Method and network entity for verifying data of AI / ML model

Abstract

A method for training an artificial intelligence / machine learning (AI / ML) model in a wireless communication network includes verifying authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; the AI / ML model is trained using the positioning data based on verification results indicating that the positioning data is authorized for training.

Description

Technical Field

[0001] The present disclosure relates to the field of wireless communications, and more particularly to a method for training an artificial intelligence / machine learning (AI / ML) model, and more particularly to determining whether specific data is usable for such training. Furthermore, the present disclosure relates to a network entity operating in a wireless communications network. Background Art

[0002] In a wireless communication system or network, such as a 3GPP network, one or more terminal devices, such as user equipment (UE), operate. The terminal devices are connected to a radio access network (RAN), such as a base station, via a radio or wireless connection. The RAN is connected to a core network (CN), which implements core network functions (Core Network Functions) in one or more network entities to control overall network operations, such as protocols, network interfaces, and services. In such a wireless communication network, it may be desirable to determine the location of one or more devices or entities, such as UEs.

[0003] Positioning in wireless communication networks can be performed with the support of artificial intelligence / machine learning AI / ML models, while other positioning methods can be performed without such AI / ML models. Summary of the Invention

[0004] One of the objects of the present disclosure is to provide a method and a network entity or device that improves the positioning process, in particular the positioning process based on the use of AI / ML models.

[0005] This object is achieved by the subject matter defined in the independent claim. Advantageous further developments are defined in the dependent claims.

[0006] The present disclosure provides a method for training an artificial intelligence / machine learning (AI / ML) model in a wireless communication network. The method includes verifying authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not. The method also includes training the AI / ML model using the positioning data based on the verification result indicating that the positioning data is authorized for training.

[0007] According to an embodiment, verifying the authorization is performed with the first network function NF, and wherein the training is performed at the first NF or the second NF.

[0008] According to an embodiment, the NF includes a location management function LMF; and wherein the second NF includes at least one of a model training logic function MTLF and a network data analysis function NWDAF.

[0009] The present disclosure provides a method that includes authorization associated with a device associated with a location and indicating whether information obtained with respect to the device is authorized for training.

[0010] According to an embodiment, a method of training an AI / ML model includes excluding positioning data from training the AI / ML model based on a verification result indicating that the positioning data is not authorized for training.

[0011] According to an embodiment, a method for training an AI / ML model is performed such that authorization is correlated with at least one of reliability of positioning data, accuracy of positioning data, an identifier of a device associated with the positioning data, etc. Further examples relate to a deployment scenario from which positioning data is obtained, a density of predefined data points, a number of sources of positioning data, and accuracy of positioning data.

[0012] According to an embodiment, verification of the positioning data is performed to obtain authorized positioning data to form a training dataset consisting of data of a single deployment scenario.

[0013] According to an embodiment, verification of the positioning data is performed to obtain authorized positioning data to form a training dataset consisting of multiple deployment scenarios.

[0014] According to an embodiment, the training of the AI / ML model is repeated over multiple iterations, so that authorization is verified for each or a subset of the multiple iterations. For example, the multiple iterations may be related to model updates, which can be implemented repeatedly or iteratively. For each or a subset of the authorizations for multiple updates of location data, data visibility can be checked or verified separately.

[0015] According to an embodiment, the method described herein includes collecting positioning data related to a location measurement process performed in a wireless communication network in a location management function (LMF). The method includes forwarding the positioning data to a model training entity based on a verification result indicating that the positioning data is authorized for training. The method further includes training an AI / ML model in the model training entity.

[0016] According to an embodiment, the method is performed such that the model training entity is an entity including a model training logic function MTLF, for example, a network data analysis function NWDAF may include the MTLF.

[0017] According to an embodiment, a method includes transmitting a data subscription request for data collection for training an AI / ML model to a location management function (LMF), i.e., a model training entity transmitting the request. The method further includes validating the data subscription request at the LMF and / or at a network data analysis function (NWDAF) to verify authorization of location data related to the location of the device, thereby obtaining at least a portion of a validation result. The method includes determining the location data of the device using the LMF, and using the location data for the model training entity based on the validation result.

[0018] According to an embodiment, the method includes providing a verification result to the model training entity; or providing the verification result and the positioning data to the model training entity. The verification result indicates that at least a portion of the positioning data is authorized for training.

[0019] According to an embodiment, a method includes transmitting, by a model inference entity, a model subscription request to a model training entity for obtaining an AI / ML model, e.g., after training. The method includes validating the model subscription request at the model training entity to obtain a subscription validation result. The method includes training the AI / ML model at the model training entity to obtain a trained model, and providing the trained model to the model inference entity based on the subscription validation result.

[0020] According to an embodiment, the model reasoning entity is or comprises a Location Management Function LMF.

[0021] According to an embodiment, the method described herein comprises obtaining, at a Location Management Function LMF, positioning data related to a location measurement procedure performed in a wireless communication network.

[0022] The method includes using the positioning data for model training in the LMF based on a verification result indicating that the positioning data is authorized for training, so as to train the AI / ML model using the LMF.

[0023] According to an embodiment, the method includes training an AI / ML model using a model training logic function (MTLF) of an LMF.

[0024] According to an embodiment, the method is performed such that training an AI / ML model results in a newly generated AI / ML model.

