Mechanisms for secured machine learning (ML) model training at user equipment (UE)

Abstract

A method includes receiving, by a service authorization entity of a wireless communication system from an ML model training service consumer entity, a service request for accessing an ML model training service of a UE, where the service request comprises information associated with the requested ML model training service; verifying, by the service authorization entity, the service request, where the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; and transmitting, by the service authorization entity to the ML model training service consumer, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service.

Description

MECHANISMS FOR SECURED MACHINE LEARNING (ML) MODEL TRAINING AT USER EQUIPMENT (UE)FIELD

[0001] Various example embodiments relate generally to wireless networks and, more particularly, to machine learning (ML) model authentication and training.BACKGROUND

[0002] Wireless networking provides significant advantages for user mobility. A user’s ability to remain connected to a wireless network while on the move provides advantages not only for the user, but also provides greater efficiency and productivity for society as a whole. As user expectations for, e.g., security and usage of ML-assisted operations, become more demanding, technology for wireless networking must also keep pace with such expectations. Accordingly, there is continuing interest in improving wireless networking technology.SUMMARY

[0003] As used herein, the term “entity” may refer to a distinct component of an architecture or device, a network device, a network node, a network function, or any other devices (physical or virtual) in a communication network.

[0004] In accordance with aspects of the present disclosure, a method includes transmitting, by a user equipment (UE) to a network, at least one UE capability associated with a machine learning (ML) model training service provided by the UE; receiving, by the UE, an instruction for providing the ML model training service; and providing, by the UE, the ML model training service.

[0005] In an aspect of the method, the instruction is received based on transmitting the at least one UE capability.

[0006] In an aspect of the method, the providing the ML model training service includes receiving, by the UE, a request for providing the ML model training service; determining to provide, by the UE, the requested ML model training service, where the determining is based on the received request and on the received instruction; and providing, by the UE, based on the determining, the requested ML model training service.

[0007] In an aspect of the method, the request for providing the ML model training service includes one or more parameters of the requested ML model training service; and the determining is further based on at least one of the one or more parameters.

[0008] In an aspect of the method, the one or more parameters of the ML model training service include at least one of the following: one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers.

[0009] In an aspect of the method, the receiving the instruction includes receiving the instruction including a policy, where the policy includes at least one of the following: an indication of one or more first ML models allowed to be trained at the UE, an indication of one or more second ML models not allowed to be trained at the UE, an indication of one or more first categories of ML models allowed to be trained at the UE, an indication of one or more second categories of ML models not allowed to be trained at the UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the UE, an indication of one or more first types of data of the UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the UE not allowed to be used for training the one or more second ML models.

[0010] In an aspect of the method, the receiving the instruction includes receiving the instruction including an access token, where the access token includes token claims for at least one the following service consumer information indicating an ML model service consumer entity allowed to use the ML model training service provided by one or more UEs, service provider information indicating one or more UE identifiers, where each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the ML model training service, and where a first UE identifier of the one or more UE identifiers identifies the UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the ML model training service: a model identifier of the ML model, a model category of the ML model, one or more analytics identifiers, where each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model, or one or more data type identifiers, where each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

[0011] In an aspect of the method, the access token includes a JavaScript object notation (JSON) web token (JWT).

[0012] In an aspect of the method, the first UE identifier identifying the UE includes an address for communicating with the UE.

[0013] In an aspect of the method, the address for communicating with the UE includes a user plane address of the UE.

[0014] In an aspect of the method, a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information on a wireless network interface, received signal strength on a wireless network interface, location data, photos, videos, or sensor data.

[0015] In accordance with aspects of the present disclosure, a method includes receiving, by a service authorization entity of a wireless communication system from a machine learning (ML) model training service consumer entity, a service request for accessing an ML model training service of a user equipment (UE), where the service request includes information associated with the requested ML model training service; verifying, by the service authorization entity, the service request, where the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; and transmitting, by the service authorization entity to the ML model training service consumer, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service.

[0016] In an aspect of the method, the information associated with the requested ML model training service includes at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

[0017] In an aspect of the method, the method further includes determining, based the one or more filtering criteria, the at least one UE that allows the requested ML model training service.

[0018] In an aspect of the method, the ML model training service consumer entity includes at least one of the following: an access network entity of the wireless communication system, a core network entity of the wireless communication system, or an application function entity.

[0019] In an aspect of the method, the core network entity is a network function.

[0020] In an aspect of the method, the network function is a network data analytics function entity.

[0021] In an aspect of the method, the network data analytics function entity includes a model training logical function.

[0022] In an aspect of the method, the verifying the service request is further based on an authentication of the ML model training service consumer entity.

[0023] In an aspect of the method, the method further includes transmitting, by the service authorization entity to the at least one UE, based on the verifying, a policy including at least one of the following: an indication of one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first categories of ML models allowed to be trained at the at least one UE, an indication of one or more second categories of ML models not allowed to be trained at the at least one UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models.

[0024] In an aspect of the method, the method further includes transmitting, by the service authorization entity to an access network node associated with the at least one UE, based on the verifying, a policy including at least one of the following: an indication of one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first categories of ML models allowed to be trained at the at least one UE, an indication of one or more second categories of ML models not allowed to be trained at the at least one UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models.

[0025] In an aspect of the method, the response indicating the at least one UE that allows the requested ML model training service includes an access token.

[0026] In an aspect of the method, the access token includes token claims for at least one the following service consumer information indicating the ML model training service consumer entity allowed to utilize the requested ML model training service, service provider information indicating one or more UE identifiers, where each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model trainingservice, and where a first UE identifier of the one or more UE identifiers identifies the at least one UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the requested ML model training service: a model identifier of the ML model, a model category of the ML model, one or more analytics identifiers, where each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model, or one or more data type identifiers, where each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

[0027] In an aspect of the method, the method further includes generating, by the service authorization entity based on the verifying, the access token.

[0028] In an aspect of the method, the access token includes a JavaScript object notation (JSON) web token (JWT).

[0029] In an aspect of the method, the first UE identifier identifying the at least one UE includes an address for communicating with the at least one UE.

[0030] In an aspect of the method, the address for communicating with the at least one UE includes a user plane address of the at least one UE.

[0031] In an aspect of the method, a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information on a wireless network interface, received signal strength on a wireless network interface, location data, photos, videos, or sensor data.

[0032] In an aspect of the method, the service authorization entity of a wireless communication system is a network function.

[0033] In an aspect of the method, the network function is a unified data management entity.

[0034] In accordance with aspects of the present disclosure, a method includes transmitting, by a machine learning (ML) model training service consumer entity, to a service authorization entity of a wireless communication system, a service request for accessing an ML model training service of a user equipment (UE), where the service request includes information associated with the requested ML model training service; receiving, by the ML model training service consumer entity from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and transmitting, by the ML model training service consumer entity to the at least one UE, based on the receiving the response, a training request including information associated with an ML model to be trained at the at least one UE.

[0035] In an aspect of the method, the information associated with the requested ML model training service includes at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

[0036] In an aspect of the method, the ML model training service consumer entity includes at least one of the following: an access network entity of the wireless communication system, a core network entity of the wireless communication system, or an application function entity.

[0037] In an aspect of the method, the response indicating the at least one UE that allows the requested ML model training service includes an access token.

[0038] In an aspect of the method, the access token includes token claims for at least one the following service consumer information indicating the ML model training service consumer entity allowed to utilize the requested ML model training service provided by one or more UEs, service provider information indicating one or more UE identifiers, where each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model training service, and where a first UE identifier of the one or more UE identifiers identifies the at least one UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the requested ML model training service: a model identifier of an ML model, a model category of the ML model, one or more analytics identifiers, where each of the one or more analytics identifiers is associated with an analytics operation allowed for the ML model, or one or more data type identifiers, where each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

[0039] In an aspect of the method, the first UE identifier identifying the at least one UE includes an address for communicating with the at least one UE.

[0040] In an aspect of the method, the address for communicating with the at least one UE is a user plane address of the at least one UE.

[0041] In an aspect of the method, the training request transmitted to the at least one UE further includes the access token.

[0042] In an aspect of the method, the access token includes a JavaScript object notation (JSON) web token (JWT).

[0043] In an aspect of the method, the information associated with the ML model in the training request includes at least one of one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers.

[0044] According to some aspects, there is provided the subject matter of the independent claims. Some further aspects are defined in the dependent claims.BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Some example embodiments will now be described with reference to the accompanying drawings.

[0046] FIG. 1 is a diagram of an example embodiment of wireless networking between a network system and a user equipment (UE), according to one illustrated aspect of the disclosure;

[0047] FIG. 2 is a diagram of example components of a network system, according to one illustrated aspect of the disclosure;

[0048] FIG. 3 is a diagram of an example embodiment of operations for provisioning a secured machine learning (ML) model training at a UE, according to one illustrated aspect of the present disclosure;

[0049] FIG. 4 is a diagram of an example embodiment of operations for provisioning a secured ML model training at a UE, according to one illustrated aspect of the present disclosure;

[0050] FIG. 5 is a flow diagram of example operations of a UE providing an ML model training service, according to one illustrated aspect of the present disclosure;

[0051] FIG. 6 is a flow diagram of example operations of a network apparatus provisioning a secured ML model training at a UE, according to one illustrated aspect of the present disclosure;

[0052] FIG. 7 is a flow diagram of example operations of a network apparatus requesting an ML model training service provided by a UE, according to one illustrated aspect of the present disclosure; and

[0053] FIG. 8 is a diagram of example embodiment of components of a UE or of a network apparatus, according to one illustrated aspect of the present disclosure.DETAILED DESCRIPTION

[0054] In the following description, certain specific details are set forth in order to provide a thorough understanding of disclosed aspects. However, one skilled in the relevant art willrecognize that aspects may be practiced without one or more of these specific details or with other methods, components, materials, etc. In other instances, well-known structures associated with transmitters, receivers, or transceivers have not been shown or described in detail to avoid unnecessarily obscuring descriptions of the aspects.

[0055] Reference throughout this specification to “one aspect” or “an aspect” means that a particular feature, structure, or characteristic described in connection with the aspect is included in at least one aspect. Thus, the appearances of the phrases “in one aspect” or “in an aspect” in various places throughout this specification are not necessarily all referring to the same aspect. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more aspects.

[0056] Embodiments described in the present disclosure may be implemented in wireless networking apparatuses, such as, without limitation, apparatuses utilizing Worldwide Interoperability for Microwave Access (WiMAX), Global System for Mobile communications (GSM, 2G), GSM EDGE radio access Network (GERAN), General Packet Radio Service (GRPS), Universal Mobile Telecommunication System (UMTS, 3G) based on basic wideband- code division multiple access (W-CDMA), high-speed packet access (HSPA), Long Term Evolution (LTE), LTE-Advanced, enhanced LTE (eLTE), 5G New Radio (5G NR), 5G Advance, 6G (and beyond) and 802.1 lax (Wi-Fi 6), among other wireless networking systems. The term ‘eLTE’ here denotes the LTE evolution that connects to a 5G core. LTE is also known as evolved UMTS terrestrial radio access (EUTRA) or as evolved UMTS terrestrial radio access network (EUTRAN).

