Data collection for AI / ML models

Abstract

A method for implementing data collection involving tasks of artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE) is described. In one example, a data collection request may be received from a UE. The received data collection request may include an identifier of a task for which data is to be collected involving the AI / ML model provided in the UE. Thereafter, a measurement configuration corresponding to the identifier of the task may be determined. A response message may then be sent to the UE. The response message may include a measurement identifier corresponding to the measurement configuration.

Description

Technical Field

[0001] Methods and examples for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models are described. Background Art

[0002] In communication networks such as fifth generation (5G) networks, various methods may be used to improve the efficiency of various network and / or communication functions that may be implemented by entities within the telecommunications network. Through control functions implemented by logical instructions, efficient and high performance operation of communication systems is possible. Recently, such control has been implemented through the use of models enabled by artificial intelligence. Such models include various data-driven functions that use machine learning techniques to generate a set of outputs that include prediction information and / or decision parameters.

[0003] Artificial intelligence / machine learning models (referred to as AI / ML models) can be trained through online or offline processes. The training involves subjecting the AI / ML models to various training data sets that enable the AI / ML models to learn features and patterns. In the context of a communication network, the trained AI / ML models can then be used to implement different network operations to achieve the functionality of the communication network. When a user equipment (i.e., UE) interacts with any one or more network devices within the network, the training data set can be acquired by collecting data of different network operations. Summary of the invention

[0004] Aspects of the present subject matter relate to methods of data collection for implementing tasks involving artificial intelligence / machine learning (AI / ML) models.

[0005] In one example, a network device for implementing data collection for a task involving an artificial intelligence / machine learning (AI / ML) model present in a user equipment (UE) is described. The network device may include at least one processor and at least one memory. The at least one memory may store instructions that can be executed by at least one processor. When the instructions are executed by at least one processor, the network device may cause the network device to perform one or more functions. For example, when the instructions are executed, the network device may receive a data collection request from a user equipment (UE). The received data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. The AI / ML model may be provided in the UE. The network device may then determine a measurement configuration corresponding to the identifier of the task. Thereafter, a response message may be sent to the UE. The response message may include a measurement identifier corresponding to the measurement configuration.

[0006] In one example, based on the measurement configuration corresponding to the measurement identifier, the network device may also reconfigure a radio communication link associated with the UE.

[0007] In another example, the data collection request may include at least one of a first set of attributes and a request type. The first set of attributes may be used for data collection corresponding to the data collection request. In addition, the type of request may indicate one of initiation or termination of data collection.

[0008] In another example, the first set of attributes may include at least one of a time instance at which data collection is initiated, a time instance at which data collection is terminated, a periodicity of data collection, a sensitivity of data requirements, or a characteristic of a geographic region for performing data collection.

[0009] In another example, the response message may further include: a list of a second set of attributes determined using the first set of attributes.

[0010] In another example, the network device may also retrieve a mapping between the plurality of tasks and the plurality of measurement configurations. Based on the mapping and the identifier of the task obtained from the data collection request, the network device may determine the measurement configuration.

[0011] In another example, the network device may also determine a first configuration corresponding to an earlier configured radio communication link associated with the UE. Thereafter, the network device may evaluate the first configuration based on the determined measurement configuration to determine whether the first configuration can be used to configure the radio communication link associated with the UE for data collection. Upon confirming that the first configuration can be used, the network device may use the first configuration as the measurement configuration to configure the radio communication link associated with the UE for data collection.

[0012] In another example, based on the evaluation of the first configuration, the network device may also confirm that the first configuration cannot be used to configure the radio communication link associated with the UE for data collection. The network device may then determine a second configuration. Thereafter, the network device may use the second configuration as a measurement configuration for configuring the radio communication link associated with the UE for data collection.

[0013] In another example, the network device may process a data collection request received from a UE. A radio communication link associated with the UE may be previously configured. Thereafter, based on previously configured measurements of the UE, the network device may determine a measurement configuration. Subsequently, the network device may send a response message to the UE. The response message may include: a measurement identifier corresponding to the measurement configuration to configure a portion of the previously configured radio communication link associated with the UE.

[0014] In another example, the network device may process a data collection termination request received from the UE. The termination request may include: a log related to data collection performed according to the data collection request. Upon determining that the termination request meets the predefined condition, the network device may then send a control instruction to the UE. The control instruction is executable by the UE to reconfigure a radio communication link associated with the UE to terminate data collection performed by the UE.

[0015] In another example, a user equipment (UE) is described, which is used to implement data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in the UE. The network device may include at least one processor and at least one memory. The at least one memory may store instructions that can be executed by at least one processor. When the instruction is executed by at least one processor, the user equipment (UE) may perform one or more functions. For example, after the execution of the instruction, the UE may send a data collection request. The data collection request may include: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. The AI / ML model may be provided in the UE. Then, the UE may receive a response message. The response message may include a measurement identifier of a measurement configuration corresponding to the identifier of the task. Thereafter, based on the measurement configuration, the UE may collect data for tasks involving the AI / ML model.

[0016] In one example, based on the response message, the task may include at least one of: training of an AI / ML model, retraining of an AI / ML model, monitoring of an AI / ML model, monitoring of functions performed by the UE, or reasoning using an AI / ML model.

[0017] In another example, the UE may further include at least one of a first set of attributes and a request type. The first set of attributes may be used for data collection corresponding to the data collection request. The type of request may indicate one of initiation or termination of data collection.

[0018] In another example, the UE may also send a termination request. The termination request may include: a log related to data collection performed according to the received response message. Thereafter, the UE may execute a control instruction received from the network device in response to the termination request. When the control instruction is executed, the radio communication link may be further reconfigured to terminate the data collection performed by the UE.

