Training data collection method and device, communication equipment and readable storage medium

Abstract

The invention provides a training data collection method and device, communication equipment and a readable storage medium, and belongs to the technical field of communication, the training data collection method is applied to first equipment, and the method comprises the following steps: sending data request information; and executing at least one of the following steps: acquiring sample data; obtaining an artificial intelligence AI model or a machine learning ML model; wherein the sample data is obtained according to first label data and first training data; the artificial intelligence AI model or the machine learning ML model is obtained based on the sample data training. According to the scheme of the invention, interaction of the collection process of the training sample data of the AI / ML model in NG-RAN positioning can be realized.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a training data collection method, apparatus, communication equipment and readable storage medium. Background Art

[0002] In R17 / R18, the traditional Next Generation Radio Access Network (NG-RAN) positioning method often relies on the Line of Sight (LOS) environment. The positioning accuracy will drop sharply in the Non Line of Sight (NLOS) environment. R18 began to study the introduction of Artificial Intelligence (AI) / Machine Learning (ML) models in positioning to improve positioning accuracy.

[0003] Introducing AI / ML models in NG-RAN positioning can effectively solve problems such as reduced positioning accuracy in NLOS channel environments. When introducing AI / ML models in positioning technology, the first thing to consider is the problem of model training. The collection of data for model training needs to be completed through the air interface, but the existing positioning-related protocols do not have a design for the collection process of training sample data for AI / ML models. Summary of the invention

[0004] Embodiments of the present invention provide a training data collection method, apparatus, communication device, and readable storage medium to solve the problem of interactive design of a collection process of training sample data without an AI / ML model in existing NG-RAN positioning.

[0005] In order to solve the above technical problems, the embodiments of the present invention provide the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a training data collection method, which is applied to a first device, and the method includes:

[0007] Send data request information;

[0008] And do at least one of the following:

[0009] Get sample data;

[0010] Obtain artificial intelligence AI model or machine learning ML model;

[0011] The sample data is obtained according to the first label data and the first training data; the artificial intelligence AI model or the machine learning ML model is trained based on the sample data.

[0012] In a second aspect, an embodiment of the present invention provides a training data collection method, which is applied to a second device, and the method includes:

[0013] Generate sample data;

[0014] The sample data is obtained according to the acquired first label data and first training data;

[0015] The sample data is used to train an artificial intelligence (AI) model or a machine learning (ML) model.

[0016] In a third aspect, an embodiment of the present invention provides a training data collection method, which is applied to a third device, and the method includes:

[0017] receiving data request information;

[0018] According to the data request information, type information of the first training data and / or first label data are sent.

[0019] In a fourth aspect, an embodiment of the present invention provides a training data collection device, applied to a first device, the device comprising:

[0020] A first sending module, used for sending data request information;

[0021] and a first processing module, the first processing module being configured to perform at least one of the following:

[0022] Get sample data;

[0023] Obtain artificial intelligence AI model or machine learning ML model;

[0024] The sample data is obtained according to the first label data and the first training data; the artificial intelligence AI model or the machine learning ML model is trained based on the sample data.

[0025] In a fifth aspect, an embodiment of the present invention provides a training data collection device, applied to a second device, the device comprising:

[0026] A first generating module, used for generating sample data;

[0027] The sample data is obtained according to the acquired first label data and first training data;

[0028] The sample data is used to train an artificial intelligence (AI) model or a machine learning (ML) model.

[0029] In a sixth aspect, an embodiment of the present invention provides a training data collection device, applied to a third device, the device comprising:

[0030] A first receiving module, used for receiving data request information;

[0031] The second sending module is used to send the type information of the first training data and / or the first label data according to the data request information.

[0032] In a seventh aspect, an embodiment of the present invention provides a communication device, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the training data collection method described in any one of the first aspects, or, when executed by the processor, implements the steps of the training data collection method described in any one of the second aspects, or, when executed by the processor, implements the steps of the training data collection method described in any one of the third aspects.

[0033] In an eighth aspect, an embodiment of the present invention provides a readable storage medium, on which a program is stored. When the program is executed by a processor, the program implements the steps of the training data collection method as described in any one of the first aspect, or implements the steps of the training data collection method as described in any one of the second aspect, or implements the steps of the training data collection method as described in any one of the third aspect.

[0034] The beneficial effects of the present invention are:

[0035] The training data collection method provided by the solution of the present invention can realize the interaction of the collection process of training sample data of the AI / ML model in NG-RAN positioning by sending data request information through the first device and executing at least one of obtaining sample data and obtaining an artificial intelligence AI model or a machine learning ML model. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A schematic diagram showing the auxiliary positioning provided by the present invention;

[0037] Figure 2 A schematic diagram showing the direct positioning provided by the present invention;

[0038] Figure 3 A flowchart showing a training data collection method applied to a first device provided by an embodiment of the present invention;

[0039] Figure 4 A flowchart showing a training data collection method applied to a second device provided by an embodiment of the present invention;

[0040] Figure 5 One of the schematic diagrams showing obtaining sample data provided by an embodiment of the present invention;

[0041] Figure 6A second schematic diagram showing obtaining sample data provided by an embodiment of the present invention;

[0042] Figure 7 A flowchart showing a training data collection method applied to a third device provided by an embodiment of the present invention;

[0043] Figure 8 It represents one of the flow charts of obtaining the first real label as TOA provided by an embodiment of the present invention;

[0044] Fig. 9 The second flowchart of obtaining the first real label as TOA provided by the embodiment of the present invention;

[0045] Fig.10 An overall flow chart showing a training data collection method provided by an embodiment of the present invention;

[0046] Fig.11 A schematic diagram showing the structure of a training data collection device applied to a first device provided in an embodiment of the present invention;

[0047] Fig.12 A schematic diagram showing the structure of a training data collection device applied to a second device provided in an embodiment of the present invention;

[0048] Fig.13 A schematic diagram showing the structure of a training data collection device applied to a third device provided in an embodiment of the present invention;

[0049] Fig.14 A schematic diagram showing the structure of a communication device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the technical problems, technical solutions and advantages to be solved by the application clearer, the following will be described in detail in conjunction with the accompanying drawings and specific embodiments. In the following description, specific details such as specific configurations and components are provided only to help fully understand the embodiments of the application. Therefore, it should be clear to those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the application. In addition, for clarity and brevity, the description of known functions and structures has been omitted.

[0051] It should be understood that the references to "one embodiment" or "an embodiment" throughout the specification mean that the specific features, structures, or characteristics associated with the embodiment are included in at least one embodiment of the present application. Therefore, the references to "in one embodiment" or "in an embodiment" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0052] In the various embodiments of the present application, it should be understood that the size of the serial numbers of the following processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0053] In the embodiments provided in the present application, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0054] In the embodiments of the present invention, the term "and / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent three situations: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0055] Before describing the specific embodiments of the present invention, the following description is first given:

[0056] A new research project "Study on Artificial Intelligence (AI) / Machine Learning (ML) for NR Air Interface" was introduced at the 3GPP RAN#94 plenary meeting, and positioning accuracy enhancement was selected as one of the research use cases. Positioning accuracy enhancement based on AI / ML models requires model training, and the training data is annotated data with real labels.

[0057] Positioning technologies based on AI / ML models are divided into two categories: assisted positioning and direct positioning, such as Figure 1 As shown, assisted positioning is to input channel information into the AI / ML model, the AI / ML model outputs the positioning intermediate variable, and then uses the traditional positioning method to calculate the UE location information; Figure 2 As shown, direct positioning is to input channel information into the AI / ML model, and the AI / ML model directly outputs UE location information.

[0058] The training samples required for AIML model training are {training data, true label}, and each training data needs to correspond to a true label. For direct positioning, the true label is the location information (coordinates) of the UE; for assisted positioning, the true label is the intermediate variable required in positioning, such as Reference Signal Time Difference (RSTD), Time of Arrival (TOA), LOS / NLOS, etc.

[0059] In order to solve the problem of interactive design of the collection process of training sample data without AI / ML model in the existing NG-RAN positioning, the embodiments of the present invention provide a training data collection method, apparatus, communication device and readable storage medium.

[0060] like Figure 3 As shown, an embodiment of the present invention provides a training data collection method, which is applied to a first device, and the method includes:

[0061] Step 301: Send data request information;

[0062] And do at least one of the following:

[0063] Get sample data;

[0064] Obtain artificial intelligence AI model or machine learning ML model;

[0065] The sample data is obtained according to the first label data and the first training data; the artificial intelligence AI model or the machine learning ML model is trained based on the sample data.

[0066] In this step, “the first device obtains sample data” may include at least one of the following:

[0067] receiving sample data sent by the second device;

[0068] Receive sample data sent by other data collection entities;

[0069] The first device generates sample data, specifically, generates (obtains) sample data from first label data and first training data.

[0070] “Obtaining an artificial intelligence (AI) model or a machine learning (ML) model” may include at least one of the following understandings:

[0071] The first device acquires sample data, and performs model training according to the sample data to obtain an AI model ML model.

