element collects candidate training data for the AI ​​model based on the updated air interface transmission configuration. When the maximum number of validity evaluations k is reached, and the first network element determines, based on the first information, that the result of the kth validity evaluation is invalid, the first network element stops collecting candidate training data for the AI ​​model.

[0056] In this implementation, based on the constraint of the maximum number of times k that validity can be determined, it is possible to prevent the collection process of the first network element from entering an infinite loop, thereby avoiding resource occupation and resource waste.

[0057] With respect to the first embodiment, some implementations of the first embodiment further include the following:

[0058] If, before the maximum number of validity determinations k is exceeded, the first network element determines, based on the first information, that the result of the j-th validity determination is valid, the first network element sends fourth information to the second network element, which includes second training data, indicates that the result of the j-th validity determination is valid, the second training data includes valid data from the candidate training data on which the j-th validity determination was performed, j is less than or equal to k, and j is a positive integer.

[0059] With respect to the first embodiment, in some implementations of the first embodiment, the first network element collecting candidate training data for an AI model includes the following:

[0060] The first network element measures a reference signal from the second network element to obtain one or more measurement results, and the candidate training data for the AI ​​model includes one or more measurement results, or The first network element measures a reference signal from the third network element to obtain one or more measurement results, and the candidate training data for the AI ​​model includes one or more measurement results.

[0061] In this implementation, depending on the application scenario, the first network element collects candidate training data for an AI model, which may be measurements obtained by measuring a reference signal sent by a second or third network element. Optionally, there may be one or more measurements. For example, the first network element may obtain one measurement by measuring the reference signal once, or multiple measurements by measuring the reference signal multiple times, or multiple measurements by measuring the reference signal once. This is not limited to these. In these implementations, the candidate training data may include one or more measurements.

[0062] In the first embodiment, in some implementations of the first embodiment, the first network element is a terminal device, the second network element is an access network device, and the first network element measures a reference signal from the second network element to obtain one or more measurement results.

[0063] Optionally, the signals from the second network element include one or more of the following: channel state information-reference signal (CSI-RS), positioning reference signal (PRS), synchronization signal and physical broadcast channel block (SSB), and / or signals on the physical broadcast channel.

[0064] In this implementation, the application of AI models is applicable to application scenarios such as AI model-based CSI feedback or CSI prediction, and AI model-based beam management, solving problems such as CSI feedback or prediction and beam management, and improving the performance of the air interface in these application scenarios.

[0065] Regarding the first embodiment, in some implementations of the first embodiment, the first training data further includes reference signal information corresponding to K optimal measurement results from one or more measurement results, where K is an integer of 1 or greater. It should be understood that when there is one measurement result, K is equal to 1, and when there are V measurement results, K is less than or equal to V, and K is 1 or greater, where V is an integer of 2 or greater.

[0066] In this implementation, the AI ​​model is applicable to beam management scenarios. In this case, the first training data further includes reference signal information corresponding to K optimal measurement results, and this information is used as labels for the AI ​​model.

[0067] With respect to the first embodiment, in some implementations of the first embodiment, the first network element is an access network device, and the second network element is a positioning device. The first network element measures a sounding reference signal from the third network element and obtains one or more measurement results. The first training data further includes location information for the third network element.

[0068] In this implementation, the AI ​​model is applicable to uplink positioning scenarios. In this case, the first network element measures the sounding reference signal of the third network element to obtain candidate training data, which includes the location information of the third network element. When the candidate training data is valid, the first network element provides the positioning device with the valid candidate training data (i.e., the first training data) and the corresponding location information of the third network element to train or update the AI ​​model. The location information of the third network element is used as the label for the AI ​​model.

[0069] With respect to the first embodiment, in some implementations of the first embodiment, the first network element is a terminal device, and the second network element is a positioning device. The first network element measures a positioning reference signal from the third network element and obtains one or more measurement results, and the third network element is an access network device. The first training data further includes location information for the first network element.

[0070] In this implementation, the AI ​​model is applicable to downlink positioning scenarios. If the candidate training data collected by the first network element (e.g., a location reference device) is valid, the first training data provided by the first network element to the second network element (i.e., the positioning device) further includes the location information of the first network element, which is used as the label for the AI ​​model.

[0071] According to a second aspect, a method is provided for acquiring training data in AI model training, the method which can be applied to AI model training network elements, such as access network devices or positioning devices. The method includes the following:

[0072] The second network element sends first information to the first network element, and the first information is of the AI ​​model and is used to determine the effectiveness of the candidate training data collected by the first network element, and the effectiveness determination result includes valid or invalid. The second network element receives second information from the first network element, and this second information indicates the result of the validity determination.

[0073] For a description of the first information, please refer to the description of the first aspect. Further details are not provided herein.

[0074] With respect to the second aspect, in some implementations of the second aspect, the second information includes the first training data, the second information indicates that the candidate training data collected by the first network element is valid, and the first training data is valid data in the candidate training data.

[0075] With respect to the second aspect, in some implementations of the second aspect, the second information indicates that the candidate training data collected by the first network element is invalid.

[0076] With respect to the second aspect, in some implementations of the second aspect, the first information is for determining constraints for determining the validity of candidate training data collected by the first network element.

[0077] With respect to the second aspect, in some implementations of the second aspect, if the candidate training data includes first training data that satisfies the constraints, the candidate training data is valid, or If the candidate training data does not include the first training data that satisfies the constraints, the candidate training data is invalid.

[0078] With respect to the second aspect, in some implementations of the second aspect, when the second information indicates that the candidate training data collected by the first network element is invalid, the method further includes the following:

[0079] The second network element sends third information to the first network element, which in turn tells the first network element to recollect candidate training data for the AI ​​model.

[0080] With respect to the second aspect, some implementations of the second aspect further include the following:

[0081] The second network element determines the air interface transmission configuration information, which corresponds to the updated air interface transmission configuration, and the first network element indicates that the air interface transmission configuration information collects candidate training data for the AI ​​model based on the updated air interface transmission configuration.

[0082] Information regarding the updated air interface transmission configuration is available at: Transmission power of the reference signal, The number of antenna ports used for the reference signal, Reference signal bandwidth, The frequency domain density of the reference signal, or Reference signal duration This includes one or more of the following updates.

[0083] With respect to the second aspect, in some implementations of the second aspect, the third information further indicates the maximum number of times k to determine validity, where k is a positive integer.

[0084] With respect to the second aspect, in some implementations of the second aspect, the first information further indicates the maximum number of times k to determine validity, where k is a positive integer.

[0085] With respect to the second aspect, some implementations of the second aspect further include the following:

[0086] The second network element receives fourth information from the first network element, the fourth information includes the second training data, the fourth information indicates that the j-th validity judgment performed by the first network element is valid, the second training data is valid data from the candidate training data on which the j-th validity judgment was performed, j is less than or equal to k, and j is a positive integer.

[0087] With respect to the second aspect, in some implementations of the second aspect, the second network element is an access network device and the first network element is a terminal device, and the method further includes the following:

[0088] A second network element sends a reference signal to a first network element, which is used by the first network element to obtain one or more measurement results corresponding to the reference signal, and the candidate training data for the AI ​​model includes one or more measurement results.

[0089] With respect to the second aspect, in some implementations of the second aspect, the first training data further includes reference signals corresponding to K optimal measurement results from one or more measurement results, where K is an integer of 1 or more.

[0090] In a second embodiment, in some implementations of the second embodiment, the second network element is a positioning device, and the first network element is an access network device. Candidate training data for an AI model includes one or more measurement results and location information of a third network element, the one or more measurement results being acquired by the first network element by measuring a sounding reference signal sent by the third network element.

[0091] In the second embodiment, in some implementations of the second embodiment, the second network element is a positioning device, and the first network element is a terminal device. Candidate training data for the AI ​​model includes one or more measurement results and location information of the first network element, the one or more measurement results being based on measurements of a positioning reference signal sent by a third network element, the third network element being an access network device. Optionally, the measurements are performed by the first network element.

[0092] Regarding the second aspect, in some implementations of the second aspect, the constraints are: Thresholds for quality indicators and criteria for determining quality indicators, Thresholds for the amount of training data that meets the quality indicator criteria, and criteria for determining the amount of training data, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. Includes one or more of the following.

[0093] With respect to the second aspect, in some implementations of the second aspect, the first information is, Thresholds for quality metrics, quality indicator criteria, Threshold for the amount of training data that satisfies the quality indicator criteria, Criteria for determining the amount of training data that meets the quality indicator criteria, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. Show one or more of them.

[0094] Optionally, the quality metrics in the above implementation include one or more quality metrics, for example, one or more of the quality metrics of the AI ​​model's label or the quality metrics of the measurement results of a reference signal.

[0095] In some implementations of the first or second embodiment, the constraints are based on the application scenario of the AI ​​model, and the application scenario of the AI ​​model is AI model-based CSI feedback or prediction, AI model-based positioning, or AI model-based beam management Includes one or more of the following.

[0096] According to a third aspect, the present application provides a communication device. The communication device may be a terminal device, a device, module, chip, etc., disposed within a terminal device, or a device that can be used together with a terminal device. In the design, the communication device may include modules configured to perform the methods / operations / steps / actions described in the first aspect, in one-to-one correspondence with them. The modules may be hardware circuits, software, or implemented by hardware circuits combined with software. In the design, the communication device may include a processing module and a communication module.

[0097] According to a fourth aspect, the present application provides a communication device. In the design, the communication device may include modules configured to perform methods / operations / steps / actions described in the second aspect, in one-to-one correspondence with them. The modules may be hardware circuits, software, or implemented by hardware circuits combined with software. In the design, the communication device may include a processing module and a communication module. In an example, the communication device is an access network device or a positioning device, and the positioning device may be, for example, an LMF network element.

[0098] According to a fifth aspect, the present application provides a communication device. The communication device includes a processor configured to implement a method according to either the first aspect or an implementation of the first aspect. The processor is coupled to memory. The memory is configured to store instructions and data. When the processor executes an instruction stored in memory, the method described in either the first aspect or an implementation of the first aspect can be implemented. Optionally, the communication device may further include memory. Optionally, the communication device may further include a communication interface. The communication interface is used by the device to communicate with another device. For example, the communication interface may be a transceiver, hardware circuitry, a bus, a module, a pin, or other type of communication interface. In the example, the communication device may be a terminal device, a device, module, chip, etc., disposed within a terminal device, or a device that can be used with a terminal device.

[0099] According to a sixth aspect, the present application provides a communication device. The communication device includes a processor configured to implement a method according to either the second aspect or an implementation of the second aspect. The processor is coupled to memory. The memory is configured to store instructions and data. When the processor executes an instruction stored in memory, the method described in either the second aspect or an implementation of the second aspect can be implemented. Optionally, the communication device may further include memory. Optionally, the communication device may further include a communication interface. The communication interface is used by the device to communicate with another device. For example, the communication interface may be a transceiver, hardware circuitry, a bus, a module, a pin, or other type of communication interface. In an example, the communication device may be an access network device, a device, module, chip, etc., located within an access network device, or a device that can be used with an access network device. In another example, the communication device may be a positioning device, a device, module, chip, etc., located within a positioning device, or a device that can be used with a positioning device.

[0100] According to the seventh aspect, the present application provides a communication system including a first network element and a second network element. For example, the interaction between the first network element and the second network element is as follows:

[0101] The second network element sends first information to the first network element, and the first information is used to determine the effectiveness of the candidate training data collected by the first network element, and the effectiveness determination result includes valid or invalid. The first network element receives the first information from the second network element. The first network element collects candidate training data for the AI ​​model. The first network element sends second information to the second network element based on the candidate training data and the first information, and the second information indicates the effectiveness determination result. The second network element receives the second piece of information from the first network element.

[0102] In particular, the solution on the first network element side should be understood by referring to the implementation in the first embodiment, and the solution on the second network element side should be understood by referring to the implementation in the second embodiment. Details are not described again herein. For example, the communication system includes terminal devices and access network devices. Optionally, the communication system includes terminal devices, access network devices, and positioning devices. Optionally, the terminal device is a location reference device, and the positioning device is an LMF network element.

[0103] According to the eighth aspect, the present application provides a communication system including a communication device described in the third or fifth aspect and a communication device described in the fourth or sixth aspect.

[0104] According to the ninth aspect, the present application further provides a computer program. When the computer program is executed on a computer, the computer is enabled to implement a method according to the first aspect, the second aspect, or any one of the implementations of the first or second aspect.

[0105] According to the tenth aspect, the application further provides a computer program product including instructions. When the instructions are executed on a computer, the computer is enabled to implement a method according to the first aspect, the second aspect, or any one of the implementations of the first or second aspect.

[0106] According to the eleventh aspect, the present application further provides a computer-readable storage medium for storing computer programs or instructions. When the computer program or instructions are executed on a computer, the computer is enabled to implement a method according to the first aspect, the second aspect, or any one of the implementations of the first or second aspect.

[0107] According to the twelfth aspect, the present application further provides a chip configured to read a computer program stored in memory and perform a method according to any one of the first aspect, the second aspect, or an implementation of the first or second aspect, or the chip includes a circuit configured to perform a method according to any one of the first aspect, the second aspect, or an implementation of the first or second aspect.

[0108] According to a thirteenth aspect, the present application further provides a chip system. The chip system includes a processor configured to support the device when implementing a method according to the first aspect, the second aspect, or any one of the implementations of the first or second aspect. In possible designs, the chip system further includes memory, which is configured to store programs and data required by the device. The chip system may include a chip, or it may include a chip and other separate components.

[0109] For the technical effects of the solutions provided in any one of the second through thirteenth embodiments or implementations of the second through thirteenth embodiments, please refer to the corresponding description in the first embodiment. Further details will not be provided again. [Brief explanation of the drawing]

[0110] [Figure 1] This is a diagram illustrating the neural network iteration process. [Figure 2] This is a diagram showing the architecture of a communication system to which the embodiments of this application can be applied. [Figure 3] This is a schematic flowchart of the method for obtaining training data in AI model training according to this application. [Figure 4] This is a diagram of an AI model-based CSI feedback mechanism. [Figure 5] This figure shows an example of acquiring training data in an AI model-based CSI feedback system according to this application. [Figure 6] This is a diagram illustrating the technical solution in the AI ​​model-based uplink positioning scenario described in this application. [Figure 7] This figure shows an example of acquiring training data for AI model-based uplink positioning according to this application. [Figure 8] This diagram illustrates the technical solution in the AI ​​model-based downlink positioning scenario described in this application. [Figure 9] This figure shows an example of acquiring training data for AI model-based downlink positioning according to this application. [Figure 10] This is a diagram of AI-assisted sparse beam scanning processing. [Figure 11] This figure shows an example of acquiring training data in AI model-based beam management according to this application. [Figure 12] This is a diagram showing the configuration of the communication device according to this application. [Figure 13] This is a diagram showing the configuration of the communication device according to this application. [Modes for carrying out the invention]

[0111] The technical solution of this application will be described below with reference to the attached drawings.