[0019] In another example, a method for implementing data collection for a task involving an artificial intelligence / machine learning (AI / ML) model present in a user equipment (UE) is described. A network device may receive a data collection request from a user equipment (UE). The received data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. In addition, the AI / ML model may be provided in the UE. Thereafter, the network device may determine a measurement configuration corresponding to the identifier of the task. Thereafter, the network device may send a response message to the UE. The response message may include a measurement identifier corresponding to the measurement configuration.

[0020] In another example, a method for implementing data collection for a task involving an artificial intelligence / machine learning (AI / ML) model present in a user equipment (UE) is described. A user equipment (UE) may send a data collection request. The data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. In addition, the AI / ML model may be provided in the UE. Thereafter, the UE may receive a response message. The response message may include an identifier of a measurement configuration corresponding to the identifier of the task. Thereafter, the UE may collect data for the task involving the AI / ML model based on the measurement configuration.

[0021] In one example, the task may include at least one of: training of an AI / ML model, retraining of an AI / ML model, monitoring of an AI / ML model, monitoring of a function performed by a UE, or reasoning using an AI / ML model.

[0022] In another example, the data collection request may include one of the first set of attributes and the request type. The first set of attributes may be used for data collection corresponding to the data collection request. In addition, the type of request may indicate one of initiation or termination of data collection.

[0023] In another example, the first set of attributes may include at least one of a time instance at which data collection is initiated, a time instance at which data collection is terminated, a periodicity of data collection, a sensitivity of data requirements, or a characteristic of a geographic region for performing data collection.

[0024] In another example, a non-transitory computer-readable medium is described. The non-transitory computer-readable medium may include computer-readable instructions. When the instructions are executed by a processor, the network device may implement data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE). The instructions may cause the network device to receive a data collection request from a user equipment (UE). The received data collection request may include: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. The AI / ML model may be provided in the UE. Thereafter, the instruction may also cause a measurement configuration corresponding to the identifier of the task to be determined. The instruction may also cause a response message to be sent to the UE. The response message may include a measurement identifier corresponding to the measurement configuration.

[0025] In one example, the instruction may also cause reconfiguration of a radio communication link associated with the UE based on the measurement configuration corresponding to the measurement identifier.

[0026] In another example, a non-transitory computer-readable medium is described. The non-transitory computer-readable medium may include computer-readable instructions. When the instructions are executed by a processor, a user equipment (UE) may implement data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in the UE. These instructions may cause the UE to send a data collection request. The data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. The AI / ML model may be provided in the UE. Thereafter, the instruction may also cause the UE to receive a response message. The response message may include a measurement identifier of a measurement configuration corresponding to an identifier of the task. The instruction may also cause the UE to collect data for tasks involving the AI / ML model based on the measurement configuration. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The following detailed description refers to the accompanying drawings, in which:

[0028] Figure 1 illustrates a network apparatus for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE) in accordance with implementations of the present subject matter;

[0029] Figure 2 Illustrated is a user equipment (UE) for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present within the UE in accordance with implementations of the present subject matter;

[0030] Figure 3A network environment according to an implementation of the present subject matter is illustrated, the network environment including a network device for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE);

[0031] Figure 4 is a block diagram of a network device for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE) according to an implementation of the present subject matter;

[0032] Figure 5 is an example call flow diagram depicting various functions between a user equipment (UE) and a network device for implementing data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in the UE in accordance with an implementation of the present subject matter;

[0033] Figure 6 is an example call flow diagram depicting various functions between a user equipment (UE) and a network device for implementing data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in the UE according to another implementation of the present subject matter;

[0034] Figure 7 is an example call flow diagram depicting various functions between a user equipment (UE) and a network device for implementing data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in the UE according to another implementation of the present subject matter;

[0035] Figure 8 is an example call flow diagram depicting various functions between a user equipment (UE) and a network device for terminating data collection according to implementations of the present subject matter;

[0036] Fig. 9 is a flow chart depicting an example method to be implemented in a network device according to an implementation of the present subject matter for implementing data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in a UE;

[0037] Fig.10 is a flow chart depicting an example method to be implemented in a user equipment (UE) for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in the UE in accordance with implementations of the present subject matter;

[0038] Fig.11 is a flow chart depicting an example method for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE) according to implementations of the present subject matter;

[0039] Fig.12 A non-transitory computer-readable medium for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE) according to implementations of the present subject matter is illustrated; and

[0040] Fig.13 A non-transitory computer-readable medium for implementing data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in a user equipment (UE) according to another implementation of the present subject matter is illustrated.

[0041] Throughout the drawings, the same reference numerals denote similar, but not necessarily identical, elements. The drawings are not necessarily drawn to scale, and the dimensions of certain parts may be exaggerated to more clearly illustrate the examples shown. In addition, the drawings provide examples and / or implementations consistent with the description; however, the description is not limited to the examples and / or implementations provided in the drawings. DETAILED DESCRIPTION

[0042] A communication network may generally involve multiple user devices that may interact with one or more network devices. Each device may perform one or more network operations through its corresponding functions. Within the network, different devices may communicate with each other through the communication network, and data exchange may be facilitated. In addition, the manner in which communication between different devices may be achieved may be based on corresponding network protocols. In the context of the fifth generation (5G) network, all communications between different networks or user devices may be based on the network protocols specified in the 5G protocol stack.

[0043] Artificial intelligence / machine learning (AI / ML) models and techniques have been used to improve the operational efficiency of communication networks. For example, AI / ML-based techniques have been applied to network energy saving, load balancing, and mobility optimization. It can be noted that the above examples are only indicative and do not represent all network functions or operations that can be implemented using AI / ML models and / or techniques. The AI / ML model used to implement such techniques can be implemented in the radio access network or in the UE.