[0072] The first device receives sample data sent by other devices (such as sample data sent by the second device or other data collection entities), and the first device itself performs model training based on the sample data to obtain an AI model or an ML model;

[0073] The first device acquires sample data and receives sample data sent by other devices, and performs model training according to the sample data to obtain an AI model or an ML model.

[0074] The first device receives an AI model or an ML model trained by another device, wherein the trained AI model or the ML model is obtained by training the other device using sample data acquired by the other device itself;

[0075] The first device receives an AI model or an ML model trained by other devices, wherein the trained AI model or the ML model is trained by sample data sent by a device other than the other devices.

[0076] The first device receives an AI model or an ML model trained by other devices, wherein the trained AI model or the ML model is trained by sample data acquired by the other devices themselves and sent by devices other than the other devices.

[0077] The data collection entity includes at least one of the following:

[0078] First device;

[0079] Network side equipment (including at least one of LMF and gNB);

[0080] Third Party Data Collection Entities.

[0081] In this step, the data request information is sent to the third device, wherein the third device is a positioning service module, which can be used to schedule any device in the first device and the second device, collect sample data, and configure and activate the transmission of the target reference signal. The third device includes: Location Management Function (LMF).

[0082] The first device includes at least one of the following:

[0083] Terminal UE; base station gNB.

[0084] The second device includes at least one of the following:

[0085] Terminal UE; base station gNB; positioning reference unit (PRU); road side unit (RSU).

[0086] It can be understood that the second device is a known terminal device or a terminal device that can obtain a position through a positioning method, and can also be a device that provides sample data for the first device.

[0087] Optionally, the data request information includes at least one of the following:

[0088] The type of the first true label;

[0089] The number of sample data required, where the number of sample data required can be understood as the number of sample data required to train the AI / ML model;

[0090] location-related information of the first device;

[0091] Type information of the first training data;

[0092] The generalization requirement of the sample data, wherein the generalization requirement of the sample data refers to the richness of the sample data. The richer the scenarios between the entities providing the sample data, the better the generalization of the sample data. The generalization requirement of the sample data can be reflected by the number of target devices and / or the number of types of target devices. The target device is a device that transmits a target reference signal. The more the number of target devices and / or the more the number of types of target devices, the more likely the generalization of the sample data will be improved. Specifically, the generalization requirement of the sample data may include at least one of the following: an upper limit value and / or a lower limit value of the number of target devices, an upper limit value and / or a lower limit value of the number of types of target devices;

[0093] The accuracy requirement of the sample data, wherein the accuracy requirement of the sample data can also be understood as the quality requirement of the real label in the first training data, and the accuracy requirement of the sample data can be reflected by at least one of the following: the type of the second device and / or the target device (the target device is the device that transmits the target reference signal), confidence-related information, it should be noted that the accuracy of the sample data provided by the device with a fixed position is relatively high, such as the PRU with a fixed position, and the confidence-related information such as the confidence of the device position, etc.;

[0094] The requirement information for configuration of a target reference signal for obtaining the first training data, the requirement information for configuration of the target reference signal is used to determine the configuration of the target reference signal, the target reference signal is used to obtain the first training data, it should be noted that the target reference signal is a reference signal used for positioning measurements between target entities, considering the type of equipment transmitting the target reference signal, the target reference signal includes at least one of the following: a downlink positioning reference signal (DownLink Positioning Reference Signal, DL-PRS), an uplink channel sounding reference signal (UpwnLinkSounding Reference Signal, UL-SRS) and a sidelink positioning reference signal (SidewnLinkPositioning Reference Signal, SL-PRS).

[0095] Optionally, the type of the first real label includes at least one of the following:

[0096] Position information, which may be pre-calibrated position information or position information obtained using a pre-set positioning method, and is used for direct positioning;

[0097] First arrival time TOA value, the first TOA value is the propagation time of the signal between the source and the receiving end. In AI-assisted positioning, the TOA tag (first TOA value) is the signal transmission time under the LOS path, which can be obtained by transmitting the location information of the target reference signal device. The first TOA value is used to assist positioning;

[0098] Line-of-sight LOS indication or non-line-of-sight NLOS indication (i.e., LOS / NLOS indicator), this label indicates whether there is a visible path between the entities sending and receiving the target reference signal. It can be obtained through the relationship between the actual propagation time of the target reference signal and the above-mentioned TOA label. The LOS / NLOS indicator is used to assist positioning.

[0099] Optionally, the location-related information of the first device includes at least one of the following:

[0100] location information of the first device;

[0101] First timestamp information related to the location information of the first device, wherein the first timestamp information is used for subsequent matching of first label data and first training data to obtain sample data;

[0102] Confidence information related to the location of the first device.

[0103] Optionally, the type information of the first training data includes at least one of the following:

[0104] The type of the first measurement quantity in the first training data, such as a channel impulse response (CIR), a power delay profile (PDP), etc.;

[0105] The dimension (size) of the first measurement quantity in the first training data, wherein when the second device collects sample data for the first device, the first device needs to indicate the dimension (size) of the first measurement quantity. The first measurement quantity may be in matrix form, and the dimension of the first measurement quantity represents the size of the matrix. For example, the matrix of the first measurement quantity is 256*2, wherein 256 represents the number of time domain sampling points of the channel estimation contained in the first measurement quantity.

[0106] The first device may have corresponding requirements for the configuration of the target reference signal according to the needs of the AI / ML model (because different target reference signal configurations may obtain different sample data, resulting in different AI / ML models at the training location). Optionally, the requirement information for the configuration of the target reference signal (also referred to as information related to the configuration of the target reference signal) includes at least one of the following:

[0107] a bandwidth for transmitting the target reference signal;

[0108] The time-frequency density information of the target reference signal, wherein the time-frequency density information of the target reference signal includes the number of symbols occupied by the target reference signal in the time domain or the comb in the frequency domain.

[0109] Optionally, the first training data is obtained by measuring according to the data request information and the target reference signal, and the first training data includes at least one of the following:

[0110] A first measurement quantity, wherein the first measurement quantity is obtained based on channel estimation or channel estimation transformation;

[0111] second timestamp information related to the first measurement;

[0112] Device identification information of a first target device for sending a target reference signal, wherein the device identification information may include an ID or identity information of the device, and the device identification information is used for subsequent matching of first label data and first training data to obtain sample data. The first target device is a device for sending a target reference signal, and may include at least one of the following: gNB, UE, PRU, RSU;

[0113] In an optional embodiment, the method further includes:

[0114] When the first condition is met, data termination information is sent, where the data termination information is used to instruct the third device to perform a preset operation, where the preset operation includes at least one of the following:

[0115] Stop sending the first tag data;

[0116] A device for transmitting a target reference signal is instructed to stop transmission of the target reference signal.

[0117] The first condition includes at least one of the following:

[0118] The sample data meets the required quantity of sample data.

[0119] The first device stops collecting sample data due to other reasons;

[0120] like Figure 4As shown, an embodiment of the present invention further provides a training data collection method, which is applied to a second device, and the method includes:

[0121] Step 401: Generate sample data;

[0122] Wherein, the sample data is obtained according to the acquired first label data and first training data;

[0123] The sample data is used to train an artificial intelligence (AI) model or a machine learning (ML) model.

[0124] In this embodiment, the first label data and the first training data are matched to obtain sample data.

[0125] Before generating the sample data, the method further includes:

[0126] In an optional embodiment, configuration information for generating sample data is obtained, where the configuration information for generating sample data includes:

[0127] Type information of the first training data;

[0128] The type information of the first training data includes at least one of the following:

[0129] a type of a first measurement quantity in the first training data;

[0130] The dimension of the first measurement quantity in the first training data.

[0131] In an optional embodiment, after the second device generates the sample data, the method further includes:

[0132] The sample data is sent.

[0133] It should be noted that, in this optional embodiment, the second device sends the sample data to the first device and / or other data collection entities.

[0134] The first device includes at least one of the following:

[0135] Terminal UE; base station gNB.

[0136] The second device includes at least one of the following:

[0137] Terminal UE; base station gNB; PRU; RSU.

[0138] It should be noted that, when the first device is a terminal UE and the second device is also a terminal UE, or when the first device is a base station gNB and the second device is also a base station gNB, it can be understood that all embodiments of the training data collection method applied to the first device side provided in the embodiments of the present invention can also be executed by the second device, and all embodiments of the training data collection method applied to the second device side can also be executed by the first device, that is, any terminal UE or any base station gNB can be used as the first device, and can also be used as the second device to execute the training data collection method on the corresponding side, and can achieve the same or similar technical effects. In the case where the second device is a PRU or RSU, the first device is a terminal UE or a base station gNB, then all embodiments of the training data collection method applied to the first device side can only be implemented by the terminal UE or the base station gNB, and all embodiments of the training data collection method applied to the second device side can only be implemented by the PRU or RSU, that is, the PRU or RSU cannot implement all embodiments of the training data collection method applied to the first device side, but the terminal UE or the base station gNB can implement all embodiments of the training data collection method applied to the first device and the second device.

[0139] The first tag data is response information to the data request information, that is, the first tag data is obtained by the third device according to the data request information sent by the first device.