[0025] According to an embodiment, training of an AI / ML model may result in an updated, retrained, or refined model from a previous model.

[0026] According to an embodiment, the positioning dataset comprises a synthetic dataset.

[0027] According to an embodiment, the synthetic data set is generated based on a statistical channel model.

[0028] Depending on the embodiment, the dataset authorized for model training is obtained based on model generalization or without considering generalization.

[0029] The present disclosure provides a computer-readable storage medium storing instructions, which, when executed, can enable the method described herein to be performed by a location management function LMF and / or a model training entity of a wireless communication system.

[0030] The present disclosure further provides a network entity, such as a LMF, for operating in a wireless communication network, the network entity being configured to verify authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not. The network entity is configured to provide the positioning data to a model training entity or use the positioning data for model training based on the verification result indicating that the positioning data is authorized for training.

[0031] The present disclosure further provides a system for training an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system comprising a first network function NF and a second NF, the system being configured to verify, with the second NF, authorization of positioning data related to a location of a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; transmit the verification result and the positioning data to the first NF; transmit the verification result and the positioning data to the first NF;

[0032] According to an embodiment, the positioning data is transmitted to the first NF, thereby indicating that the positioning data is authorized for training.

[0033] According to an embodiment, the first NF includes a model training logic function MTLF and / or a network data analysis function NWDAF.

[0034] According to an embodiment, a system is configured for training an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system including at least a first network function NF, the system being configured to: verify, using the first NF, authorization of positioning data associated with a p-device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; and train the AI / ML model with the first NF using the positioning data based on the verification result indicating that the positioning data is authorized for training.

[0035] According to an embodiment, the first NF is adapted to store a plurality of positioning data and exclude positioning data that is not authorized for training from training.

[0036] According to an embodiment, the first NF includes a Location Management Function LMF.

[0037] According to the embodiment

[0038] The technical solutions provided by embodiments of the present disclosure have the following beneficial effects. The present disclosure is advantageous because it allows determining and thereby selecting whether positioning data associated with a device is used to train an AI / ML model. In addition, embodiments allow maintaining the privacy of the UE's positioning data.

[0039] It should be understood that the content described in this section is not intended to identify the key or key features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure can be easily understood from the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, explain the principles of the present disclosure. The accompanying drawings are illustrative and serve to explain the present disclosure and are not to be construed as limiting the present disclosure to the embodiments shown.

[0041] Figure 1 This is a flowchart of the training process for AI / ML direct positioning using the LMF side model;

[0042] Figure 2 A flowchart of a method according to an embodiment of the present disclosure is shown;

[0043] Figure 3 A schematic flow chart illustrating at least a portion of a method according to an embodiment of the present disclosure;

[0044] Figure 4 shows a flowchart related to training an AI / ML model according to an embodiment of the present disclosure;

[0045] Figure 5 A flowchart for forwarding measurement data for training AI / ML models according to an embodiment of the present disclosure is shown;

[0046] Figure 6 shows a flow chart according to a further embodiment of the present invention;

[0047] Figure 7 shows a detailed flowchart related to the training of AI / ML models according to an embodiment of the present disclosure;

[0048] Figure 8 shows a flow chart of a method according to an embodiment, wherein the training of the AI / ML model is performed at the LMF;

[0049] Figure 9 A flowchart further defining a method for training an AI / ML model using LMF according to an embodiment of the present disclosure is shown;

[0050] Figure 10 A flowchart related to a method for defining a purpose for training an AI / ML model according to an embodiment of the present disclosure is shown;

[0051] Figure 11 is a schematic block diagram of the system operation according to an embodiment;

[0052] Figure 12is a schematic diagram of signaling in a wireless communication network according to the first solution proposed in the present disclosure;

[0053] Figure 13 A schematic diagram illustrating signaling in a wireless communication network according to a second solution provided by an embodiment of the present disclosure; and

[0054] Figure 14 A block diagram is shown, which illustrates an electronic device configured to implement an image processing method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0055] Illustrative embodiments of the present invention are described below with reference to the accompanying drawings. Various details of the embodiments of the present invention are included to facilitate understanding and should be considered as illustrative only. Accordingly, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present invention. In addition, for the sake of clarity and brevity, descriptions of well-known functions and structures are omitted from the following description.

[0056] In the present invention, the term “and / or” is intended to cover all possible combinations and subcombinations of the listed elements, including any one of the listed elements alone, any subcombination of elements, or all of them, and does not necessarily exclude other elements.

[0057] In the present invention, the phrase "at least one of..." is intended to cover any one or more of the listed elements, including any one of the listed elements alone, any subcombination of the elements, or all of them, without necessarily excluding any other elements and without necessarily requiring all elements.

[0058] The terms used in the embodiments of the present invention are intended to describe specific embodiments and should not be construed as limiting the present invention. In the present invention and the appended claims, the singular forms "a / an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0059] It should be understood that although terms such as "first," "second," and "third" may be used to describe various information in embodiments of the present invention, such information should not be limited by these terms. These terms are used only to distinguish between information of the same type. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information without departing from the scope of embodiments of the present invention. Depending on the context, the term "if" as used herein may be interpreted as "when," "upon," or "in response to a determination."

[0060] Illustrative embodiments of the present invention are described below with reference to the accompanying drawings. Various details of the embodiments of the present invention are included to facilitate understanding and should be regarded as illustrative only. Accordingly, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present invention. In addition, for the sake of clarity and brevity, descriptions of well-known functions and structures are omitted from the following description.