[0112] First, the relevant concepts and technologies in the embodiments of this application will be briefly described.

[0113] An AI model is a functional model that maps inputs of a specific dimension to outputs of a specific dimension. The model parameters of an AI model are obtained through machine learning training. For example, f(x) = ax 2 +b is a quadratic function model, which may be considered an AI model, where a and b are parameters of the AI ​​model, which can be obtained through machine learning training. For example, the AI ​​models described in the following embodiments of this application are not limited to neural networks, linear regression models, decision tree models, support vector machines (SVMs), Bayesian networks, Q-learning models, or other machine learning (ML) models.

[0114] A training dataset is the data used for training, validating, and testing a model in machine learning. The quantity and quality of the data affect the effectiveness of machine learning. Training data may include the inputs of an AI model, or it may include both the inputs and target outputs of an AI model. The target output is the target value that the AI ​​model outputs, and is also sometimes called a truth value, output truth value, label, or label sample.

[0115] Model training is the process of training model parameters by selecting an appropriate loss function and using an optimization algorithm so that the value of the loss function becomes smaller than a threshold or satisfies the target requirements.

[0116] AI model design primarily consists of a data collection phase (e.g., collecting training data and / or inference data), a model training phase, and a model inference phase. An inference result application phase may also be included. In the data collection phase, a data source is used to provide training datasets and inference data. In the model training phase, the training data provided by the data source is analyzed or trained to obtain an AI model. The AI ​​model represents the mapping relationship between the model's inputs and outputs. Obtaining an AI model through learning using the model training node is equivalent to obtaining the mapping relationship between the model's inputs and outputs through learning using the training data. In the model inference phase, the AI ​​model obtained through training in the model training phase is used to perform inference based on the inference data provided by the data source to obtain the inference result. This phase may also be understood as follows: The inference data is input into the AI ​​model, and the output is obtained through the AI ​​model. This output is the inference result. The inference result may indicate configuration parameters used (acted upon) by the execution object, and / or actions performed by the execution object. The inference result is released in the inference result application phase. For example, inference results may be planned in a unified manner by an execution (actor) entity. For instance, an execution entity may send inference results to one or more execution objects (e.g., a core network device, an access network device, or a terminal device) for execution. In another example, an execution entity may further feed back the model's performance to a data source to facilitate subsequent model update training.

[0117] The loss function is used to measure the difference between the model's predicted values ​​and its truth values.

[0118] Model application involves using a pre-trained model to solve real-world problems.

[0119] Machine learning (ML) is a crucial technological approach for realizing artificial intelligence (AI). Machine learning can be divided into supervised learning, unsupervised learning, and reinforcement learning.

[0120] For example, in supervised learning, a mapping relationship between sample values ​​and sample labels is learned by using a machine learning algorithm based on collected sample values ​​and sample labels, and the learned mapping relationship is represented by using a machine learning model. The process of training a machine learning model is the process of learning the mapping relationship. For example, in signal detection, a received signal containing noise is the sample, and the actual constellation points corresponding to the signal are the labels. In machine learning, it is expected that the mapping relationship between samples and labels will be learned through training, that is, that the machine learning model will be able to learn the signal detector. During training, model parameters are optimized by calculating the error between the model's predicted values ​​and the actual labels. Once the mapping relationship is learned, it is possible to predict the sample label for each new sample by using the learned mapping relationship. The mapping relationship learned through supervised learning may include linear and nonlinear mappings. Learning tasks may be classified into classification tasks and regression tasks based on the type of label.

[0121] Please refer to Figure 1. Figure 1 is a diagram of the neural network iteration process. As shown in Figure 1, n samples are selected to form a batch, and then the batch is fed into the neural network to obtain the output result. Next, the output result and sample labels are fed into the loss function to calculate the loss in the current round. Finally, the respective derivative of each parameter is used together with the step parameter to perform parameter updates. This is the iteration in the training process. A batch means a "group," indicating that the neural network processes data in batches. The batch size indicates the amount of samples in each batch. Therefore, a sample quantity with an appropriate size can usually be used for parallel computing to accelerate the training speed. The amount of data processed at one time should not be excessively large.

[0122] A training dataset is a set of training samples. Each training sample is an input to a neural network. The training dataset is used for model training. The training dataset is one of the most important parts of machine learning. The training process in machine learning essentially involves learning several features of the training dataset so that the difference between the neural network's output and the ideal target value (i.e., the label or output truth value) is minimized. Typically, even when the same network structure is used, the weights and outputs of trained neural networks will differ by using different training datasets. Therefore, the composition and selection of the training dataset determine, to some extent, the performance of the trained neural network.

[0123] When AI models are applied to air interface technology, data from the actual deployed network must be collected to form the necessary training dataset for model updates / training, regardless of whether the model updates / training are offline or online. A good training dataset helps in designing wireless communication AI algorithms to achieve greater performance gains and improve the generalization ability and robustness of the final design algorithm across multiple scenarios.

[0124] When an AI model is applied to several application scenarios of air interface technology, if the AI ​​model training network elements and the training data acquisition network elements are not in the same network, the training data needs to be exchanged between the AI ​​model training network elements and the training data acquisition network elements. Based on the current technological situation, since the training data is usually exchanged periodically or continuously, this is prone to wasting air interface resources.

[0125] Furthermore, the process by which training data collection network elements collect training data is not limited by the requirements of the AI ​​model training network elements, and invalid collection often occurs. For example, the training data collected by the training data collection network elements may not be the training data that the AI ​​model training network elements actually need, leading to invalid interactions and wasting air interface resources. In addition, if the AI ​​model training network elements use the training data to train the AI ​​model, the AI ​​model's training dataset becomes contaminated, resulting in several problems, including inaccurate gain estimation, model overfitting, weak generalization ability, and poor scenario adaptability.

[0126] Regarding the above-mentioned problems, this application provides a method for acquiring training data in AI model training, thereby helping to solve or improve the above-mentioned problems.

[0127] The technical solutions provided in this application can be applied to a variety of communication systems. For example, the communication systems may be fourth-generation (4G) communication systems (e.g., long-term evolution (LTE) systems), fifth-generation (5G) communication systems, worldwide interoperability for microwave access (WiMAX) or wireless local area network (WLAN) systems, satellite communication systems, or future communication systems, such as 6G communication systems, or converged systems of multiple systems. 5G communication systems are also sometimes called new radio (NR) systems.

[0128] Network elements in a communication system may send signals to or receive signals from other network elements. These signals may include information, signaling, data, etc. Network elements may be replaced by entities, network entities, devices, communication devices, communication modules, nodes, communication nodes, etc. In this application, network elements are used as illustrative examples.

[0129] The communication system to which this application is applicable includes a first network element and a second network element, and may optionally further include a third network element. The quantities of the first network element, the second network element, and the third network element are not limited.

[0130] Please refer to Figure 2. Figure 2 is a diagram of the architecture of a communication system to which embodiments of the present application are applicable. Figure 2(a) is a diagram of the architecture of a communication system to which embodiments of the present application are applicable. For example, the communication system includes a network device 110, a terminal device 120, and a terminal device 130. Terminal devices 120 and 130 may access and communicate with the network device 110. Optionally, the network device 110 may be an access network device. In an implementation, the communication system may further include an AI entity. The network device may transfer data reported by the terminal devices and relating to an AI model to the AI ​​entity, the AI ​​entity performs AI-related operations such as training dataset construction and model training, and provides the network device with outputs of AI-related operations such as a trained AI model, model evaluation, and test results. In another implementation, the AI ​​entity may also be located inside the network device 110, i.e., in a module of the network device 110. Figure 2(b) is another diagram of the architecture of a communication system to which embodiments of the present application are applicable. The communication system includes a network device 110, a terminal device 120, a terminal device 130, and a positioning device 140. The positioning device 140 and the network device 110 may communicate with each other using interface messages. For example, the positioning device 140 may be a location management function (LMF), and the network device 110 may be an access network device, such as a gNB or eNB. This is not limited to these. For example, if the access network device 110 is a gNB, the gNB and the LMF may exchange information using NR positioning protocol A (NRPPa) messages.If the access network device 110 is an eNB, the eNB and LMF may exchange information using LTE positioning protocol (LPP) messages. Optionally, a terminal device may communicate directly with the positioning device 140, for example, the interaction between terminal device 130 and positioning device 140 shown in Figure 2(b). In Figure 2, the AI ​​entity may be configured within the positioning device 140 or disposed separately from the positioning device 140. This is not limited. Optionally, the positioning device and network device may be different modules of the same device or different, separate devices.

[0131] In actual applications, one network device may service one or more terminal devices. A terminal device may also access one or more network devices. The number of terminal devices and network devices included in a wireless communication system is not limited to the embodiments of this application. In addition, the positioning device 140 in Figure 2(b) is not limited to LMF network elements, but may instead be another network element having positioning capabilities, and the number of positioning devices is also not limited.

[0132] For example, the network device may be a device that has wireless transceiver functionality. Network devices are devices that provide wireless communication functionality and are typically located on the network side, but are not limited to, next-generation node B (gNodeB, gNB) in fifth-generation (5G) communication systems, base stations in sixth-generation (6G) mobile communication systems, base stations in future mobile communication systems, or access points (AP) in wireless fidelity (Wi-Fi) systems, evolved node B (eNB) in long-term evolution (LTE) systems, radio network controllers (RNC), node B (NodeB, NB), base station controllers (BSC), home base stations (e.g., home evolved node B, or home node B, HNB), baseband units (BBU), transmission reception points (TRP), transmitting points (TP), base transceiver stations, This includes BTS, satellites, and unmanned aerial vehicles. In the network structure, network devices may include central unit (CU) nodes, distributed unit (DU) nodes, RAN devices including CU nodes and DU nodes, or RAN devices including CU control plane nodes, CU user plane nodes, and DU nodes. Alternatively, network devices may include radio controllers, relay stations, in-vehicle devices, and wearable devices in a cloud radio access network (CRAN) scenario.In addition, a base station may be a macro base station, a micro base station, a relay node, a donor node, or a combination thereof. Alternatively, a base station may be a communication module, modem, or chip deployed in the above-mentioned devices or equipment. Alternatively, a base station may be a mobile switching center, a device performing base station functions in device-to-device (D2D), vehicle-to-everything (V2X), or machine-to-machine (M2M) communications, a network-side device in a 6G network, or a device performing base station functions in a future communication system. A base station may support networks with the same or different access technologies, but is not limited to these.

[0133] Network devices may be fixed or mobile. For example, access network device 110 is static and is responsible for wireless transmission and reception from terminal devices 120 and 130 in one or more cells. Access network device 110 may also be mobile. For example, a helicopter or unmanned aerial vehicle may be configured as a mobile base station, and one or more cells may move based on the location of the mobile base station. In another example, a helicopter or unmanned aerial vehicle may be configured as a device that communicates with base station 110.

[0134] In this application, a communication device configured to implement the functions of an access network may be an access network device, a network device having some of the 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, hardware circuitry, software modules, or a combination of hardware circuitry and software modules. This device may be installed in or used together with an access network device. In the methods of this application, an example in which the communication device configured to implement the functions of an access network device is an access network device is used for illustrative purposes.

[0135] A terminal device may be an entity configured to receive or transmit signals on the user side, such as a mobile phone. Terminal devices include handheld devices with wireless connectivity, other processing devices connected to a wireless modem, and in-vehicle devices. Terminal devices may be portable, pocket-sized, handheld, computer-integrated, or in-vehicle mobile devices. Terminal devices 120 can be widely used in various scenarios, such as cellular communications, Wi-Fi systems, D2D, V2X, peer-to-peer (P2P), M2M, machine-type communication (MTC), Internet of Things (IoT), virtual reality (VR), augmented reality (AR), industrial control, autonomous driving, telemedicine, smart grids, smart furniture, smart offices, smart wearables, smart cities, unmanned aerial vehicles, robots, remote sensing, passive sensing, positioning, navigation and tracking, automated delivery and mobility.Some examples of communication devices 120 include 3GPP standard user equipment (UE), Wi-Fi system stations (STA), fixed devices, mobile devices, handheld devices, wearable devices, cellular 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, airplanes, ships, remote control devices, smart home devices, industrial devices, personal communication services (PCS), telephones, wireless local loop (WLL) stations, personal digital assistants (PDAs), wireless network cameras, tablet computers, palmtop computers, mobile internet devices (MIDs), wearable devices such as smartwatches, virtual reality (VR) devices, augmented reality (AR) devices, and industrial control devices. These include wireless terminals in control systems, terminals in vehicle internet systems, wireless terminals in unmanned (autonomous) driving, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities such as smart fuel supply systems, terminal devices for high-speed rail, and wireless terminals for smart homes such as smart speakers, smart coffee machines, and smart printers.The terminal device 120 may be a wireless device or a device installed within a wireless device in the various scenarios described above, for example, a communication module, modem, or chip within the device. The terminal device may also be called a terminal, user equipment (UE), mobile station (MS), mobile terminal (MT), etc. Alternatively, the terminal device may be a terminal device in a future wireless communication system. In addition, the terminal device may further include a location reference device, for example, an automated guided vehicle (AGV) or a device having a similar function. The specific technologies and specific device forms used by the terminal device are not limited to the embodiments of this application.

[0136] In this application, a communication device configured to implement terminal device functions may be a terminal device, a terminal device having some of the functions of the above communication device, or a device capable of supporting the implementation of the above terminal device functions, such as a chip system. This device may be installed in or used together with the terminal device. In this application, a chip system may include a chip, or include a chip and other individual components.

[0137] The number of devices and the types of devices shown in Figure 2 are for illustrative purposes only and should be understood as not being limited to this application. In actual applications, the communication system may further include more terminal devices, more access network devices, and more positioning devices, and may further include other network elements, such as a core network device and / or network element configured to implement artificial intelligence functions.

[0138] The technical solutions provided in this application are described below.

[0139] Please refer to Figure 3. Figure 3 is a schematic flowchart of a method for acquiring training data in AI model training according to this application. In the method of Figure 3, the first network element may be a network element that collects training data for the AI ​​model, and the second network element is an AI model training network element. Optionally, the second network element may be an AI model training network element and also a network element on which AI inference is performed. The first and second network elements may be logically deployed separately. In different implementations, the first and second network elements may be physically deployed in the same network element or in different network elements. This is not limited to these.

[0140] 310: The first network element receives first information from the second network element, and the first information determines the validity of the candidate training data collected by the first network element, and the validity determination result includes valid or invalid.