[0057] Model training via network data analytics function (NWDAF) including a model training logical function (MTLF) has been described for example in clauses 5.1, 7.5 and 7.6 of 3rd Generation Partnership Project (3GPP) technical document TS 23.288 VI 8.3.0 (2023-09) (“TS 23.288 document”). In addition, various studies have been conducted on cross-domain artificial intelligence machine learning (Al ML), where a user equipment (UE) and a corresponding radio access network (RAN) and / or a UE and a corresponding core network may collaboratively or jointly train an ML model. Cross-domain Al ML may be an important aspect not only for 5G, but also for 6G and beyond. Cross-domain Al ML can be used to assist various operations in a wireless communication system, for example, including, but not limited to, network planning, automation of network operations, network slicing, and improving quality of service and user experience, etc.

[0058] For joint ML model training by a UE and a corresponding network (e.g., an access network and / or a core network), an ML model training process can be executed at the UE totrain at least a portion of an ML model. In some examples, the UE may partially train an ML model and provide the partially trained mode to the network for further training. In other examples, the UE may receive an ML model that is partially trained by the network and may further train the partially trained ML model at the UE. Currently, there is no mechanism for identifying and / or indicating the type(s) of ML models that are approved by the UE for training at the UE or identifying and / or indicating the type(s) of UE data that are allowed to be used for training an ML model at the UE.

[0059] Further, in some scenarios, a malicious node can be present in a network. In such scenarios, the malicious mode can transmit an unauthorized request to a UE to initiate an ML model training at the UE that may not be allowed by the UE and can additionally consume UE data that may not allowed by the UE. Currently, when a UE receives an ML model training request, the UE has no means to verify the purpose of the ML model training and / or the type of UE data that will be used to facilitate the ML model training.

[0060] The issues discussed above may become more severe in the case of inter-public land mobile network (PLMN) communications. For example, in some instances, a home public mobile land network (HPLMN) of the UE may have knowledge about ML models and / or analytics that the UE has agreed to train. However, when the UE is in a visiting public mobile land network (VPLMN), the VPLMN may not have such knowledge.

[0061] As used herein, and unless indicated otherwise, the terms “network system”, “network side”, and “network” (where used alone) may include both access network components and core network components.

[0062] As used herein, the term “ML model training service consumer entity” may refer to an entity that accesses or utilizes an ML model training service provided by a UE. That is, the UE may operate as an ML model training service provider entity. Examples of an ML model training service consumer entity may include an access network entity (e.g., a RAN model training management function), a core network entity (e.g., a 5G core model training management function, such as an NWDAF), or an application function (which may be part of a core network or outside of a core network). In the present disclosure, the terms “ML model training service consumer entity” and “MTLF entity” may be used interchangeably herein such that a description referring to one of the terms shall be treated as though the description also referred to the other term. In an example embodiment, “ML model training service consumer entity” or “MTLF entity” may refer to a MTLF of a network data analytics function (NWDAF) in a 5G network, as specified for example in clauses 5.1, 7.5 and 7.6 of 3GPP TS 23.288VI 8.3.0 (2023-09). In another example embodiment, these terms may refer to a network entity in a 6G network, or in later generations of 3GPP networks.

[0063] As used herein, the terms “ML model training service” or “ML model training process” may refer to performing operations for training an ML model, where the operations may include providing an ML algorithm with training data to find patterns and make predictions in achieving a particular task (e.g., associated with a network operation, a network service, or any processing associated with communications between a UE and a corresponding network). At the end of the training, the ML model may be referred to as a trained ML model that can perform the particular task. In the present disclosure, the terms “ML model training service” or “ML model training process” may be used interchangeably herein, such that a description referring to one of the terms shall be treated as though the description also referred to the other term.

[0064] Aspects of the present disclosure provide techniques for provisioning a secured ML model training at a UE, where the UE may verify (e.g., authenticate and / or authorize) an ML training request received from an ML model training service consumer entity.

[0065] According to aspects of the present disclosure, a secured ML model training at a UE can be provisioned via a policy or an access token. For instance, in one aspect, a policy may be pushed from a network (e.g., by a service authorization entity of the network, such as a user data management (UDM) entity, or any suitable network function entity) to the UE. The policy may indicate one or more ML model training service consumer entities that are permitted to request training of certain ML model(s) at the UE, the categories (or purposes) of ML models allowed to be trained at the UE, and / or the types of data stored at the UE allowed to be used during ML model training at the UE. As such, upon the UE receiving a training request to execute a ML model training process at the UE, the UE may verify the training request against the policy and determine to accept or reject the training request based on the verification. In another example, the policy may be transmitted to an access network node serving the UE instead of directly to the UE. In this way, the access network node may accept or reject the training request before the training request reaches the UE.

[0066] In another aspect, upon a service authorization entity receiving a service request from an ML model training service consumer entity for initiating an ML model training at a UE, the service authorization entity may verify the service request based on information indicate din the service request and information related to ML model training services provided by UE(s). Upon a successful verification, the service authorization entity may generate an access token with token claims corresponding to results of the verification and transmit thegenerated access token to the ML model training service consumer entity. Subsequently, when the ML model training service consumer entity requests the ML model training from the UE, the ML model training service consumer entity may include the access token along with request. Upon the UE receiving the training request, the UE may verify the access token (e.g., based on an authorization server key configured at the UE) and determine whether to accept or reject the training request based on the verification.

[0067] According to an aspect of the present disclosure, a service authorization entity of a wireless communication system may receive, from one or more UEs, a UE capability associated with an ML model training service provided by a respective UE. The service authorization entity may be a UDM entity, an existing network function (NF) entity of a 5G core network, or any suitable NF entity. The service authorization entity may further receive, from an ML model service consumer entity, a service request for accessing (or utilizing) an ML model training service of a UE. The service request may include information associated with the requested ML model training service. The service authorization entity may verify the service request. As part of verifying the service request, the service authorization entity may authenticate the ML model service consumer entity. Upon a successful authentication, the service authorization entity may compare the information associated with the requested ML model training service (included in the received service request) and the ML model training capabilities of the one or more UEs. For instance, the information associated with the requested ML model training service may match an ML model training capability of at least one UE of the one or more UEs. Based on the verifying, the service authorization entity may further transmit, to the ML model training service consumer entity, a response indicating the at least one UE that allows the requested ML model training service.

[0068] In aspects, the information associated with the requested ML model training service may include at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria. The filtering criteria may be associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

[0069] In aspects, the ML model training capabilities of the one or more UEs may include at least one of the following: an indication of an ML model allowed be trained at a respective UE of the one or more UEs, an indication of data of a respective UE allowed for training the ML model, or an indication of an allowable analytics operation associated with the ML model.

[0070] In aspects, the service authorization entity may be a UDM entity of a core network of the wireless communication system. In aspects, the ML training service consumer entity may include at least one of the following: an access network entity (e.g., RAN node model training management function) of the wireless communication system, a core network entity (e.g., 5G core model training management function such as a NWDAF) of the wireless communication system, or an application function (AF) entity. In aspects, the core network entity is a network function. In aspects, the network function is a network data analytics function entity. In aspects, the network data analytics function entity comprises a model training logical function.

[0071] In one aspect, the service authorization entity may transmit, based on the verifying, a policy to the at least one UE or an access network node associated with the at least one UE. The policy may include an indication of one or more first ML models allowed to be trained at the at least one UE. Additionally or alternatively, the policy may include an indication of one or more second ML models not allowed to be trained at the at least one UE. Additionally or alternatively, the policy may include an indication of one or more first categories of ML models allowed to be trained at the at least one UE. Additionally or alternatively, the policy may include an indication of one or more second categories of ML models not allowed to be trained at the at least one UE. Additionally or alternatively, the policy may include an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE. Additionally or alternatively, the policy may include an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE. Additionally or alternatively, the policy may include an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models. Additionally or alternatively, the policy may include an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models.

[0072] In another aspect, the service authorization entity may include an access token (e.g., a signed access token) in the response transmitted to the ML model service consumer entity. To that end, based on the verifying, the service authorization entity may generate the access token. The access token may include token claims for service consumer information, service provider information, and / or service information. For instance, the service consumer information may indicate the ML model service consumer entity allowed to utilize the requested ML model training service. The service provider information may indicate one or more UE identifiers. Each of the one or more UE identifiers identifies a respective one of theone or more UEs providing the requested ML model training service. A first UE identifier of the one or more UE identifiers identifies the at least one UE. In an example, the first UE identifier may include a user plane address (e.g., an Internet protocol (IP) address) of the at least one UE. The service information may indicate one or more parameters of an ML model allowed to be trained using the requested ML model training service. For instance, the one or more parameters may include a model identifier of the ML model. Additionally or alternatively, the one or more parameters may include a model category of the ML model. Additionally or alternatively, the one or more parameters may include one or more analytics identifiers, where each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model. Additionally or alternatively, the one or more parameters may include one or more data type identifiers. Some examples of data types may include, but are not limited to, channel state information (CSI) on a wireless network interface, received signal strength on a wireless network interface, photos, videos, sensor data, and / or location data,. A data type identifier for photos or videos may be further specialized into a data type identifier for photos or videos comprising potentially personally identifiable information, such as metadata indicating a timestamp and / or coordinates of a geographic location associated with capturing the photo or video. A data type of sensor data may be further specialized into data types according to the type of sensor, such temperature sensor, acceleration sensor, gyroscope sensor, ambient light sensor, proximity sensor, magnetism sensor or magnetoelectric sensor. In aspects, the access token includes a JavaScript object notation (JSON) web token (JWT).

[0073] In aspects, upon the ML model service consumer entity receiving the response (indicating the at least one UE that allows the requested ML model training service) from the service authorization entity, the ML model service consumer entity may transmit, to the at least one UE, based on the receiving the response, a training request including information associated with an ML model to be trained at the at least one UE. In an aspect, the information associated with the ML model included in the training request may include at least one of a model identifier of an ML model or an indication of data of the at least one UE to be used for training the ML model. In aspects, the training request may further include the access token discussed above.

[0074] In aspects, upon the UE receiving the training request to execute the ML model training service at the UE, the UE may determine whether to accept or reject the training request based on a verification of the training request. For instance, if the UE has received the policy discussed above, the UE may verify the training request against the policy. Alternatively, if the training request includes the access token discussed above, the UE may verify the access token(e.g., based on an authorization server key configured at the UE). Such and other aspects will be described in more detail later herein.

[0075] Aspects of the present disclosure may provide several benefits. For example, providing a UE with a policy for ML model training at the UE may allow the UE to verify an ML model training request against the policy prior to initiating an ML model training as requested. The alternative approach that includes an access token in an ML model training request to a UE may allow the UE to verify access token prior to initiating an ML model training as requested. Accordingly, the UE may avoid accepting an unauthorized ML model training request (e.g., transmitted by a malicious node in the network). Further, the use of the policy configuration or the access token may provide a network with flexibility in controlling ML model trainings at UEs.

[0076] The present disclosure may use the term “serving network device” to refer to a network node or network device (or a portion thereof) that services a UE. As used herein, the terms “transmit to,” “receive from,” and “cooperate with,” (and their variations) include communications that may or may not involve communications through one or more intermediate devices or nodes. The term “acquire” (and its variations) includes acquiring in the first instance or reacquiring after the first instance. The term “connection” may mean a physical connection or a logical connection.

[0077] The present disclosure uses 5G NR as an example of a wireless network and may use smartphones and / or uncrewed aerial vehicles as an example of UEs. It is intended and shall be understood that such examples are merely illustrative, and the present disclosure is applicable to other wireless networks and user equipment.