[0140] Optionally, the first tag data includes at least one of the following:

[0141] First true label;

[0142] Device identification information of a second target device for sending a target reference signal, wherein the device identification information includes a device ID and / or device identity information, and the device identification information is used for subsequent matching of the first label data and the first training number to obtain sample data. The second target device is a device for sending a target reference signal, and may include at least one of the following: gNB, UE, PRU, RSU;

[0143] The third timestamp information related to the first real label is used for subsequent matching of the first label data and the first training data to obtain sample data.

[0144] The first training data is obtained by measuring according to the data request information and the target reference signal.

[0145] Optionally, the first training data includes at least one of the following:

[0146] A first measurement quantity, where the first measurement quantity is obtained based on channel estimation or channel estimation transformation;

[0147] second timestamp information related to the first measurement;

[0148] Device identification information of a first target device used to send a target reference signal;

[0149] Second time of arrival TOA value.

[0150] In an optional embodiment, the process of obtaining the first training data includes:

[0151] A target reference signal is received, where the target reference signal is a reference signal sent by a first target device and used to obtain first training data, and the target reference signal is measured according to type information of the first training data to obtain the first training data.

[0152] The first measurement quantity includes at least one of the following:

[0153] Channel estimation value H;

[0154] Time information of channel estimation;

[0155] Power information of channel estimation;

[0156] Phase information for channel estimation.

[0157] In an optional embodiment, the first training data includes a second TOA value;

[0158] The second TOA value is obtained by at least one of the following methods:

[0159] It is obtained by measuring the target reference signal, that is, by non-AI / ML model methods;

[0160] It is obtained by using a trained target model and a first measurement quantity, wherein the target model includes an AI model and / or an ML model, and the first measurement quantity is obtained by measuring the target reference signal according to the type information of the first training data, that is, the second TOA value is obtained by the AI / ML model.

[0161] In an optional embodiment, when the data request information includes the type of the first real label, and the type of the first real label is a line-of-sight LOS indication or a non-line-of-sight NLOS indication (LOS / NLOS indicator), the device receiving the target reference signal (i.e., the second device in this embodiment) obtains the LOS indication or the NLOS indication based on the first TOA value and the second TOA value; wherein the first TOA value is obtained based on the location information of the second device and the location information of the first target device used to send the target reference signal. Specifically, the second device obtains the LOS / NLOS indicator label based on the relationship between the first TOA value and the second TOA value, merges the LOS / NLOS indicator label with other information in the first label data, and obtains the first label data {LOS / NLOS indicator, device identification information, third timestamp information} required by the second device.

[0162] It should be noted that, in this embodiment, the data request information is sent by the first device to the third device, and the third device configures the second device to perform the operation of this optional embodiment.

[0163] In an optional embodiment, generating sample data includes:

[0164] According to the device identification information of the first target device in the first training data and the device identification information of the second target device in the first label data, and / or, the second timestamp information in the first training data and the third timestamp information in the first label data, the first label data and the first training data are matched to obtain the sample data, that is, according to the device identification information and / or timestamp information in the first training data and the first label data, the first training data and the first label data are paired to obtain sample data {first measurement quantity, first true label}, and the obtained sample data is the data for training the AI / ML model.

[0165] In this optional embodiment, the following implementation modes are provided, that is, according to the device identification information of the first target device in the first training data and the device identification information of the second target device in the first label data, and / or, the second timestamp information in the first training data and the third timestamp information in the first label data, the first label data and the first training data are matched to obtain the sample data, including at least one of the following:

[0166] Embodiment 1: When the positions of the device sending the target reference signal and the device receiving the target reference signal are both fixed, the first label data and the first training data are matched according to the device identification information of the first target device and the device identification information of the second target device to obtain the sample data;

[0167] Specifically, when the device identification information of the first target device is consistent with the device identification information of the second target device, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

[0168] In this embodiment, the entities that send and receive the target reference signal are fixed, and the second device matches the device identification information of the first target device with the device identification information of the second target device. If the device identification information is the same, sample data {first measurement quantity, first true label} is obtained.

[0169] Embodiment 2: When the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, receive the updated first label data sent by the third device according to the position information of the third target device corresponding to the second timestamp information, and match the updated first label data with the first training data according to the device identification information of the second target device and the device identification information of the first target device in the updated first label data, and / or, according to the third timestamp information in the updated first label data and the second timestamp information, to obtain the sample data; wherein the position information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the position information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a pre-set positioning method, and the third target device is a device with a not fixed position among the device sending the target reference signal and the device receiving the target reference signal;

[0170] Specifically, when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device, and when the third timestamp information in the updated first label data is consistent with the second timestamp information, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

[0171] Exemplarily, if the position of the third target device is not fixed, the third target device needs to report the position information of the third target device at the moment of sending and / or receiving the target reference signal (i.e., the second timestamp information), and then receive the updated first label data {first real label, device identification information, third timestamp information} sent by the third device according to the position information of the third target device corresponding to the second timestamp information, and then the second device matches the device identification information of the second target device in the updated first label data with the device identification information of the first target device, and matches the third timestamp information in the updated first label data with the second timestamp information. If both matches are successful, the sample data {first measurement quantity, first real label} is obtained.

[0172] The specific method is as follows:

[0173] Method 1: When the third target device is a UE, the second device obtains the first training data {first measurement quantity, device identification information, second timestamp information Tb}, the third target device updates the location information corresponding to the second timestamp information Tb reported, the third device obtains the updated first label data {first true label, set identification information, third timestamp information Tb} according to the location information corresponding to the updated second timestamp information Tb reported, and provides feedback, the second device matches the device identification information of the second target device in the updated first label data with the device identification information of the first target device, and matches the third timestamp information in the updated first label data with the second timestamp information. If both are successful, the sample data {first measurement quantity, first true label} is obtained.

[0174] Method 2: When the second device is a gNB or RSU, the process is as follows Figure 5 As shown, the second device takes gNB as an example, and other third target devices (including UE and / or PRU, Figure 5 Taking UE as an example), a target reference signal UL-PRS is sent to the third target device (sent at time Tb), and the corresponding updated location information is reported at the same time, or the third device obtains the location information (x2, y2) and timestamp information Tb through a positioning method. The third device obtains the first label data {first true label (TOA2), device identification information (UE ID), second timestamp information Tb} according to (x2, y2) and the location information of the second device, and feeds it back to the second device. The second device obtains the first training data according to the target reference signal. The second device matches the device identification information in the first label data with the device identification information in the first training data, and matches the timestamp information in the first label data with the timestamp information in the first training data. If both matches are successful, the sample data {first measurement quantity, first true label} is obtained.

[0175] Implementation method three: When the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, the first label data and the first training data are matched according to at least one item of the device identification information of the second target device, the second timestamp information, the device identification information of the first target device and the third timestamp information to obtain the sample data.

[0176] Specifically, implementation mode 3 includes at least one of the following methods:

[0177] Mode 1: When the device identification information of the second target device is consistent with the device identification information of the first target device, and the time indicated by the second timestamp information is later than the time indicated by the third timestamp information, the first true label in the first label data and the first measurement in the first training data are matched to obtain the sample data;

[0178] Specifically, the second device pairs the first training data {first measurement quantity B1, device identification information, second timestamp information Tb} with the first label data {first true label A1, device identification information, third timestamp information Ta}. If the device identification information is the same and Tb≥Ta, the sample data {first measurement quantity B1, first true label A1} is obtained.

[0179] Method 2: determining the changed position information of the third target device, sending the changed position information to the third device, receiving the updated first label data sent by the third device according to the changed position information, and matching the first true label in the first label data with the first measurement in the first training data when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the updated first label data, to obtain the sample data;

[0180] Specifically, when the third target device is a UE, the changed location information may be location information that has been reported in the data request and obtained after updating. When the third target device is a UE or a PRU, if no data request is initiated, the changed location information may be re-reported location information. The third device sends updated first label data {first true label A2, device identification information, third timestamp information Ta'} based on the changed location information. The second device pairs the first training data {first measurement quantity B1, device identification information, second timestamp information Tb} with the updated first label data {first true label A2, device identification information, third timestamp information Ta'}. If the device identification information is the same and Tb≥Ta', the sample data {first measurement quantity B1, first true label A2} is obtained.

[0181] Method three: determine the change position information of the third target device, send the change position information to the third device, receive the updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is later than or equal to the time indicated by the second timestamp information in the first training data and the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the first label data, match the first true label in the first label data and the first measurement in the first training data to obtain the sample data;

[0182] Specifically, the first label data {first real label A1, device identification information, third timestamp information Ta} sent by the third device is received, the change position information of the third target device is determined, and the updated first label data {first real label A2, device identification information, third timestamp information Ta'} sent by the third device according to the change position information is received. The second device pairs the first training data {first measurement quantity B1, device identification information, second timestamp information Tb}, the updated first label data {first real label A2, device identification information, third timestamp information Ta'} and the first label data {first real label A1, device identification information, third timestamp information Ta}. If the device identification information is the same and Ta<=Tb<=Ta', the first measurement quantity B1 is paired with the first real label A1 to obtain the sample data {first measurement quantity B1, first real label A1}.