[0141] The first network element can determine the validity of the collected candidate training data based on the first information. In other words, the first network element can determine whether the collected candidate training data is valid based on the first information.

[0142] In particular, if the first network element determines, based on the first information, that the collected candidate training data includes valid training data, the first network element determines that the collected candidate training data is valid. If the first network element determines, based on the first information, that the collected candidate training data does not include valid candidate training data, the first network element determines that the collected candidate training data is invalid. In other words, if some of the candidate training data in the candidate training data collected by the first network element is valid, the determination result is valid. If the determination result is valid, the valid candidate training data becomes the training data collected by the first network element. For example, valid candidate training data (which may be abbreviated as "valid data" below) may be some or all of the collected candidate training data, and are collectively referred to as the first training data in this specification. If the candidate training data collected by the first network element does not include valid candidate training data, the determination result is invalid.

[0143] In this embodiment of the present application, it can be seen that after collecting candidate training data, the first network element determines its effectiveness once on its own.

[0144] In the example, the first piece of information represents a constraint, which is used by the first network element to determine the effectiveness of the collected candidate training data.

[0145] For example, constraints are, Thresholds for quality indicators and criteria for determining quality indicators, Thresholds for the amount of training data that meets the quality indicator criteria, and criteria for determining the amount of training data, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. Includes one or more of the following.

[0146] Optionally, the first piece of information is used to determine the constraints.

[0147] For example, the first piece of information is: Thresholds for quality metrics, quality indicator criteria, Threshold for the amount of training data that satisfies the quality indicator criteria, Criteria for determining the amount of training data that meets the quality indicator criteria, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. This indicates one or more of the following pieces of information.

[0148] It is assumed that the constraint includes one or more quality metrics. When the first piece of information is used to determine the constraint, multiple implementations may exist.

[0149] Optionally, in this example, the first piece of information indicates thresholds for one or more quality metrics, and the criteria for one or more quality metrics are predefined in the protocol. In this example, the first network element determines the constraints based on the first piece of information and the protocol predefined criteria.

[0150] Optionally, in another example, the first piece of information represents thresholds for one or more quality metrics and criteria for one or more quality metrics. In this example, the first network element determines the constraints based on the first piece of information.

[0151] Optionally, in yet another example, the constraint includes multiple quality metrics, and the first piece of information indicates thresholds for some of the multiple quality metrics, while the thresholds for the other quality metrics and the criteria for determining the multiple quality metrics are predefined in the protocol. In this example, the first network element determines the constraint based on the first piece of information and the protocol predefined.

[0152] Optionally, in yet another example, the first piece of information might represent thresholds for some quality metrics and one index piece of information, which is used to determine the criteria for some quality metrics, the thresholds for other quality metrics in the constraints, and the criteria for other quality metrics. In this example, the first network element determines the constraints based on the first piece of information and the index piece of information.

[0153] Optionally, in yet another example, the first piece of information represents a single index piece of information used to determine thresholds for one or more quality indicators and criteria for one or more quality indicators. In this example, the first network element determines constraints based on the index piece of information.

[0154] In addition, optionally, the "protocol pre-definition" in the above example may also be another implementation, such as pre-configuration or pre-storage. This is not limited to this.

[0155] In addition, for specific implementations where the first information is used to determine constraints, please refer to the corresponding implementation in the content section of the Invention Summary. Further details are not described herein.

[0156] In the example, the first piece of information further indicates the maximum number of times k is used to determine effectiveness, where k is a positive integer.

[0157] When the first piece of information includes multiple pieces of the above information, it may be understood that these pieces of information may be carried in a single message or carried separately in multiple messages. In other words, the first piece of information may be carried in a single message or carried in multiple messages.

[0158] The maximum duration for collecting candidate training data corresponding to a single effectiveness assessment is denoted as Z below, where Z is a number greater than 0. The maximum duration for collecting candidate training data is also the maximum duration for which the candidate training data can be used for effectiveness assessment. When the retention duration of candidate training data exceeds the maximum duration Z, the candidate training data becomes invalid and is no longer used for effectiveness assessment. Optionally, the maximum duration Z may be the same as the interval between two adjacent effectiveness assessments, or it may be greater than or less than the interval between two adjacent effectiveness assessments. If the maximum duration is greater than the interval between two adjacent effectiveness assessments, the candidate training data corresponding to a single effectiveness assessment may include all or part of the candidate training data corresponding to one or more effectiveness assessments prior to the current effectiveness assessment. The interval between two adjacent effectiveness assessments may be fixed, specifically, effectiveness assessments may be performed periodically within a specific time frame, or they may be variable, specifically, the time of effectiveness assessments may not be fixed. For example, effectiveness assessment counting is performed on candidate training data that meet threshold requirements. When the amount of candidate training data that meets the threshold requirement is met, the effectiveness assessment is complete and the result becomes valid. When the amount of candidate training data that meets the threshold requirement does not meet the requirement and exceeds the maximum effectiveness assessment interval T (i.e., the preset interval threshold), or when the amount of collected candidate training data exceeds the preset threshold (i.e., the maximum amount of candidate training data collected), the effectiveness assessment is also complete and the result becomes invalid. One or more of the specific temporal relationship information of the effectiveness assessment, such as the assessment time, the start time or duration of periodic assessments, the maximum interval T, and the maximum amount of candidate training data collected, may be predefined in whole or in part in the protocol, or may be based on configuration.

[0159] From the above description, it is known that in this application, a second network element sends first information to a first network element, and the first information is for determining the validity of the candidate training data collected by the first network element. In practice, the second network element indicates the requirements for training data to the second network element by using the first information. In other words, only candidate training data that meets the requirements can be used as training data for training or updating an AI model. Only after "filtering" has been performed on the candidate training data collected by the first network element can candidate training data that meets the requirements be used as training data, and it can be seen that the first network element provides the training data to the second network element for use. Therefore, after collecting candidate training data, the first network element determines the validity of the collected candidate training data based on the first information. If the determination result is invalid, it indicates that the candidate training data collected this time is not required by the second network element, that is, the candidate training data collected this time does not contain candidate training data that meets the requirements. In this case, the first network element may recollect the training data. Therefore, in the process where the first network element collects training data for the AI ​​model, the training data may not be obtained through a single collection. In a particular implementation, the maximum interval T for a single validity check is equivalent to specifying how often the first network element performs a validity check.

[0160] For example, if the result of the i-th validity check is invalid, the first network element recollects the candidate training data. After a time period, the first network element performs a validity check on the candidate training data collected for the (i+1)th time, where i is a positive integer. Thus, in this embodiment of the present application, the number of times the first network element collects candidate training data for the AI ​​model corresponds to or is equal to the number of times the first network element performs a validity check. In other words, each time the first network element performs a validity check, it indicates that the candidate training data was collected once prior to the current check. To clarify the description of the technical solution, the single candidate training data collection just before the i-th check is referred to herein as the i-th collection. As there are different implementations, in this example, the validity check performed for the (i+1)th time may be on the candidate training data obtained through the (i+1)th collection, or on the candidate training data obtained through the (i+1)th collection and one or more collections prior to the (i+1)th collection. This is not limited to this. In this implementation, after a single validation check, the issue of how the first network element processes the candidate training data collected before the current validation check may arise.

[0161] In this example, after starting the collection of candidate training data for the i-th time, the first network element performs an effectiveness check on the candidate training data collected for the i-th time after an interval T0 (less than or equal to the maximum interval T), i.e., performs the i-th effectiveness check. The result of the i-th effectiveness check is assumed to be invalid. If the maximum number of effectiveness checks k cannot be exceeded, the first network element may discard the candidate training data collected for the i-th time and perform the (i+1)th collection. In this example, each effectiveness check is performed only on candidate training data collected within interval T0. If the current collection is invalid, the candidate training data collected this time is discarded. In another example, a single effectiveness check may be performed on candidate training data collected within multiple intervals T0. Alternatively, candidate training data for which a single effectiveness check has been performed may be collected multiple times. The result of the i-th effectiveness check is assumed to be invalid. If the maximum number of effectiveness checks k cannot be exceeded, the first network element may retain some of the candidate training data collected for the i-th time. For example, the first network element holds a portion of the candidate training data collected in the i-th iteration that satisfies certain quality indicator criteria in the constraints. Then, the (i+1)th collection is performed. After interval T0, the first network element performs a validity check on the candidate training data collected in the (i+1)th iteration and the portion of the candidate training data collected in the i-th iteration whose retention duration does not exceed the maximum duration Z, i.e., performs the (i+1)th validity check. Optionally, if the result of the validity check is always invalid before the maximum number of validity checks k is exceeded, the first network element may hold the candidate training data that satisfies certain quality indicator criteria and is included in the candidate training data collected each time, and after completing a new collection, perform a validity check together on the retained historical candidate training data that satisfies certain quality indicator criteria and the newly collected candidate training data.In a single effectiveness assessment, candidate training data deemed invalid is for the purpose of that assessment only, and this does not mean that candidate training data deemed invalid in that assessment can never be used as training data. These specific implementations are not limited to this application.

[0162] In the example, the constraints are based on the application scenario of the AI ​​model. For example, the application scenario of the AI ​​model is not limited, AI model-based CSI feedback or prediction, AI model-based positioning, or AI model-based beam management This includes the following scenario.

[0163] In different application scenarios, one or more of the quality metrics, quality metric thresholds, quality metric criteria, training data quantity thresholds for satisfying the quality metric criteria, and training data quantity criteria in constraints may differ. Examples of different application scenarios are provided separately below.

[0164] 320: The first network element collects candidate training data for the AI ​​model.

[0165] The first network element collects candidate training data for the AI ​​model. Optionally, the first network element may begin collecting candidate training data for the AI ​​model after receiving the first information, i.e., based on the trigger of the first information. Optionally, the first network element may begin collecting candidate training data for the AI ​​model before or at the time of receiving the first information. In other words, the order of steps 310 and 320 is not limited.

[0166] Optionally, the training data requirements for the second network element may change. When the second network element's requirements for training data change, the second network element may send updated first information to the first network element. Updating the first information primarily refers to updating the constraints determined based on the first information. Upon receiving the updated first information, the first network element determines the validity of the collected candidate training data based on the constraints determined by the updated first information. In the following embodiments, only the first information received by the first network element at a specific time is used as an example to illustrate the validity determination and subsequent procedures.

[0167] 330: The first network element sends second information to the second network element, and the second information indicates the result of the effectiveness assessment of the candidate training data collected by the first network element.

[0168] If the first network element determines, based on the first information, that the collected candidate training data is valid, the first network element sends the second information to the second network element, and the second information indicates that the candidate training data collected by the first network element is valid.

[0169] Optionally, in this example, the first network element sends the first training data to the second network element, and the first training data implicitly indicates that the candidate training data collected by the first network element is valid. Optionally, in another example, the first network element sends the first training data to the second network element. In this case, the first network element further sends information to the second network element indicating that the candidate training data collected by the first network element is valid, for example, information a. In this example, the first network element sends the first training data and information a to the second network element, and this information indicates that the current collection is valid.

[0170] If the first network element determines, based on the first information, that the collected candidate training data is invalid, the first network element sends the second information to the second network element, which indicates that the candidate training data collected by the first network element is invalid. If the candidate training data collected by the first network element is invalid, the first network element sends only an indication to the second network element that the collected candidate training data is invalid, and does not send the collected invalid candidate training data, thereby reducing the waste of air interface resources.

[0171] In the example, if the first network element determines that the collected candidate training data is invalid, the first network element discards the candidate training data collected this time. Alternatively, in some implementations described above, invalid candidate training data in a single validity check may be retained for subsequent validity checks. Furthermore, if the second network element indicates to the first network element that it wants to recollect the training data for the AI ​​model, the first network element recollects the candidate training data for the AI ​​model.

[0172] Optionally, in step 320, the first network element may collect candidate training data for an AI model, in particular by measuring a reference signal from a second or third network element to obtain candidate training data for an AI model. In other words, the candidate training data includes the measurement results obtained by the first network element by measuring the reference signal. In this application, the reference signal is typically a signal for channel measurement. Channel measurement may be used for one or more functions such as channel state information feedback, beam management, or positioning. The reference signal may include one or more of the following: channel state information reference signal, synchronization signals such as primary and / or secondary synchronization signals, physical broadcast signals, synchronization signals and physical broadcast channel blocks (SSBs), demodulation reference signals, phase tracking reference signals, or positioning reference signals. The reference signal may differ when the AI ​​model is applied to different scenarios. Different application scenarios are described separately by using examples in the following embodiments. In addition, the third network element is a different network element from the second network element.

[0173] In this example, the first network element may be understood to measure a reference signal from the second network element and obtain a measurement result. Optionally, there may be one or more measurement results. The candidate training data for the AI ​​model, collected by the first network element, includes one or more measurement results. In this example, before the second network element sends a reference signal to the first network element, the second network element sends air interface transmit configuration information to the first network element. The air interface transmit configuration information corresponds to an air interface transmit configuration, and the air interface transmit configuration information indicates to the first network element that it will collect candidate training data for the AI ​​model based on the air interface transmit configuration. In other words, the second network element sends a reference signal based on the air interface transmit configuration, and the first network element measures the reference signal from the second network element, obtains a measurement result, and collects candidate training data based on the air interface transmit configuration. Optionally, the first network element is a UE, and the second network element is an access network device, such as a base station.

[0174] In another example, the first network element measures a signal from the third network element and obtains a measurement result. Optionally, there may be one or more measurement results. Candidate training data for the AI ​​model, collected by the first network element, includes one or more measurement results. Optionally, the first network element is an access network device, e.g., a base station, and the third network element is a UE. In this example, before measuring the reference signal from the third network element, the first network element configures the third network element to send a reference signal. Specifically, the first network element sends air interface transmit configuration information to the third network element, which corresponds to an air interface transmit configuration. Similar to the example above, the third network element sends a reference signal based on the air interface transmit configuration, and the first network element measures the reference signal from the third network element, obtains a measurement result, and obtains candidate training data based on the air interface transmit configuration.

[0175] In the example above, the air interface transmission configuration is optional. Transmission power of the reference signal, The number of antenna ports used for the reference signal, Reference signal bandwidth, The frequency domain density of the reference signal, or Reference signal duration It may include one or more of the following.

[0176] From the procedure shown in Figure 3, it can be seen that the first network element collects candidate training data for the AI ​​model and determines the validity of the candidate training data based on the first information. This can also be understood as filtering the collected candidate training data. When the collected candidate training data is valid, the first network element sends the valid candidate training data to the second network element, thereby training or updating the AI ​​model. In this case, the valid candidate training data is the training data. When the candidate training data collected by the first network element is invalid, the first network element indicates to the second network element that the current collection of candidate training data is invalid.