[0078] FIG. 1 is a diagram depicting an example of wireless networking between a network system 100 and a user equipment (UE) 150. The network system 100 may include one or more network nodes 120, one or more servers 110, and / or one or more network equipment 130 (e.g., test equipment). The network nodes 120 will be described in more detail below. As used herein, the term “network apparatus” may refer to any component of the network system 100, such as the server 110, the network node 120, the network equipment 130, any component(s) of the foregoing, and / or any other component(s) of the network system 100. Examples of network apparatuses include, without limitation, apparatuses implementing aspects of 5G NR, among others. The present disclosure describes embodiments related to 5G NR and embodiments that involve aspects defined by 3GPP. However, it is contemplated that embodiments relating to other wireless networking technologies are encompassed within the scope of the present disclosure.

[0079] The following description provides further details of examples of network nodes. In a 5G NR network, a gNodeB (also known as gNB) may include, e.g., a node that provides NR user plane and control plane protocol terminations towards the UE and that is connected via a NG interface to the 5G core (5GC), e.g., according to 3GPP TS 38.300 V16.6.0 (2021- 06) section 3.2, which is hereby incorporated by reference herein.

[0080] A gNB supports various protocol layers, e.g., Layer 1 (LI) - physical layer, Layer 2 (L2), and Layer 3 (L3).

[0081] The layer 2 (L2) of NR is split into the following sublayers: Medium Access Control (MAC), Radio Link Control (RLC), Packet Data Convergence Protocol (PDCP) and Service Data Adaptation Protocol (SDAP), where, e.g.: o The physical layer offers to the MAC sublayer transport channels; o The MAC sublayer offers to the RLC sublayer logical channels; o The RLC sublayer offers to the PDCP sublayer RLC channels; o The PDCP sublayer offers to the SDAP sublayer radio bearers; o The SDAP sublayer offers to 5GC quality of service (QoS) flows; o Control channels include broadcast control channel (BCCH) and physical control channel (PCCH).

[0082] Layer 3 (L3) includes, e.g., radio resource control (RRC), e.g., according to 3GPP TS 38.300 V16.6.0 (2021-06) section 6, which is hereby incorporated by reference herein.

[0083] A gNB central unit (gNB-CU) includes, e.g., a logical node hosting, e.g., RRC, SDAP, and PDCP protocols of the gNB or RRC and PDCP protocols of the en-gNB, that controls the operation of one or more gNB distributed units (gNB-DUs). The gNB-CU terminates the Fl interface connected with the gNB -DU. A gNB-CU may also be referred to herein as a CU, a central unit, a centralized unit, or a control unit.

[0084] A gNB Distributed Unit (gNB-DU) includes, e.g., a logical node hosting, e.g., RLC, media access control (MAC), and physical (PHY) layers of the gNB or en-gNB, and its operation is partly controlled by the gNB-CU. One gNB-DU supports one or multiple cells. One cell is supported by only one gNB-DU. The gNB-DU terminates the Fl interface connected with the gNB-CU. A gNB-DU may also be referred to herein as DU or a distributed unit.

[0085] As used herein, the term “network node” may refer to any of a gNB, a gNB-CU, or a gNB-DU, or any combination of them. A RAN node or network node such as, e.g., a gNB, gNB-CU, or gNB-DU, or parts thereof, may be implemented using, e.g., an apparatus with at least one processor and / or at least one memory with processor-readable instructions(“program”) configured to support and / or provision and / or process CU and / or DU related functionality and / or features, and / or at least one protocol (sub-)layer of a RAN, e.g., layer 2 and / or layer 3. Different functional splits between the central and distributed unit are possible. An example of such an apparatus and components will be described in connection with FIG. 8 below.

[0086] The gNB-CU and gNB-DU parts may, e.g., be co-located or physically separated. The gNB-DU may even be split further, e.g., into two parts, e.g., one including processing equipment and one including an antenna. A central unit (CU) may also be called baseband unit / radio equipment controller / cloud-RAN / virtual-RAN (BBU / REC / C-RAN / V-RAN), open- RAN (O-RAN), or part thereof. A distributed unit (DU) may also be called remote radio head / remote radio unit / radio equipment / radio unit (RRH / RRU / RE / RU), or part thereof. Hereinafter, in various example embodiments of the present disclosure, a network node, which supports at least one of central unit functionality or a layer 3 protocol of a radio access network, may be, e.g., a gNB-CU. Similarly, a network node, which supports at least one of distributed unit functionality or a layer 2 protocol of the radio access network, may be, e.g., a gNB-DU.

[0087] A gNB-CU may support one or multiple gNB-DUs. A gNB-DU may support one or multiple cells and, thus, could support a serving cell for a user equipment (UE) or support a candidate cell for handover, dual connectivity, and / or carrier aggregation, among other procedures.

[0088] The UE 150 may be or include a wireless or mobile device, an apparatus with a radio interface to interact with a RAN, a smartphone, an in-vehicle apparatus, an Internet of Things (loT) device, or a machine-to-machine (M2M) device, among other types of user equipment. Such UE 150 may include: at least one processor; and at least one memory including program code; where the at least one memory and the computer program code are configured to, with the at least one processor, cause the apparatus at least to perform certain operations, such as, e.g., RRC connection to the RAN. An example of components of a UE will be described in connection with FIG. 9. In embodiments, the UE 150 may be configured to generate a message (e.g., including a cell ID) to be transmitted via radio towards a RAN (e.g., to reach and communicate with a serving cell). In embodiments, the UE 150 may generate and transmit and receive RRC messages containing one or more RRC PDUs (packet data units). Persons skilled in the art will understand RRC protocol as well as other procedures a UE may perform.

[0089] With continuing reference to FIG. 1, in the example of a 5G NR network, the network system 100 provides one or more cells, which define a coverage area of the networksystem 100. As described above, the network system 100 may include a gNB of a 5G NR network or may include any other apparatus configured to control radio communication and manage radio resources within a cell. As used herein, the term “resource” may refer to radio resources, such as a resource block (RB), a physical resource block (PRB), a radio frame, a subframe, a time slot, a sub-band, a frequency region, a sub-carrier, a beam, etc. In embodiments, the network node 120 may be called a base station.

[0090] FIG. 1 provides an example and is merely illustrative of a network system 100 and a UE 150. Persons skilled in the art will understand that the network system 100 includes components not illustrated in FIG. 1 and will understand that other user equipment may be in communication with the network system 100.

[0091] FIG. 2 is a block diagram of example components of the network system 100 of FIG. 1. A 5G NR network may be described as an example of the network system 100, and it is intended that aspects of the following description shall be applicable to other types of network systems, as well. The network system may operate in accordance with the signals and connections shown in FIG. 1 such that the UE 150 is in communication with the network system 100 through the RAN 225. Additionally, the network system may be divided into user plane components and functions and control plane components and functions, as shown and described herein. Unless indicated otherwise, the terms “component”, “function”, and “service” may be used interchangeably herein, and they may refer to and be implemented by instructions executed by one or more processors.

[0092] Example functions of the components are described below. The example functions are merely illustrative, and it shall be understood that additional operations and functions may be performed by the components described herein. Additionally, the connections between components may be virtual connections over service-based interfaces such that any component may communicate with any other component. In this manner, any component may act as a service “producer,” for any other component that is a service “consumer,” to provide services for network functions.

[0093] For example, a control network (CN) 210 is described in the control plane of the network system. The CN 210 includes an authentication server function (AUSF) 211, an access and mobility function (AMF) 212, and a session management function (SMF) 213. The CN 210 also includes a network slice selection function (NSSF) 214, a NEF 215, a network repository function (NRF) 216, and a UDM 217, which may include UDR 224.

[0094] Additional components and functions of the CN 210 include an AF 218, a PCF 219, and an NWDAF 220.

[0095] The user plane includes the UE 150, a RAN 225, a user plane function (UPF) 226, and a data network (DN) 227. The RAN 225 may include one or more components described in connection with FIG. 1, such as one or more network nodes. However, the RAN 225 may not be limited to such components. The UPF 226 provides connection for data being transmitted over the RAN 225. The DN 227 identifies services from service providers, Internet access, and third-party services, for example.

[0096] The AMF 212 processes connection and mobility tasks. The AUSF 211 receives authentication requests from the AMF 212 and interacts with UDM 217 to authenticate and validate network responses for determination of successful authentication. The SMF 213 conducts PDU session management, as well as manages session context with the UPF 226.

[0097] The NSSF 214 may select a network slicing instance (NSI) and determine the allowed network slice selection assistance information (NSSAI). This selection and determination are utilized to set the AMF 212 to provide service to the UE 150. The NEF 215 secures access to network services for third parties to create specialized network services. The NRF 216 acts as a repository to store network functions to allow the functions to register with and discover each other.

[0098] The UDM 217 generates authentication vectors for use by the AUSF 211 and AMF 212 and provides user identification handling. The UDM 217 may be connected to the UDR 224 which stores data associated with authentication, applications, or the like. The AF 218 provides application services to a user (e.g., streaming services, etc.). The PCF 219 provides policy control functionality. For example, the PCF 219 may assist in network slicing and mobility management, as well as provide QoS and charging functionality. In some instances, the UDM 217 and / or UDR 224 may be referred to as a network data component.

[0099] The NWDAF 220 collects data (e.g., from the UE 150 and the network system) to perform network analytics and provide insight to functions that utilize the analytics in the providing of services.

[0100] In some examples, the NWDAF 220 may include an ML model training management function entity that may request the UE 150 to train an ML model and / or jointly train an ML model with the UE 150. In some examples, the RAN 225 may include an ML model training management function entity that may request the UE 150 to train an ML. model and / or jointly train an ML model with the UE 150. In some examples, the AF 218 may request the UE 150 to train an ML model and / or jointly train an ML model with the UE 150. In general, the ML model training management function entity (e.g., a core network entity) at the NWDAF 220, the ME model training management function entity (e.g., an access network entity) in theRAN 225, the AF 218 may operate as an ML model training service consumer entity utilizing an ML model training service provided by the UE 150.

[0101] As further shown in FIG. 2, the network system may further include a ML model training service consumer entity 230. The ML model training service consumer entity 230 may utilize an ML model training service provided by the UE 150. In some examples, the ML model training service consumer entity 230 may be an AF similar to the AF 218 but may be outside of the core network.

[0102] FIG. 2 is merely an example of components of a network system, and variations are contemplated to be within the scope of the present disclosure. In embodiments, the network system may include other components not illustrated in FIG. 2. In embodiments, the network system may not include every component illustrated in FIG. 2. In embodiments, the components and connections may be implemented with different connections than those illustrated in FIG. 2. Such and other embodiments are contemplated to be within the scope of the present disclosure.

[0103] The following will describe example operations for provisioning a secured ML model training at a UE. FIG. 3 shows example signals and operations for provisioning a secure ML model training at a UE via a policy. FIG. 4 shows example signals and operations for provisioning a secure ML model training at a UE via an access token. FIG. 5 shows example operations of a UE, FIG. 6 shows example operations of a network apparatus (e.g., a service authorization entity), and FIG. 7 shows example operations of a network apparatus (e.g., an ML model training service consumer entity or an ML model service consumer entity).

[0104] FIG. 3 is a diagram of an example embodiment of operations for provisioning a secured ML model training at a UE, according to one illustrated aspect of the present disclosure. It will be understood that a described signal may have associated operations and a described operation may have associated signals. Accordingly, a described signal may also involve an operation and a described operation may also involve a signal. Additionally, FIG. 3 will describe signals between and operations performed by various network components, such as those shown at the top of FIG. 3. Specifically, the components include a UE 310 (e.g., the UE 150 of FIG. 1), a RAN node (e.g., the network node 120 of FIG. 1), a RAN node model training management function entity (e.g., the network node 120 of FIG. 1), an AMF (e.g., the AMF 212 of FIG. 2), a 5G core model training management function (e.g., the NWDAF 220 of FIG. 2), a UDM (e.g., the UDM 217 of FIG. 2), and an AF (e.g., the AF 218 or the ML model training service consumer entity 230 of FIG. 2). The network components are illustrative, and it is contemplated that other components may be involved in the signals or mayperform the operations. In some examples, each of the network components may implement the operations using an apparatus with components as shown in FIG. 8. One or more of the following operations may be implemented in connection with the operations of the present disclosure, such as the examples discussed above with reference to FIGS. 1-2.