[0061] The inclusion of any one, any subcombination, or all of the listed elements is not necessarily exclusive of other elements.

[0062] In the present invention, the phrase "at least one of..." is intended to cover any one or more of the listed elements, including any one of the listed elements alone, any subcombination of the elements, or all of them, without necessarily excluding any other elements and without necessarily requiring all elements.

[0063] The terms used in the embodiments of the present invention are intended to describe specific embodiments and should not be construed as limiting the present invention. In the present invention and the appended claims, the singular forms "a / an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0064] It should be understood that although terms such as "first," "second," and "third" may be used to describe various information in embodiments of the present invention, such information should not be limited by these terms. These terms are used only to distinguish between information of the same type. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information without departing from the scope of embodiments of the present invention. Depending on the context, the term "if" as used herein may be interpreted as "when," "upon," or "in response to a determination."

[0065] Figure 1 This is a schematic flow chart of the training process for AI / ML direct positioning using the LMF side model known in 3GPP TR 38.843. Figure 1 The training process for AI / ML direct positioning using the LMF side model corresponds to TR 38.843 Figure 6 .2.2.1-1. In 102, the location management function LMF subscribes to training data from the network data analysis function NWDAF for direct positioning of AI / ML models. The training data subscription can be a request for analysis using an existing service or a request for models or other types of data related to training. Several optional parameters defined in TS23.288 can be provided as service operation inputs. The training data subscription can be triggered by the LMF receiving a previous request from another network entity. The trigger itself is not in Figure 1 and can be considered not to form part of the training process itself.

[0066] At 104, data collection is performed to train a model for direct localization by the training network function NF. If the model is trained by the LMF, the LMF can use measurements already available to the LMF or new measurements, as well as possible additional data such as analytical data, to train the model. If the model is trained by the NWDAF, the NWDAF can collect the necessary data from the NF to train the model.

[0067] In conditional action 106 , if the subscription to training data from the LMF in 102 requires training a model, the NWDAF containing the model training logic function MTLF can train the AI / ML model.

[0068] In 108, the NWDAF provides a training data notification to the LMF. Based on the request in 102, if 106 is executed, the notification may include the requested trained AI / ML model for direct positioning, or the NWDAF may provide the requested analysis or data in the notification to the LMF for 112 to be executed.

[0069] In conditional action 112 , if the AI / ML model for positioning has not been trained and is not provided 104 , the LMF trains the AI / ML model in 112 .

[0070] Figure 2 A flow chart of a method according to an embodiment of the present invention is shown. The authorization of the positioning data is verified to determine whether the positioning data is used to train the AI / ML model.

[0071] According to an embodiment, a method for training an artificial intelligence / machine learning (AI / ML) model in a wireless communication network includes verifying 202 authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not. For example, the device may be implemented as a user equipment (UE), i.e., a subscriber to the network, which may be interpreted as a device having a SIM card or an eSIM card.

[0072] The method includes training 200 an AI / ML model using the positioning data based on a validation result indicating that the positioning data is authorized for training.

[0073] According to an embodiment, verifying the authorization is performed with the first network function NF, and wherein the training is performed at the first NF or the second NF.

[0074] According to an embodiment, the NF includes a location management function LMF; and wherein the second NF includes at least one of a model training logic function MTLF and a network data analysis function NWDAF.

[0075] According to an embodiment, verification of the positioning data is performed to obtain authorized positioning data, thereby forming a training data set consisting of data of a single deployment scenario.

[0076] According to an embodiment, verification of the positioning data is performed to obtain authorized positioning data, thereby forming a training data set consisting of multiple deployment scenarios.

[0077] In conjunction with the embodiments, the positioning data may include measurement data related to the location of the device reported from the user equipment UE and / or the base station gNB.

[0078] According to an embodiment, positioning data may be considered to include one or more measurement results related to a determined final position of each UE, where the final position is a position calculated by the Location Management Function (LMF) based on measurement data related to the location of the device. Such measurement results may be reported by the user equipment (UE) and / or the base station (gNB).

[0079] Figure 3 A flow chart of a method related to a case where the authorization in step 204 is unsuccessful according to a further embodiment of the present invention is shown.

[0080] According to an embodiment, the method includes excluding positioning data from training an AI / ML model based on a verification result indicating that the positioning data is not authorized for training. According to some embodiments, authorization is associated with a device associated with a location and indicates whether information obtained about the device is authorized for training. According to some embodiments, authorization is associated with at least one of the reliability of the positioning data, the accuracy of the positioning data, an identifier of the device associated with the positioning data, etc. That is, the positioning data may be considered valid or useful, that is, authorized or not with respect to the device, the type of the device, the location of the device, the user of the device, or the like, but for example, may be considered authorized or unauthorized based on the accuracy of the positioning data, and a method is implemented for obtaining the positioning data, that is, a specific downlink-based method or process and / or a specific uplink-based process. By selecting the positioning data used to train the model, it can be ensured that information that is invalid or that would degrade the model can be excluded from model training, thereby allowing for high quality of the AI / ML model.

[0081] Figure 4 A flowchart illustrating at least a portion of a method according to an embodiment of the present invention is shown. The AI / ML model can be repeatedly trained or repeatedly updated.