[0177] If candidate training data collected once is invalid, it may also be understood that this indicates that no training data was collected this time, and if candidate training data collected once is valid, it may also be understood that no training data was collected this time. In this case, the valid candidate training data becomes training data, which is referred to herein as the first training data and is provided to the second network element by the first network element.

[0178] In the embodiments of this application, it should be understood that "the first network element recollects training data for the AI ​​model" also means "the first network element recollects candidate training data for the AI ​​model." The first network element attempts to recollect training data only when it has not collected training data. The purpose of recollection is to expect to collect training data for the AI ​​model. However, in the process of collecting training data for the AI ​​model, candidate training data is collected first, and then training data is obtained from the candidate training data through filtering.

[0179] After receiving the second information from the first network element, if the second network element determines, based on the second information, that the current collection by the first network element is invalid, then, if possible, the second network element determines that the AI ​​model training data needs to be collected again.

[0180] In the example, the second network element sends third information to the first network element, which indicates to the first network element that it should recollect training data for the AI ​​model. Optionally, in the example, the third information indicates the maximum number of validity checks k. Optionally, each validity check involves collecting a new batch of training data, i.e., a new training dataset. Thus, the maximum number of validity checks is also sometimes called the maximum number of times training dataset collections will be performed.

[0181] In this example, when the second network element receives information from the first network element indicating that the candidate training data collected by the first network element is invalid, the second network element sends a third piece of information to the first network element indicating that it should recollect the training data for the AI ​​model. In addition, the third piece of information indicates the maximum number of validity checks, k, where k is a positive integer. If the candidate training data corresponding to the first validity check is invalid, the first network element recollects the candidate training data or continues collecting it based on the third piece of information. Recollection or continued collection may occur multiple times. After each recollection or continued collection, a validity check is performed. If the result of the check is invalid, the first network element may continue the next recollection and the next validity check until the maximum number of validity checks, k, is reached. If all validity checks from the first to the (k-1)th check are invalid, and the result of the kth validity check is still invalid, the first network element stops collecting training data.

[0182] In this example, when the second network element indicates to the first network element that training data should be recollected by using a third piece of information, the second network element may further indicate to the first network element the maximum number of validity checks k. In other words, the maximum number of validity checks k is sent after the second network element has determined that training data needs to be recollected. The maximum number of validity checks k may be included in the third piece of information or carried in other information other than the third piece of information. Optionally, in another example, as described in step 501, the second network element indicates the maximum number of validity checks k in the first piece of information sent to the first network element. The second network element indicates the maximum number of validity checks k to the first network element to limit the first network element's process of recollecting candidate training data, and therefore, if no training data (i.e., valid candidate training data) is collected, the first network element will not enter a recollection cycle without a time limit, but will stop collecting after reaching the maximum number of validity checks k, regardless of whether training data has been collected or not.

[0183] Before the maximum number of validity checks k is exceeded, if the first network element determines, based on the first information, that the result of the j-th validity check is valid, that is, if it determines that the candidate training data on which the j-th validity check was performed contains valid candidate training data, the first network element sends fourth information to the second network element. The fourth information includes the second training data and indicates that the result of the j-th validity check is valid. The second training data may specifically include valid candidate training data from the candidate training data on which the j-th validity check was performed. j is less than or equal to k, and j is a positive integer.

[0184] The j-th effectiveness evaluation may be considered an effectiveness evaluation performed on a set of candidate training data, and it should be noted that all candidate training data included in the set are training data on which the j-th effectiveness evaluation has been performed. The training data on which the j-th effectiveness evaluation has been performed is not limited to candidate training data collected in the j-th evaluation, but may also include candidate training data obtained in one or more collections prior to the j-th collection. This is not limited to the j-th evaluation.

[0185] Optionally, if the j-th validity determination result is valid, the first network element sends the second training data and information a to the second network element, and information a indicates that the current collection is valid.

[0186] In addition, in the embodiments of this application, the maximum number of validity checks k corresponds to the start moment, and the start moment is understood as the start moment of the training data collection process corresponding to the maximum number of validity checks k. For example, the start moment may be the moment when the first network element receives the first or third piece of information. In other words, the first network element begins collecting training data for the AI ​​model from the moment the first or third piece of information is received. Optionally, the moment of termination of the collection process is indeterminate. For example, if the result of the j-th validity check is valid before the maximum number of validity checks k is exceeded, the collection process terminates, and the first network element sends valid candidate training data (i.e., the first training data) to the second network element, where j is less than or equal to k, and j is a positive integer. However, if all validity checks from the first to the kth check are invalid, the moment when the result of the k-th validity check is invalid is determined to be the moment of termination of the collection process.

[0187] According to the method for acquiring training data provided in this application, the training data collection network element performs a validity check on the collected candidate training data and provides the valid candidate training data to the AI ​​model training network element, ensuring that the collection network element provides only training data that meets the requirements of the training network element. Since training data that does not meet the requirements is filtered out and removed on the collection network element side, the exchange of invalid training data is omitted. This not only saves air interface resources but also prevents contamination of the training dataset on the training network element side and avoids other adverse effects.

[0188] The above describes in detail the main steps for obtaining training data for an AI model. Below, using examples, we will explain how to obtain training data when applying an AI model to different scenarios.

[0189] Application Scenario 1 AI model-based channel state information (CSI) feedback or CSI prediction. For example, in application scenario 1, AI model training or updating is deployed on the access network device side. The access network device sends a downlink reference signal to the UE, and the UE obtains a measurement result by measuring the downlink reference signal, which is candidate training data. The UE determines the validity of the obtained candidate training data and provides valid candidate training data to the access network device side for AI model training or updating. For example, in this application scenario, the downlink reference signal may be CSI-RS in particular. The valid candidate training data provided to the access network device by the UE is the AI ​​model label, specifically CSI.

[0190] In many application scenarios, the access network needs to acquire a downlink CSI to determine one or more of the following configurations for scheduling the UE's downlink data channel: resources, modulation and coding scheme (MCS), and precoding. In time division duplex (TDD) systems, due to the reciprocity between the uplink and downlink channels, the access network device can acquire the uplink CSI by measuring the uplink reference signal, infer the downlink CSI, and, for example, use the uplink CSI as the downlink CSI. In frequency division duplex (FDD) systems, the reciprocity between uplink channels cannot be guaranteed, and the downlink CSI is acquired by the UE by measuring the downlink reference signal. For example, the UE acquires the downlink CSI by measuring signals such as CSI-RS or synchronization signal and physical broadcast channel block (SSB). The UE generates a CSI report in a predefined manner within the protocol or in a manner pre-configured by the access network device, and feeds back the downlink CSI to the access network device using the CSI report, thereby allowing the access network device to obtain the downlink CSI.

[0191] Please refer to Figure 4. Figure 4 is a diagram of an AI model-based CSI feedback mechanism. As shown in Figure 4, an autoencoder (AE) model includes two submodels: an encoder and a decoder, and AE generally refers to a network structure that includes these two submodels. AE models are also sometimes called bidirectional models, bidirectional models, or cooperative models. The encoder and decoder of an AE are typically trained together and may be used together. CSI feedback can be implemented based on the AI ​​model of the AE. For example, the UE side measures the downlink reference signal sent by the base station to obtain the measured CSI. The UE compresses and quantizes the CSI obtained through the measurement using the encoder and feeds back the compressed and quantized information, for example, the "feedback CSI" shown in Figure 4, to the base station. The base station reconstructs the "feedback CSI" using the decoder to obtain the reconstructed CSI. For the base station, the decoder input is information about the CSI that is fed back by the UE, and the CSI obtained by the UE through measurement needs to be used as the truth value (or label) of the reconstructed CSI for decoder training.

[0192] In application scenario 1, the AI ​​model deployed on the access network device side may be the decoder shown in Figure 4.

[0193] Please refer to Figure 5. Figure 5 shows an example of acquiring training data in an AI model-based CSI feedback according to this application.

[0194] 501: Optionally, the access network device determines that AI model training data needs to be collected.

[0195] 502: The access network device sends a first piece of information to the UE, which is used by the UE to determine the validity of the candidate training data collected. Optionally, the validity determination may be valid or invalid.

[0196] For example, the first piece of information indicates constraints for determining the effectiveness of the candidate training data collected by the UE.

[0197] For information regarding the first point, limitations, etc., please refer to the relevant explanation in step 310. Further details will not be provided again in this specification.

[0198] In application scenario 1, the example might show that the quality indicators of the measurement results are the SINR of the training data and the amount of training data. The first information indicates the SINR threshold Q and the training data amount threshold N. The SINR criteria and the training data amount criteria (e.g., SINR is greater than or equal to Q and the amount of training data is greater than or equal to N) may be predefined in the protocol. In another example, the first information indicates the SINR threshold Q, the training data amount threshold N, the SINR criteria, and the training data amount criteria. For example, the first information indicates thresholds Q and N, and includes an information field indicating the criteria. For example, the information field contains 1 bit, which corresponds to the SINR criteria and the training data amount criteria. For example, a 1-bit value of "1" indicates that "the SINR of the training data is greater than or equal to Q and the amount of training data is greater than or equal to N," and a 1-bit value of "0" indicates that "the SINR of the training data is greater than Q and the amount of training data is greater than N." For example, the information field contains two bits b1 and b0, where b1 corresponds to the SINR criterion and b0 corresponds to the training data quantity criterion. For example, when the value of b1 is 1, it indicates that "the SINR of the training data is greater than or equal to Q", and when the value of b1 is 0, it indicates that "the SINR of the training data is less than Q". Similarly, b0 indicates the training data quantity criterion. Details are not explained again. In yet another example, the first information indicates the threshold Q for the SINR of the training data and the threshold N for the training data quantity. In addition, the first information indicates some of the criteria for the quality indicator, while the criteria for other parts of the quality indicator are predefined in the protocol. For example, the first information indicates thresholds Q and N. In addition, the first information further includes a 1-bit information field. When the value of the 1-bit is 1, it indicates that "the SINR is greater than or equal to Q". When the value of the 1-bit is 0, it indicates that "the SINR is less than Q".The criteria for determining the amount of training data are predefined in the protocol, for example, "the amount of training data is at least N." It should be understood that the above implementation is merely an example of how the first piece of information determines the constraint; it is not limited to this.

[0199] For example, N could be an integer multiple of the batches used during AI model training, or the amount of training data required for the AI ​​model to converge.

[0200] 503: The access network device sends a reference signal to the UE.

[0201] The UE obtains one or more measurement results by measuring a reference signal of an access network device. In this application, the measurement results may be expressed as, alternatively, a reference signal measurement result or a channel measurement result. The alternative expressions are also applicable to embodiments in other application scenarios. Details are not described below.

[0202] Optionally, the measurement results may include channel responses, such as a channel response matrix.

[0203] In addition, optionally, the UE may obtain one measurement result through a single measurement. In this case, the candidate training data will contain one measurement result. Optionally, the UE may obtain multiple measurement results through multiple measurements. In this case, the candidate training data will contain multiple measurement results.

[0204] For example, quality metrics of measurement results may include, but are not limited to, one or more of the following: first-path power, first-path arrival delay, timing error group (TEG), average power at time-domain sampling points, phase difference between antenna ports, full-band or sub-band equivalent SINR, full-band or sub-band interference level, line-of-light (LOS) probability, inter-site synchronization error, or confidence in the measurement results. These metrics are applicable to application scenario 1 or other application scenarios described below. It should be understood that these quality metrics can be obtained by performing corresponding processing on the measurement results of a reference signal. Specific processing processes are not limited herein and may include, for example, any known or future processing.

[0205] It should be understood that before an access network device sends a reference signal to the UE, the access network device further sends air interface transmit configuration information corresponding to the reference signal to the UE. The air interface transmit configuration information indicates the relevant air interface configuration for the access network device to send the reference signal. For example, the air interface transmit configuration information may include, but is not limited to, one or more of the following: the transmit power of the reference signal, the number of antenna ports used by the access network device to send the reference signal, the bandwidth of the reference signal, the frequency domain density of the reference signal, and the duration of the reference signal. Those skilled in the art will understand that the air interface transmit configuration information may further include other relevant information, which is not listed here one by one.

[0206] In this scenario, the candidate training data collected by the UE consists of one or more measurement results or one or more channel measurement results obtained by measuring a reference signal from an access network device.

[0207] For example, in a 5G system, the reference signal may be a channel state information reference signal (CSI-RS).

[0208] 504:UE determines the effectiveness of the collected candidate training data based on the first piece of information.

[0209] For example, candidate training data is assumed to be multiple measurement results obtained by the UE by measuring a reference signal. These multiple measurement results constitute the candidate training data. Based on the constraints, the UE determines whether the multiple measurement results contain valid candidate training data (or valid measurement results). In the example above, if the constraint is "the quality metric (e.g., SINR) is greater than or equal to threshold Q, and the amount of candidate training data where the quality metric is greater than or equal to threshold Q is at least N", then the UE determines whether the multiple collected measurement results contain measurement results where the quality is greater than or equal to threshold Q. For simplicity, measurement results where the quality metric is greater than or equal to threshold Q will be referred to as measurement result 1 below. If the UE determines that the multiple collected measurement results contain measurement result 1, it further needs to determine whether the amount of measurement result 1 reaches N. Based on the constraints, if it determines that valid measurement results have been collected, the UE determines that the candidate training data collected is valid. Valid candidate training data (i.e., the first training data, also referred to below as valid data) is a portion of the measurement results that satisfy the constraints. For example, if the quantity of measurement result 1 is P, and P is an integer greater than or equal to N, then these P measurement results 1 are the valid data collected in this instance, i.e., the first training data.

[0210] Conversely, if the UE determines that none of the collected measurement results satisfy the constraints, for example, if the collected measurement results include measurement result 1 where the SINR is greater than or equal to the threshold Q, but the amount of measurement result 1 is less than N, or if the SINR of all of the collected measurement results is less than the threshold Q, the UE determines that the candidate training data collected this time is invalid.

[0211] 505:UE sends second information to the access network device based on the effectiveness assessment of the collected candidate training data, and this second information indicates the effectiveness assessment result.

[0212] Where possible, the second piece of information indicates that the candidate training data collected by the UE is valid. In the example, the second piece of information might be valid collected candidate training data, for example, P measurement results 1 in the example above. In this example, the P measurement results 1 are valid candidate training data, and the P measurement results 1 also implicitly indicate that the candidate training data collected by the UE is valid. In another example, the UE sends the second piece of information and valid candidate training data. In this example, the second piece of information indicates that the candidate training data collected by the UE is valid. For example, the second piece of information might contain 1 bit. When the value of the 1 bit is "1", it indicates that the candidate training data collected by the UE is valid. In addition, the UE sends the valid candidate training data to the access network device. In contrast, the previous example can further reduce signaling overhead, assuming that it is possible to indicate that the candidate training data collected by the UE is valid.