[0105] As shown in FIG. 3, the RAN node and the RAN node model training management function may be part of a RAN 350 of a wireless communication system (e.g., the network system 100 discussed above). The AMF, the 5G Core model training management function, and the UDM may be part of a core network 352 of the wireless communication system. The AF may be outside of the core network 352 as shown or part of the core network 352. In the illustrated example of FIG. 3, each of the RAN node model training management function, the 5G core model training management function, and the AF may operate as an ML model training service consumer entity requesting an ML model training service provided by a UE. However, in general, one or more of the RAN node model training management function, the 5G core model training management function, and the AF may request an ML model training service provided by one or more UEs. Further, the UDM may operate as a service authorization entity. However, in general, any other authorization entity (e.g., an authorization server, an NRF similar to the NRF 216) in the network system may operate as a service authorization entity. Further, each of the components may communicate with the UE 310 via the RAN node that serves the UE 310. For brevity, the following description may not describe communication detail between the UE and the RAN node. Persons skilled in the art will understand protocols and / or procedures between the UE and the RAN node.

[0106] At a high level, the UDM may provision a secured ML model training at a UE via a policy configuration at the UE.

[0107] At operation 301, the UE 310 transmits, and the UDM receives, information regarding an ML model training capability of the UE 310. In an example, the UE 310 may transmit the ML model training capability information as part of a registration procedure performed with the core network 352. In general, the UE 310 may provide the ML model training capability information via any suitable mechanism (e.g., via short message service (SMS), customer care calling, or portal). The ML model training capability information may indicate if a certain ML model training service is to be enabled or not. For instance, if the UE provides (or allows) a certain ML model training service, the UE may include an indication of that ML model training service as enabled in the ML model training capability information. If, however, the UE does not provide (or allow) a certain ML model training service, the UE may include an indication of that ML model training service as disabled in the ML model trainingcapability information. The ML model training capability may further provide information regarding a category of ML models that may be trained by the UE 310. In an example, the ML model training capability may include an indication of an ML model allowed to be trained at the UE 310, an indication of data of the UE 310 allowed to be used for training the ML model, or an indication of an allowable analytics operation associated with ML model. Examples of the model category may include, but are not limited to, advertisement, location, CSI compression, etc.

[0108] At operation 302, the UDM stores the ML model training capability information received from the UE 310 and transmits to the UE 310, a confirmation response to the UE 310’s ML model training capability information.. In some examples, the confirmation may additionally include information regarding categories of ML model trainings that are configured in the network system. For instance, the network system may be configured with one or more analytics identifiers. Each analytics identifier may identify an analytics operation or analytics service (e.g., network planning, network slicing, etc.). Further, the network system may be configured with a mapping between ML model trainings and analytics identifier. For instance, an ML model A may be mapped to an analytics identifier for an analytics operation A. That is, ML model trainings can be categorized according to the analytics operations (e.g., purposes for the training). In general, one or more ML models can be mapped to one analytics identifier or one ML model can be mapped to one or more analytics identifiers. While FIG. 3 illustrates the UDM receiving ML model training capability information from one UE, the UDM may receive ML model training capability information from a plurality of UEs (e.g., 2, 3, 4, 5, 10, 100 or more).

[0109] At operation 303a, the RAN node model training management function transmits a service request to the UDM via the AMF. The service request may include information associated with a requested ML model training service of a UE. For instance, the service request information may include a UE identifier list and / or one or more filter criteria. The UE identifier list may include one or more UE identifiers, each identifying a respective UE requested for the respective ML model training service. In an example, the UE identifier list may include an identifier of the UE 310. The filtering criteria may include a model identifier of a requested ML model to be trained at the respective UE, a model category of the requested ML model, and / or an analytics identifier of an analytics operation associated with the requested ML model.

[0110] At operation 304a, the UDM verifies the service request received at operation 303a. The verification may include authenticating the RAN node model training managementfunction, and upon a successful authentication, the UDM verifies if the requested ML model identifier and / or the ML model category match the ML model training capability of a UE. As discussed above, the UDM may have received ML model training capability information from a plurality of UEs (e.g., including the UE 310). Thus, the UDM may compare the ML model identifier, the ML model category, and / or the analytics identifier indicated by the service request to ML model training capability information of each UE. If the verification of the service request is successful, the following operations may be performed. Otherwise, the following operations may not be performed.

[0111] At operation 304b, the UDM prepares a list of UEs that are allowed to be used for training the ML model and / or the ML model category indicated by the service request. The preparation may be based on the filter criteria received at operation 303a. In the illustrated example, the UDM may determine, based on the filter criteria, that the UE 310 allows the requested ML model training service, and thus the list may include the UE 310’s identifier.

[0112] At operation 305, based on the verification at operation 304a is a success, the UDM pushes a policy to the UE 310. In some instances, the UDM may transmit, and the UE 310 may receive the policy via UE parameter update (UPU) signaling. In some instances, the UDM may transmit, and the UE 310 may receive the policy via steering of roaming (SoR) signaling. In some instances, the UDM may transmit, and the UE 310 may receive the policy via non-access stratum (NAS) signaling. In general, the policy may be transmitted via any suitable signaling.

[0113] The policy may be configured in any suitable format. In an example, the policy may include a first policy indicating an allowed model identifier and / or an allowed analytics identifier and a second policy indicating that Data 1 and Data 2 of the UE are allowed to be used for training an ML model at the UE.

[0114] In an aspect, the policy may include an indication of one or more first ML models allowed to be trained at the UE 310. Additionally or alternatively, the policy may include an indication of one or more second ML models not allowed to be trained at the UE 310. Additionally or alternatively, the policy may include an indication of one or more first categories of ML models allowed to be trained at the UE 310. Additionally or alternatively, the policy may include an indication of one or more second categories of ML models not allowed to be trained at the UE 310. Additionally or alternatively, the policy may include an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the UE 310. Additionally or alternatively, the policy may include an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the UE 310. Additionally or alternatively, the policymay include an indication of one or more first types of data of the UE 310 allowed to be used for training the one or more first ML models. Additionally or alternatively, the policy may include an indication of one or more second types of data of the UE 310 not allowed to be used for training the one or more second ML models.

[0115] Stated differently, the policy may include at least one of the following: 1) one or more first ML model identifiers, where an ML model identified by one of the one or more first ML model identifier is allowed to be trained at the UE 310; 2) one or more second ML model identifiers, where an ML model identified by one of the one or more second ML model identifier is not allowed to be trained at the UE 310; 3) one or more first ML model categories, where an ML model associated with one of the one or more first ML model categories is allowed to be trained at the UE 310; 4) one or more second ML model categories, where an ML model associated with one of the one or more second ML model categories is not allowed to be trained at the UE 310; 5) one or more first analytics identifiers, where an ML model associated with one of the one or more first analytics identifiers is allowed to be trained at the UE; 6) one or more second analytics identifiers, where an ML model associated with one of the one or more second analytics identifiers is not allowed to be trained at the UE; 7) one or more first data type identifiers, where data stored at the UE 310 of a type associated with one of the one or more first UE data type identifiers is allowed to be used for training the ML model; or 8) one or more second UE data type identifiers, where data at the UE 310 of a type associated with one of the one or more second UE data type identifiers is not allowed to be used for training the ML model.

[0116] At operation 306a, in response to the service request received at operation 303a, the UDM transmits a response to the RAN node model training management function via the AMF. The response may indicate the list of UE(s) that allow the ML model training service requested by the RAN node model training management function.

[0117] At operation 307a, upon receiving the response at operation 306a, the RAN node model training management function transmits a training request to the UE 310. The training request may indicate an ML model (e.g., as authorized by the UDM) to be trained at the UE 310. The training request may also indicate the type(s) of data (e.g., Data 1 and Data 2) required for training the ML model. Some examples of data types may include channel state information on a wireless network interface, received signal strength on a wireless network interface, location data, photos, videos, and / or sensor data. In an aspect, the training request may include one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, and / or one or more UE data type identifiers. In general, the RAN node modeltraining management function may transmit, to each UE indicated in the list (received at operation 306a), a request for the ML model training service according to the response received from the UDM at operation 306a.

[0118] At operation 308, upon receiving the training request at operation 307a, the UE 310 runs a policy check on the training request. To that end, the UE 310 may verify the training request using the policy information received at operation 305. If the verification is a success, the UE 310 accepts the training request and trains the ML model or a portion of the ML model. If, however, the verification fails, the UE 310 rejects the training request. In an example, as part of training the ML model, the UE may process an ML algorithm with training data to find patterns and make predictions in achieving a particular task. The training data may include input data and expected output data. The input data may be used as inputs to the ML algorithms, the expected output data may operate as ground truth, and the training may be process until the output of the ML algorithm is substantially close to the expected output data (e.g., the ML algorithm output is within an acceptable error limit).

[0119] In general, the training request at operation 307a can request training of one of more ML models. In an aspect, the training request may include one or more parameters of an ML model training service including at least one of the following: one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers, where one of the one or more UE data type identifiers is associated with a type of data at the UE 310. The UE 310 may determine to provide the ML model training service based on the one or more parameters and the policy.

[0120] At operation 309, after completing the training, the UE transmits the trained ML model or the partially trained ML model to the ML model training service consumer entity (e.g., the RAN node model training management function) that requested the training.

[0121] As discussed above, each of the RAN node model training management function, the 5G core model training management function, and the AF may operate as an ML model training service consumer entity requesting an ML model training service provided by one or more UEs. Generally speaking, the 5G core model training management function and the AF may perform similar operations as the RAN node model training management function to request an ML model training service provided by one or more UEs.

[0122] For instance, at operation 303b, the 5G core model training management function transmits a service request to the UDM. The service request may be as discussed above at operation 303a. In response to the service request, the UDM performs the operations 304a, 304b, and 305 as discussed above. At operation 306b, the UDM transmits a response to the 5Gcore model training management function. The response may indicate the list of UE(s) that allow the ML model training service requested by the 5G core model training management function, where the list of UE(s) includes the UE 310. At operation 307b, upon receiving the response at operation 306b, the 5G core model training management function transmits a training request to the UE 310. The training request may be as discussed above with reference to operation 307. At operation 308, the UE may accept or reject the request received from the 5G core model training management function at operation 307b.

[0123] Similarly, at operation 303c, the AF transmits a service request to the UDM. The service request may be as discussed above at operation 303a. In response to the service request, the UDM perform the operations 304a, 304b, and 305 as discussed above. At operation 306c, the UDM transmits a response to the AF. The response may indicate the list of UE(s) that allow the ML model training service requested by the AF, where the list of UE(s) includes the UE 310. At operation 307c, upon receiving the response at operation 306c, the AF transmits a training request to the UE 310. The training request may be as discussed above with reference to operation 307a. At operation 308, the UE may accept or reject the training request received from the AF at operation 307c.

[0124] In general, the UDM may configure a UE with a policy indicating one or more ML model training service consumer entities that are permitted to request training of certain ML model(s) at the UE, the categories (or purposes) of ML models allowed to be trained at the UE, and / or the types of data stored at the UE allowed to be used for training each ML model training at the UE.