[0044] As described above, the AI / ML model can be trained based on the collected data. In some cases, the data collection can be performed by the UE itself. In this case, the data collection is controlled or managed by the access network. Therefore, the access network configures the UE for data collection of measurements that may need to be collected. Based on the configuration provided by the access network, the UE can collect measurements, which can then be sent to the access network. Although such an approach can be followed when the AI / ML model is implemented in the access network, such an approach may not be sufficient in the case where the AI / ML model is implemented within the UE itself.

[0045] It is understood that the data collected by the UE based on the RAN configuration (also referred to as measurements) may not be sufficient for training or for configuring various other tasks involving the AI / ML models within the UE. For example, the configuration time for data collection may be different from that required by the UE. In some other cases, properties such as the periodicity of measurement collection may also be different in the RAN-based configuration. Such aspects may cause challenges, especially in a multi-vendor environment, where UEs from different vendors (each with their own AI / ML models) may have to communicate and operate with an access network from another vendor.

[0046] Furthermore, in other cases, the data and its attributes required by the UE's AI / ML model may be known only to the UE and not to the access network, so there is always a risk that the data collected by the UE (based on the configuration provided by the access network) is not sufficient to complete the task (which may involve the AI / ML model). In addition, certain types of measurements (such as inter-frequency and inter-RAT measurements) require the configuration of measurement gaps, which, if not performed, will not enable the UE to collect data. These challenges can also exacerbate the problem of inaccurate configuration, which can make data collection challenging.

[0047] Methods for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE) are described. In one example, data collection or measurement can be initiated by the UE through a data collection request. The data collection request may include an identifier of a task, which may involve an AI / ML model implemented within the UE itself and may be performed using the same. Tasks involving AI / ML models may include various tasks, examples of which include, but are not limited to: training of AI / ML models, retraining of AI / ML models, monitoring of AI / ML models, monitoring of functions performed by the UE, or reasoning using AI / ML models. It may be noted that the following examples are indicative only and are not exhaustive. Other example tasks for data collection may also be performed without departing from the scope of the present subject matter.

[0048] Returning to this example, the data collection request may include an identifier of the task for which data collection is to be performed. Thereafter, the data collection request may be shared with a network device within the access network. The network device may process the data collection request upon receipt and determine one or more measurement configurations accordingly. In one example, one or more measurement configurations may be determined based on a predefined mapping available to the network device. To this end, the network device may compare the identifier of the task provided in the data collection request with a plurality of tasks and determine the configuration of the required (multiple) measurements (i.e., the measurement configuration) accordingly. The measurement configuration may be considered to define the manner in which the UE may perform measurements, or any communication or data exchange, based on which the UE will perform the data collection process.

[0049] Once one or more measurement configurations are determined, the network device may generate a response message based on the identified one or more measurement configurations. In one example, the response message may include an identifier of the identified (multiple) measurement configurations. Thereafter, the network device may send a response message to the UE within the access network. Thereafter, the response message may be used to configure a radio communication link associated with the UE (i.e., for reconfiguring the current measurement configuration of the UE or for configuring a new measurement configuration for the UE). Once the radio communication link is configured, the UE may start data collection based on the configuration performed by the network device (according to the measurement configuration). The radio communication link may refer to an RRC connection between the UE and the network device, and the configuration or reconfiguration of the radio communication link may modify the RRC connection, including the establishment / modification / release of the measurement configuration.

[0050] In one example, different configurations may be used as measurement configurations for configuring a radio communication link associated with a UE for data collection. For example, a network device may determine a first configuration based on which configuration of the radio communication link may be performed. The first configuration may be similar to a configuration setting of an earlier configured radio communication link previously established for the UE. In another example, the network device may determine a second configuration based on which the radio communication link may be reconfigured.

[0051] It can be appreciated that the current method enables the UE to specify tasks that require data collection. Based on this, the network device configures the radio communication link for data collection according to the request received from the UE. As a result, the UE can efficiently collect data for the intended task. Since the configuration of the radio communication link is based on the requirements communicated by the UE, the data collection enabled for the UE is consistent with the requirements of the UE.

[0052] Already referenced Figure 1-Figure 13The manner of implementing the above examples is explained in detail. Although aspects of the subject matter can be implemented in various communication systems, transmission environments, and / or configurations, these implementations are described as examples in the following (multiple) system environments.

[0053] The terms "during", "while", and "at" as used herein are not precise terms, meaning that an action occurs immediately after an initiating action, but there may be some delay, such as propagation delay, between the initial action and the reaction triggered by the initial action. In addition, for clarity of description, the words "connected" and "coupled" are used throughout, and may include direct or indirect connections. Various examples of the subject matter are described below with reference to several examples.

[0054] Figure 1 A network device 100 according to an implementation of the present subject matter is illustrated, which enables data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE). In one example, the network device 100 can be implemented within an access network. In the case where the access network is a fifth generation (5G) network, the network device 100 can be implemented at the RRC (radio resource control) layer of the 5G protocol stack. In another example, the network device 100 can be one of a base station device and a gNodeB (gNB). For another example, the network device 100 can also be implemented as any hardware-based, software-based, or network-based entity. It can also be noted that such a network device 100 can be understood based on the type of access network in which the network device 100 can be implemented.

[0055] like Figure 1 As depicted in , network device 100 includes processor 102 and memory 104, which is coupled to processor 102 and accessible by processor 102. Although not shown, network device 100 may include other components, such as an interface for communicating with an external storage or computing device through an access network, a display, an input / output interface, an operating system, applications, data, and other software or hardware components (all of which are not shown for simplicity). In one example, memory 104 may also include (multiple) instructions 106.