[0213] In another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid. In the example, the second piece of information may contain one bit. When the value of this one bit is "0", it indicates that the candidate training data collected by the UE is invalid.

[0214] For example, the second piece of information may be conveyed by using uplink control information (UCI) signaling. For instance, the UCI contains one bit of information indicating whether the candidate training data collected by the UE is valid or invalid. Optionally, valid candidate training data may also be sent in the UCI if the UE implicitly indicates that the collected candidate training data is valid by using valid candidate training data, as shown in the example above. This is not limited to this.

[0215] 506: The access network device determines, based on the second piece of information, whether the candidate training data for the UE is valid.

[0216] In response to the determination result in step 505, the second piece of information, if possible, indicates that the candidate training data collected by the UE is valid. In this case, the access network device retrieves further valid candidate training data collected by the UE from the UE. Furthermore, the access network device trains or updates the AI ​​model based on the valid candidate training data, as shown in step 507.

[0217] 507: Access network device trains an AI model, retrieves an AI model, or updates an AI model.

[0218] In another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid. In a possible implementation in this case, the access network device either retains the CSI feedback for the original AI model or switches to the CSI feedback for the non-AI model. In the example, retaining the original AI model might be for a scenario where a trained AI model is deployed on the access network device and the training data is being collected now for the purpose of updating the AI ​​model. Switching to the non-AI model might be for a scenario where a trained AI model is not on the access network device and the training data is being collected now for training to acquire an AI model. In this scenario, if no valid candidate training data is acquired during the current collection, the access network device may switch to the CSI feedback for the non-AI model. Two possible cases are shown in step 508.

[0219] 508: Access network devices perform CSI feedback based on the original AI model or by switching to a non-AI model.

[0220] The training data collection procedure ends when the access network device performs step 507 or step 508.

[0221] Optionally, in another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid. After obtaining the second piece of information, the access network device determines to recollect the training data as shown in steps 509 and 510.

[0222] 509: The access network device determines that it will recollect the training data for the AI ​​model.

[0223] 510: The access network device sends a third piece of information to the UE, which indicates to the UE that it will recollect the training data for the AI ​​model.

[0224] In optional, possible implementations, a third piece of information further indicates the maximum number of validity checks k, where k is a positive integer. In optional, other possible implementations, the maximum number of validity checks k may also be indicated by the first piece of information. This is not limited to these implementations. These two implementations are described in detail in the procedure shown in Figure 3. Further details are not described again herein.

[0225] Optionally, when training data for the AI ​​model is recollected, the access network device may update the air interface transmit configuration. Correspondingly, the UE recollects candidate training data for the AI ​​model based on the updated air interface transmit configuration.

[0226] 511: Optionally, the access network device sends air interface transmit configuration information to the UE, and the air interface transmit configuration information indicates the updated air interface transmit configuration.

[0227] The air interface transmit configuration information is described above. If the air interface transmit configuration is updated, the air interface transmit configuration information in step 511 will reflect the updated air interface transmit configuration. For example, an update to the air interface transmit configuration may include, but is not limited to, updates to the reference signal transmit power, the amount of antenna ports used by the access network device to send the reference signal, the reference signal bandwidth, the reference signal frequency domain density, and the reference signal duration. For example, an update to the air interface transmit configuration may include an increase in the reference signal transmit power and an increase in the reference signal frequency domain density. In this case, the access network device attempts to send the reference signal to the UE with higher transmit power and higher frequency domain density to enable the UE to acquire candidate training data that satisfies the constraints. The air interface transmit configuration is updated to ensure the accuracy of AI model-based CSI feedback over the air interface.

[0228] Of course, when the AI ​​model training data is recollected, the original air interface transmission configuration may not be updated. In this case, the UE recollects the candidate training data with the original air interface transmission configuration, determines the validity of the recollected candidate training data based on the constraints, and displays the validity determination result to the access network device.

[0229] 512: Optionally, the UE recollects training data for the AI ​​model.

[0230] In the procedure for recollecting training data, please understand that the effectiveness assessment and indication of the assessment results for the recollected candidate training data are the same as in the procedure described above in Figure 5. Further details will not be explained again. Please understand that in the process of recollecting training data, the UE is constrained by the maximum number of effectiveness assessments, k.

[0231] Optionally, the maximum number of validity checks k may be determined by the access network device based on the urgency of training data collection. For example, urgency may be the interval from the last update of the AI ​​model to the present time. For example, if the interval from the last update of the AI ​​model to the present time is large and exceeds a threshold, the requirement to update the AI ​​model is considered urgent. A larger interval indicates a higher probability of changes in the channel environment and a potential decrease in the AI ​​model's fit to the current channel environment. Therefore, the update requirement is more urgent. In this case, the maximum number of validity checks k may be set to a correspondingly larger value, so that valid candidate training data can be obtained through multiple re-collections after one invalid collection. If the interval from the last update of the AI ​​model to the present time is very small, for example, smaller than a threshold, the update requirement is considered not urgent, and the maximum number of validity checks k may be set to a small value. Optionally, the urgency criteria may also be implemented in a different manner, but are not limited to this.

[0232] The method for acquiring training data in AI model training provided in this application can be applied to AI model-based CSI feedback or CSI prediction scenarios, thereby reducing the waste of air interface resources in the AI ​​model training procedure. In addition, when a UE sends invalid training data to an access network device, this contaminates the AI ​​model's training dataset, thereby affecting the training or updating of the AI ​​model and causing inaccurate gain evaluations. This solution avoids the above-mentioned effects.

[0233] Application Scenario 2 AI model-based positioning scenarios Since uplink positioning differs from downlink positioning, the application of the technical solution of this application to uplink positioning and downlink positioning will be described separately below.

[0234] 1. Application in uplink positioning Please refer to Figure 6. Figure 6 is a diagram of the technical solution in an AI model-based uplink positioning scenario according to the present application. As shown in Figure 6, in uplink positioning, AI model training or updating is performed on the network side. A 5G system is used as an example. The AI ​​model may be deployed in a positioning device in the core network, e.g., an LMF network element. In uplink positioning, the input to the AI ​​model is one or more channel responses (or channel measurement results) corresponding to one or more sounding reference signals, and the output to the AI ​​model is the location of the UE. There may be one or more transmitters of the one or more sounding reference signals, e.g., UEs, and one or more receivers, e.g., access network devices.

[0235] As shown in Figure 6, in uplink positioning, when training the AI ​​model used for positioning, the positioning device obtains from the access network side multiple measurement results obtained by one access network device measuring multiple sounding reference signals or by each of multiple access network devices measuring one or more sounding reference signals, along with location information of a third network element. The multiple sounding reference signals may include multiple sounding reference signals from one third network element, or one or more sounding reference signals from each of multiple third network elements. The location information of a third network element sending multiple sounding reference signals at different moments, or the location information of multiple third network elements sending one or more sounding reference signals at one or more moments, is used as the truth value (i.e., label) of the location information output by the AI ​​model.

[0236] Please refer to Figure 7. Figure 7 shows an example of acquiring training data in AI model-based uplink positioning according to this application.

[0237] 701: Optionally, the positioning device determines that it needs to collect training data for the AI ​​model.

[0238] 702: The positioning device sends a first piece of information to the access network, which is used by the access network device to determine the validity of the candidate training data collected. Optionally, the determination may be valid or invalid.

[0239] For example, the first piece of information may be carried in interface messages between a positioning device and an access network device. A 5G system is used as an example. If the positioning device is an LMF and the access network device is a gNB, the first piece of information between the LMF and the gNB may be included in NRPPa messages.

[0240] 703: An access network device sends air interface transmit configuration information (e.g., air interface transmit configuration information #1) to a third network element, which indicates an air interface transmit configuration used by the third network element to send a sounding reference signal.

[0241] In this embodiment, the third network element is a network element that can provide location information for the third network element. In the example, the third network element may be a location reference device. The location reference device may be considered a special network element and may typically be configured by a network vendor. For example, a network vendor may configure one or more of the location, transmit capability, receive capability, and processing capability of the location reference device. The location reference device may provide location information for the location reference device to an access network device. For example, the third network element may be a reference UE or an automated guided vehicle (AGV). In another example, the third network element may, alternatively, be a regular UE. In this specification, a regular UE is a location reference device. After obtaining location information for a regular UE by using some positioning method, the regular UE may provide location information to an access network device.

[0242] 704: The access network device measures a sounding reference signal from a third network element and obtains one or more measurement results.

[0243] A 5G system is used as an example. The sounding reference signal transmitted by the third network element may be an SRS (sounding reference signal).

[0244] In step 704, the access network device measures a sounding reference signal sent by the third network element and obtains one or more measurement results, and there is a correspondence between one or more measurement results and the location information of the third network element. In the example, the third network element sends a sounding reference signal at location 1, and the access network device obtains measurement result 1 by measuring the sounding reference signal, and measurement result 1 corresponds to location 1. The third network element sends a sounding reference signal at location 2, and the access network device obtains measurement result 2 by measuring the sounding reference signal, and measurement result 2 corresponds to location 2. Optionally, in another example, the absolute location of the third network element does not change, but the surrounding environment of the third network element changes at different times. The access network device measures the sounding reference signal sent by the third network element at different times, and the obtained measurement results may also change. For example, measurement result 1 acquired by the access network device at time 1 corresponds to location 1 of the third network element, and measurement result 2 acquired at time 2 corresponds to location 1 of the third network element. Optionally, in yet another example, there may be multiple third network elements in this embodiment. The access network device acquires multiple measurement results by individually measuring the sounding reference signals from the multiple third network elements. In other words, each of the multiple measurement results corresponds to the location of one of the multiple third network elements. Correspondingly, in step 705, the multiple third network elements individually provide their respective location information to the access network device or positioning device.

[0245] 705: The third network element provides location information for the third network element.

[0246] In this embodiment, one third network element is used as an example for illustrative purposes. One location information corresponds to one or more measurement results obtained by one or more access network devices by measuring one or more sounding reference signals sent by the third network element at the location corresponding to the location information. In the uplink positioning scenario, the candidate training data for the AI ​​model is one or more measurement results and the location information of the third network element corresponding to one or more measurement results. When one or more access network devices determine that the collected candidate training data (i.e., one or more measurement results) is valid, one or more access network devices individually provide the valid candidate training data to the positioning device.

[0247] Optionally, location information of a third network element corresponding to valid candidate training data may be provided to the positioning device by the third network element via at least one of one or more access network devices, as shown in step 705a. It should be understood that 705a is an implementation of step 705. The location information of the third network element may be visible or invisible to at least one access network device.

[0248] Optionally, in the implementation, a third network element directly provides subframe location information (not shown in the figure) to the positioning device. The positioning device obtains valid candidate training data and location information corresponding to the valid candidate training data from one or more access network devices. It should be understood that there are multiple valid candidate training data and also multiple location information for the third network element. The positioning device determines the correspondence between the valid candidate training data and the location information. The positioning device uses the location information as a label for the AI ​​model and trains or updates the AI ​​model, i.e., trains it in a creation process or in an update process. In this embodiment, it should be understood that the valid candidate training data for the AI ​​model obtained by the positioning device (i.e., the first training data) is among the measurement results obtained by the access network device by measuring the sounding reference signal sent by the third network element, and includes one or more measurement results that satisfy the constraints, and location information for the third network element, corresponding to each of the measurement results. The location information for the third network element is the output truth value, i.e., the label, of the AI ​​model.

[0249] 706: The access network device determines the validity of the collected candidate training data based on the first information.

[0250] In step 706, note that the access network device specifically determines the validity of the measurement results in the candidate training data.

[0251] For example, in application scenario 2, the quality indicators of the measurement results may include, but are not limited to, one or more of the following: first path power, first path arrival delay, timing error group (TEG), average power of time-domain sampling points, phase difference between antenna ports, equivalent SINR for the entire or subband, interference level information for the entire or subband, line of light (LOS) probability, indication information for inter-site synchronization error, or indication information for the reliability of the measurement results. In addition, in positioning application scenarios, the quality indicator of the label may be the distance between different sample locations.

[0252] For example, quality metrics in constraints may be the SINR and quantity of training data. In the example, the threshold for the quantity of training data may be N, where N is an integer multiple of the batch or the minimum amount of training data required for the AI ​​model to converge. Optionally, in an uplink positioning scenario, candidate training data for the AI ​​model collected by the access network device may further include labels, which are location information. For example, quality metrics in constraints may further include quality metrics for the labels. For example, the quality metric for the labels may be the distance between locations of different samples. This is not limited herein.

[0253] For information on determining effectiveness, please refer to the explanation in step 504. Further details are not provided in this specification.

[0254] 707: The access network device sends second information to the positioning device based on the effectiveness determination result of the candidate training data, and the second information indicates the effectiveness determination result.

[0255] For example, the second piece of information may be included in an interface message between the access network device and the positioning device. Alternatively, the access network device could send an interface message to the positioning device, and this interface message could contain the second piece of information.

[0256] 708: The positioning device determines, based on the second piece of information, whether the candidate training data collected by the access network device is valid.

[0257] Where possible, the second piece of information indicates that the candidate training data collected by the access network device is valid. In this case, the positioning device obtains the valid candidate training data (i.e., the first training data) collected by the access network device. In this specification, the first training data particularly includes one or more measurement results that satisfy the constraints and location information of a third network element, corresponding to each of the one or more measurement results. Furthermore, the positioning device trains or updates the AI ​​model based on the valid candidate training data, as shown in step 709.

[0258] 709: The positioning device trains an AI model and either acquires or updates the AI ​​model.

[0259] In another possible case, the second piece of information indicates that the candidate training data collected by the access network device is invalid. In a possible implementation in this case, the positioning device either retains the original AI model or switches to a non-AI model, as shown in step 710.

[0260] 710: The positioning device performs uplink positioning based on the original AI model or by switching to a non-AI model.

[0261] Optionally, in another case, the second information indicates that the candidate training data collected by the access network device is invalid. In another possible implementation, the positioning device determines to re-collect training data for the AI model. In this case, step 711 and step 712 are further included.

[0262] 711: The positioning device determines to re-collect training data.

[0263] 712: The positioning device sends third information to the access network device, where the third information indicates to the access network device that training data for the AI model is to be re-collected.

[0264] Optionally, the third information further indicates the maximum number k of validity determinations, and k is a positive integer. Optionally, the maximum number k of validity determinations may alternatively be indicated by the first information. For details, reference is made to the related description of the procedure shown in Figure 3, and no further description is provided herein.