[0125] While FIG. 3 illustrates the UDM transmitting the policy to the UE 310 at operation 305, in other aspects, the UDM may transmit the policy to the RAN node that the UE 310 is attached to. For instance, the RAN node may have the ML model training capability information of the UE 310 (e.g., the information transmitted at operation 301), and thus the RAN node may determine whether to accept or reject a request to run an ML model at the UE 310 before the request reaches the UE 310.

[0126] FIG. 3 is merely an example of signals and operations, and variations are contemplated to be within the scope of the present disclosure. In embodiments, the signals and operations may include others not illustrated in FIG. 3. In embodiments, the signals and operations may not include every signal and operation illustrated in FIG. 3. In embodiments, the signals and operations may be implemented with different connections than those illustrated in FIG. 3. Such and other embodiments are contemplated to be within the scope of the present disclosure.

[0127] FIG. 4 is a diagram of an example embodiment of operations for provisioning a secured ML model training at a UE, according to one illustrated aspect of the present disclosure. It will be understood that a described signal may have associated operations and a described operation may have associated signals. Accordingly, a described signal may also involve an operation and a described operation may also involve a signal. Additionally, FIG. 4 will describe signals between and operations performed by various network components, such as those shown at the top of FIG. 4. Specifically, the components include a UE 310 (e.g., the UE 150 of FIG. 1), a RAN node (e.g., the network node 120 of FIG. 1), a RAN node model training management function entity (e.g., the network node 120 of FIG. 1), an AMF (e.g., the AMF 212 of FIG. 2), a 5G core model training management function (e.g., the NWDAF 220 of FIG. 2), a UDM (e.g., the UDM 217 of FIG. 2), and an AF (e.g., the AF 218 or the ML model training service consumer entity 230 of FIG. 2). The network components are illustrative, and it is contemplated that other components may be involved in the signals or may perform the operations. In some examples, each of the network components may implement the operations using an apparatus with components as shown in FIG. 8. One or more of the following operations may be implemented in connection with the operations of the present disclosure, such as the examples discussed above with reference to FIGS. 1-3.

[0128] Generally speaking, the signals and operations of FIG. 4 include features similar to the operations of FIG. 3 in many respects. For example, the operations 401, 402, 403a-403c, 404a-404b, and 408 are similar to operations 301, 302, 303a-303c, 304a-304b, and 309, respectively. Accordingly, for brevity, details of those operations will not be repeated here and can be referred to the corresponding descriptions above.

[0129] Further, similar to FIG. 3, each of the RAN node model training management function, the 5G core model training management function, and the AF may operate as an ML model training service consumer entity requesting an ML model training service provided by a UE. Further, the UDM may operate as a service authorization entity. However, in general, any other authorization entity (e.g., an authorization server) in the network system may operate as a service authorization entity. Further, each of the components may communicate with the UE 310 via the RAN node that serves the UE 310. For brevity, the following description may not describe communication detail between the UE and the RAN node. Persons skilled in the art will understand protocols and / or procedures between the UE and the RAN node.

[0130] At a high level, the UDM may provision a secured ML model training at a UE via an access token. For instance, upon receiving a request from an ML model training service consumer entity for initiating a ML model training at a UE, the UDM may verify the requestas discussed above with reference to operations 304a and 304b. Instead of pushing a policy to a UE that allows the requested ML model training, the UDM generates an access token. The access token may include token claims for service consumer information, service provider information, and service information as will be discussed more fully below. The UDM may provide the access token to the ML model training service consumer entity (that requested the ML model training). When the ML model training service consumer entity requests the UE for the ML model training, the ML model training service consumer entity may include the access token as part of the request. The UE may determine whether to accept or reject the request based on a validation of the access token.

[0131] For example, at operation 403a, the RAN node model training management function transmits a service request to the UDM via the AMF as discussed above with reference to operation 303 a. In response to the service request, the UDM performs operations 404a and 404b similar to the operations 304a and 304b discussed above.

[0132] At operation 404c, in response to a successful verification of the service request from the RAN node model training management function, the UDM generates an access token (e.g., utilizing an authorization server key). The generation may be based on information registered about the ML model training service consumer entity (RAN node model training management function) that initiated the service request, the service request (e.g., the information in the service request), the ML model training capability information of the UE 310 received from UE 310 at operation 301, and / or the mapping of UE model category preferences and the analytics identifier(s) configured in the system. Stated differently, the UDM may generate the access token with the token claims including service consumer information, service provider information, and service information. The service consumer information may include the type and / or the instance of the ML model training service consumer entity that may transmit a request to the UE 310 for training an ML model at the UE 310. The service provider information may include UE identifier(s) of UE(s) that may be used to perform the ML model training, or collaborative training with the ML model training service consumer entity. In some examples, each of the UE identifier(s) may include a user plane address (e.g., IP address) of a respective UE that may be used for communication with the respective UE. The service information may include one or more analytics identifiers, one or more ML model identifiers, one or more categories of ML model to be trained. The service information may additionally include an indication of allowed UE data for training per ML model identifier. In aspects, the access token is a JWT according to 3GPP document TS 33.501 , V13.4.1. In aspects, the access token is a JWT according to Internet engineering task force(IETF) request for comments (RFC) document 7519. In aspects, the access token is used for OAuth 2.0 authorization protocol according IETF RFC document 6749. In some aspects, the access token is a signed access token (e.g., including a signature).

[0133] At 405a, in response to the service request received from the RAN node model training management function at operation 403a, the UDM transmits a response including the generated access token to the RAN node model training management function via the AMF.

[0134] At 406a, the RAN node model training management function transmits a training request to the UE 310 along with the access token received at operation 405a. The training request may indicate an ML model to be trained at the UE 310. The training request may include parameter(s) of the ML model as discussed above with reference to operation 307a. In general, the RAN node model training management function may transmit, to each UE indicated in the list (received at operation 306a), a request for the ML model training service according to the response received from the UDM at operation 405a.

[0135] At 407, the UE 310 verifies the access token received at operation 406a. To that end, the UE 310 may be configured with an authorization server key (e.g., the same as the authorization serve key used by the UDM to generate the access key at operation 405) and may verify the access token using the authorization server key. If the verification is a success, the UE 310 accepts the request and trains the ML model or a portion of the ML model. If, however, the verification fails, the UE 310 rejects the request.

[0136] As discussed above, each of the RAN node model training management function, the 5G core model training management function, and the AF may operate as an ML model training service consumer entity requesting an ML model training service provided by a UE. Generally speaking, the 5G core model training management function and the AF may perform similar operations as the RAN node model training management function to request an ML model training service of a UE.

[0137] For instance, at operation 403b, the 5G core model training management function transmits a service request to the UDM. The service request may be as discussed above at operation 403a. In response to the service request, the UDM performs the operations 404a- 404c (e.g., generate an access token when the verification of the service request is a success) as discussed above. At operation 405b, the UDM transmits the generated access token to the 5G core model training management function. At operation 406b, upon receiving the access token, the 5G core model training management function transmits a request to the UE 310 along with the access token received at operation 405b. The request may indicate an ML model (as authorized by the UDM) to be trained at the UE 310. At operation 407, the UE may acceptor reject the request received from the 5G core model training management function at operation 406b.

[0138] Similarly, at operation 403c, the AF transmits a service request to the UDM. The service request may be as discussed above at operation 403a. In response to the service request, the UDM performs the operations 404a-404c (e.g., generate an access token when the verification of the service request is a success) as discussed above. At operation 405c, the UDM transmits the generated access token to the AF. At operation 406c, upon receiving the access token, the AF transmits a request to the UE 310 along with the access token received at operation 405c. The request may indicate an ML model (as authorized by the UDM) to be trained at the UE 310. At operation 407, the UE may accept or reject the request received from the AF at operation 406c.

[0139] FIG. 4 is merely an example of signals and operations, and variations are contemplated to be within the scope of the present disclosure. In embodiments, the signals and operations may include others not illustrated in FIG. 4. In embodiments, the signals and operations may not include every signal and operation illustrated in FIG. 4. In embodiments, the signals and operations may be implemented with different connections than those illustrated in FIG. 6. Such and other embodiments are contemplated to be within the scope of the present disclosure.

[0140] FIG. 5 is a flow diagram of example operations of a UE providing an ML model training service, according to one illustrated aspect of the present disclosure. In an aspect, the UE may correspond to the UE 150 discussed above with reference to FIGS. 1 and 2 or the UE 310 discussed above with reference to FIGS. 3 and 4. In some examples, the UEs may implement the operations using an apparatus with components as shown in FIG. 8. The operations of FIG. 5 may include similar mechanisms as discussed above with reference to FIGS. 1-4.

[0141] At block 502, a UE transmits, to a network, at least one UE capability associated with an ML model training service provided by the UE.

[0142] At block 504, the UE receives an instruction for providing the ML model training service. In an aspect, the instruction is received based on transmitting the at least one UE capability.

[0143] At block 506, the UE provides the ML model training service.

[0144] In an aspect, the providing the ML model training service at block 506 includes receiving a request for providing the ML model training service. The providing the ML model training service further includes determining to provide the requested ML model trainingservice, where the determining is based on the received request and the received instruction. The providing the ML model training service further includes providing, based on the determining, the requested ML model training service. In an aspect, the request for providing the ML model training service includes one or more parameters of the requested ML model training service; and the determining is further based on at least one of the one or more parameters. In an aspect, the one or more parameters of the ML model training service include at least one of the following: one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers.

[0145] In an aspect, the receiving the instruction at block 504 includes receiving, by the UE from a service authorization entity, the instruction including a policy as discussed above with reference to FIG. 3. For instance, the policy includes at least one of the following: an indication of one or more first ML models allowed to be trained at the UE, an indication of one or more second ML models not allowed to be trained at the UE, an indication of one or more first categories of ML models allowed to be trained at the UE, an indication of one or more second categories of ML models not allowed to be trained at the UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the UE, an indication of one or more first types of data of the UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the UE not allowed to be used for training the one or more second ML models.

[0146] In an aspect, the receiving the instruction at block 504 includes receiving, by the UE from an ML model service consumer entity (e.g., an access network entity, a core network entity, or an AF entity), the instruction including an access token as discussed above with reference to FIG. 4. For instance, the access token includes token claims for service consumer information, service provider information, and / or service information. The service consumer information indicates the ML model service consumer entity allowed to use the ML model training service provided by one or more UEs. The service provider information indicates one or more UE identifiers, where each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the ML model training service, and where a first UE identifier of the one or more UE identifiers identifies the UE. In an example, the first UE identifier identifying the UE includes an address for communicating with the UE. In an example, the address for communicating with the UE is a user plane address of the UE. The service information may indicate one or more parameters of an ML model allowed to be trainedusing the requested ML model training service. For instance, the one or more parameters may include a model identifier of the ML model. Additionally or alternatively, the one or more parameters may include a model category of the ML model. Additionally or alternatively, the one or more parameters may include one or more analytics identifiers, where each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model. Additionally or alternatively, the one or more parameters may include one or more data type identifiers. Some examples of data types may include, but are not limited to, CSI, received signal strength on a wireless network interface, photos, videos, sensor data, and / or location data. In an aspect, the access token includes JWT.

[0147] The operations of FIG. 5 are merely illustrative, and variations are contemplated to be within the scope of the present disclosure. In embodiments, the operations may include other blocks not illustrated in FIG. 5. In embodiments, the operations may not include every block illustrated in FIG. 5. In embodiments, the operation may be implemented in a different order than that illustrated in FIG. 5. Such and other embodiments are contemplated to be within the scope of the present disclosure.