[0183] Method 4: Determine the change position information of the third target device, send the change position information to the third device, receive the updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is earlier than the time indicated by the second timestamp information in the first training data, match the first true label in the updated first label data and the first measurement in the first training data to obtain the sample data.

[0184] Specifically, the first label data {first real label A1, device identification information, third timestamp information Ta} sent by the third device is received, the change position information of the third target device is determined, and the updated first label data {first real label A2, device identification information, third timestamp information Ta'} sent by the third device according to the change position information is received. The second device pairs the first training data {first measurement quantity B1, device identification information, second timestamp information Tb}, the updated first label data {first real label A2, device identification information, third timestamp information Ta'} and the first label data {first real label A1, device identification information, third timestamp information Ta}. If the device identification information is the same and Tb>Ta', the first measurement quantity B1 is paired with the first real label A2 to obtain the sample data {first measurement quantity B1, first real label A2}.

[0185] The relationship between the timestamps in the above method 3 and method 4 is as follows Figure 6 As shown, when the third target device is UE as an example, and the second device is gNB as an example, the third target device UE reports location information (x1, y1) and third timestamp information Ta, the third device obtains the first label data {first true label TOA1, device identification information, third timestamp information Ta} according to the location information (x1, y1) and the location information of the second device, the third device sends the first label data to the second device, the third target device sends the target reference signal to the second device, obtains the first training data {first measurement quantity, device identification information (UE ID), second timestamp information Tb}, the third target device UE reports the updated location information (x2, y2) and third timestamp information Ta', the third device obtains the first label data {first true label TOA2, device identification information, third timestamp information Ta'} according to the location information (x2, y2) and the location information of the second device, and sends the first label data to the second device.

[0186] It should also be noted that, when the second device is a gNB, the device sending the target reference signal to the second device (gNB) includes at least one of the following:

[0187] UE;PRU.

[0188] When the second device is a UE or a PRU, the device sending the target reference signal to the second device includes at least one of the following:

[0189] UE;PRU;gNB;RSU.

[0190] In an optional embodiment, the method further includes:

[0191] The sample data is sent to a data collection entity.

[0192] like Figure 7 As shown, an embodiment of the present invention further provides a training data collection method, which is applied to a third device, and the method includes:

[0193] Step 701: Receive data request information.

[0194] In this step, the third device receives the data request information sent by the first device.

[0195] Step 702: Send type information of first training data and / or first label data according to the data request information.

[0196] In this step, the third device sends the type information of the first training data and / or the first label data to the first device and / or the second device. When the third device sends the type information of the first training data and the first label data, the type information of the first training data is sent first, and then the first label data is sent.

[0197] The data request information includes at least one of the following:

[0198] The type of the first true label;

[0199] The required quantity of sample data;

[0200] location-related information of the first device;

[0201] Type information of the first training data;

[0202] Generalization requirements of sample data;

[0203] Accuracy requirements for sample data;

[0204] The requirement information for obtaining configuration of a target reference signal for the first training data is used to determine the configuration of a target reference signal used to obtain the first training data.

[0205] The first tag data includes at least one of the following:

[0206] First true label;

[0207] Device identification information of the second target device used to send the target reference signal

[0208] The third timestamp information related to the first real label.

[0209] In an optional embodiment, after receiving the data request information, the method further includes:

[0210] According to the data request information, a second device is determined, and the transmission of a target reference signal is configured and activated; wherein the second device is a device for generating sample data, and the sample data is used to train an artificial intelligence AI model or a machine learning ML model.

[0211] In an optional embodiment, when the data request information includes the type of the first real label, and the type of the first real label is location information, the first real label in the first label data is obtained by at least one of the following methods:

[0212] Receiving location information sent by the second device, wherein the location information of the second device is reported in real time, or the location information of the second device needs to be re-reported after being updated;

[0213] The location information obtained by a preset positioning method, wherein the preset positioning method is not limited in this embodiment, and needs to be re-acquired after the location information of the second device is updated;

[0214] The location information of the second device is acquired in advance, wherein the location information of the second device (when the second device is a PRU) is known by the third device.

[0215] The first real label (the type of the first real label is location information) includes at least one of the following:

[0216] Location information;

[0217] fourth timestamp information associated with the location information;

[0218] Confidence information associated with the location information.

[0219] In an optional embodiment, when the data request information includes the type of the first real label, and the type of the first real label is a first arrival time TOA value (expressed as TOA_1), the first real label in the first label data is obtained based on the location information of the second device and the location information of the first target device; wherein the first target device is a device that sends a target reference signal.

[0220] Combine the following Figure 8 , taking the second device as gNB as an example, the overall process of obtaining the first real label as TOA is described:

[0221] 1. The first device (taking gNB as an example) sends data request information to the third device (LMF), where the data request information includes at least one of the following:

[0222] First true label type (TOA);

[0223] The required amount of sample data;

[0224] Generalization requirements of sample data;

[0225] Configuration information of the target reference signal.

[0226] 2. TRP configuration information interaction: After receiving the data request information, the LMF determines the available second device according to the data request information, configures and activates the transmission of UL-SRS;

[0227] If the second device is a UE, the location information, timestamp information related to the location information, and confidence information are reported to the LMF as needed, or the LMF obtains the location information of the UE through a positioning method;

[0228] If the second device is a PRU, proceed directly to the next step.

[0229] 3-7. Positioning capability interaction: the second device starts transmission of UL-SRS according to the resource configuration of UL-SRS; the third device sends positioning request information to the first device; the first device determines the transmission resources of UL-SRS; the first device configures UE SRS for the second device; the first device sends a positioning request information response to the third device; the third device sends positioning activation request information to the first device; the first device activates the transmission of UE SRS; the first device sends a positioning activation response to the third device.

[0230] 8. After receiving the UL-SRS sent by the second device, the gNB obtains the first training data {first measurement amount, second timestamp information Tb, device identification information of the second device} through measurement.

[0231] 9. The second device reports the location information {location coordinates, timestamp Ta} or the location information of the second device (PRU) known to LMF. LMF calculates the TOA tag between the second device and the first device (gNB) based on the location information of the second device. The TOA tag is the signal transmission time under the LOS path, that is, TOA = d / c, d is the straight-line distance between the second device and gNB, and c is the speed of light.

[0232] 10. Feedback the first tag data {first true tag (TOA), device identification information of the second device, and third timestamp Ta} to the gNB.

[0233] 11. The gNB matches the first training data and the first label data to obtain sample data {first measurement amount, TOA}.

[0234] 12. When the sample data collected by the first device meets the required quantity, the first device sends a data termination request message to the third device.

[0235] 13. After receiving the data termination request information sent by gNB, LMF sends a deactivation message to gNB or releases all UL-SRS resources.

[0236] Combine the following Fig. 9 , taking the second device as UE as an example, the overall process of obtaining the first real label as TOA is described:

[0237] 1. The first device (taking UE as an example) sends data request information to the third device (LMF), wherein the data request information includes at least one of the following:

[0238] the type of the second device;

[0239] location information of the first device, where the location information may be obtained by a New Radio Access Technology (NR-RAT) method or a non-NR-RAT method;

[0240] A timestamp Ta corresponding to the location of the first device;

[0241] confidence information corresponding to the location of the first device;

[0242] First tag type (TOA);

[0243] A lower limit (threshold) on the number of second devices;

[0244] The quantity of sample data required;

[0245] The bandwidth of the transmission target reference signal;

[0246] The configuration information of the target reference signal includes at least one of the following: comb of the target reference signal; transmission bandwidth of the target reference signal; configuration information of the target reference signal;

[0247] The type of sample data;

[0248] The size of the sample data (referring to the dimension of the first measurement quantity).

[0249] 2. TRP configuration information interaction: After receiving the data request information, the third device determines the available second device according to the request information, configures and activates the transmission of DL-PRS;

[0250] If the second device is a UE, the location information, timestamp information related to the location information, and confidence information are reported to the LMF as needed, or the LMF obtains the location information of the UE through a positioning method;

[0251] If the second device is a PRU, proceed directly to the next step.

[0252] 3-10. Positioning capability interaction; the third device sends a target reference signal transmission request to the first device; the second device reports location information to the third device; the first device determines a DL-PRS transmission resource; the first device transmits DL-PRS to the second device; the first device sends a target reference signal transmission response to the third device; the first device sends a first auxiliary information request to the third device; the first device receives the first auxiliary information sent by the third device.

[0253] 11. After receiving the DL-PRS, the first device and / or the second device obtains the first training data {first measurement value, second timestamp information Tb, device identification information} through measurement.

[0254] 12.LMF calculates the TOA tag between the first device and / or the second device and the third device based on the location information of the first device and / or the second device. The TOA tag is the signal transmission time under the LOS path (whether or not the LOS path is included), that is, TOA = d / c, d is the straight-line distance between the first device and / or the second device and the gNB, and c is the speed of light. Feedback the first tag data {first real tag (TOA), device representation information, third timestamp Ta} to the first device and / or the second device.

[0255] 13. The first device and / or the second device matches the first training data and the first label data to obtain sample data {first measurement quantity, first true label}.