[0265] In this application scenario, the maximum number of validity checks k may be configured by the positioning device based on the urgency of the requirements for the AI ​​model's training data. This is similar to that in application scenario 1. In the example, the urgency criterion may be determined based on the resulting error of the current AI model's estimation of the location of the third network element, or on the interval from the last update of the AI ​​model to the present time. For example, if the resulting error of the positioning device's estimation of the location of the third network element based on the current AI model is large, for example, if it is above a specified threshold, the requirement may be determined to be urgent. In this case, the maximum number of validity checks k may be set to a large value. Conversely, if the resulting error of the current AI model's estimation of the location of the third network element is small, for example, if it is below a specified threshold, the requirement may be determined to be not urgent. In this case, the maximum number of validity checks k may be set to a small value. The resulting error of the AI ​​model's estimation is determined by splitting the training data collected in previous AI model training into a training set and a validation set. Since the error in the training set is already extremely low, the error in the validation set's estimation results is used as a criterion for determining whether the AI ​​model has become significantly ineffective. Additionally, this criterion may also be set based on the interval between the last update of the AI ​​model and the present time. See the explanation for Application Scenario 1 for details. Further details will not be explained again.

[0266] If the positioning device optionally decides to recollect training data for the AI ​​model, it may indicate to the access network device that it will update the air interface transmission configuration used by the access network device to collect the training data, as shown in step 713.

[0267] 713: The access network device sends air interface transmit configuration information (for example, air interface transmit configuration information #2) to a third network element, and the air interface transmit configuration information indicates the updated air interface transmit configuration.

[0268] It should be understood that the update of the air interface transmit configuration in step 713 is an update of the air interface transmit configuration in step 703. For example, the update may include, but is not limited to, an increase in the transmit power of the sounding reference signal or an increase in the frequency domain density of the sounding reference signal. It should be understood that the purpose of updating the air interface transmit configuration is to allow the access network device to collect candidate training data that meets the constraints and to provide that candidate training data to the positioning device.

[0269] 714: The access network device recollects the training data for the AI ​​model.

[0270] The method for acquiring training data in AI model training provided in this application is applicable to AI model-based uplink positioning scenarios, and it can be seen that this makes it possible to reduce the waste of air interface resources in the AI ​​model training procedure. In addition, access network devices send invalid training data to positioning devices, which causes contamination of the AI ​​model training dataset, thereby affecting the training or updating of the AI ​​model and causing inaccurate gain evaluation. This solution avoids the above effects.

[0271] In the uplink positioning scenario, it should be understood that the access network device is an example of a training data collection network element, and the positioning device is an example of an AI model training network element.

[0272] 2. Application in Downlink Positioning Please refer to Figure 8. Figure 8 is a diagram of the technical solution in an AI model-based downlink positioning scenario according to this application. As shown in Figure 8, in downlink positioning, AI model inference is deployed on the UE side, while AI model training is deployed on the network side positioning device, e.g., an LMF network element. The AI ​​model deployed on the positioning device uses the corresponding channel response obtained by the UE by measuring a reference signal as input and the UE's location as output. A 5G system is used as an example. The reference signal may be a positioning reference signal and may be sent to the UE by one or more base stations (BSs).

[0273] Please refer to Figure 9. Figure 9 shows an example of acquiring training data in AI model-based uplink positioning according to this application.

[0274] 801: Optionally, the positioning device determines that training data for the AI ​​model needs to be collected.

[0275] The training data for the AI ​​model may come from measurements performed by a single UE on multiple reference signals, or from measurements performed by each of multiple UEs on one or more reference signals. The multiple reference signals may come from one or more access network devices. This embodiment will be described in terms of communication between the positioning device and a specific UE among the UE or one or more UEs.

[0276] 802: The positioning device sends a first piece of information to the UE, which is used to determine the validity of the candidate training data collected by the UE. Optionally, the validity determination may be valid or invalid.

[0277] Optionally, in the example, the positioning device sends the first information to the UE via the access network device, as shown in steps 802a and 802b of Figure 9. In another example, the positioning device may, alternatively, send the first information directly to the UE through the interface between the positioning device and the UE. The positioning device sends information #1 to the access network device, which indicates to the access network device that it will send a positioning reference signal to the UE. Optionally, information #1 may, alternatively, be the first information. The access network device sends a positioning reference signal to the UE based on the trigger of information #1. The implementation shown in Figure 9 is only an example.

[0278] 803: The access network device sends a positioning reference signal to the UE.

[0279] The UE obtains candidate training data by measuring positioning reference signals from an access network device, or from an access network device and another access network device, which in turn consists of one or more measurement results of the positioning reference signals and the UE's location information corresponding to one or more measurement results.

[0280] Please understand that before sending the PRS to the UE, the access network device further sends PRS air interface transmit configuration information to the UE to indicate the PRS air interface transmit configuration.

[0281] A 5G system is used as an example. The positioning reference signal sent to the UE by the access network device may be a PRS. The candidate training data consists of one or more measurement results obtained by the UE by measuring the PRS, and the location information of the UE corresponding to one or more measurement results.

[0282] 804:UE determines the effectiveness of the collected candidate training data based on the first piece of information.

[0283] For example, the UE determines the validity of the collected candidate training data based on the constraint indicated by the first information. In this specification, in particular, the UE determines the validity of one or more measurement results. Similar to step 504, reference is made to step 504 for understanding, and a detailed description is omitted herein. In addition, for examples of quality indicators included in constraints in a downlink positioning scenario, reference is made to the description of the uplink positioning scenario, and details will not be repeated herein.

[0284] 805: The UE sends second information to a positioning device, wherein the second information indicates the validity determination result.

[0285] Where possible, the second information indicates that the candidate training data collected by the UE is valid. In an example, the second information may be first training data among the candidate training data collected by the UE, and the first training data includes location information of the UE. Alternatively, the first training data is specifically a measurement result satisfying the constraint and the location information of the UE corresponding to the measurement result. In another possible case, the second information indicates that the candidate training data collected by the UE is invalid.

[0286] Optionally, in step 805, the UE may directly send the second information to the positioning device through the interface between the UE and the positioning device, as shown in Figure 9. Alternatively, the UE may send the second information to an access network device, which then sends the second information to the positioning device. Alternatively, when the candidate training data collected by the UE is valid, the UE sends any information contained in the second information, for example, measurement results contained in the first training data that satisfy the constraints (i.e., valid measurement results), to the access network device and the UE's location information to the positioning device. The access network device then sends the measurement results that satisfy the constraints to the positioning device. Thus, the positioning device obtains the first training data, which includes valid measurement results and the UE's location information corresponding to the valid measurement results. This is not limited to this.

[0287] 806: The positioning device determines, based on the second piece of information, whether the collection of candidate training data collected by the UE is valid.

[0288] Where possible, the second piece of information indicates that the candidate training data collected by the UE is valid. In this case, the second piece of information may include the first training data, which includes the UE's location information. Furthermore, the positioning device trains or updates the AI ​​model based on the first training data, as shown in step 807.

[0289] 807: The positioning device trains an AI model and either acquires or updates the AI ​​model.

[0290] In another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid. In a possible implementation in this case, the access network device either maintains beam management of the original AI model, as shown in step 808, or switches to a non-AI model to perform downlink positioning.

[0291] 808: The positioning device performs positioning by maintaining the original AI model or by switching to a non-AI model.

[0292] Optionally, in another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid. In a possible implementation, the positioning device determines to recollect the training data, as shown in step 809.

[0293] 809: The positioning device determines that it will collect training data again.

[0294] 810: The positioning device sends a third piece of information to the UE, which indicates to the UE that it will recollect training data for the AI ​​model.

[0295] Optionally, a third piece of information may further indicate the maximum number of effectiveness evaluations k. Alternatively, the first piece of information indicates the maximum number of effectiveness evaluations k.

[0296] Optionally, when training data for the AI ​​model is recollected, the positioning device may indicate to the access network device that it will update the air interface transmit configuration. For example, the positioning device sends information #2 to the access network device, which indicates to the access network device that it will recollect training data. In response, the access network device sends air interface transmit configuration information corresponding to the updated air interface transmit configuration to the UE, as shown in step 811.

[0297] 811: The access network device sends air interface transmit configuration information to the UE, and the air interface transmit configuration information indicates the updated air interface transmit configuration.

[0298] Based on the updated air interface transmission configuration, the UE measures the positioning reference signal sent by the access network device to recollect training data for the AI ​​model.

[0299] 812:UE recollects the training data for the AI ​​model.

[0300] Optionally, the UE in this embodiment may be a location-based device or a regular UE. This is not limited to this. For location-based devices or regular UEs, see the description in step 703. Further details are not provided.

[0301] Furthermore, it should be understood that in this embodiment, the UE is an example of a training data collection network element, and the positioning device is an example of an AI model training network element.

[0302] The method for acquiring training data in AI model training provided in this application is applicable to AI model-based downlink positioning scenarios, and it can be seen that this makes it possible to reduce the waste of air interface resources in the AI ​​model training procedure. In addition, terminal devices (e.g., location reference devices) send invalid training data to positioning devices, which causes contamination of the AI ​​model training dataset, thereby affecting the training or updating of the AI ​​model and causing inaccurate gain evaluations. This solution avoids the above effects.

[0303] Application Scenario 3 AI model-based beam management For example, in application scenario 3, AI model training is deployed on the access network device side. The access network device needs to acquire training data obtained by the UE side (e.g., one or more measurement results of the reference signal) by measuring a reference signal, and use the training data acquired from the UE side to train or update the AI ​​model. In this application scenario, the AI ​​model labels are reference signal information corresponding to K optimal measurement results. Alternatively, in application scenario 3, the reference signal information corresponding to K optimal measurement results may relate to K beams and be replaced with information corresponding to K optimal measurement results, for example, the IDs of K beams.

[0304] In 5G systems, it can be seen that high-frequency bands above 6 GHz are introduced for data communication. Compared to the mid-to-low frequency bands below 6 GHz, the continuously available bandwidth of the high-frequency band spectrum is larger, and the center frequency is higher. Therefore, it is possible to obtain higher transmission speeds and greater system capacity. However, the propagation distance of high-frequency signals (e.g., millimeter waves) is limited and the path fading effect is strong, resulting in poor coverage. Thanks to large-scale antenna technology, high-frequency communication systems typically use a large number of antennas for beamforming, thereby obtaining clear beam gain and compensating for the limited propagation distance caused by high-frequency propagation characteristics. However, when accurate beamforming is designed, the base station needs to obtain accurate channel information from the terminal. Obtaining channel information from such a large-scale antenna array requires consuming a large amount of air interface overhead, which is unacceptable in a real system. Experiments have shown that high-frequency radio channels have clear sparsity, meaning that the main energy of the channel is concentrated in a limited amount of path. For example, when there is an unobstructed line-of-sight (LoS) path between the transmitter and receiver of a signal, the primary energy between the receiver and transmitter is concentrated in the LoS path. When there is a non-line-of-sight (NLOS) obstacle between the transmitter and receiver, the primary energy is concentrated in the path that can be reached through a single reflection. Typically, each path has a different angle of incidence and exit. Therefore, the transmitter and receiver of a high-frequency communication system only need to match their beam directions to the angle of incidence and exit of the primary path of the channel in order to obtain the most channel transmission energy and complete the communication.

[0305] For a high-frequency communication system, it is assumed that the transmitter has a total of S antennas and the receiver has R antennas, and the antenna configurations may include linear antennas or planar array antennas. The transmitter and receiver perform multiplication on their antennas using different precoding weights to precode the transmitted signal, thereby causing the transmitted signal to have a beamforming effect. For example, in a downlink signal transmission model, this is as follows:

[0306] Y=VHWX+N The receiver's receive precoding matrix is ​​V, and its channel response is H. The transmitter's transmit precoding matrix is ​​W, the transmitted signal is X, and the noise is N. The signal received by the receiver is Y. The form of the transmitter's transmit precoding matrix is ​​W = [W1, W2, ... W] M ] and W i V is the precoding weight of the i-th antenna of the transmitter. Similarly, the form of the receiver's received precoding matrix is ​​V=[V1,V2,…V R ] and V i is the precoding weight for the i-th antenna of the receiver. The signal obtained by precoding the transmitted signal X using W is WX, and WX is the final transmitter signal of the transmitter. WX has a beamforming effect in space. WX can be classified into a reference signal and a data signal depending on the difference between the information carried on X. Typically, the reference signal is sent during beam management processing. Possible types of reference signals include SSB, CSI-RS, SRS, phase-tracking reference signal (PTRS), demodulation reference signal (DMRS), etc.

[0307] The angle of the main path of a channel can be within a wide range, for example, from 0 to 360 degrees. Each precoding matrix W can cover only a specific angular range in space and corresponds to one beamformed beam. Therefore, multiple precoding matrices W need to be designed to ensure good signal coverage. Multiple precoding matrices W with different beam angles form a single codebook. Both the transmitter and receiver maintain their own codebooks. In beam management processing, the transmitter and receiver implement angular matching between them by traversing and scanning their own codebooks. For example, the transmitter's codebook contains 64 precoding matrices corresponding to 64 beamformed beams. The receiver's codebook contains 4 precoding matrices corresponding to 4 beamformed beams. Therefore, determining the optimal pair of beamformed beams for the receiver and transmitter requires a total of 256 (64*4) scans, resulting in extremely large scanning overhead and delay.

[0308] In beam management technology based on AI models, sparse beam scanning can be implemented, and the overhead and delay of beam scanning can be significantly reduced. For a high-frequency communication system, it is assumed that the transmitter has a total of S antennas, the receiver has R antennas, the transmitter's codebook has S precoding matrices (S beamformed beams), and the receiver's codebook has R precoding matrices (R beamformed beams). For each received beamformed beam, any transmitted beamformed beam can form a transmit / receive beam pair with a received beamformed beam. Therefore, the process of determining the optimal receive / receive beam pair can be split as follows: Transmitter beam scanning is performed on the received beams to determine the best matched transmit beam, and then this process is repeated for the remaining R-1 received beams to determine the globally best transmit / receive beam pair. Similarly, for each transmit beam, any received beam can form a transmit / receive beam pair with a transmit beam. Therefore, the process of determining the optimal transmit / receive beam pair can be split as follows: Receiver beam scanning is performed on the transmit beam to determine the best matched transmit beam, and then this process is repeated for the remaining S-1 transmit beams to determine the globally best transmit / receive beam pair. Therefore, transmitter beam scanning will be used as an example below for explanation.

[0309] Please refer to Figure 10. Figure 10 is a diagram of AI-assisted sparse beam scanning processing.