[0056] The processor 102 may be implemented as a dedicated processor, a shared processor, or multiple individual processors, some of which may be shared. The memory 104 may be communicatively connected to the processor 102. Among other functions, the processor 102 may also retrieve and execute computer-readable instructions stored in the memory 104, including (multiple) instructions 106. The memory 104 may be implemented as, for example, a volatile memory such as a RAM (random access memory) or a non-volatile memory such as an EPROM (erasable programmable read-only memory), flash memory, etc. The (multiple) instructions 106 may be executed to enable data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE).

[0057] Although not shown, the network device 100 can communicate with multiple user equipment or UEs via a communication link (wired to wireless) implemented within the access network. In the context of this example, each UE can include a corresponding artificial intelligence / machine learning (AI / ML) model, which has been explained in the previous description.

[0058] In operation, the processor 102 may retrieve and execute (multiple) instructions 106 to perform one or more actions, as further discussed. For example, as a result of the execution of the instructions 108, the network device 100 may receive a data collection request from a user equipment (UE). The received data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. As previously described, the AI / ML model may be provided in the UE. Thereafter, the instructions 110 may be executed to determine a measurement configuration corresponding to the identifier of the task.

[0059] Continuing, then, instructions 112 may be executed to send a response message to the UE. The response message may include a measurement identifier of the measurement configuration. In another example, based on the measurement configuration corresponding to the measurement identifier, instructions 106 may also cause reconfiguration of a radio communication link associated with the UE.

[0060] Combination Figure 2 The manner in which the UE receives a response message in response to a data collection request provided by the UE is explained in more detail. Figure 2 A user device 200 according to an implementation of the present subject matter is illustrated, which is capable of implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in the UE. Examples of UE 200 may include, but are not limited to, a mobile phone, a personal computer (PC), a portable computer, a laptop, a tablet computer, a wearable device, or any other device used by an end user. It may be noted that such an example of UE 200 is merely illustrative, and any other examples will also fall within the scope of the present subject matter.

[0061] like Figure 2 As depicted in , UE 200 includes a processor 202 and a memory 204, which is coupled to the processor 202 and accessible by the processor 202. Although not shown, UE 200 may include other components, such as an interface for communicating with an external storage or computing device through an access network, a display, an input / output interface, an operating system, applications, data, and other software or hardware components (all of which are not shown for simplicity). In one example, the memory 204 may also include (multiple) instructions 206.

[0062] The processor 202 and the memory 204 may be implemented in a manner similar to the processor 102 and the memory 104. For the sake of brevity, such explanation is not provided here. The memory 204 may be communicatively connected to the processor 202. Among other functions, the processor 202 may also retrieve and execute computer-readable instructions stored in the memory 204, including (multiple) instructions 206. (Multiple) instructions 206 may be executed to enable data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in the UE 200.

[0063] In operation, processor 202 may retrieve and execute instruction(s) 206 to perform one or more actions, as discussed further. For example, as a result of execution of instruction 208, UE 200 may send a data collection request. The data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. As previously described, the AI / ML model ( Figure 2 2) may be provided in UE 200. Thereafter, instruction 210 may be executed to receive a response message. The response message may include a measurement identifier of the measurement configuration corresponding to the identifier of the task.

[0064] Continuing, then, based on the measurement configuration, instructions 212 may be executed to collect data for tasks involving the AI / ML model.

[0065] The above functions performed as a result of the execution of instructions 106 and 206 may be performed by different programmable entities. Figure 3-Figure 13 The manner in which the network apparatus 100 and the user equipment 200 enable data collection for tasks involving AI / ML models present in the UE is explained in more detail.

[0066] Figure 3A network environment 300 for implementing data collection for tasks involving AI / ML models present in a UE according to an implementation of the present subject matter is illustrated. The network environment 300 (referred to as environment 300) can be part of any access network. In another example, the network environment 300 can also be part of any other communication network such as a fifth generation (5G) network. In this case, the entity implemented within the 5G network can operate based on the 5G protocol stack. It can be noted that the present method can be implemented to any access network or any communication network. All such examples are within the scope of the present subject matter.

[0067] Environment 300 may include network device 302. In one example, network device 302 may be similar to Figure 1 The network device 100 described above. In the context of the environment 300 implemented in a fifth generation (5G) network, the network device 302 may be implemented with an RRC (radio resource control) layer of a 5G protocol stack. In another example, the network device 302 may be one of a base station device and a gNodeB (gNB). The network device 302 may include a configuration engine 304. In one example, the configuration engine 304 may be implemented as a processing resource.

[0068] like Figure 3 As further depicted in FIG. 1 , the network environment 300 may include a plurality of user equipments (UEs) 306-1, 306-2, ..., 306-N (collectively referred to as UE 306 and individually as UE 306), which in turn may communicate with a network device 302. In one example, the UE 306 may be similar to a combination of Figure 2 UE 200 described in . Each of the UEs 306 may include an artificial intelligence / machine learning (AI / ML) model 308 and a collection engine 310. In one example, the collection engine 310 may be implemented as a processing resource. Although the AI / ML model 308 and the collection engine 310 are only depicted in UE 306-1, it may be noted that this is only for clarity. All other UEs may also include their corresponding AI / ML models and collection engines.

[0069] In one example, the network device 302 may be adapted to operate for 5G-NR (i.e., new radio), which in turn may interface with the core network via one or more backhaul links. The network device 302 may perform various functions, including, but not limited to, transmission of user data, wireless channel encryption and decryption, integrity protection, header compression, mobility control functions (e.g., handover, dual connectivity), paging, positioning, and delivery of warning messages, among other functions. The network device 302 may communicate directly or indirectly with each other similar base station device via a backhaul link, which in turn may be wired or wireless.