[0256] 14. The second device transmits the sample data to the first device or other data collection entity.

[0257] 15. The first device sends data termination request information to the second device.

[0258] 16. The second device receives the data termination request information and sends a target reference signal transmission request to the first device, wherein the target reference signal transmission request includes at least one of the following:

[0259] PRS Configuration Request Type("config" or "off")

[0260] PRS TRP Item: TRP ID and / or Requested DL PRS TransmissionCharacteristics.

[0261] The first device stops reference signal transmission according to the target reference signal transmission request.

[0262] In an optional embodiment, when the data request information includes the type of the first real label, and the type of the first real label is a line-of-sight LOS indication or a non-line-of-sight NLOS indication (LOS / NLOS indicator), the device receiving the target reference signal (i.e., the second device in this embodiment) obtains the LOS indication or the NLOS indication based on the first TOA value and the second TOA value; wherein the first TOA value is obtained based on the location information of the second device and the location information of the first target device used to send the target reference signal. Specifically, the second device obtains the LOS / NLOS indicator label based on the relationship between the first TOA value and the second TOA value, merges the LOS / NLOS indicator label with other information in the first label data, and obtains the first label data {LOS / NLOS indicator, device identification information, third timestamp information} required by the second device.

[0263] The following takes the first device as a gNB as an example to specifically describe the process when the first real tag type is LOS / NLOS:

[0264] The first device (gNB) sends data request information to the third device (LMF), where the data request information includes at least one of the following:

[0265] First real label type (LOS / NLOS indicator);

[0266] Second, the number of equipment required;

[0267] The required quantity of sample data;

[0268] Information related to the target reference signal configuration.

[0269] After the LMF receives the data request message, it determines the available second device according to the data request message, configures and activates the transmission of UL-SRS;

[0270] Among them, if the second device is a UE, it is necessary to report location information, timestamp information related to the location information, and confidence information to the LMF, or the LMF obtains the location information of the UE through a positioning method;

[0271] If the second device is a PRU, directly execute the next step.

[0272] The second device starts the transmission of UL-SRS according to the resource configuration of UL-SRS;

[0273] After the first device receives the UL-SRS sent by the second device, it measures and obtains the first training data {first measurement quantity, second timestamp information Tb, device identification information, TOA information (TOA_1)};

[0274] Among them, TOA_1 can be obtained through an AI / ML model; and / or,

[0275] TOA_1 can be obtained through a non-AI / ML method.

[0276] The LMF calculates the TOA between the second device and the first device according to the location information {location coordinates, timestamp Ta} reported by the second device or the location information of the second device (PRU) known to the LMF. The TOA label is the signal transmission time under the LOS path, that is, TOA = d / c, where d is the straight-line distance between the fourth device and the gNB, and c is the speed of light. And the first label data {first true label (TOA_2), fourth device ID or identity information, timestamp Ta} is fed back to the first device.

[0277] The second device pairs according to the timestamp information and / or device identification information in the first training data set {first measurement quantity, second timestamp information Tb, device ID or identity identification information, TOA information (TOA_1)} and the first label data {first true label (TOA_2), device identification information, third timestamp information Ta}. If the pairing is successful, sample data is obtained, and then:

[0278] The LOS / NLOS indicator is obtained by comparing TOA_1 and TOA_2, including but not limited to the following method: TOA_1 - TOA_2 < TH (threshold), it is determined as LOS, otherwise it is determined as NLOS. The TH value depends on the implementation.

[0279] The first device obtains updated first label data {LOS / NLOS indicator, device identification information, third timestamp information Ta} according to the first label data and the LOS / NLOS determination result.

[0280] In an optional embodiment, the method further includes:

[0281] Sending updated first label data to the second device according to the location information of the third target device corresponding to the second timestamp information in the first training data; wherein the location information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the location information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a preset positioning method; the first training data is obtained by measuring according to a target reference signal; the third target device is a device with a non-fixed position between a device sending the target reference signal and a device receiving the target reference signal;

[0282] In an optional embodiment, the method further includes:

[0283] Receive the change location information sent by the third target device

[0284] Sending updated first tag data according to the changed position information;

[0285] The third target device is a device whose position is not fixed among the device that sends the target reference signal and the device that receives the target reference signal.

[0286] Combine the following Fig.10 , specifically describes the overall process of the training data collection method provided by the embodiment of the present invention:

[0287] The first device sends data request response information to the third device;

[0288] The third device determines the second device according to the data request response information, and configures target reference signal transmission;

[0289] The third device sends configuration information (measurement information) of the target reference signal to the first device and / or the second device;

[0290] The first device and / or the second device performs measurement according to the target reference signal to obtain first training data;

[0291] The third device sends response information of the data request information to the first device and / or the second device, wherein the response information includes the first tag data;

[0292] The first device and / or the second device performs data annotation (data matching) according to the first training data and the first label data to obtain sample data.

[0293] The sample data of the second device is sent to the first device. After the sample data collected by the first device meets the requirements, the data termination request information is sent to the third device. The third device feeds back the response information of the data termination request information and stops the transmission of the target reference signal.

[0294] The embodiment of the present invention utilizes the signaling and processes in the existing air interface to implement the labeling of AI / ML assisted positioning model training data, with small computational complexity and low complexity. It clarifies the process of collecting AI / ML model training data and provides a feasible solution for introducing AI / ML into air interface positioning.

[0295] like Fig.11 As shown, an embodiment of the present invention further provides a training data collection device, which is applied to a first device, and the device includes:

[0296] The first sending module 1101 is used to send data request information;

[0297] and a first processing module 1102, the first processing module being configured to perform at least one of the following:

[0298] Get sample data;

[0299] Obtain artificial intelligence AI model or machine learning ML model;

[0300] The sample data is obtained according to the first label data and the first training data; the artificial intelligence AI model or the machine learning ML model is trained based on the sample data.

[0301] Optionally, the data request information includes at least one of the following:

[0302] The type of the first true label;

[0303] The required quantity of sample data;

[0304] location-related information of the first device;

[0305] Type information of the first training data;

[0306] Generalization requirements of sample data;

[0307] Accuracy requirements for sample data;

[0308] The requirement information for obtaining configuration of a target reference signal for the first training data is used to determine the configuration of a target reference signal used to obtain the first training data.

[0309] Optionally, the type of the first real label includes at least one of the following:

[0310] Location information;

[0311] First arrival time TOA value;

[0312] Line-of-sight (LOS) indication or non-line-of-sight (NLOS) indication.

[0313] Optionally, the location-related information of the first device includes at least one of the following:

[0314] location information of the first device;

[0315] first timestamp information related to the location information of the first device;

[0316] Confidence information related to the location of the first device.

[0317] Optionally, the type information of the first training data includes at least one of the following:

[0318] a type of a first measurement quantity in the first training data;

[0319] The dimension of the first measurement quantity in the first training data.

[0320] Optionally, the first training data includes at least one of the following:

[0321] A first measurement quantity;

[0322] second timestamp information related to the first measurement;

[0323] Device identification information of a first target device used to send a target reference signal;

[0324] The second TOA value.

[0325] It should be noted that the training data collection device applied to the first device provided in an embodiment of the present invention is a device capable of executing the above-mentioned training data collection method applied to the first device. All embodiments of the above-mentioned training data collection method applied to the first device are applicable to the device and can achieve the same or similar technical effects.

[0326] like Fig.12 As shown, an embodiment of the present invention further provides a training data collection device, which is applied to a second device, and the device includes:

[0327] The first generating module 1201 is used to generate sample data;

[0328] The sample data is obtained according to the acquired first label data and first training data;

[0329] The sample data is used to train an artificial intelligence (AI) model or a machine learning (ML) model.

[0330] Optionally, the first training data includes at least one of the following:

[0331] A first measurement quantity;

[0332] second timestamp information related to the first measurement;

[0333] Device identification information of a first target device used to send a target reference signal;

[0334] Second time of arrival TOA value.

[0335] Optionally, the device further comprises:

[0336] A first signal receiving module, configured to receive a target reference signal, where the target reference signal is a reference signal sent by a first target device and used to obtain first training data;

[0337] The first measurement module is used to measure the target reference signal according to the type information of the first training data to obtain the first training data.

[0338] Optionally, the first measurement includes at least one of the following:

[0339] channel estimation value;

[0340] Time information of channel estimation;

[0341] Power information of channel estimation;

[0342] Phase information for channel estimation.

[0343] Optionally, the first training data includes a second TOA value;

[0344] The second TOA value is obtained by at least one of the following methods:

[0345] Measured based on a target reference signal;

[0346] obtained by using a trained target model and a first measurement amount, wherein the target model includes an AI model and / or an ML model, and the first measurement amount is obtained by measuring the target reference signal according to type information of first training data;

[0347] Optionally, the device further comprises:

[0348] A second processing module, configured to obtain the LOS indication or the NLOS indication according to the first TOA value and the second TOA value when the data request information includes the type of the first real label and the type of the first real label is a line-of-sight LOS indication or a non-line-of-sight NLOS indication;

[0349] The first TOA value is obtained according to the location information of the second device and the location information of the first target device used to send the target reference signal.