[0310] The precoding matrix in the transmitter's codebook is assumed to correspond to primary beams, for example, 64 beamformed beams. In conventional solutions, all 64 beams need to be scanned to determine the optimal beam. However, in AI-assisted (i.e., AI model-based) sparse beam scanning solutions, only a few beams in the codebook need to be scanned, for example, secondary beams marked in Figure 10, for example, 16 beams. There can be many ways to select some beams from the codebook to scan, and each combination of selected beams is called a sparse beam pattern. The transmitter performs precoding by using the precoding matrix in the sparse beam pattern and sends a reference signal. The receiver inputs the measurement results of the reference signal into a neural network for AI beam prediction, and the neural network outputs an index of K beams, also called the index of the top K beams. Note that the K beams are K of all 64 beamformed beams in the codebook, and are not limited to K of the beamformed beams contained in the sparse beam pattern. The receiver feeds back the indices of the top K beams to the transmitter. Optionally, if K > 1, the transmitter scans only the K beamformed beams and sends a beamformed reference signal. The receiver measures the energy of the K reference signals using an energy sensing method and selects the beam with the strongest energy as the optimal beam.

[0311] The following describes the application of the method for acquiring training data in AI model training provided in this application to beam management based on sparse beam scanning technology. In this application, the transmitter is an access network device, such as a base station. The receiver is an UE. The UE measures a reference signal from the access network device, acquires multiple measurement results, determines the top K beam indices, and feeds back the multiple measurement results and the top K beam indices to the access network device.

[0312] Please refer to Figure 11. Figure 11 shows an example of acquiring training data in AI model-based beam management according to this application.

[0313] 901: Optionally, the access network device determines that AI model training data needs to be collected.

[0314] 902: The access network device sends a first piece of information to the UE, which is used by the UE to determine the validity of the candidate training data collected. Optionally, the validity determination may be valid or invalid.

[0315] 903: An access network device sends multiple reference signals to the UE.

[0316] The UE measures multiple reference signals from the access network device to obtain multiple measurement results, i.e., candidate training data. Optionally, the multiple reference signals correspond to the first value described above, for example, 64 beamformed beams.

[0317] Optionally, in application scenario 3, the reference signal may be CSI-RS and / or SSB. In this example, CSI-RS and / or SSB are used by the UE to perform channel measurements.

[0318] 904:UE determines the effectiveness of the collected candidate training data based on the first piece of information.

[0319] 905:UE sends a second piece of information to the access network device, which indicates the result of the validity determination.

[0320] Where possible, the second piece of information indicates that the candidate training data collected by the UE is valid. In an example, the second piece of information may be the first training data, which includes reference signal information corresponding to K optimal measurement results from one or more measurement results, where K is an integer greater than or equal to 1. For example, the reference signal information corresponding to K optimal measurement results may be understood as the indices of the top K beams in Figure 10. In another example, the second piece of information may be the first training data, which includes information on L measurement results from multiple measurement results and L reference signals corresponding to L measurement results. The L measurement results are valid measurement results, i.e., measurement results that satisfy the constraints. In yet another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid.

[0321] In this application scenario, the reference signal may also be replaced with a "beam." This is not limited to this.

[0322] 906: The access network device determines, based on the second piece of information, whether the candidate training data collected by the UE is valid.

[0323] Where possible, the second piece of information indicates that the candidate training data collected by the UE is valid. In this case, the access network device retrieves the first training data from the UE. Furthermore, the access network device trains or updates the AI ​​model based on the first training data, as shown in step 907.

[0324] 907: An access network device trains an AI model based on the first training data and either acquires or updates the AI ​​model.

[0325] In another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid. In a possible implementation in this case, the access network device either maintains beam management for the original AI model or switches to a non-AI model and performs beam management, as shown in step 908.

[0326] 908: The access network device either maintains the original AI model which remains unchanged, or switches to a non-AI model to perform beam management.

[0327] Optionally, in another possible case, the second piece of information indicates that the candidate training data collected by the UE is invalid. In a possible implementation, the access network device determines to recollect the training data, as shown in step 909.

[0328] 909: The access network device determines that it will recollect the training data.

[0329] 910: The access network device sends a third piece of information to the UE, which indicates to the UE that it will recollect the training data for the AI ​​model.

[0330] Optionally, a third piece of information further indicates the maximum number of validity tests k. Alternatively, the maximum number of validity tests k may also be indicated by the first piece of information. This is not limited to this.

[0331] Optionally, when training data for the AI ​​model is recollected, the access network device may update the air interface transmit configuration. Correspondingly, the UE recollects candidate training data for the AI ​​model based on the updated air interface transmit configuration.

[0332] 911: The access network device sends air interface transmit configuration information to the UE, and the air interface transmit configuration information indicates the updated air interface transmit configuration.

[0333] 912:UE recollects the training data for the AI ​​model.

[0334] Similar to the two application scenarios described above, in the process of recollecting candidate training data, the UE is limited by a maximum number of effectiveness checks k. Optionally, the maximum number of effectiveness checks k may be determined by the access network device based on the urgency of training data collection. For example, the urgency criterion may be the time period from the last update of the AI ​​model to the present, or it may be set by the access network device based on the current AI model due to the error in the result of estimating the optimal measurement (or optimal beam). The error in the result of estimating the AI ​​model is determined by splitting the training data collected in previous AI model training into a training set and a validation set. Since the error in the training set is already very low, the error in the estimated result of the validation set is used as a criterion for determining whether the AI ​​model has become significantly invalid. For example, if the error in the result of the access network device predicting information about the reference signal corresponding to the optimal measurement when the UE receives the reference signal, based on the current AI model, is large, for example, above a specified threshold, the requirement may be determined to be urgent. In this case, the maximum number of effectiveness checks k may be set to a large value. Conversely, if the error in the prediction of information about the reference signal corresponding to the optimal measurement result when the UE receives the reference signal, based on the current AI model, is small—for example, smaller than a specified threshold—the requirement may be determined to be non-urgent. In this case, the maximum number of validity determinations k may be set to a small value. In addition, this determination criterion may also be set based on the interval from the last update of the AI ​​model to the present time. For details, please refer to the description of Application Scenario 1. Further details will not be explained again.

[0335] The method for acquiring training data in AI model training provided in this application is applicable to AI model-based beam management scenarios, and it can be seen that this makes it possible to reduce the waste of air interface resources in the AI ​​model training procedure. In addition, this solution avoids contamination of the AI ​​model training dataset caused by the UE by sending invalid training data to the access network device.

[0336] In the method procedure diagrams in Figures 3 to 11, the step sequence numbers are intended merely to illustrate the technical solution of this application and should not constitute a limitation to any particular implementation of the method. These steps may be extended to more steps or combined to fewer steps depending on different specific implementations. This is not limited to these steps. In addition, steps indicated by dashed lines in Figures 3 to 11 are optional steps.

[0337] The above describes in detail the method for acquiring training data in the AI ​​model training provided in this application. Refer to Figure 12 for the same technical concept. This application provides a communication device 1000.

[0338] As shown in Figure 12, the communication device 1000 includes a processing module 1001 and a communication module 1002. The communication device 1000 may be a terminal device, or a communication device used in or by matching with a terminal device that can implement a communication method performed on the terminal device side, such as a chip or circuit. Alternatively, the communication device 1000 may be a network device, or a communication device used on or by matching with a network device that can implement a communication method performed on the network device side, such as a chip or circuit. For example, the network device side may be, for example, an access network device or a positioning device in the method embodiment of this application.

[0339] Communication modules are also sometimes called transceiver modules, transceivers, transceiver machines, or transceiver devices. Processing modules are also sometimes called processors, processing boards, processing units, or processing units. Optionally, communication modules are configured to perform transmit and receive operations on the terminal device side or the network device side in the manner described above. Components configured to implement receive functionality in a communication module may be considered receive units, and components configured to implement transmit functionality in a communication module may be considered transmit units. In other words, a communication module includes both receive and transmit units.

[0340] When the communication device 1000 is used in a terminal device, the processing module 1001 may be configured to implement the processing functions of the terminal device in the embodiments shown in Figures 3 to 11, and the communication module 1002 may be configured to implement the receiving and transmitting functions of the terminal device in the embodiments shown in Figures 3 to 11. Alternatively, the communication device may be understood by referring to the third aspect and possible designs of the third aspect in the summary of the invention.

[0341] When the communication device 1000 is used in a network device, the processing module 1001 may be configured to implement the processing functions of the network device (e.g., an access network device or a positioning device) in the embodiments of Figures 3 to 11, and the communication module 1002 may be configured to implement the receiving and transmitting functions of the network device in the embodiments of Figures 3 to 11. Alternatively, the communication device may be understood by referring to the fourth aspect and possible designs of the fourth aspect in the summary of the invention.

[0342] It should be noted that the first or second network element shown in Figure 3 is specifically a terminal device or network device (e.g., an access network device or a positioning device), which are described in detail in the above method embodiments in various different application scenarios. The fact that the first network element is a terminal device or network device may be understood by referring to specific embodiments, which are not described again in detail herein.

[0343] In addition, it should be noted that communication modules and / or processing modules may be implemented using virtual modules. For example, processing modules may be implemented using software function units or virtual devices, and communication modules may be implemented using software functions or virtual devices. Alternatively, processing modules or communication modules may be implemented using entity devices. For example, if the device is implemented using chip / chip circuitry, the communication module may be input / output circuitry and / or a communication interface, performing input operations (corresponding to the receive operations described above) and output operations (corresponding to the send operations described above). Processing modules are integrated processors, microprocessors, or integrated circuits.

[0344] The modularization in this application is illustrative and merely represents a division into logical functions; actual implementations may involve other divisions. In addition, the functional modules in this example may be integrated into a single processor, and each module may exist physically independently, or two or more modules may be integrated into a single module. The integrated module may be implemented in hardware form or in the form of a software functional module.

[0345] Please refer to Figure 13, based on the same technical concept. This application further provides a communication device 1100. Optionally, the communication device 1100 may be a chip or a chip system. Optionally, in this application, a chip system may include a chip or include a chip and other individual components.

[0346] The communication device 1100 may be configured to implement the functionality of any network element in the communication system described in the above example. The communication device 1100 may include at least one processor 1110. Optionally, the processor 1110 is coupled to memory. The memory may be located within the device. Alternatively, the memory may be integrated with the processor. Alternatively, the memory may be located outside the device. For example, the communication device 1100 may further include at least one memory 1120. The memory 1120 stores computer programs, computer programs or instructions, and / or data necessary to implement any one of the above examples. The processor 1110 may execute the computer programs stored in the memory 1120 to complete the method in any one of the above examples.

[0347] The communication device 1100 may further include a communication interface 1130, which may exchange information with another device via the communication interface 1130. For example, the communication interface 1130 may be a transceiver, a circuit, a bus, a module, a pin, or another type of communication interface. When the communication device 1100 is a chip-type device or circuit, the communication interface 1130 in the device 1100 may alternatively be an input / output circuit that may input information (or referred to as received information) and output information (or referred to as transmitted information). The processor may be an integrated processor, a microprocessor, an integrated circuit, or a logic circuit. The processor may determine output information based on input information.

[0348] In this application, coupling may be an indirect coupling or communication connection between devices, units, or modules in an electrical, mechanical, or other form, used for information exchange between devices, units, or modules. The processor 1110 may operate in cooperation with the memory 1120 and the communication interface 1130. The specific connection medium between the processor 1110, the memory 1120, and the communication interface 1130 is not limited in this application.

[0349] Optionally, as shown in Figure 13, the processor 1110, memory 1120, and communication interface 1130 are connected to each other by using a bus 1140. Bus 1140 may be a peripheral component interconnect (PCI) bus, an extended industry standard architecture (EISA) bus, etc. Buses may be classified as address buses, data buses, control buses, etc. For ease of representation, only one thick line is used to represent buses in Figure 13, but this does not mean that there is only one bus or only one type of bus.

[0350] In this application, the processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array or another programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component that can implement or carry out the methods, steps, and logic block diagrams disclosed herein. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed with reference to this application may be implemented directly by a hardware processor or by a combination of hardware and software modules in the processor.

[0351] In this application, memory may be non-volatile memory, such as a hard disk drive (HDD) or solid-state drive (SSD), or volatile memory, such as random access memory (RAM). Memory may be any other medium that can carry or store expected program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to such medium. Alternatively, memory in this application may be a circuit or any other device that can implement a storage function and is configured to store program instructions and / or data.

[0352] In possible implementations, the communication device 1100 may be used on the network device side, for example, in an access network device or positioning device in the embodiments of this application. In particular, the communication device 1100 may be a network device or a device capable of supporting a network device when implementing the corresponding function on the network device side in any one of the above examples. Memory 1120 stores computer programs (or instructions) and / or data for implementing the function on the network device side in any one of the above examples. Processor 1110 may execute the computer programs stored in memory 1120 to complete the method performed by the network device side in any one of the above examples. When the communication device is used in an access network device, the communication interface in the communication device 1100 may be configured to interact with a terminal device to send information to or receive information from a terminal device. In addition, optionally, the communication interface in the communication device 1100 may be further configured to interact with a core network device, for example, a positioning device (e.g., an LMF network element), to send information to or receive information from the positioning device.

[0353] In another possible implementation, the communication device 1100 may be used as a terminal device. In particular, the communication device 1100 may be a terminal device or a device that can support a terminal device when implementing the functions of a terminal device in any one of the above examples. Memory 1120 stores computer programs (or instructions) and / or data for implementing the functions of a terminal device in any one of the above examples. Processor 1110 may execute the computer programs stored in memory 1120 to complete the methods performed by the terminal device in any one of the above examples. When the communication device is used as a terminal device, the communication interface in the communication device 1100 may be configured to interact with a network device (e.g., an access network device) and to send information to or receive information from the network device.

[0354] The communication device 1100 provided in this example may be used on the network device side (for example, an access network device or a positioning device) to complete a method performed by the network device side, or it may be used on a terminal device to complete a method performed by the terminal device. Therefore, for the technical effects that can be achieved by this embodiment, please refer to the method embodiment described above. Details will not be described again herein.

[0355] Based on the above examples, this application provides a communication system including a network device and a terminal device. In one example, the communication system includes an access network device and a terminal device. In another example, the communication system includes a positioning device, an access network device, and a terminal device. The access network device and terminal device, or the positioning device, access network device, and terminal device, may implement the communication method provided in the examples shown in Figures 3 to 11.

[0356] All or part of the technical solutions provided in this application may be implemented using software, hardware, firmware, or any combination thereof. When software is used to implement an embodiment, all or part of the embodiment may be implemented in the form of a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded onto a computer and executed, all or part of the procedures or functions according to this application are generated. The computer may be a general-purpose computer, a dedicated computer, a computer network, a terminal device, an access network device, or another programmable device. The computer instructions may be stored on a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions may be transmitted wired (e.g., coaxial cable, fiber optic cable, or digital subscriber line (DSL)) or wirelessly (e.g., infrared, radio, or microwave) from one website, computer, server, or data center to another. Computer-readable storage media can be any available medium accessible to a computer, or a data storage device that integrates one or more available media, such as a server or data center. Available media can be magnetic media (e.g., floppy disks, hard disk drives, or magnetic tapes), optical media (e.g., digital video discs (DVDs)), semiconductor media, and so on.