[0148] FIG. 6 is a flow diagram of example operations of a network apparatus provisioning a secured ML model training at a UE, according to one illustrated aspect of the present disclosure. In an aspect, the network apparatus may be an apparatus or network node of a service authorization entity of a wireless communication system. In an aspect, the service authorization entity may correspond to the UDM discussed above with reference to FIGS. 2-4, and the wireless communication system may be a network system as discussed above with reference to FIGS. 1-2. In some examples, the network apparatus implement the operations using an apparatus with components as shown in FIG. 8. The operations of FIG. 6 may include similar mechanisms as discussed above with reference to FIGS. 1-4.

[0149] At block 602, a service authorization entity of a wireless communication system receives, from an ML model service consumer entity, a service request for accessing an ML model training service of a UE (e.g., the UE 150 or 310), where the service request includes information associated with the requested ML model training service.

[0150] In an aspect, the ML model service consumer entity includes an access network entity of the wireless communication system, a core network entity of the wireless communication system, and / or an AF entity.

[0151] In an aspect, the information associated with the requested ML model training service includes at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteriaassociated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

[0152] At block 604, the service authorization entity verifies the service request, where the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs. In an aspect, the verifying the service request is further based on an authentication of the ML model service consumer entity.

[0153] At block 606, the service authorization entity transmits, to the ML model service consumer entity, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service.

[0154] In an aspect, the service authorization entity further prepares the at least one UE that allows the requested ML model training service.

[0155] In an aspect, the service authorization entity further transmits, based on the verifying at block 604, a policy including at least one of the following: an indication of one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first categories of ML models allowed to be trained at the at least one UE, an indication of one or more second categories of ML models not allowed to be trained at the at least one UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models. In one aspect, the service authorization entity transmits the policy to the at least one UE. In another aspect, the service authorization entity transmits the policy to an access network node (e.g., a gNB, the RAN node of FIGS. 3-4) associated with the at least one UE.

[0156] In an aspect, the response indicating the at least one UE that allows the requested ML model training service at block 604 includes an access token. In an aspect, the access token includes token claims for service consumer information, service provider information, and / or service information. The service consumer information indicates the ML model service consumer entity allowed to utilize the requested ML model training service. The serviceprovider information indicates one or more UE identifiers, where each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model training service, and where a first UE identifier of the one or more UE identifiers identifies the at least one UE. In an example, the first UE identifier identifying the at least one UE includes an address for communicating with the at least one UE. In an example, the address for communicating with the at least one UE is a user plane address of the at least one UE. The service information indicates one or more parameters of an ML model allowed to be trained using the requested ML model training service. For instance, the one or more parameters may include a model identifier of the ML model. Additionally or alternatively, the one or more parameters may include a model category of the ML model. Additionally or alternatively, the one or more parameters may include one or more analytics identifiers, where each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model. Additionally or alternatively, the one or more parameters may include one or more data type identifiers, where each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model. In an example, a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: CSI, received signal strength on a wireless network interface, photos, videos, sensor data, or location data. In an aspect, the service authorization entity further generates, based on the verifying, the access token. In an aspect, the access token includes a JWT.

[0157] In an aspect, the ML model training capabilities of the one or more UEs at block 604 includes at least one of the following an indication of an ML model allowed be trained at the at least one UE, an indication of data of the at least one UE allowed for training the ML model, or an indication of an allowable analytics operation associated with ML model.

[0158] The operations of FIG. 6 are merely illustrative, and variations are contemplated to be within the scope of the present disclosure. In embodiments, the operations may include other blocks not illustrated in FIG. 6. In embodiments, the operations may not include every block illustrated in FIG. 6. In embodiments, the operation may be implemented in a different order than that illustrated in FIG. 6. Such and other embodiments are contemplated to be within the scope of the present disclosure.

[0159] FIG. 7 is a flow diagram of example operations of a network apparatus requesting an ML model training service provided by a UE, according to one illustrated aspect of the present disclosure. In an aspect, the network apparatus may be an apparatus or network node of an ML model service consumer entity. In an aspect, the ML model service consumer entitymay correspond to the RAN node model training management function, the 5G core model training management function, or the AF discussed above with reference to FIGS. 2-4. In some examples, the network apparatus may implement the operations using an apparatus with components as shown in FIG. 8. The operations of FIG. 7 may include similar mechanisms as discussed above with reference to FIGS. 1-11.

[0160] At block 702, an ML model service consumer entity transmits, to a service authorization entity of a wireless communication system, a service request for accessing an ML model training service of a UE (e.g., the UE 150 or 310), where the service request includes information associated with the requested ML model training service. In an aspect, the ML model service consumer entity includes at least one of the following an access network entity of the wireless communication system, a core network entity of the wireless communication system, or an AF entity.

[0161] In an aspect, the information associated with the requested ML model training service includes at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

[0162] At block 704, the ML model service consumer entity receives, from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service.

[0163] In an aspect, the response indicating the at least one UE that allows the requested ML model training service at block 704 includes an access token. In an aspect, the access token includes token claims for service consumer information, service provider information, and / or service information. The service consumer information indicates the ML model service consumer entity allowed to utilize the requested ML model training service. The service provider information indicates one or more UE identifiers, where each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model training service, and where a first UE identifier of the one or more UE identifiers identifies the at least one UE. In an example, the first UE identifier identifying the at least one UE includes an address for communicating with the at least one UE. In an example, the address for communicating with the at least one UE is a user plane address of the at least one UE. The service information indicates one or more parameters of an ML model allowed to be trained using the requested ML model training service. For instance, the one or more parameters mayinclude a model identifier of the ML model. Additionally or alternatively, the one or more parameters may include a model category of the ML model. Additionally or alternatively, the one or more parameters may include one or more analytics identifiers, where each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model. Additionally or alternatively, the one or more parameters may include one or more data type identifiers, where each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model. In an example, a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: CSI, received signal strength on a wireless network interface, photos, videos, sensor data, or location data. In an aspect, the access token includes a JWT.

[0164] At block 706, the ML model service consumer entity transmits, to the at least one UE, based on the receiving the response, a training request including information associated with an ML model to be trained at the at least one UE. In an aspect, the information associated with the ML model in the training request includes at least one of one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers, where one of the one or more UE data type identifiers is associated with a type of data at the at least one UE. In an aspect, the training request transmitted to the at least one UE at block 706 further includes the access token.

[0165] The operations of FIG. 7 are merely illustrative, and variations are contemplated to be within the scope of the present disclosure. In embodiments, the operations may include other blocks not illustrated in FIG. 7. In embodiments, the operations may not include every block illustrated in FIG. 7. In embodiments, the operation may be implemented in a different order than that illustrated in FIG. 7. Such and other embodiments are contemplated to be within the scope of the present disclosure.

[0166] Referring now to FIG. 8, there is shown a block diagram of example components of a UE or a network apparatus. The apparatus includes an electronic storage 810, a processor 820, a memory 850, and a network interface 840. The various components may be communicatively coupled with each other. The processor 820 may be and may include any type of processor, such as a single-core central processing unit (CPU), a multi-core CPU, a microprocessor, a digital signal processor (DSP), a System-on-Chip (SoC), or any other type of processor. The memory 850 may be a volatile type of memory, e.g., RAM, or a non-volatile type of memory, e.g., NAND flash memory. The memory 850 includes processor-readable instructions that are executable by the processor 820 to cause the apparatus to perform various operations, including those mentioned herein, such as the operations of FIGS. 1-7.

[0167] The electronic storage 810 may be and include any type of electronic storage used for storing data, such as hard disk drive, solid state drive, and / or optical disc, among other types of electronic storage. The electronic storage 810 stores processor-readable instructions for causing the apparatus to perform its operations and stores data associated with such operations, such as storing data relating to 5G NR standards, among other data. The network interface 840 may implement wireless networking technologies such as 5G NR and / or other wireless networking technologies.

[0168] The components shown in FIG. 8 are merely examples, and persons skilled in the art will understand that an apparatus includes other components not illustrated and may include multiples of any of the illustrated components. Such and other embodiments are contemplated to be within the scope of the present disclosure.

[0169] Further embodiments of the present disclosure include the following examples.

[0170] Example 1.1. A method comprising: transmitting, by a user equipment (UE) to a network, at least one UE capability associated with a machine learning (ML) model training service provided by the UE; receiving, by the UE, an instruction for providing the ML model training service; and providing, by the UE, the ML model training service.

[0171] Example 1.2. The method of Example 1.1, wherein the instruction is received based on transmitting the at least one UE capability.

[0172] Example 1.3. The method of Example 1.1 or 1.2, wherein the providing the ML model training service comprises: receiving, by the UE, a request for providing the ML model training service; determining to provide, by the UE, the requested ML model training service, wherein the determining is based on the received request and on the received instruction; and providing, by the UE, based on the determining, the requested ML model training service.

[0173] Example 1.4. The method of Example 1.3, wherein: the request for providing the ML model training service comprises one or more parameters of the requested ML model training service; and the determining is further based on at least one of the one or more parameters.

[0174] Example 1.5. The method of Example 1.4, wherein the one or more parameters of the ML model training service comprise at least one of the following: one or more ML model identifiers, one or more ML model categories,one or more analytics identifiers, or one or more UE data type identifiers.

[0175] Example 1.6. The method of any one of Examples 1.1-1.5, wherein the receiving the instruction comprises: receiving the instruction comprising a policy, wherein the policy comprises at least one of the following: an indication of one or more first ML models allowed to be trained at the UE, an indication of one or more second ML models not allowed to be trained at the UE, an indication of one or more first categories of ML models allowed to be trained at the UE, an indication of one or more second categories of ML models not allowed to be trained at the UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the UE, an indication of one or more first types of data of the UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the UE not allowed to be used for training the one or more second ML models.

[0176] Example 1.7. The method of any one of Examples 1.1-1.5, wherein the receiving the instruction comprises: receiving the instruction comprising an access token, wherein the access token comprises token Examples for at least one the following: service consumer information indicating an ML model service consumer entity allowed to use the ML model training service provided by one or more UEs, service provider information indicating one or more UE identifiers, wherein each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the ML model training service, and wherein a first UE identifier of the one or more UE identifiers identifies the UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the ML model training service: a model identifier of the ML model,a model category of the ML model, one or more analytics identifiers, wherein each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model, or one or more data type identifiers, wherein each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

[0177] Example 1.8. The method of Example 1.7, wherein the access token comprises a JavaScript object notation (JSON) web token (JWT).

[0178] Example 1.9. The method of Example 1.7 or Example 1.8, wherein the first UE identifier identifying the UE comprises an address for communicating with the UE.

[0179] Example 1.10. The method of Example 1.9, wherein the address for communicating with the UE comprises a user plane address of the UE.

[0180] Example 1.11. The method of any one of Examples 1.7-1.10, wherein a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information on a wireless network interface, received signal strength on a wireless network interface, location data, photos, videos, or sensor data.

[0181] Example 2.1. A method comprising: receiving, by a service authorization entity of a wireless communication system from a machine learning (ML) model training service consumer entity, a service request for accessing an ML model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service; verifying, by the service authorization entity, the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; and transmitting, by the service authorization entity to the ML model training service consumer, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service.

[0182] Example 2.2. The method of Example 2.1, wherein the information associated with the requested ML model training service comprises at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

[0183] Example 2.3. The method of Example 2.2, further comprising: determining, based the one or more filtering criteria, the at least one UE that allows the requested ML model training service.