[0082] According to an embodiment, the method includes repeatedly training the AI / ML model 402 over multiple iterations, such that authorization is verified for each or a subset of the multiple iterations. For example, each time location data of a device is obtained, it may be verified whether the data, information, or measurement results are provided to the model training entity. To save some verification work, it may be considered to skip or avoid such verification, for example, every two or three times, or the like, or vice versa, to verify authorization from time to time.

[0083] Figure 5 A flow chart of a method according to an embodiment of the present disclosure is shown. Figure 5 A first solution involves a location management function and a model training entity based on and communicating with the location management function.

[0084] According to an embodiment, a method includes obtaining 502 measurement data from a location management function (LMF), the measurement data relating to a location measurement procedure performed in a wireless communication network. The measurement data from the LMF may include one or both of the following: one portion is measurement data from the gNB / UE. The other portion is measurement results from the LMF, which calculates the final location of the UE based on the measurement data. The method includes forwarding 405 the positioning data and / or the measurement data as at least part of the positioning data to a model training entity based on a verification result indicating that the positioning data is authorized for training. The method further includes extracting 506 an AI / ML model with the model training entity. For example, the model training entity is an entity that includes a model training logic function (MTLF). For example, the model training entity may be or may include a network data analysis function (NWDAF) of the wireless communication network.

[0085] Figure 6 A flow chart of a method according to a further embodiment of the present disclosure is shown. Thus, a subscription request may be provided to the LMF to receive positioning data.

[0086] According to an embodiment, the method includes transmitting 602 a data subscription request from a model training entity for data collection for AI / ML model training to a location management function (LMF). That is, the model training entity may request location data or the results of the data collection. The method may include validating 604 the data subscription request at the LMF to verify authorization for location data related to the location of the device, thereby obtaining at least a partial validation result. The method may include determining 606 the location data of the device using the LMF, and may include providing 608 the location data to the model training entity based on the validation result.

[0087] According to some embodiments, the model inference entity is or may include a location management function, one or more of which may be operable in a wireless communication network.

[0088] According to an embodiment, the method includes providing a verification result to the model training entity. Alternatively, the method may include providing the verification result and the positioning data to the model training entity. The method may be adapted such that the verification result indicates that the positioning data is authorized for training.

[0089] Figure 7 A flow chart according to a further embodiment of the present invention and relating to a verification performed according to the first solution is shown.

[0090] According to an embodiment, the method includes transmitting 702 a model subscription request for obtaining an AI / ML model to a model training entity using a model reasoning entity. For example, the model reasoning entity can be an LMF, an access and mobility function AMF, or other entity using an AI / ML model. The method may include validating 704 the model subscription request at the model training entity to obtain a subscription validation result. The method may further include training 706 the AI / ML model at the model training entity to obtain a trained model. The method may include providing 708 the trained model to the model reasoning entity based on the subscription validation result. That is, the model reasoning entity may request submission of the trained model and may receive the model based on successful validation of the request.

[0091] Figure 8 A flowchart of a method according to a further embodiment of the present disclosure and related to the second solution provided by the embodiments described herein is shown. According to this solution, LMF can train AI / ML models.

[0092] According to an embodiment, a method includes obtaining 802, at a location management function (LMF), positioning data related to a location measurement process performed in a wireless communication network. The method may include using 804 the positioning data for model training in the LMF based on a verification result indicating that the positioning data is authorized for training, so as to train an AI / ML model using the LMF.

[0093] According to an embodiment, the LMF may include a model training logic function MTLF.

[0094] Figure 9 A flow chart of the method according to this embodiment is shown.

[0095] According to an embodiment, the method may include using the model training logic function MTLF of 902LMF to train the AI / ML model.

[0096] Figure 10 A flow chart of a method according to a further embodiment of the present invention and relating to the proposal of an AI / ML model is shown.

[0097] According to an embodiment, the method includes obtaining 1002 a newly generated AI / ML model or an updated model obtained from a previous model by training the AI / ML model. In other words, the positioning data can be used to generate a new model or update an existing model. Accordingly, a trained AI / ML model can be obtained.

[0098] According to an embodiment, a computer-readable storage medium stores instructions which, when executed, cause one of the methods described herein to be performed by a location management function LMF and / or a model training entity of a wireless communication system.

[0099] Figure 11 11 is a block diagram illustrating system operation according to an embodiment. During system operation, UE 1101 may provide 1121 positioning data or a basis thereof, for example, by providing measurement results. System 1103 may include at least one network function (NF). In the illustrated example, system 1103 includes two NFs 1105 and 1107. In other embodiments, system 1103 may include only a single NF or more than two NFs. NF 1105 may request 1123 that NF 1107 provide a training dataset or positioning data. In response to verification 1125, NF 1107, with the assistance of UDM 1114, may provide 1127 authorized positioning data to NF 1105 to enable model training by NF 1105. While NF 1107 may be capable of model training, system 1103 may be implemented without NF 1105. Embodiments may be implemented and / or executed in an operational wireless communication system, but may also be used in a training system, such as an offline or non-operating wireless communication system.

[0100] According to an embodiment, a wireless communication system is configured to perform a method for training an artificial intelligence / machine learning (AI / ML) model in a wireless communication network. The method is performed by a system including a first network function (NF) and a second NF. The method includes: verifying, using the second NF, authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not. The method includes transmitting the verification result to the first NF and, based on the verification result indicating that the positioning data is authorized for training, training the AI / ML model using the positioning data with the first NF.