[0070] Returning to the present example, in operation, the network device 302 may reconfigure the radio communication link associated with the UE 306-1 to enable data collection for tasks involving the AI / ML model 308 present in the UE 306-1. Figure 4 306-1, the operations of the different elements within the environment 300 are also explained with respect to the logical components of the network device 302. It can also be noted that although the foregoing description has been explained with respect to the reconfiguration of the radio communication link of the UE 306-1, this should not be interpreted as limiting the scope of the present subject matter in any way. The proposed method can be implemented to all other UEs in the environment 300, and all such examples are within the scope of the present subject matter.

[0071] Figure 4 1 is a block diagram of an example network device 302 to be implemented in a network environment 300 according to an example of the present subject matter. The network device 302 may be similar to Figure 1 The network device 100 described in .

[0072] In this example, the network device 302 includes (multiple) processors 402, memory 404, (multiple) interfaces 406, and transceiver 408. (Multiple) processors 402 can also be implemented as (multiple) signal processors, (multiple) state machines, and / or any other device or component that manipulates signals based on operating instructions. The memory 404 can store one or more executable instructions that can be retrieved and executed to perform one or more operations to enable the UE to collect data for a task. The memory 404 can also be used to store data generated or used during the operation of the network device 302. The memory 404 can be a non-transitory computer-readable medium, including, for example, a volatile memory (such as RAM) or a non-volatile memory (such as EPROM, flash memory, etc.).

[0073] The interface(s) 406 may include various interfaces, for example, interfaces for data input and output, and interfaces for exchanging various operation instructions between other devices (such as devices within the environment 300). Figure 3 , (multiple) interfaces 406 may also be used to enable communication between network device 302 and user equipment (UE) 306. (Multiple) interfaces 406 may be implemented as hardware or software. Transceiver 408 may be used by network device 302 to send or receive data or signals while communicating with other components in environment 300.

[0074] The network device 302 may also include (multiple) engines 410 and data 412. (Multiple) engines 410 may be implemented as a combination of hardware and programs (e.g., program instructions) to implement one or more functions of the network device 302. In the examples described herein, such a combination of hardware and programs may be implemented in several different ways. For example, the program of (multiple) engines 410 may be executable instructions. Such instructions may be stored on a non-transitory machine-readable storage medium, which may be directly or indirectly (e.g., via a networking mode) coupled to the network device 302. In one example, (multiple) engines 410 may include processing resources, such as a single processor, or a combination of multiple processors, to execute such instructions. In this example, the non-transitory machine-readable storage medium may store instructions that implement (multiple) engines 410 when executed by the processing resources. In other examples, (multiple) engines 410 may be implemented as electronic circuit systems.

[0075] In one example, the engine(s) 410 may include the configuration engine 304 and the other engine(s) 414. The other engine(s) 414 may implement one or more additional functions of the network device 302. The data 412 includes information that may be utilized or generated by the engine(s) 410 during operation of the network device 302. In one example, the data 412 includes a data collection request 416, measurement configuration(s) 418, a response message 420, and other data 422. In addition, the data collection request may include task(s) identifier 424, first attribute(s) 426, and a request type 428. The response message 420 may include measurement configuration(s) identifier 430 and second attribute(s) 432.

[0076] Please note that combined Figure 3-Figure 4 The above description, combined with Figure 5 The call flow diagram shown in further explains the manner in which data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in UE 306-1 is implemented. Figure 5 An example call flow diagram 500 is illustrated that depicts various functions between a UE 306-1 and a network device 302 for implementing data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in the UE 306-1 according to an implementation of the present subject matter. As depicted, according to one example, the call flow diagram 500 provides a series of exchanges between the UE 306-1 and the network device 302. In one example, the messaging between the UE 306-1 and the network device 302 can be based on signaling based on the RRC protocol.

[0077] In operation, a plurality of user equipments (UEs) 306 may communicate with a network device 302 in an access network. Each UE 306 may include an artificial intelligence / machine learning (AI / ML) model 308. The AI / ML model 308 may be used to improve the efficiency of the operation of the corresponding UE 306 in which the AI / ML model 308 may be provided. As previously described, some functions to be enabled by the AI / ML model 308 may be specific to a given UE, such as UE 306-1, and the functions to be enabled by the AI / ML model 308 implemented in another UE (UE 306-2) may be different. As a result, the data to be collected by different UEs 306 may also be different.

[0078] The foregoing example is described from the perspective of UE 306-1. It may be noted that the method explained in conjunction with this example is also applicable to other UEs 306. Such examples also belong to the scope of the present subject matter.

[0079] Utilizing the AI / ML model 308 to implement various UE or network-related functions may require training of the AI / ML model 308. Once trained, the AI / ML model 308 can be implemented within the UE 306-1 to implement such functions. During its implementation, the performance of the AI / ML model 308 can be monitored, and it can be determined to perform additional training on the AI / ML model 308. To this end, the collection engine 310 of the UE 306-1 collects data that can be used for various tasks involving the AI / ML model 308. It can be noted that examples of such tasks involving the AI / ML model 308 can include, but are not limited to, training of the AI / ML model 308, retraining of the AI / ML model 308, monitoring of the AI / ML model 308, monitoring of functions performed by the UE 306-1, or reasoning using the AI / ML model 308. It can be noted that the following examples are merely indicative and not exhaustive. Other example tasks are also possible without departing from the scope of the present subject matter. It may also be noted that tasks involving the AI / ML model 308 may also include certain other functions, which may be performed directly or may be caused to be performed by the AI / ML model 308 .