[0184] Example 2.4. The method any one of Examples 2.1-2.3, wherein the ML model training service consumer entity comprises at least one of the following: an access network entity of the wireless communication system, a core network entity of the wireless communication system, or an application function entity.

[0185] Example 2.5. The method of Example 2.4, wherein the core network entity is a network function.

[0186] Example 2.6. The method of Example 2.5, wherein the network function is a network data analytics function entity.

[0187] Example 2.7. The method of Example 2.6, wherein the network data analytics function entity comprises a model training logical function.

[0188] Example 2.8. The method of any one of Examples 2.1-2.7, wherein the verifying the service request is further based on an authentication of the ML model training service consumer entity.

[0189] Example 2.9. The method of any one of Examples 2.1-2.8, further comprising: transmitting, by the service authorization entity to the at least one UE, based on the verifying, a policy comprising at least one of the following: an indication of one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second ML models not allowed to be trained at the at least one UE,an indication of one or more first categories of ML models allowed to be trained at the at least one UE, an indication of one or more second categories of ML models not allowed to be trained at the at least one UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models.

[0190] Example 2.10. The method of any one of Examples 2.1-2.8, further comprising: transmitting, by the service authorization entity to an access network node associated with the at least one UE, based on the verifying, a policy comprising at least one of the following: an indication of one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first categories of ML models allowed to be trained at the at least one UE, an indication of one or more second categories of ML models not allowed to be trained at the at least one UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models.

[0191] Example 2.11. The method of any one of Examples 2.1-2.8, wherein the response indicating the at least one UE that allows the requested ML model training service comprises an access token.

[0192] Example 2.12. The method of Example 2.11, wherein the access token comprises token Examples for at least one the following: service consumer information indicating the ML model training service consumer entity allowed to utilize the requested ML model training service, service provider information indicating one or more UE identifiers, wherein each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model training service, and wherein a first UE identifier of the one or more UE identifiers identifies the at least one UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the requested ML model training service: a model identifier of the ML model, a model category of the ML model, one or more analytics identifiers, wherein each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model, or one or more data type identifiers, wherein each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

[0193] Example 2.13. The method of Example 2.11 or Example 2.12, further comprising: generating, by the service authorization entity based on the verifying, the access token.

[0194] Example 2.14. The method of any one of Examples 2.11-2.13, wherein the access token comprises a JavaScript object notation (JSON) web token (JWT).

[0195] Example 2.15. The method of any one of Examples 2.12-2.14, wherein the first UE identifier identifying the at least one UE comprises an address for communicating with the at least one UE.

[0196] Example 2.16. The method of Example 2.15, wherein the address for communicating with the at least one UE comprises a user plane address of the at least one UE.

[0197] Example 2.17. The method of any one of Examples 2.12-2.16, wherein a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information on a wireless network interface, received signal strength on a wireless network interface,location data, photos, videos, or sensor data.

[0198] Example 2.18. The method of any one of Examples 2.1-2.17, wherein the ML model training capabilities of the one or more UEs comprises at least one of the following: an indication of an ML model allowed be trained at the at least one UE, an indication of data of the at least one UE allowed for training the ML model, or an indication of an allowable analytics operation associated with ML model.

[0199] Example 2.19. The method of any one of Examples 2.1-2.18, wherein the service authorization entity of a wireless communication system is a network function.

[0200] Example 2.20. The method of Example 2.19, wherein the network function is a unified data management entity.

[0201] Example 3.1. A method comprising: transmitting, by a machine learning (ML) model training service consumer entity, to a service authorization entity of a wireless communication system, a service request for accessing an ML model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service; receiving, by the ML model training service consumer entity from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and transmitting, by the ML model training service consumer entity to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained at the at least one UE.

[0202] Example 3.2. The method of Example 3.1, wherein the information associated with the requested ML model training service comprises at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

[0203] Example 3.3. The method of Example 3.1 or Example 3.2, wherein the ML model training service consumer entity comprises at least one of the following: an access network entity of the wireless communication system, a core network entity of the wireless communication system, or an application function entity.

[0204] Example 3.4. The method of any one of Examples 3.1-3.3, wherein the response indicating the at least one UE that allows the requested ML model training service comprises an access token.

[0205] Example 3.5. The method of Example 3.4, wherein the access token comprises token Examples for at least one the following: service consumer information indicating the ML model training service consumer entity allowed to utilize the requested ML model training service provided by one or more UEs, service provider information indicating one or more UE identifiers, wherein each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model training service, and wherein a first UE identifier of the one or more UE identifiers identifies the at least one UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the requested ML model training service: a model identifier of an ML model, a model category of the ML model, one or more analytics identifiers, wherein each of the one or more analytics identifiers is associated with an analytics operation allowed for the ML model, or one or more data type identifiers, wherein each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

[0206] Example 3.6. The method of Example 3.5, wherein the first UE identifier identifying the at least one UE comprises an address for communicating with the at least one UE.

[0207] Example 3.7. The method of Example 3.6, wherein the address for communicating with the at least one UE is a user plane address of the at least one UE.

[0208] Example 3.8. The method of Example 3.5, wherein a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information,received signal strength on a wireless network interface, location data, photos, videos, or sensor data.

[0209] Example 3.9. The method of any one of Examples 3.4-3.8, wherein the training request transmitted to the at least one UE further comprises the access token.

[0210] Example 3.10. The method of any one of Examples 3.4-3.9, wherein the access token comprises a JavaScript object notation (JSON) web token (JWT).

[0211] Example 3.11. The method of any one of Examples 3.1-3.10, wherein the information associated with the ML model in the training request comprises at least one of: one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers.

[0212] Example 4. A user equipment (UE) comprising: at least one processor; and at least one memory storing instructions which, when executed by the processor, cause the UE at least to perform the method of any one of Examples 1.1-1.11.

[0213] Example 5. An apparatus comprising: at least one processor; and at least one memory storing instructions which, when executed by the processor, cause the apparatus at least to perform the method of any one of Examples 2.1-2.20.

[0214] Example 6. An apparatus comprising: at least one processor; and at least one memory storing instructions which, when executed by the processor, cause the apparatus at least to perform the method of any one of Examples 3.1-3.11.

[0215] Example 7. A user equipment (UE) comprising: means for transmitting, to a network, at least one UE capability associated with a machine learning (ML) model training service provided by the UE; means for receiving an instruction for providing the ML model training service; and means for providing the ML model training service.

[0216] Example 8. An apparatus comprising:means for receiving, from a machine learning (ML) model training service consumer entity, a service request for accessing an ML model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service; means for verifying the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; and means for transmitting, to the ML model training service consumer entity, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service.

[0217] Example 9. An apparatus comprising: means for transmitting, to a service authorization entity of a wireless communication system, a service request for accessing a machine learning (ML) model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service; means for receiving, from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and means for transmitting, to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained at the at least one UE.

[0218] Example 10. A system comprising: at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the system at least to: transmit, to a network, at least one user equipment (UE) capability associated with a machine learning (ML) model training service provided by the UE; receive an instruction for providing the ML model training service; provide the ML model training service; receive, from an ML model training service consumer entity, a service request for accessing an ML model training service of a UE, wherein the service request comprises information associated with the requested ML model training service; verify the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs;transmit, to the ML model training service consumer entity, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service. transmit, to a service authorization entity of a wireless communication system, an service request for accessing a ML model training service of a UE, wherein the service request comprises information associated with the requested ML model training service; receive, from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and transmit, to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained at the at least one UE.

[0219] Example 11. A system comprising: means for transmitting, to a network, at least one user equipment (UE) capability associated with a machine learning (ML) model training service provided by the UE; means for receiving an instruction for providing the ML model training service; means for providing the ML model training service; means for receiving, from an ML model training service consumer entity, a service request for accessing an ML model training service of a UE, wherein the service request comprises information associated with the requested ML model training service; means for verifying the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; means for transmitting, to the ML model training service consumer entity, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service; means for transmitting, to a service authorization entity of a wireless communication system, a service request for accessing a ML model training service of a UE, wherein the service request comprises information associated with the requested ML model training service; means for receiving, from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and means for transmitting, to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained atthe at least one UE.

[0220] Example 12. A computer readable non-transitory medium comprising program instructions stored thereon for performing at least the method of any one of Examples 1.1-1.11, 2.1-2.20, and 3.1-3.11.

[0221] Example 13. A computer readable medium comprising program instructions stored thereon for performing at least the method of any one of Examples 1.1-1.11, 2.1-2.20, and 3.1- 3.11.

[0222] Example 14. A computer program comprising instructions for performing at least the method of any one of Examples 1.1-1.11, 2.1-2.20, and 3.1-3.11.

[0223] The embodiments and aspects disclosed herein are examples of the present disclosure and may be embodied in various forms. For instance, although certain embodiments herein are described as separate embodiments, each of the embodiments herein may be combined with one or more of the other embodiments herein. Specific structural and functional details disclosed herein are not to be interpreted as limiting, but as a basis for the claims and as a representative basis for teaching one skilled in the art to variously employ the present disclosure in virtually any appropriately detailed structure. Like reference numerals may refer to similar or identical elements throughout the description of the figures.

[0224] The phrases “in an aspect,” “in aspects,” “in various aspects,” “in some aspects,” or “in other aspects” may each refer to one or more of the same or different aspects in accordance with this present disclosure. The phrase “a plurality of’ may refer to two or more.

[0225] The phrases “in an embodiment,” “in embodiments,” “in various embodiments,” “in some embodiments,” or “in other embodiments” may each refer to one or more of the same or different embodiments in accordance with the present disclosure. A phrase in the form “A or B” means “(A), (B), or (A and B).” A phrase in the form “at least one of A, B, or C” means “(A); (B); (C); (A and B); (A and C); (B and C); or (A, B, and Q.”

[0226] Any of the herein described methods, programs, algorithms or codes may be converted to, or expressed in, a programming language or computer program. The terms “programming language” and “computer program,” as used herein, each include any language used to specify instructions to a computer, and include (but is not limited to) the following languages and their derivatives: Assembler, Basic, Batch files, BCPL, C, C+, C++, Delphi, Fortran, Java, JavaScript, machine code, operating system command languages, Pascal, Perl, PL1, Python, scripting languages, Visual Basic, metalanguages which themselves specify programs, and all first, second, third, fourth, fifth, or further generation computer languages. Also included are database and other data schemas, and any other meta-languages. Nodistinction is made between languages which are interpreted, compiled, or use both compiled and interpreted approaches. No distinction is made between compiled and source versions of a program. Thus, reference to a program, where the programming language could exist in more than one state (such as source, compiled, object, or linked) is a reference to any and all such states. Reference to a program may encompass the actual instructions and / or the intent of those instructions.

[0227] While aspects of the present disclosure have been shown in the drawings, it is not intended that the present disclosure be limited thereto, as it is intended that the present disclosure be as broad in scope as the art will allow and that the specification be read likewise. Therefore, the above description should not be construed as limiting, but merely as exemplifications of particular aspects. Those skilled in the art will envision other modifications within the scope and spirit of the claims appended hereto.

Claims

WE CLAIM:

1. A method comprising: transmitting, by a user equipment (UE) to a network, at least one UE capability associated with a machine learning (ML) model training service provided by the UE; receiving, by the UE, an instruction for providing the ML model training service; and providing, by the UE, the ML model training service.