[0101] According to an embodiment, a system is configured to: train an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system including a first network function NF and a second NF, the system being configured to: verify, using the second NF, the authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; transmit the verification result and the positioning data to the first NF; and, based on the verification result indicating that the positioning data is authorized for training, train the AI / ML model using the positioning data with the first NF.

[0102] According to an embodiment, the system is implemented to transmit positioning data to the first NF, thereby indicating that the positioning data is authorized for training.

[0103] According to an embodiment, the first NF includes a model training logic function MTLF and / or a network data analysis function NWDAF.

[0104] According to an embodiment, a system is configured to: train an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system including at least a first network function NF, the system being configured to: use the first NF to verify the authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; based on the verification result indicating that the positioning data is authorized for training, use the positioning data to train the AI / ML model with the first NF.

[0105] According to an embodiment, the system is implemented such that the first NF is adapted to store a plurality of positioning data and exclude positioning data that is not authorized for training from training.

[0106] According to an embodiment, the first NF includes a Location Management Function LMF.

[0107] According to an embodiment, a wireless communication system is configured for a method of training an artificial intelligence / machine learning AI / ML model in a wireless communication network, the method being performed by a system including a first network function NF and a second NF, the method including: using the second NF to verify the authorization of positioning data related to a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; sending the verification result to the first NF; and based on the verification result indicating that the positioning data is authorized for training, using the positioning data to train the AI / ML model with the first NF.

[0108] According to an embodiment, a system is configured to: train an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system including a first network function NF and a second NF, the system being configured to: use the second NF to verify the authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; transmit the verification result and the positioning data to the first NF; and based on the verification result indicating that the positioning data is authorized for training, use the positioning data to train the AI / ML model with the first NF.

[0109] According to an embodiment, the positioning data is transmitted to the first NF, thereby indicating that the positioning data is authorized for training.

[0110] According to an embodiment, the first NF includes a model training logic function MTLF and / or a network data analysis function NWDAF.

[0111] According to an embodiment, a system is configured to: train an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system including at least a first network function NF, the system being configured to: use the first NF to verify the authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; based on the verification result indicating that the positioning data is authorized for training, use the positioning data to train the AI / ML model with the first NF.

[0112] According to an embodiment, the first NF is adapted to store a plurality of positioning data and exclude positioning data that is not authorized for training from training.

[0113] According to an embodiment, the first NF includes a Location Management Function LMF.

[0114] Figure 12 A block diagram illustrates signaling that may be implemented in a wireless communication network according to an embodiment and a first solution. For example, a user equipment (UE) 1102, a base station (gNB) 1104, an AMF 1106, a NWDAF 1108, a LMF 1112, a unified data management (UDM) 1114, and an LCS client 1116 may operate in the network. With respect to the present disclosure, the AMF 1106, the UDM 1114, and / or the LCS client 1116 may be optional, among other things.

[0115] According to the first solution, NWDAF 1108 is assumed to be the AI ​​model training entity responsible for collecting data for AI / ML model training. This embodiment addresses the definition of which entity trains the model for direct AI / ML targeting, as well as the situation where the entity training the model and the model consumer are different, and how the model consumer obtains or receives the trained AI / ML model. A method for training AI / ML models based on authorization is proposed.

[0116] For example, while the UE 1102 and gNB 1104 may include one or more antennas for wirelessly transmitting and / or receiving signals in a wireless communication network, other entities may be implemented as an option without a wireless interface, but this is not excluded. While the AMF 1106, NWDAF 1108, LMF 1112, UDM 1114, and / or LCS client 1116 may implement this, one or more of them may also be connected to the wireless communication network via optical fiber and / or wired connections, or a combination thereof.

[0117] In 1122, when LMF 1112 performs positioning services, an AI model training entity, such as NWDAF 1108, subscribes to data collection for AI model training services, such as LMF 1112, to collect positioning data, such as measurement data from UE and / or gNB 1102 / 1104. This subscription can be based on Figure 6 If the subscription request is accepted, LMF 1112 may send the positioning data to the requesting entity, such as NWDAF 1108, for use in AI model training when performing positioning services.

[0118] In 1124, when a new / updated AI model is generated after AI / ML model training, the AI ​​model training entity, such as LMF 1112, can subscribe to the AI ​​model service provided by the model training entity, such as NWDAF 1108, to obtain the newly generated AI / ML model. In the case of accepting the subscription request, NWDAF 1108 can send the updated / trained AI model to LMF 1112. This can be done according to the Figure 7 Relevant disclosure.

[0119] In 1126, the AMF 1106 may receive an LCS service request, e.g., from the UE, e.g., as a Mobile Originated Location Request (MOLR), and / or from a different entity, such as the LCS Client 1116, e.g., as a Mobile Terminating Location Request (LR), or the AMF 1106 itself.

[0120] In 1128, the AMF 1106 may select the LMF 1112, for example, select one from a plurality of LMFs, and send a positioning request to the LMF 1112 for positioning.

[0121] In 1129, LMF 1112 may check or verify whether the UE's positioning data is authorized for AI / ML model training using UDM 1114. For example, this may include verifying whether there is a new indication in the UE subscription data to indicate whether it is authorized or unauthorized.