[0080] Returning to this example, UE 306-1 may need to collect certain data for a task. Each task involving AI / ML model 308 may require different types of data that can be obtained from different kinds of measurements. It is understood that UE 306-1 can communicate with other network entities in the network using its corresponding radio communication link. Depending on the type of data to be collected, the radio communication link between UE 306-1 and network device 302 can be configured.

[0081] It may be noted that the method of the present subject matter may be implemented to reconfigure any type of radio communication link associated with UE 306-1, wherein UE 306-1 may operate in a network in any manner. For example, the radio communication link of UE 306-1 may be configured so that UE 306-1 may perform radio resource management (RRM) L3 measurements. For another example, this may be done so that UE 306-1 may perform L1 measurements. All such and other examples are within the scope of the present subject matter.

[0082] It may also be noted that the methods of the present subject matter may be implemented and applied to both single-sided AI / ML models and double-sided AI / ML models. It may be appreciated that a single-sided AI / ML model may be trained at one of the UE or the network device. On the other hand, a double-sided AI / ML model may be trained at both the UE and the network device. Both models will be within the scope of the present subject matter.

[0083] In operation, according to an example of the present subject matter, the UE 306-1 may send a data collection request 416 to the network device 302 (represented by step 502 in the call flow diagram 500). The data collection request 416 may include an identifier 424 of a task for which data is to be collected (referred to as a task identifier 424), which task may involve the AI / ML model 308 and may be performed using it. It should be noted and understood that the AI / ML model 308 may be implemented within the UE 306-1 itself. For example, the task identifier 424 may uniquely identify any one of the tasks of training the AI / ML model 308, retraining the AI / ML model 308, monitoring the AI / ML model 308, monitoring of a function performed by the UE 306-1, or reasoning performed using the AI / ML model 308. Although explained by the task identifier 424, a similar identifier may be indicated in the data collection request 416, thereby indicating that data collection is sought for more than one task.

[0084] In addition to the identifier 424, the data collection request 416 may also include additional information. For example, the data collection request may include a first set of attributes (referred to as first attributes 426), a type of request (referred to as request type 428), or other information. The first attributes 426 may define the conditions based on which the data collection corresponding to the data collection request 416 is to be performed. Examples of such first attributes 426 may include, but are not limited to: a time instance for initiating data collection, a time instance for terminating data collection, a periodicity for data collection, a sensitivity of data requirements, or characteristics of a geographic area for performing data collection. It may be noted that such first attributes 426 are merely exemplary and should not be interpreted as limiting the scope of the present subject matter in any way. Any other type of attribute for defining the conditions based on which data collection is to be performed may also be used and are within the scope of the present subject matter.

[0085] In addition, the request type 428 may indicate whether the request is for initiation or termination of data collection. For example, the foregoing description has been explained for the data collection request 416 as a data collection initiation request. As will be explained in the foregoing description, after receiving the data collection initiation request, the network device 302 may process the request and may reconfigure the radio communication link associated with the UE 306-1 for data collection.

[0086] Upon receiving the data collection request 416, the configuration engine 304 of the network device 302 may process it and accordingly determine a measurement configuration 418 corresponding to the task identifier 424 (represented by step 504 in the call flow diagram 500). In one example, the configuration engine 304 may determine the measurement configuration 418 based on a predefined mapping available to the network device 302. To this end, the configuration engine 304 may retrieve a mapping between a plurality of tasks and a plurality of measurement configurations. In one example, the predefined mapping may be stored in other data 422. Based on the mapping, and the task identifier 424 obtained from the data collection request 416, the configuration engine 304 may determine the measurement configuration 418. It may be noted that the measurement configuration 418 may be regarded as defining the manner in which the UE may perform measurements, any communication or data exchange, based on which the UE 306-1 will perform the data collection process. The measurement configuration 418 may indicate characteristics of a communication link associated with the UE 306-1. The measurement configuration 418 may indicate a single characteristic or a plurality of characteristics based on which the radio communication link of the UE may be configured.

[0087] An exemplary table is provided below that illustrates various measurement configurations 418 and corresponding characteristics based on which a communication link associated with UE 306-1 may be configured. It may be noted that the characteristics provided in the following table are exemplary only and other examples are within the scope of the present subject matter.

[0088]

[0089] In another example, the configuration engine 304 can use various different configuration settings to configure the radio communication link associated with the UE 306-1 for data collection. In this case, a specific configuration setting can be used as a measurement configuration 418 for reconfiguring the radio communication link associated with the UE 306-1. For example, the configuration engine 304 can determine a first configuration, based on which the configuration of the radio communication link can be performed. The first configuration can correspond to an earlier or currently configured radio communication link associated with the UE 306-1. The configuration engine 304 can evaluate the first configuration based on the determined measurement configuration 418 to determine whether the first configuration can be used to configure the radio communication link associated with the UE 306-1. In one example, after confirming that the first configuration can be used, the configuration engine 304 can use the first configuration as (multiple) measurement configurations 418 for configuring the radio communication link associated with the UE 306-1 for data collection. It can be understood that reusing the configuration corresponding to the earlier or currently configured radio communication link can reduce signal overhead.

[0090] However, after evaluating the first configuration, if the configuration engine 304 determines that the first configuration cannot be used to configure the radio communication link associated with the UE 306-1 for data collection, the configuration engine 304 may determine a second configuration different from the first configuration. Thereafter, the configuration engine 304 may use the second configuration as (multiple) measurement configurations 418 for configuring the radio communication link associated with the UE 306-1 for data collection. Once the measurement configuration 418 is determined, the configuration engine 304 may generate a response message 420 based on the identified measurement configuration 418. In one example, the response message 420 may include an identifier of the identified measurement configuration 418 (referred to as a measurement configuration identifier 430 or identifier 430). In another example, the response message 420 may also include a second configuration, and a corresponding identifier 430. When the UE 306-1 may not know the configuration that may need to be performed based on the measurement configuration identifier 430, it may be useful to include the measurement configuration 418 in the response message 420.