[0350] Optionally, the first generating module 1201 includes:

[0351] A first processing unit is used to match the first label data and the first training data according to the device identification information of the first target device in the first training data and the device identification information of the second target device in the first label data, and / or the second timestamp information in the first training data and the third timestamp information in the first label data, so as to obtain the sample data.

[0352] Optionally, the first processing unit is specifically used for at least one of the following:

[0353] In a case where the positions of the device sending the target reference signal and the device receiving the target reference signal are both fixed, matching the first label data with the first training data according to the device identification information of the first target device and the device identification information of the second target device to obtain the sample data;

[0354] In a case where the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, receiving the updated first label data sent by the third device according to the position information of the third target device corresponding to the second timestamp information, matching the updated first label data with the first training data according to the device identification information of the second target device and the device identification information of the first target device in the updated first label data, and / or, according to the third timestamp information in the updated first label data and the second timestamp information, to obtain the sample data; wherein the position information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the position information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a preset positioning method, and the third target device is a device with a non-fixed position among the device sending the target reference signal and the device receiving the target reference signal;

[0355] When the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, the first label data and the first training data are matched according to at least one item of the device identification information of the second target device, the second timestamp information, the device identification information of the first target device and the third timestamp information to obtain the sample data.

[0356] Optionally, the first processing unit is specifically configured to:

[0357] When the device identification information of the first target device is consistent with the device identification information of the second target device, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

[0358] Optionally, the first processing unit is specifically configured to:

[0359] When the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device, and / or when the third timestamp information in the updated first label data is consistent with the second timestamp information, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

[0360] Optionally, the first processing unit is specifically used for at least one of the following:

[0361] When the device identification information of the second target device is consistent with the device identification information of the first target device, and the time indicated by the second timestamp information is later than the time indicated by the third timestamp, matching the first true label in the first label data and the first measurement in the first training data to obtain the sample data;

[0362] determining change position information of a third target device, sending the change position information to the third device, receiving updated first label data sent by the third device according to the change position information, and matching a first true label in the first label data with a first measurement in the first training data when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the updated first label data, to obtain the sample data;

[0363] Determine change position information of a third target device, send the change position information to the third device, receive updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is later than or equal to the time indicated by the second timestamp information in the first training data, and the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the first label data, match the first true label in the first label data and the first measurement in the first training data to obtain the sample data;

[0364] Determine the change position information of the third target device, send the change position information to the third device, receive the updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is earlier than the time indicated by the second timestamp information in the first training data, match the first true label in the updated first label data and the first measurement in the first training data to obtain the sample data.

[0365] It should be noted that the training data collection device applied to the second device provided in the embodiment of the present invention is a device capable of executing the above-mentioned training data collection method applied to the second device. All embodiments of the above-mentioned training data collection method applied to the first device are applicable to the device and can achieve the same or similar technical effects.

[0366] like Fig.13 As shown, an embodiment of the present invention further provides a training data collection device, which is applied to a third device, and the device includes:

[0367] The first receiving module 1301 is used to receive data request information;

[0368] The second sending module 1302 is configured to send type information of the first training data and / or first label data according to the data request information.

[0369] Optionally, the data request information includes at least one of the following:

[0370] The type of the first true label;

[0371] The required quantity of sample data;

[0372] location-related information of the first device;

[0373] Type information of the first training data;

[0374] Generalization requirements of sample data;

[0375] Accuracy requirements for sample data;

[0376] The requirement information for obtaining configuration of a target reference signal for the first training data is used to determine the configuration of a target reference signal used to obtain the first training data.

[0377] Optionally, the first tag data includes at least one of the following:

[0378] First true label;

[0379] Device identification information of the second target device used to send the target reference signal

[0380] The third timestamp information related to the first real label.

[0381] Optionally, the device further comprises:

[0382] A determination module, configured to determine a second device according to the data request information;

[0383] Among them, the second device is a device for generating sample data, and the sample data is used to train an artificial intelligence AI model or a machine learning ML model.

[0384] Optionally, when the data request information includes the type of the first real label, and the type of the first real label is location information, the first real label in the first label data is obtained by at least one of the following methods:

[0385] receiving location information sent by a second device;

[0386] Location information obtained by a pre-set positioning method;

[0387] The location information of the second device is acquired in advance.

[0388] Optionally, the first real label includes at least one of the following:

[0389] Location information;

[0390] fourth timestamp information associated with the location information;

[0391] Confidence information associated with the location information.

[0392] Optionally, when the data request information includes the type of the first real label, and the type of the first real label is a first arrival time TOA value, the first real label in the first label data is obtained based on the location information of the second device and the location information of the first target device; wherein the first target device is a device that sends a target reference signal.

[0393] Optionally, the device further comprises:

[0394] A first information sending module is used to send updated first label data to a second device according to the position information of a third target device corresponding to the second timestamp information in the first training data; wherein the position information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the position information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a preset positioning method; the first training data is obtained by measuring according to a target reference signal; the third target device is a device with a non-fixed position between a device sending the target reference signal and a device receiving the target reference signal;

[0395] Optionally, the device further comprises:

[0396] The first information receiving module is used to receive the change position information sent by the third target device

[0397] A second information sending module, used for sending updated first tag data according to the changed position information;

[0398] The third target device is a device whose position is not fixed among the device that sends the target reference signal and the device that receives the target reference signal.

[0399] It should be noted that the training data collection device applied to the third device provided in the embodiment of the present invention is a device capable of executing the above-mentioned training data collection method applied to the third device. All embodiments of the above-mentioned training data collection method applied to the third device are applicable to the device and can achieve the same or similar technical effects.

[0400] like Fig.14 As shown, an embodiment of the present invention also provides a communication device, which is a first device, including: a processor 1401; and a memory 1403 connected to the processor 1401 through a bus interface 1402, the memory 1403 is used to store programs and data used by the processor 1401 when performing operations, and the processor 1401 calls and executes the programs and data stored in the memory 1403.

[0401] The transceiver 1404 is connected to the bus interface 1402 and is used to receive and send data under the control of the processor 1401. Specifically, the processor 1401 is used to read the program in the memory 1403, and the transceiver 1404 executes:

[0402] Send data request information;

[0403] And, the processor 1401 executes: obtaining sample data; and / or, the transceiver 1404 executes: obtaining an artificial intelligence AI model or a machine learning ML model;

[0404] The sample data is obtained according to the first label data and the first training data; the artificial intelligence AI model or the machine learning ML model is trained based on the sample data.

[0405] Optionally, the data request information includes at least one of the following:

[0406] The type of the first true label;

[0407] The required quantity of sample data;

[0408] location-related information of the first device;

[0409] Type information of the first training data;

[0410] Generalization requirements of sample data;

[0411] Accuracy requirements for sample data;

[0412] The requirement information for obtaining configuration of a target reference signal for the first training data is used to determine the configuration of a target reference signal used to obtain the first training data.

[0413] Optionally, the type of the first real label includes at least one of the following:

[0414] Location information;

[0415] First arrival time TOA value;

[0416] Line-of-sight (LOS) indication or non-line-of-sight (NLOS) indication.

[0417] Optionally, the location-related information of the first device includes at least one of the following:

[0418] location information of the first device;

[0419] first timestamp information related to the location information of the first device;

[0420] Confidence information related to the location of the first device.

[0421] Optionally, the type information of the first training data includes at least one of the following:

[0422] a type of a first measurement quantity in the first training data;

[0423] The dimension of the first measurement quantity in the first training data.

[0424] Optionally, the first training data includes at least one of the following:

[0425] A first measurement quantity;

[0426] second timestamp information related to the first measurement;

[0427] Device identification information of a first target device used to send a target reference signal;

[0428] The second TOA value.

[0429] Among them, Fig.14 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically linking together various circuits of one or more processors represented by processor 1401 and memory represented by memory 1403. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 1404 may be a plurality of components, i.e., including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium. The processor 1401 is responsible for managing the bus architecture and general processing, and the memory 1403 may store data used by the processor 1401 when performing operations.

[0430] An embodiment of the present invention also provides a communication device, which is a second device and includes: a processor; and a memory connected to the processor via a bus interface, the memory being used to store programs and data used by the processor when performing operations, and the processor calling and executing the programs and data stored in the memory.

[0431] The transceiver is connected to the bus interface and is used to receive and send data under the control of the processor. Specifically, the processor is used to read the program in the memory.

[0432] The structure of the second device is similar to Fig.14 The structures of the communication devices shown are basically the same and will not be described in detail here.

[0433] The processor performs:

[0434] Generate sample data;

[0435] The sample data is obtained according to the acquired first label data and first training data;

[0436] The sample data is used to train an artificial intelligence (AI) model or a machine learning (ML) model.

[0437] Optionally, the processor is further configured to:

[0438] Obtain configuration information for generating sample data, wherein the configuration information for generating sample data includes:

[0439] Type information of the first training data;

[0440] The type information of the first training data includes at least one of the following:

[0441] a type of a first measurement quantity in the first training data;

[0442] The dimension of the first measurement quantity in the first training data.

[0443] Optionally, the transceiver is used for:

[0444] The sample data is sent.