[0122] In 1132, UE 1102 and / or gNB 1104 can report positioning measurements, for example, to LMF 1112, and LMF 1112 can calculate the UE final position, i.e., the position can be determined based on the measurement results.

[0123] In 1134, if the NWDAF 1108 has subscribed to data collection for AI model training services, see 1122 and / or 1124, the LMF 1112 may send measurement data and / or other location-related information to the NWDAF 1108. In 1136, the NWDAF 1108 may perform AI / ML model training by using the measurement data / positioning data, such as received from the LMF 1112.

[0124] In 1138, LMF 1112 may send the determined or final UE location to, for example, AMF 1106.

[0125] In 1142, the AMF 1106 may send the UE location to an LCS consumer, such as the UE 1102, gNB 1104, and / or LCS client 1106, or other entities. In other words, Figure 12 AI model training based on subscription service is shown.

[0126] Figure 13 A schematic block diagram illustrating signaling in a wireless communication network according to an embodiment of the second solution is shown. According to the second solution, it is assumed that the LMF 1112 is an AI model training entity and can collect data from the NWDAF 1108 for AI model training. Figure 12 In contrast, 1122, 1124 and / or 1134 can be skipped. In addition, Figure 12 Compared to training the AI ​​model in 1136 at NWDAF 1108 , model training can be performed in 1236 at LMF 1112 .

[0127] The embodiments point out that known solutions only suggest that the NWDAF / LMF acts as an AI model training entity by collecting the necessary data. However, the present disclosure recognizes that this does not take into account the authorization from the UE / network side, i.e. whether the UE's measurement data is allowed / authorized to be used for AI / ML model training. The embodiments suggest that before the NWDAF / LMF starts training the AI ​​model or updating it, an authorization check is performed using, for example, the UDM, for example, when performing AI-based positioning, by defining a new indication in the UE subscription data in the UDM to indicate whether the measurement data can be used for AI model training. The UDM may already include or define a set of UE subscription data. The embodiments may include defining a new indicator, for example, for the UE, to indicate whether positioning data can be used for AI model training.

[0128] To train the AI / ML model for positioning, embodiments may use UE measurement information as model input and / or base station / gNB measurement information as model input.

[0129] The AI / ML models mentioned herein can be used for, for example, direct AI / ML positioning; assisted AI / ML positioning; assisted AI / ML positioning using a multi-TRP configuration; assisted positioning using a single model with a single TRP configuration and N TRPs; and assisted positioning using N models with a single TRP configuration and N TRPs. The embodiments described herein relate not only to model generation but also to model monitoring and refinement. Depending on the embodiment, training an AI / ML model results in an updated, retrained, or refined model from the previous model.

[0130] According to an embodiment, the positioning dataset comprises a synthetic dataset.

[0131] According to an embodiment, the synthetic data set is generated based on a statistical channel model.

[0132] Depending on the embodiment, the dataset authorized for model training is obtained based on model generalization or without considering generalization.

[0133] According to some embodiments, the training data used to train the model may include using, among other things, synthetic datasets, such as synthetic datasets generated based on statistical channel models. For example, such data can provide assistance for model training, validation, and testing.

[0134] For the evaluation of the AI / ML-assisted localization process, one or more of the following intermediate performance indicators can be used:

[0135] -LOS classification accuracy, if the model output includes hard-valued (non-)line-of-sight LOS / NLOS indicators, where the LOS / NLOS indicators are generated for the link between the UE and the TRP;

[0136] - Timing estimate accuracy (expressed as a metric), if the model output includes timing estimates (e.g., ToA, RSTD).

[0137] - Angle estimation accuracy (in degrees), if the model output includes angle estimates (e.g. AoA, AoD).

[0138] According to embodiments, by selecting, controlling, and / or validating training data, the reliability of the data in terms of ground truth associated with the model training data may be improved.

[0139] For direct AI / ML targeting, the performance of model monitoring methods may be based on, for example, label-based methods and / or label-free methods.

[0140] For AI / ML-assisted positioning, it can be found that the label-based model monitoring method uses the TOA and / or LOS / NLOS indicator as the model output, and the estimated true value label (i.e., TOA and / or LOS / NLOS indicator) is provided by the position estimate of the associated conventional positioning method. The associated conventional positioning method refers to the method that uses the AI / ML model output to determine the target UE location.

[0141] Embodiments allow for improved localization accuracy on test datasets through better construction of training datasets. For example, a training dataset may consist of data from multiple deployment scenarios, including data from the same deployment scenario as the test dataset. Some embodiments further incorporate model fine-tuning / retraining, where the model can be retrained / fine-tuned using datasets from the same deployment scenario as the test dataset.

[0142] The embodiments utilize the finding that when an AI / ML model is trained using a poor-quality dataset, such as one (single) deployment scenario, and tested using a dataset from a different deployment scenario, localization accuracy may deteriorate, whereas localization accuracy on the test dataset can be improved by using a better training dataset construction and / or model fine-tuning / retraining. This improvement can be achieved by licensing data from this dataset according to the embodiments.

[0143] For example, better training dataset construction can be incorporated to ensure that the training dataset consists of data from multiple deployment scenarios, including data from the same deployment scenarios as the test dataset.

[0144] When using model fine-tuning / retraining, the model can be retrained / fine-tuned using a dataset from the same deployment scenario as the test dataset. For this purpose, embodiments can be used to select / validate the training data for retraining / fine-tuning.