2. The method of claim 1, wherein the instruction is received based on transmitting the at least one UE capability.

3. The method of claim 1 or 2, wherein the providing the ML model training service comprises: receiving, by the UE, a request for providing the ML model training service; determining to provide, by the UE, the requested ML model training service, wherein the determining is based on the received request and on the received instruction; and providing, by the UE, based on the determining, the requested ML model training service.

4. The method of claim 3, wherein: the request for providing the ML model training service comprises one or more parameters of the requested ML model training service; and the determining is further based on at least one of the one or more parameters.

5. The method of claim 4, wherein the one or more parameters of the ML model training service comprise at least one of the following: one or more ML model identifiers, one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers.

6. The method of any one of claims 1-5, wherein the receiving the instruction comprises: receiving the instruction comprising a policy, wherein the policy comprises at least one of the following:an indication of one or more first ML models allowed to be trained at the UE, an indication of one or more second ML models not allowed to be trained at the UE, an indication of one or more first categories of ML models allowed to be trained at the UE, an indication of one or more second categories of ML models not allowed to be trained at the UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the UE, an indication of one or more first types of data of the UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the UE not allowed to be used for training the one or more second ML models.

7. The method of any one of claims 1-5, wherein the receiving the instruction comprises: receiving the instruction comprising an access token, wherein the access token comprises token claims for at least one the following: service consumer information indicating an ML model service consumer entity allowed to use the ML model training service provided by one or more UEs, service provider information indicating one or more UE identifiers, wherein each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the ML model training service, and wherein a first UE identifier of the one or more UE identifiers identifies the UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the ML model training service: a model identifier of the ML model, a model category of the ML model, one or more analytics identifiers, wherein each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model, orone or more data type identifiers, wherein each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

8. The method of claim 7, wherein the access token comprises a JavaScript object notation (JSON) web token (JWT).

9. The method of claim 7 or claim 8, wherein the first UE identifier identifying the UE comprises an address for communicating with the UE.

10. The method of claim 9, wherein the address for communicating with the UE comprises a user plane address of the UE.

11. The method of any one of claims 7-10, wherein a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information on a wireless network interface, received signal strength on a wireless network interface, location data, photos, videos, or sensor data.

12. A method comprising: receiving, by a service authorization entity of a wireless communication system from a machine learning (ML) model training service consumer entity, a service request for accessing an ML model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service; verifying, by the service authorization entity, the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; and transmitting, by the service authorization entity to the ML model training service consumer, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service.

13. The method of claim 12, wherein the information associated with the requested ML model training service comprises at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

14. The method of claim 13, further comprising: determining, based the one or more filtering criteria, the at least one UE that allows the requested ML model training service.

15. The method any one of claims 12-14, wherein the ML model training service consumer entity comprises at least one of the following: an access network entity of the wireless communication system, a core network entity of the wireless communication system, or an application function entity.

16. The method of claim 15, wherein the core network entity is a network function.

17. The method of claim 16, wherein the network function is a network data analytics function entity.

18. The method of claim 17, wherein the network data analytics function entity comprises a model training logical function.

19. The method of any one of claims 12-18, wherein the verifying the service request is further based on an authentication of the ML model training service consumer entity.

20. The method of any one of claims 12-19, further comprising:transmitting, by the service authorization entity to the at least one UE, based on the verifying, a policy comprising at least one of the following: an indication of one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first categories of ML models allowed to be trained at the at least one UE, an indication of one or more second categories of ML models not allowed to be trained at the at least one UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models.

21. The method of any one of claims 12-19, further comprising: transmitting, by the service authorization entity to an access network node associated with the at least one UE, based on the verifying, a policy comprising at least one of the following: an indication of one or more first ML models allowed to be trained at the at least one UE, an indication of one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first categories of ML models allowed to be trained at the at least one UE, an indication of one or more second categories of ML models not allowed to be trained at the at least one UE, an indication of one or more first analytic operations associated with the one or more first ML models allowed to be trained at the at least one UE,an indication of one or more second analytic operations associated with the one or more second ML models not allowed to be trained at the at least one UE, an indication of one or more first types of data of the at least one UE allowed to be used for training the one or more first ML models, or an indication of one or more second types of data of the at least one UE not allowed to be used for training the one or more second ML models.

22. The method of any one of claims 12-19, wherein the response indicating the at least one UE that allows the requested ML model training service comprises an access token.

23. The method of claim 22, wherein the access token comprises token claims for at least one the following: service consumer information indicating the ML model training service consumer entity allowed to utilize the requested ML model training service, service provider information indicating one or more UE identifiers, wherein each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model training service, and wherein a first UE identifier of the one or more UE identifiers identifies the at least one UE, or service information indicating at least one of the following parameters of an ML model allowed to be trained using the requested ML model training service: a model identifier of the ML model, a model category of the ML model, one or more analytics identifiers, wherein each of the one or more analytics identifiers is associated with an analytics operation allowable for the ML model, or one or more data type identifiers, wherein each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

24. The method of claim 22 or claim 23, further comprising: generating, by the service authorization entity based on the verifying, the access token.

25. The method of any one of claims 22-24, wherein the access token comprises a JavaScript object notation (JSON) web token (JWT).

26. The method of any one of claims 23 to 25, wherein the first UE identifier identifying the at least one UE comprises an address for communicating with the at least one UE.

27. The method of claim 26, wherein the address for communicating with the at least one UE comprises a user plane address of the at least one UE.

28. The method of any one of claims 23 to 27, wherein a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information on a wireless network interface, received signal strength on a wireless network interface, location data, photos, videos, or sensor data.

29. The method of any one of claims 12-28, wherein the ML model training capabilities of the one or more UEs comprises at least one of the following: an indication of an ML model allowed be trained at the at least one UE, an indication of data of the at least one UE allowed for training the ML model, or an indication of an allowable analytics operation associated with ML model.

30. The method of any one of claims 12-29, wherein the service authorization entity of a wireless communication system is a network function.

31. The method of claim 30, wherein the network function is a unified data management entity.

32. A method comprising: transmitting, by a machine learning (ML) model training service consumer entity, to a service authorization entity of a wireless communication system, a service request for accessing an ML model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service;receiving, by the ML model training service consumer entity from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and transmitting, by the ML model training service consumer entity to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained at the at least one UE.

33. The method of claim 32, wherein the information associated with the requested ML model training service comprises at least one of the following: one or more UE identifiers, each identifying a respective UE requested for the ML model training service, or one or more filtering criteria associated with at least one of the following: a model identifier of a requested ML model, a model category of the requested ML model, or an analytics identifier of an analytics operation associated with the requested ML model.

34. The method of claim 32 or claim 33, wherein the ML model training service consumer entity comprises at least one of the following: an access network entity of the wireless communication system, a core network entity of the wireless communication system, or an application function entity.

35. The method of any one of claims 32-34, wherein the response indicating the at least one UE that allows the requested ML model training service comprises an access token.

36. The method of claim 35, wherein the access token comprises token claims for at least one the following: service consumer information indicating the ML model training service consumer entity allowed to utilize the requested ML model training service provided by one or more UEs, service provider information indicating one or more UE identifiers, wherein each of the one or more UE identifiers identifies a respective one of the one or more UEs providing the requested ML model training service, and wherein a first UE identifier of the one or more UE identifiers identifies the at least one UE, orservice information indicating at least one of the following parameters of an ML model allowed to be trained using the requested ML model training service: a model identifier of an ML model, a model category of the ML model, one or more analytics identifiers, wherein each of the one or more analytics identifiers is associated with an analytics operation allowed for the ML model, or one or more data type identifiers, wherein each of the one or more data type identifiers identifies a type of data at a respective one of the one or more UEs allowable for training the ML model.

37. The method of claim 36, wherein the first UE identifier identifying the at least one UE comprises an address for communicating with the at least one UE.

38. The method of claim 37, wherein the address for communicating with the at least one UE is a user plane address of the at least one UE.

39. The method of claim 36, wherein a data type identifier of the one or more data type identifiers identifies a type of data associated with at least one of the following: channel state information, received signal strength on a wireless network interface, location data, photos, videos, or sensor data.

40. The method of any one of claims 35-39, wherein the training request transmitted to the at least one UE further comprises the access token.

41. The method of any one of claims 35-40, wherein the access token comprises a JavaScript object notation (JSON) web token (JWT).

42. The method of any one of claims 32-41, wherein the information associated with the ML model in the training request comprises at least one of: one or more ML model identifiers,one or more ML model categories, one or more analytics identifiers, or one or more UE data type identifiers.

43. A user equipment (UE) comprising: at least one processor; and at least one memory storing instructions which, when executed by the processor, cause the UE at least to perform the method of any one of claims 1-11.

44. An apparatus comprising: at least one processor; and at least one memory storing instructions which, when executed by the processor, cause the apparatus at least to perform the method of any one of claims 12-31.

45. An apparatus comprising: at least one processor; and at least one memory storing instructions which, when executed by the processor, cause the apparatus at least to perform the method of any one of claims 32-42.

46. A user equipment (UE) comprising: means for transmitting, to a network, at least one UE capability associated with a machine learning (ML) model training service provided by the UE; means for receiving an instruction for providing the ML model training service; and means for providing the ML model training service.

47. An apparatus comprising: means for receiving, from a machine learning (ML) model training service consumer entity, a service request for accessing an ML model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service; means for verifying the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; andmeans for transmitting, to the ML model training service consumer entity, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service.

48. An apparatus comprising: means for transmitting, to a service authorization entity of a wireless communication system, a service request for accessing a machine learning (ML) model training service of a user equipment (UE), wherein the service request comprises information associated with the requested ML model training service; means for receiving, from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and means for transmitting, to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained at the at least one UE.

49. A system comprising: at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the system at least to: transmit, to a network, at least one user equipment (UE) capability associated with a machine learning (ML) model training service provided by the UE; receive an instruction for providing the ML model training service; provide the ML model training service; receive, from an ML model training service consumer entity, a service request for accessing an ML model training service of a UE, wherein the service request comprises information associated with the requested ML model training service; verify the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; transmit, to the ML model training service consumer entity, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service. transmit, to a service authorization entity of a wireless communication system, an service request for accessing a ML model training service of a UE, wherein theservice request comprises information associated with the requested ML model training service; receive, from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and transmit, to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained at the at least one UE.

50. A system comprising: means for transmitting, to a network, at least one user equipment (UE) capability associated with a machine learning (ML) model training service provided by the UE; means for receiving an instruction for providing the ML model training service; means for providing the ML model training service; means for receiving, from an ML model training service consumer entity, a service request for accessing an ML model training service of a UE, wherein the service request comprises information associated with the requested ML model training service; means for verifying the service request, wherein the verifying is based at least in part on a comparison of the information associated with the requested ML model training service and ML model training capabilities of one or more UEs; means for transmitting, to the ML model training service consumer entity, based on the verifying, a response indicating at least one UE of the one or more UEs that allows the requested ML model training service; means for transmitting, to a service authorization entity of a wireless communication system, a service request for accessing a ML model training service of a UE, wherein the service request comprises information associated with the requested ML model training service; means for receiving, from the service authorization entity, a response indicating at least one UE that allows the requested ML model training service; and means for transmitting, to the at least one UE, based on the receiving the response, a training request comprising information associated with an ML model to be trained at the at least one UE.

51. A computer readable non-transitory medium comprising program instructions stored thereon for performing at least the method of any one of claims 1-11, 12-31, or 32-42.

52. A computer readable medium comprising program instructions stored thereon for performing at least the method of any one of claims 1-11, 12-31, or 32-42.

53. A computer program comprising instructions for performing at least the method of any one of claims 1-11, 12-31, or 32-42.

Images