[0445] Optionally, the first tag data includes at least one of the following:

[0446] First true label;

[0447] Device identification information of the second target device used to send the target reference signal

[0448] The third timestamp information related to the first real label.

[0449] Optionally, the first training data includes at least one of the following:

[0450] A first measurement quantity;

[0451] second timestamp information related to the first measurement;

[0452] Device identification information of a first target device used to send a target reference signal;

[0453] Second time of arrival TOA value.

[0454] Optionally, the transceiver is further used for:

[0455] receiving a target reference signal, where the target reference signal is a reference signal sent by a first target device and used to obtain first training data;

[0456] The processor is further configured to:

[0457] The target reference signal is measured according to the type information of the first training data to obtain the first training data.

[0458] Optionally, the first measurement includes at least one of the following:

[0459] channel estimation value;

[0460] Time information of channel estimation;

[0461] Power information of channel estimation;

[0462] Phase information for channel estimation.

[0463] Optionally, the first training data includes a second TOA value;

[0464] The second TOA value is obtained by at least one of the following methods:

[0465] Measured based on a target reference signal;

[0466] obtained by using a trained target model and a first measurement amount, wherein the target model includes an AI model and / or an ML model, and the first measurement amount is obtained by measuring the target reference signal according to type information of first training data;

[0467] Optionally, the processor is further configured to:

[0468] When the data request information includes the type of the first real label, and the type of the first real label is a line-of-sight (LOS) indication or a non-line-of-sight (NLOS) indication, obtaining the LOS indication or the NLOS indication according to the first TOA value and the second TOA value;

[0469] The first TOA value is obtained according to the location information of the second device and the location information of the first target device used to send the target reference signal.

[0470] Optionally, the processor is specifically configured to:

[0471] According to the device identification information of the first target device in the first training data and the device identification information of the second target device in the first label data, and / or, the second timestamp information in the first training data and the third timestamp information in the first label data, the first label data and the first training data are matched to obtain the sample data.

[0472] Optionally, the processor is specifically configured to perform at least one of the following:

[0473] In a case where the positions of the device sending the target reference signal and the device receiving the target reference signal are both fixed, matching the first label data with the first training data according to the device identification information of the first target device and the device identification information of the second target device to obtain the sample data;

[0474] In a case where the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, receiving the updated first label data sent by the third device according to the position information of the third target device corresponding to the second timestamp information, matching the updated first label data with the first training data according to the device identification information of the second target device and the device identification information of the first target device in the updated first label data, and / or, according to the third timestamp information in the updated first label data and the second timestamp information, to obtain the sample data; wherein the position information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the position information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a preset positioning method, and the third target device is a device with a non-fixed position among the device sending the target reference signal and the device receiving the target reference signal;

[0475] When the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, the first label data and the first training data are matched according to at least one item of the device identification information of the second target device, the second timestamp information, the device identification information of the first target device and the third timestamp information to obtain the sample data.

[0476] Optionally, the processor is specifically configured to:

[0477] When the device identification information of the first target device is consistent with the device identification information of the second target device, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

[0478] Optionally, the processor is specifically configured to:

[0479] When the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device, and / or when the third timestamp information in the updated first label data is consistent with the second timestamp information, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

[0480] Optionally, the processor is specifically configured to perform at least one of the following:

[0481] When the device identification information of the second target device is consistent with the device identification information of the first target device, and the time indicated by the second timestamp information is later than the time indicated by the third timestamp, matching the first true label in the first label data and the first measurement in the first training data to obtain the sample data;

[0482] determining change position information of a third target device, sending the change position information to the third device, receiving updated first label data sent by the third device according to the change position information, and matching a first true label in the first label data with a first measurement in the first training data when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the updated first label data, to obtain the sample data;

[0483] Determine change position information of a third target device, send the change position information to the third device, receive updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is later than or equal to the time indicated by the second timestamp information in the first training data, and the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the first label data, match the first true label in the first label data and the first measurement in the first training data to obtain the sample data;

[0484] Determine the change position information of the third target device, send the change position information to the third device, receive the updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is earlier than the time indicated by the second timestamp information in the first training data, match the first true label in the updated first label data and the first measurement in the first training data to obtain the sample data.

[0485] An embodiment of the present invention also provides a communication device, which is a third device and includes: a processor; and a memory connected to the processor via a bus interface, the memory being used to store programs and data used by the processor when performing operations, and the processor calling and executing the programs and data stored in the memory.

[0486] The transceiver is connected to the bus interface and is used to receive and send data under the control of the processor. Specifically, the processor is used to read the program in the memory.

[0487] The structure of the third device is similar to Fig.14 The structures of the communication devices shown are basically the same and will not be described in detail here.

[0488] The transceiver performs at least one of the following:

[0489] receiving data request information;

[0490] According to the data request information, type information of the first training data and / or first label data are sent.

[0491] Optionally, the data request information includes at least one of the following:

[0492] The type of the first true label;

[0493] The required quantity of sample data;

[0494] location-related information of the first device;

[0495] Type information of the first training data;

[0496] Generalization requirements of sample data;

[0497] Accuracy requirements for sample data;

[0498] The requirement information for obtaining configuration of a target reference signal for the first training data is used to determine the configuration of a target reference signal used to obtain the first training data.

[0499] Optionally, the first tag data includes at least one of the following:

[0500] First true label;

[0501] Device identification information of the second target device used to send the target reference signal

[0502] The third timestamp information related to the first real label.

[0503] Optionally, the processor is further configured to:

[0504] Determining a second device according to the data request information;

[0505] Among them, the second device is a device for generating sample data, and the sample data is used to train an artificial intelligence AI model or a machine learning ML model.

[0506] Optionally, when the data request information includes the type of the first real label, and the type of the first real label is location information, the first real label in the first label data is obtained by at least one of the following methods:

[0507] receiving location information sent by a second device;

[0508] Location information obtained by a pre-set positioning method;

[0509] The location information of the second device is acquired in advance.

[0510] Optionally, the first real label includes at least one of the following:

[0511] Location information;

[0512] fourth timestamp information associated with the location information;

[0513] Confidence information associated with the location information.

[0514] Optionally, when the data request information includes the type of the first real label, and the type of the first real label is a first arrival time TOA value, the first real label in the first label data is obtained based on the location information of the second device and the location information of the first target device; wherein the first target device is a device that sends a target reference signal.

[0515] Optionally, the transceiver is further used for:

[0516] Sending updated first label data to the second device according to the location information of the third target device corresponding to the second timestamp information in the first training data; wherein the location information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the location information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a preset positioning method; the first training data is obtained by measuring according to a target reference signal; the third target device is a device with a non-fixed position between a device sending the target reference signal and a device receiving the target reference signal;

[0517] Optionally, the transceiver is further used for:

[0518] Receive the change location information sent by the third target device

[0519] Sending updated first tag data according to the changed position information;

[0520] The third target device is a device whose position is not fixed among the device that sends the target reference signal and the device that receives the target reference signal.

[0521] In addition, a specific embodiment of the present invention further provides a readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the steps of the training data collection method applied to the first device as described in any one of the above are implemented, or the steps of training data collection applied to the second device as described in any one of the above are implemented, or the steps of training data collection applied to the third device as described in any one of the above are implemented.

[0522] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0523] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may be physically included separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0524] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform some steps of the resource selection method described in each embodiment of the present invention, or to perform some steps of the information sending method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store program codes.

[0525] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary personnel in this technical field, several improvements and modifications can be made without departing from the principles described in the present invention. These improvements and modifications are also within the scope of protection of the present invention.

Claims

1. A training data collection method, characterized in that: Applied to a first device, the method includes: Send data request information; And do at least one of the following: Get sample data; Obtain artificial intelligence AI model or machine learning ML model; The sample data is obtained according to the first label data and the first training data; the artificial intelligence AI model or the machine learning ML model is trained based on the sample data.

2. The method according to claim 1, characterized in that The data request information includes at least one of the following: The type of the first true label; The required quantity of sample data; location-related information of the first device; Type information of the first training data; Generalization requirements of sample data; Accuracy requirements for sample data; The requirement information for obtaining configuration of a target reference signal for the first training data is used to determine the configuration of a target reference signal used to obtain the first training data.

3. The method according to claim 2, characterized in that The type of the first real label includes at least one of the following: Location information; First arrival time TOA value; Line-of-sight (LOS) indication or non-line-of-sight (NLOS) indication.

4. The method according to claim 2, characterized in that: The location-related information of the first device includes at least one of the following: location information of the first device; first timestamp information related to the location information of the first device; Confidence information related to the location of the first device.

5. The method according to claim 2, characterized in that: The type information of the first training data includes at least one of the following: a type of a first measurement quantity in the first training data; The dimension of the first measurement quantity in the first training data.

6. The method according to claim 1, characterized in that The first training data includes at least one of the following: A first measurement quantity; second timestamp information related to the first measurement; Device identification information of a first target device used to send a target reference signal; The second TOA value.

7. A training data collection method, characterized in that: Applied to the second device, the method includes: Generate sample data; The sample data is obtained according to the acquired first label data and first training data; The sample data is used to train an artificial intelligence (AI) model or a machine learning (ML) model.