[0091] In another example, the response message 420 may also include a list of a second set of attributes (referred to as second attributes 432). The second attributes 432 may correspond to a set of attributes determined using the first attributes 426 or other criteria configured in the radio access network (RAN), based on which data collection may be allowed and data collection may be performed once the radio communication link associated with the UE 306-1 is configured. In one example, the second attribute 432 may be determined using the first attribute 426. For example, the first attribute 426 may require or suggest several time windows for the UE 306-1 to perform data collection. The (multiple) second attributes 432 may include one of the time windows, so that the UE 306-1 may perform data collection during the time window of the (multiple) second attributes 432.

[0092] In another example, the value provided in the first attribute 426 may not be suitable for data collection. In this case, the second set of attributes 432 can be determined using other criteria configured in the radio access network (RAN). The configuration engine 304 can provide additional attributes as the second set of attributes 432, and once the radio communication link associated with the UE 306-1 is configured, data collection can be performed based on these attributes. For example, the first attribute 426 can provide the characteristics of the geographical area in which the UE 306-1 can be operating and data collection is to be performed. If the geographical area provided in the first attribute 426 is not suitable for data collection, the configuration engine 304 can suggest a new area in which the UE 306-1 is to be enabled for data collection in the second set of attributes 432. In another example, the second attribute 432 can also include an indication that data collection cannot be implemented using the first set of attributes 426 and that the additional attributes have been provided as the second set of attributes 432.

[0093] In another example, the response message 420 may also include a report notification, which, when received by the UE, may notify the UE 306-1 to report additional information to the network device 302. Examples of such additional information may include, but are not limited to, the number of data collection sessions, the corresponding data collection timestamp, and the number of collected data samples. It may be noted that such a list of examples is not exhaustive, and any other examples will also be within the scope of the present subject matter.

[0094] Thereafter, a response message 420 may be sent to the UE 306-1 within the access network (represented by step 506 in the call flow diagram 500). The UE 306-1 may receive the response message 420. After receiving the response message 420, a radio communication link associated with the UE 306-1 may be configured (represented by step 508 in the call flow diagram 500). In one example, the configuration of the radio communication link may be performed through RRC reconfiguration.

[0095] In another example, after receiving the response message 420, the UE 306-1 may send a confirmation message (represented by step 510 in the call flow diagram 500) to the network device 302. Such a confirmation message may indicate the successful or unsuccessful receipt of the measurement configuration 418 in the received response message 420.

[0096] It should be noted that the radio communication link associated with UE 306 - 1 may be configured based on the measurement configuration corresponding to the measurement configuration identifier 430 received in the response message 420 .

[0097] Once the radio communication link is configured, the collection engine 310 may collect data based on the configuration performed by the network device 302 (represented by step 512 in the call flow diagram 500). As previously described, data is to be collected for tasks involving the AI / ML model 308. Examples of such tasks include, but are not limited to, training of the AI / ML model 308, retraining of the AI / ML model 308, monitoring of the AI / ML model 308, monitoring of functions performed by the UE 306-1, reasoning performed using the AI / ML model 308, or a combination thereof. Such tasks involving the AI / ML model 308 will be well understood by those skilled in the art, and for the sake of brevity, an explanation thereof is not provided herein. It may be noted again that data may also be collected for any other tasks. All such additional examples are within the scope of the present subject matter.

[0098] It may be noted that, as previously described, different types of configuration settings may be used for configuration of a radio communication link associated with UE 306-1.Some exemplary use cases are described in detail herein.

[0099] In a first example, a radio communication link associated with UE 306-1 may be configured based on configuration settings that are similar to characteristics of an earlier configured radio communication link that may have been previously established by UE 306-1.Such an example has been described in conjunction with call flow diagram 600. Figure 6An example call flow diagram 600 is illustrated depicting an example where a radio communication link associated with UE 306-1 may be configured based on similar configuration settings of an earlier configured radio communication link associated with UE 306-1. As previously described, such configuration settings are referred to as a first configuration.

[0100] In this case, UE 306-1 may send a data collection request 416 to network device 302 (represented by step 602 in call flow diagram 600). Data collection request 416 may include an identifier 424 of a task for which data may be collected, which may involve AI / ML model 308. An explanation has been provided previously and will not be repeated here for the sake of brevity.

[0101] Upon receiving the data collection request 416, the configuration engine 304 of the network device 302 may process it. The configuration engine 304 may determine a first configuration, based on which the configuration of the radio communication link may be performed. The first configuration may correspond to the characteristics of the earlier configured radio communication link associated with the UE 306-1. The configuration engine 304 may evaluate the first configuration based on the determined measurement configuration 418 to determine whether the first configuration can be used to configure the radio communication link associated with the UE 306-1 for data collection. After confirming that the first configuration can be used, the configuration engine 304 may use the first configuration as a measurement configuration 418 for configuring the radio communication link associated with the UE 306-1 for data collection (represented by step 604 in the call flow diagram 600). It will be appreciated that reusing the configuration corresponding to the earlier configured radio communication link may reduce signal overhead.

[0102] Once the measurement configuration 418 is determined, the configuration engine 304 can generate a response message 420 based on the identified measurement configuration 418 and send it to the UE 306-1 (represented by step 606 in the call flow diagram 600). In one example, the response message 420 can include an identifier of the identified measurement configuration 418 (referred to as measurement configuration identifier 430 or identifier 430).