[0145] For example, when using models for direct AI / ML and / or AI / ML-assisted positioning, the dataset can be constructed based on predefined data points or density of points, number of sources used, accuracy of results and / or information sources, where the predefined values ​​can include predefined absolute or relative thresholds or threshold ranges.

[0146] Embodiments allow for the generation and / or construction of datasets for model training using or without the concept of model generalization, where AI / ML models are trained and tested using datasets from the same deployment scenario. This is because, by verifying that the measurement information in the network is authorized, datasets can be selected based on specific requirements, where these requirements may change over time and / or may account for changes in the measurement data, such as over time. Alternatively or in addition, training datasets can be collected for the same or different models based on the positioning method used for model inference.

[0147] According to embodiments, the effect of training data sample density (i.e., the size of the training data set for a given evaluation area) can be compensated for through authorization and / or active settings. For example, evaluation using a uniform UE distribution can show that larger training data set sizes (i.e., higher sample density) result in smaller positioning errors (in meters) until a saturation point is reached, at which point additional training data does not lead to further improvements in positioning accuracy.

[0148] Embodiments allow obtaining a dataset for model training to obtain an improved training dataset construction (i.e., a hybrid dataset), where the training dataset consists of data from multiple deployment scenarios, including data from the same deployment scenario as the test dataset.

[0149] Some embodiments also involve fine-tuning / retraining, where the model is retrained / fine-tuned using a dataset from the same deployment scenario as the test dataset.

[0150] Figure 14 is a block diagram illustrating an electronic device 1300 according to an embodiment of the present invention.

[0151] Electronic device is used to represent various forms of digital computers, such as laptops, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. Electronic device may also represent various forms of CN network entities. The components shown herein, their connections and relationships, and their functions are described by way of example only and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0152] refer to Figure 14 , device 1300 includes a computing unit 1301 that performs various appropriate operations and processes according to computer program instructions stored in a read-only memory (ROM) 1302 or loaded from a storage unit 1308 into a random access memory (RAM) 1303. Various programs and data for the operation of the memory device 1300 may also be stored in the RAM 1303. The computing unit 1301, ROM 1302, and RAM 1303 are interconnected via a bus 1304. An input / output (I / O) interface 1305 is also connected to the bus 1304.

[0153] The components in device 1300 are connected to I / O interface 1305, including: input unit 1306, such as a keyboard and mouse; output unit 1307, such as various types of displays and speakers; storage unit 1308, such as a magnetic disk and optical disk; and communication unit 1309, such as a network card, modem, wireless communication transceiver, etc. Communication unit 1309 allows device 1300 to exchange information / data with other devices via computer networks such as the Internet and various telecommunication networks.

[0154] The computing unit 1301 may be composed of various general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 1301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1301 performs the various methods and processes described above, such as the image processing method. For example, in some embodiments, the image processing method may be implemented as a computer software program, which may be tangibly embodied on a machine-readable medium, such as the storage unit 1308. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 1300 via the ROM 1302 and / or the communication unit 1309. When the computer program is loaded into the RAM 1303 and executed by the computing unit 1301, one or more steps of the image processing method described above may be performed. In some embodiments, the computing unit 1301 may be configured to perform the image processing method in any other suitable manner (e.g., via firmware).

[0155] Various implementations of the above-described systems and techniques can be implemented in digital electronic circuits, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can be implemented in one or more computer programs executable and / or interpretable on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor and can receive data and instructions from a storage system, at least one input device, and at least one output device, and can transmit data and instructions to the storage system, at least one input device, and at least one output device.

[0156] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the processor or controller executes the program code, the functions and / or operations specified in the flowcharts and / or block diagrams are performed. The program code can be executed entirely or partially on the machine, and can be executed partially on the machine as a stand-alone software package, partially on a remote machine, or entirely on a remote machine or server.

[0157] In the context of the present invention, a machine-readable medium may be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0158] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer. The computer has a display device (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD)) to display information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user. For example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0159] The systems and techniques described herein can be implemented on a computing system that includes a back-end component (e.g., as a data server), or a computing system that includes a middleware component (e.g., an application server), or a computing system that includes a front-end component (e.g., a user computer with a graphical user interface or web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computer system that includes such back-end components, middleware components, front-end components, or any combination thereof. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0160] A computer system may include a client and a server. The client and server are typically remote from each other and typically interact via a communications network. The client-server relationship arises from computer programs running on their respective computers and establishing a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within a cloud computing service system that addresses the management difficulties and weak business development capabilities of traditional physical hosts and virtual private servers (VPSs). The server may also be a server in a distributed system or a server integrated with a blockchain.

[0161] It should be understood that the various forms of flows described above can be used to reorder, add, or delete steps. For example, as long as the expected results of the technical solutions in the present invention can be achieved, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, and no limitations are imposed herein.

[0162] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and replacements may be made based on design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the scope of the principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for training an artificial intelligence / machine learning (AI / ML) model in a wireless communication network, the method comprising: verifying authorization of positioning data associated with a device in a wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not authorized for training; The AI / ML model is trained using the positioning data based on the verification result indicating that the positioning data is authorized for training.

2. The method of claim 1 , wherein verifying the authorization is performed with a first network function NF, and wherein training is performed at the first NF or the second NF.