[0357] In this application, cross-referencing is permitted between examples, provided there is no logical inconsistency. For example, cross-referencing may occur between methods and / or terms in method embodiments, between functions and / or terms in apparatus embodiments, and between functions and / or terms in examples of apparatus and examples of methods.

[0358] As used herein, terms such as “component,” “module,” and “system” are used to describe computer-related entities, hardware, firmware, combinations of hardware and software, software, or running software. For example, a component may be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program, and / or a computer. As illustrated by the use of diagrams, both a computing device and an application running on a computing device can be components. One or more components may reside within a process and / or an execution thread, and components may be located on one computer and / or distributed across two or more computers. In addition, these components may be executed from various computer-readable media that store various data structures. For example, components may communicate by using local and / or remote processing, as well as on signals, for example, one or more data packets (e.g., data from two components interacting with another component in a local system, data from two components interacting with another component in a distributed system, and / or data from two components interacting with another component over a network such as the Internet, which interacts with other systems by using signals).

[0359] Those skilled in the art will notice, in combination with the examples described in the embodiments disclosed herein, that units and algorithmic steps may be implemented by electronic hardware or by a combination of computer software and electronic hardware. Whether the function is implemented by hardware or by software depends on the specific application and the design constraints of the technical solution. Those skilled in the art may use different methods to implement the functions described for each specific application, but the implementation should not be considered to be beyond the scope of this application.

[0360] For the sake of a convenient and brief explanation, it will be obvious to those skilled in the art that detailed working procedures of the above systems, apparatus, and units should be referred to in the corresponding procedures in the above-described method embodiments. Further details are not described herein.

[0361] In some embodiments provided in this application, it should be understood that the systems, apparatus, and methods disclosed may be implemented in other ways. For example, the apparatus embodiments described above are schematic. For example, unit division is merely a logical functional division, and other divisions may be possible in actual implementations. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not implemented. In addition, mutual coupling, direct coupling, or communication connection shown or discussed may be implemented by using some interfaces. Indirect coupling or communication connection between apparatus or units may be implemented electronically, mechanically, or in other forms.

[0362] Units described as separate parts may or may not be physically separate, and parts shown as units may or may not be physical units, and may be located in one location or distributed across multiple network units. Some or all of the units may be selected based on the actual requirements to achieve the objectives of the solution of the embodiment.

[0363] In addition, the functional units in the embodiments of this application may be integrated into a single processing unit, or each unit may exist physically independently, or two or more units may be integrated into a single unit.

[0364] When a function is implemented in the form of a software function unit and sold or used as an independent product, the function may be stored on a computer-readable storage medium. Based on such understanding, the technical solutions of this application, or parts of them that contribute to the present art, or parts of the technical solutions, may be implemented in the form of a software product. A computer software product includes several instructions that instruct a computer device (which may be a personal computer, server, or network device) to carry out all or part of the steps of the methods described in embodiments of this application, and which are stored on a storage medium. The storage medium includes any medium capable of storing program code, such as a USB flash drive, a removable hard disk, read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0365] The above description is merely a specific implementation of this application and is not intended to limit the scope of protection of this application. Any modification or substitution readily understood by a person skilled in the art within the scope of the art disclosed in this application falls within the scope of protection of this application. Therefore, the scope of protection of this application must be subject to the scope of protection of the claims.

Claims

1. A method for acquiring training data in artificial intelligence (AI) model training, wherein the method is carried out by a first network element or a chip used in the first network element, A step of receiving first information from a second network element, wherein the first information is for determining the effectiveness of the collected candidate training data, and the result of the effectiveness determination includes valid or invalid. Steps include collecting candidate training data for the AI ​​model, A step of sending second information to the second network element based on the candidate training data and the first information, wherein the second information indicates the determination result of the effectiveness. Methods that include...

2. The method according to claim 1, wherein the second information includes first training data, the second information indicates that the collected candidate training data is valid, and the first training data is valid data in the candidate training data.

3. The method according to claim 1, wherein the second information indicates that the collected candidate training data is invalid.

4. The method according to claim 1, wherein the first information is for determining constraints for determining the effectiveness of the collected candidate training data.

5. The aforementioned method, If it is determined that the candidate training data includes first training data that satisfies the constraints, the step is to determine that the candidate training data is valid, or If it is determined that the candidate training data does not include the first training data that satisfies the constraints, the step of determining that the candidate training data is invalid. The method according to claim 4, further comprising:

6. If the collected candidate training data is invalid, the method A step of receiving third information from the second network element, wherein the third information indicates that candidate training data for the AI ​​model is to be collected again. The method according to claim 3, further comprising:

7. The aforementioned method, A step of determining air interface transmission configuration information, further comprising the step of indicating that the air interface transmission configuration information corresponds to an updated air interface transmission configuration, and that the air interface transmission configuration information collects the candidate training data of the AI ​​model based on the updated air interface transmission configuration, The information regarding the updated air interface transmission configuration is as follows: Transmission power of the reference signal, The number of antenna ports used for the reference signal, Reference signal bandwidth, The frequency domain density of the reference signal, or Reference signal duration The method according to claim 6, which includes one or more of the updates.

8. The method according to claim 6, wherein the third piece of information further indicates the maximum number of times k used to determine the effectiveness, where k is a positive integer.

9. The method according to claim 6, wherein the first information further indicates the maximum number of times k for determining the effectiveness, where k is a positive integer.

10. The aforementioned method, The steps include: collecting candidate training data for the AI ​​model based on the updated air interface transmission configuration; If the maximum number of times k for determining effectiveness is reached, and the result of the k-th effectiveness determination is determined to be invalid based on the first information, the step of collecting the candidate training data for the AI ​​model is stopped. The method according to claim 8, further comprising

11. The aforementioned method, Step 1: If, before the maximum number of times k for determining effectiveness is exceeded, the result of the j-th effectiveness determination is determined to be effective based on the first information, send fourth information to the second network element, wherein the fourth information includes second training data, the fourth information indicates that the result of the j-th effectiveness determination is effective, the second training data includes effective data from the candidate training data on which the j-th effectiveness determination was performed, j is less than or equal to k, and j is a positive integer. The method according to claim 10, further comprising:

12. The step of collecting the candidate training data for the AI ​​model is: A step of measuring a reference signal from the second network element and obtaining one or more measurement results, wherein the candidate training data of the AI ​​model includes the one or more measurement results, or A step of measuring a reference signal from a third network element and obtaining one or more measurement results, wherein the candidate training data of the AI ​​model includes the one or more measurement results. The method according to claim 2, including the method described in claim 2.

13. The first network element is a terminal device or a chip used in the terminal device, and the second network element is an access network device or a chip used in the access network device. The first network element acquires the one or more measurement results when it measures the reference signal from the second network element. The method according to claim 12.

14. The method according to claim 13, wherein the first training data further includes reference signal information or beam information corresponding to K optimal measurement results from among the one or more measurement results, where K is an integer of 1 or more.

15. The first network element is an access network device or a chip used in the access network device, and the second network element is a positioning device or a chip used in the positioning device. When the first network element measures a sounding reference signal from the third network element and obtains one or more measurement results, The first training data further includes location information for one or more of the third network elements. The method according to claim 12.

16. The first network element is a terminal device or a chip used in the terminal device, and the second network element is a positioning device or a chip used in the positioning device. The first network element measures a positioning reference signal from the third network element and obtains one or more measurement results, and the third network element is an access network device. The first training data further includes location information of one or more of the first network elements. The method according to claim 12.

17. The aforementioned constraints Thresholds for quality indicators and criteria for determining the quality indicators, A threshold for the amount of training data that satisfies the quality indicator judgment criteria, and a judgment criterion for the amount of said training data, or Indication information for the maximum duration of training data collection corresponding to a single effectiveness assessment. The method according to claim 4, comprising one or more of the above.

18. The first information described above is Thresholds for quality metrics, quality indicator criteria, Threshold for the amount of training data that satisfies quality indicator criteria, Criteria for determining the amount of training data that meets the quality indicator criteria, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. The method according to claim 4, wherein one or more of the above are represented.

19. The aforementioned constraints are based on the application scenario of the AI ​​model, and the application scenario of the AI ​​model is, AI model-based CSI feedback or CSI prediction, AI model-based positioning, or AI model-based beam management The method according to claim 4, comprising one or more of the above.

20. A method for acquiring training data in AI model training, wherein the method is carried out by a second network element or a chip used in the second network element, A step of sending first information to a first network element, wherein the first information is of an AI model and is for determining the effectiveness of candidate training data collected by the first network element, and the result of the effectiveness determination includes valid or invalid. A step of receiving second information from the first network element, wherein the second information indicates the determination result of the effectiveness, and Methods that include...

21. The method according to claim 20, wherein the second information includes first training data, the second information indicates that the candidate training data collected by the first network element is valid, and the first training data is valid data in the candidate training data.

22. The method according to claim 20, wherein the second information indicates that the candidate training data collected by the first network element is invalid.

23. The method according to claim 20, wherein the first information is for determining constraints for determining the effectiveness of the candidate training data collected by the first network element.

24. If the candidate training data includes first training data that satisfies the constraints, then the candidate training data is valid, or If the candidate training data does not include the first training data that satisfies the constraints, the candidate training data is invalid. The method according to claim 23.

25. If the second information indicates that the candidate training data collected by the first network element is invalid, the method, A step of sending third information to the first network element, wherein the third information indicates to the first network element that candidate training data for the AI ​​model should be collected again. The method according to claim 22, further comprising:

26. The aforementioned method, A step of determining air interface transmission configuration information, further comprising the step of indicating the first network element that the air interface transmission configuration information corresponds to an updated air interface transmission configuration and that the air interface transmission configuration information collects the candidate training data of the AI ​​model based on the updated air interface transmission configuration, The information regarding the updated air interface transmission configuration is as follows: Transmission power of the reference signal, The number of antenna ports used for the reference signal, Reference signal bandwidth, The frequency domain density of the reference signal, or Reference signal duration The method according to claim 25, which includes one or more of the updates.

27. The method according to claim 25, wherein the third piece of information further indicates the maximum number of times k used to determine the effectiveness, where k is a positive integer.

28. The method according to claim 25, wherein the first information further indicates the maximum number of times k used to determine the effectiveness, where k is a positive integer.

29. The aforementioned method, A step of receiving fourth information from the first network element, wherein the fourth information includes second training data, the fourth information indicates that the result of the j-th validity determination performed by the first network element is valid, the second training data is valid data from the candidate training data on which the j-th validity determination was performed, j is less than or equal to k, and j is a positive integer. The method according to claim 27, further comprising:

30. The second network element is an access network device or a chip used in the access network device, and the first network element is a terminal device or a chip used in the terminal device, and the method is The method according to claim 21, further comprising the step of sending a reference signal to a first network element, the reference signal being used by the first network element to obtain one or more measurement results corresponding to the reference signal, and the candidate training data for the AI ​​model including the one or more measurement results.

31. The method according to claim 30, wherein the first training data further includes reference signals corresponding to K optimal measurement results from among the one or more measurement results, where K is an integer of 1 or more.

32. The method according to claim 20, wherein the second network element is a positioning device or a chip used in the positioning device, the first network element is an access network device or a chip used in the access network device, the candidate training data for the AI ​​model includes one or more measurement results and location information of a third network element, and the one or more measurement results are obtained by the first network element by measuring a sounding reference signal sent by the third network element.

33. The method according to claim 20, wherein the second network element is a positioning device or a chip used in the positioning device, the first network element is a terminal device or a chip used in the terminal device, the candidate training data for the AI ​​model includes one or more measurement results and location information of the first network element, the one or more measurement results are based on measurements of a positioning reference signal sent by a third network element, and the third network element is an access network device.

34. The aforementioned constraints Thresholds for quality indicators and criteria for determining the quality indicators, A threshold for the amount of training data that satisfies the quality indicator judgment criteria, and a judgment criterion for the amount of said training data, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. The method according to claim 23, comprising one or more of the above.

35. The first information described above is Thresholds for quality metrics, quality indicator criteria, Threshold for the amount of training data that satisfies quality indicator criteria, Criteria for determining the amount of training data that meets the quality indicator criteria, or Maximum duration of candidate training data collection corresponding to a single effectiveness assessment. The method according to claim 20, wherein one or more of the above are represented.

36. The aforementioned constraints are based on the application scenario of the AI ​​model, and the application scenario of the AI ​​model is, AI model-based CSI feedback or CSI prediction, AI model-based positioning, or AI model-based beam management The method according to claim 23, comprising one or more of the above.

37. A communication device comprising a module configured to implement the method described in any one of claims 1 to 19.

38. A communication device comprising a module configured to implement the method described in any one of claims 20 to 36.

39. A communication device, A processor is provided, the processor is coupled to a memory, and the processor is configured to call computer program instructions stored in the memory to carry out the method according to any one of claims 1 to 19. Communication device.

40. A communication device, A processor is provided, the processor is coupled to a memory, and the processor is configured to call computer program instructions stored in the memory to carry out the method according to any one of claims 20 to 36. Communication device.

41. A communication device comprising a processor and a communication interface, wherein the communication interface is configured to receive data and / or information and to transmit the received data and / or information to the processor, the processor processes the data and / or information, and the communication interface is further configured to output the data and / or information processed by the processor, thereby enabling the communication device to carry out the method according to any one of claims 1 to 19.

42. A communication device comprising a processor and a communication interface, wherein the communication interface is configured to receive data and / or information and to transmit the received data and / or information to the processor, the processor processes the data and / or information, and the communication interface is further configured to output the data and / or information processed by the processor, thereby enabling the communication device to carry out the method according to any one of claims 20 to 36.

43. A computer-readable storage medium, wherein the computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the computer is enabled to carry out the method according to any one of claims 1 to 19.

44. A computer-readable storage medium, wherein the computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the computer is enabled to carry out the method according to any one of claims 20 to 36.

45. A computer program, the computer program includes instructions, and when the instructions are executed on the computer, the computer is enabled to carry out the method according to any one of claims 1 to 19.

46. A computer program comprising instructions, wherein when the instructions are executed on a computer, the computer is enabled to carry out the method described in any one of claims 20 to 36.

47. A communication system comprising a communication device configured to carry out the method described in any one of claims 1 to 19.

48. A communication system comprising a communication device configured to carry out the method described in any one of claims 20 to 36.

Images