8. The method according to claim 7, characterized in that Before generating the sample data, the method further includes: Acquire configuration information for generating sample data, where the configuration information for generating sample data includes: Type information of the first training data; The type information of the first training data includes at least one of the following: a type of a first measurement quantity in the first training data; The dimension of the first measurement quantity in the first training data.

9. The method according to claim 7, characterized in that: The method further includes: sending the sample data.

10. The method according to claim 7, characterized in that The first tag data includes at least one of the following: First true label; Device identification information of the second target device used to send the target reference signal The third timestamp information related to the first real label.

11. The method according to claim 7, characterized in that The first training data includes at least one of the following: A first measurement quantity; second timestamp information related to the first measurement; Device identification information of a first target device used to send a target reference signal; Second time of arrival TOA value.

12. The method according to claim 7, characterized in that The method further comprises: receiving a target reference signal, where the target reference signal is a reference signal sent by a first target device and used to obtain first training data; The target reference signal is measured according to the type information of the first training data to obtain the first training data.

13. The method according to claim 11, characterized in that The first measurement includes at least one of the following: channel estimation value; Time information of channel estimation; Power information of channel estimation; Phase information for channel estimation.

14. The method according to claim 7, characterized in that The first training data includes a second TOA value; The second TOA value is obtained by at least one of the following methods: Measured based on a target reference signal; It is obtained by using a trained target model and a first measurement quantity, wherein the target model includes an AI model and / or an ML model, and the first measurement quantity is obtained by measuring the target reference signal according to type information of first training data.

15. The method according to claim 7, characterized in that The method further comprises: When the data request information includes the type of the first real label, and the type of the first real label is a line-of-sight (LOS) indication or a non-line-of-sight (NLOS) indication, obtaining the LOS indication or the NLOS indication according to the first TOA value and the second TOA value; The first TOA value is obtained according to the location information of the second device and the location information of the first target device used to send the target reference signal.

16. The method according to claim 7, characterized in that Generate sample data, including: According to the device identification information of the first target device in the first training data and the device identification information of the second target device in the first label data, and / or, the second timestamp information in the first training data and the third timestamp information in the first label data, the first label data and the first training data are matched to obtain the sample data.

17. The method according to claim 16, characterized in that According to the device identification information of the first target device in the first training data and the device identification information of the second target device in the first label data, and / or, the second timestamp information in the first training data and the third timestamp information in the first label data, the first label data and the first training data are matched to obtain the sample data, including at least one of the following: In a case where the positions of the device sending the target reference signal and the device receiving the target reference signal are both fixed, matching the first label data with the first training data according to the device identification information of the first target device and the device identification information of the second target device to obtain the sample data; In a case where the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, receiving the updated first label data sent by the third device according to the position information of the third target device corresponding to the second timestamp information, matching the updated first label data with the first training data according to the device identification information of the second target device and the device identification information of the first target device in the updated first label data, and / or, according to the third timestamp information in the updated first label data and the second timestamp information, to obtain the sample data; wherein the position information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the position information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a preset positioning method, and the third target device is a device with a non-fixed position among the device sending the target reference signal and the device receiving the target reference signal; When the position of at least one of the device sending the target reference signal and the device receiving the target reference signal is not fixed, the first label data and the first training data are matched according to at least one item of the device identification information of the second target device, the second timestamp information, the device identification information of the first target device and the third timestamp information to obtain the sample data.

18. The method according to claim 17, characterized in that According to the device identification information of the first target device and the device identification information of the second target device, matching the first label data with the first training data to obtain the sample data includes: When the device identification information of the first target device is consistent with the device identification information of the second target device, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

19. The method according to claim 17, characterized in that According to the device identification information of the second target device in the updated first label data and the device identification information of the first target device, and / or according to the third timestamp information and the second timestamp information in the updated first label data, matching the first label data and the first training data to obtain the sample data includes: When the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device, and when the third timestamp information in the updated first label data is consistent with the second timestamp information, the first true label in the first label data and the first measurement quantity in the first training data are matched to obtain the sample data.

20. The method according to claim 17, characterized in that According to the device identification information of the second target device, the second timestamp information, the device identification information of the first target device, and at least one of the third timestamp information, the first label data and the first training data are matched to obtain the sample data, including at least one of the following: When the device identification information of the second target device is consistent with the device identification information of the first target device, and the time indicated by the second timestamp information is later than the time indicated by the third timestamp, matching the first true label in the first label data and the first measurement in the first training data to obtain the sample data; determining change position information of a third target device, sending the change position information to the third device, receiving updated first label data sent by the third device according to the change position information, and matching a first true label in the first label data with a first measurement in the first training data when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the updated first label data, to obtain the sample data; Determine change position information of a third target device, send the change position information to the third device, receive updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is later than or equal to the time indicated by the second timestamp information in the first training data, and the time indicated by the second timestamp information in the first training data is later than or equal to the time indicated by the third timestamp information in the first label data, match the first true label in the first label data and the first measurement in the first training data to obtain the sample data; Determine the change position information of the third target device, send the change position information to the third device, receive the updated first label data sent by the third device according to the change position information, and when the device identification information of the second target device in the updated first label data is consistent with the device identification information of the first target device in the first training data, and when the time indicated by the third timestamp information in the updated first label data is earlier than the time indicated by the second timestamp information in the first training data, match the first true label in the updated first label data and the first measurement in the first training data to obtain the sample data.

21. A training data collection method, characterized in that: Applied to a third device, the method includes: receiving data request information; According to the data request information, type information of the first training data and / or first label data are sent.

22. The method according to claim 21, characterized in that The data request information includes at least one of the following: The type of the first true label; The required quantity of sample data; location-related information of the first device; Type information of the first training data; Generalization requirements of sample data; Accuracy requirements for sample data; The requirement information for obtaining configuration of a target reference signal for the first training data is used to determine the configuration of a target reference signal used to obtain the first training data.

23. The method according to claim 21, characterized in that The first tag data includes at least one of the following: First true label; Device identification information of the second target device used to send the target reference signal The third timestamp information related to the first real label.

24. The method according to claim 21, characterized in that After receiving the data request information, the method further includes: Determining a second device according to the data request information; Among them, the second device is a device for generating sample data, and the sample data is used to train an artificial intelligence AI model or a machine learning ML model.

25. The method according to claim 21, characterized in that In a case where the data request information includes the type of the first real label, and the type of the first real label is location information, the first real label in the first label data is obtained by at least one of the following methods: receiving location information sent by a second device; Location information obtained through a pre-set positioning method; The location information of the second device is acquired in advance.

26. The method according to claim 25, characterized in that The first true label includes at least one of the following: Location information; fourth timestamp information associated with the location information; Confidence information associated with the location information.

27. The method according to claim 21, characterized in that When the data request information includes the type of the first real label, and the type of the first real label is a first arrival time TOA value, the first real label in the first label data is obtained based on the location information of the second device and the location information of the first target device; wherein the first target device is a device that sends a target reference signal.

28. The method according to claim 21, characterized in that The method further comprises: Send updated first label data to the second device according to the position information of the third target device corresponding to the second timestamp information in the first training data; wherein the position information of the third target device corresponding to the second timestamp information is sent by the third target device to the third device, or the position information of the third target device corresponding to the second timestamp information is determined by the third device according to the second timestamp information and a pre-set positioning method; the first training data is measured according to the target reference signal; the third target device is a device with a non-fixed position between the device sending the target reference signal and the device receiving the target reference signal.

29. The method according to claim 21, characterized in that The method further comprises: Receive the change location information sent by the third target device Sending updated first tag data according to the changed position information; The third target device is a device whose position is not fixed among the device that sends the target reference signal and the device that receives the target reference signal.

30. A training data collection device, characterized in that: Applied to a first device, the apparatus comprises: A first sending module, used for sending data request information; and a first processing module, the first processing module being configured to perform at least one of the following: Get sample data; Obtain artificial intelligence AI model or machine learning ML model; The sample data is obtained according to the first label data and the first training data; the artificial intelligence AI model or the machine learning ML model is trained based on the sample data.

31. A training data collection device, characterized in that: Applied to a second device, the apparatus comprises: A first generating module, used for generating sample data; The sample data is obtained according to the acquired first label data and first training data; The sample data is used to train an artificial intelligence (AI) model or a machine learning (ML) model.

32. A training data collection device, characterized in that: Applied to a third device, the apparatus comprises: A first receiving module, used for receiving data request information; The second sending module is used to send the type information of the first training data and / or the first label data according to the data request information.

33. A communication device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the training data collection method as described in any one of claims 1 to 6, or, when executed by the processor, implements the steps of the training data collection method as described in any one of claims 7 to 20, or, when executed by the processor, implements the steps of the training data collection method as described in any one of claims 21 to 29.

34. A readable storage medium, characterized in that: The readable storage medium stores a program, and when the program is executed by the processor, it implements the steps in the training data collection method as described in any one of claims 1 to 6, or implements the steps in the training data collection method as described in any one of claims 7 to 20, or implements the steps in the training data collection method as described in any one of claims 21 to 29.

Images