3. The method of claim 2, wherein the NF comprises a location management function LMF; and wherein, The second NF includes a network data analysis function NWDAF. The method of claim 3 , wherein the NWDAF comprises a model training logic function (MTLF).

5. A method as claimed in any preceding claim, wherein the authorisation is associated with a device associated with the location and indicates whether information obtained in relation to the device is authorised for use in training.

6. A method as claimed in any one of the preceding claims, comprising: The positioning data is excluded from training of the AI / ML model based on a verification result indicating that the positioning data is not authorized for training.

7. A method as claimed in any one of the preceding claims, wherein the authorization is related to at least one of: Deployment scenarios for obtaining positioning data; predefined density of data points, The number of sources of location data, Accuracy of positioning data; The location management function LMF or the network data analysis function NWDAF sends an AI model training authorization request to the unified data management UDM. The UDM checks whether there is an authorization indication for AI model training in the UE subscription data.

8. A method as claimed in any one of the preceding claims, wherein the positioning data comprises measurement data related to the location of the device reported from the user equipment UE and / or the base station gNB.

9. A method as claimed in any one of the preceding claims, wherein the positioning data comprises measurement results related to the determined final position of the UE calculated by the location management function LMF based on measurement data related to the position of the device reported from the user equipment UE and / or the base station gNB.

10. A method as claimed in any one of the preceding claims, wherein the AI / ML model is repeatedly trained in a plurality of iterations, wherein authorization is verified for each or a subset of the plurality of iterations.

11. A method as claimed in any one of the preceding claims, comprising: collecting positioning data in the location management function LMF, the positioning data being related to a location measurement process performed in the wireless communication network; as well as forwarding the positioning data to a model training entity based on the verification result indicating that the positioning data is authorized for training; and Train AI / ML models in the model training entity.

12. The method of claim 11, wherein the model training entity is an entity comprising a network data analysis function (NWDAF). The method of claim 12 , wherein the NWDAF comprises a model training logic function (MTLF).

14. The method according to any one of claims 11 to 13, comprising: Transmitting a data subscription request for data collection for training AI / ML models to the location management function LMF; Verifying the data subscription request at the LMF and / or the network data analysis function NWDAF to verify authorization of the positioning data related to the location of the device, thereby obtaining a verification result; Based on the validation results, the positioning data in the model training entity is used.

15. The method according to any one of claims 11 to 14, comprising: Transmitting a model subscription request for obtaining an AI / ML model at a model inference entity and a model training entity; Training the AI / ML model at the model training entity to obtain a trained model; as well as Provide the trained model to the Model Inference entity. The method of claim 15 , wherein the model reasoning entity is or includes a location management function (LMF).

17. A method as claimed in any one of the preceding claims, comprising: obtaining, at a location management function LMF, positioning data related to a location measurement procedure performed in the wireless communication network; as well as Based on the verification result indicating that the positioning data is authorized for training, the positioning data is used for model training in the LMF, wherein the AI / ML model is trained in the LMF.

18. The method of any one of the preceding claims, wherein training the AI / ML model results in a newly generated AI / ML model.

19. A method as claimed in any one of the preceding claims, wherein the dataset authorized for model training is obtained based on model generalization or without considering generalization.

20. A computer-readable storage medium storing instructions which, when executed, cause the method of any one of the preceding claims to be performed by a location management function LMF and / or a model training entity of a wireless communication system.

21. A network entity, such as a location management function (LMF), configured to operate in a wireless communication network, the network entity being configured to: verifying authorization of positioning data associated with a device in a wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not; and Based on the verification result indicating that the positioning data is authorized for training, the positioning data is provided to the model training entity or the positioning data is used for model training.

22. A wireless communication system configured to A method for training an artificial intelligence / machine learning AI / ML model in a wireless communication network, the method being performed by a system including a first network function (NF) and a second NF, the method comprising: verifying authorization of positioning data associated with a device in a wireless communication network using a second NF to obtain a verification result indicating whether the positioning data is authorized for training or not; Transmit the verification result to the first NF; Based on the verification result indicating that the positioning data is authorized for training, the AI / ML model is trained with the first NF using the positioning data.

23. A system configured to: Training an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system comprising a first network function NF and a second NF, the system being configured to: verifying authorization of positioning data associated with a device in a wireless communication network using a second NF to obtain a verification result indicating whether the positioning data is authorized for training or not; Transmit the verification result and positioning data to the first NF; The verification result and positioning data are transmitted to the first NF.

24. The system of claim 23, wherein the positioning data is transmitted to the first NF, thereby indicating that the positioning data is authorized for training.

25. The system of claim 23 or 24, wherein the first NF comprises a model training logic function (MTLF) and / or a network data analysis function (NWDAF).

26. A system configured to: Training an artificial intelligence / machine learning AI / ML model in a wireless communication network, the system comprising at least a first network function NF, the system being configured to: verifying, using the first NF, authorization of positioning data associated with a device in the wireless communication network to obtain a verification result indicating whether the positioning data is authorized for training or not; Based on the verification result indicating that the positioning data is authorized for training, the AI / ML model is trained with the first NF using the positioning data.

27. The system of claim 26, wherein the first NF is adapted to store a plurality of positioning data and exclude positioning data that is not authorized for training from the training.

28. The system of claim 26 or 27, wherein the first NF comprises a location management function (LMF).

Images