[0103] UE 306-1 may receive response message 420. After receiving response message 420, a radio communication link associated with UE 306-1 may be configured (represented by step 608 in call flow diagram 600). It should be noted that the radio communication link associated with UE 306-1 may be configured based on the first configuration.

[0104] In another example, the radio communication link may be previously configured. In this case, the radio communication link associated with UE 306-1 may need to be configured over a previous configuration associated with UE 306-1. Figure 7 An example call flow diagram 700 is illustrated, depicting an example in which a radio communication link associated with UE 306-1 may have been previously configured, and additional configuration may be sought.

[0105] about Figure 7 , UE 306-1 may send a data collection request 702 to network device 302 (represented by step 704 in call flow diagram 700). In one example, data collection request 702 may include an identifier 424 of a task for which data is to be collected.

[0106] Upon receiving the data collection request 702, the configuration engine 304 of the network device 302 may process it. Based on the received data collection request 702 and the task identifier 424, the configuration engine 304 may determine a measurement configuration 418 (represented by step 706 in the call flow diagram 700). It should be understood that such a measurement configuration 418 may be determined based on additional measurements that the UE 306-1 may require on top of previously configured measurements. In one example, the measurement configuration 418 may include configurations that may be completely different to affect data collection by the UE 306-1.

[0107] Once the measurement configuration 418 is determined, the configuration engine 304 may generate a response message 420 based on the identified measurement configuration 418 and send it to the UE 306-1 (represented by step 708 in the call flow diagram 700). In one example, the response message 420 may include an identifier (referred to as a measurement configuration identifier 430 or identifier 430) of the identified measurement configuration 418. In another example, the response message 420 may also include: the measurement configuration 418, and the corresponding identifier 430.

[0108] UE 306-1 may receive response message 420. After receiving response message 420, a radio communication link associated with UE 306-1 may be configured (represented by step 710 in call flow diagram 700). It should be noted that the radio communication link associated with UE 306-1 may thus be configured in addition to a previously configured radio communication link.

[0109] It may be noted that the above use cases and corresponding call flow diagrams are merely exemplary and should not be construed as limiting the scope of the present subject matter in any way. The radio communication link associated with UE 306-1 may be configured based on any combination of the above, or based on a measurement configuration corresponding to the measurement configuration identifier 430 received in the response message 420. For example, the radio communication link associated with UE 306-1 may be configured in part based on settings similar to the earlier configured radio communication link, and in part based on additional measurements provided by the network device 302. All such examples are within the scope of the present subject matter.

[0110] Once the radio communication link is configured for data collection, UE 306-1 may terminate data collection. Figure 8 An example call flow diagram 800 is illustrated, depicting an example in which UE 306-1 may terminate data collection.

[0111] To terminate the data collection process, UE 306-1 may send a termination request 802 to network device 302 (represented by step 804 in call flow diagram 800), wherein UE 306-1 may indicate that data collection is to be terminated. In one example, termination request 802 may include logs related to data collection performed according to the data collection request (as described above).

[0112] The network device 302 may receive the termination request 802. Based on the received termination request 802, the configuration engine 304 may then generate a control instruction (represented by step 806 in the call flow diagram 800). In one example, the configuration engine 304 may generate the control instruction based on a determination that the termination request 802 meets the predefined condition. The control instruction may then be sent to the UE 306-1 in the form of a response message 420 (represented by step 808 in the call flow diagram 800). After receiving the response message 420, the UE 306-1 may execute the control instruction. When executed, the control instruction may also reconfigure a radio communication link associated with the UE 306-1 to terminate data collection performed by the UE 306-1 (represented by step 810 in the call flow diagram 800). In another example, after reconfiguring the radio communication link to terminate data collection, the UE 306-1 may send a confirmation message to the network device 302 (represented by step 812 in the call flow diagram 800). Such a confirmation message may inform network device 302 that the data collection termination process has been completed.

[0113] Although not depicted, termination of data collection may also be initiated by network device 302 without receiving any such termination request 802 from UE 306-1. For example, based on one or more predefined conditions, network device 302 may generate a control instruction for reconfiguring the radio communication link to terminate data collection. Such examples are also within the scope of the claimed subject matter.

[0114] Fig. 9 An example method 900 to be implemented in a network device according to an implementation of the present subject matter is illustrated, the method being used to implement data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a UE. The order in which the method is described should not be construed as limiting, and any number of the described method blocks may be combined in any order to implement the above method or an alternative method. In addition, the method 900 may be implemented by any suitable hardware, non-transitory machine-readable program instructions, or a combination thereof, or by a logic circuit system through a processing resource or (multiple) computing devices.

[0115] It is also understood that method 900 may be performed by programmed and / or configured network devices present within the network, wherein such devices include, for example, Figure 3-Figure 8 900 . In addition, in some cases, program instructions stored in a non-transitory computer-readable medium can implement method 900 when executed, which will be readily appreciated. For example, a non-transitory computer-readable medium can include a digital memory, a magnetic storage medium (such as one or more disks and tapes), a hard drive, or an optically readable digital data storage medium. Although method 900 is described below with reference to network device 302, as described above, other suitable systems for performing the method can be utilized. Furthermore, implementation of the method is not limited to such examples.

[0116] At block 902, a network device may receive a data collection request from a user equipment (UE). The data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. The AI / ML model may be provided in the UE. For example, a plurality of user equipments (UEs) 306 may communicate with a network device 302 in an access network. Each UE 306 may include an artificial intelligence / machine learning (AI / ML) model 308. In order to utilize the AI / ML model 308 to implement various UE-related functions performed by a UE (e.g., UE 306-1), various tasks may need to be performed, such as training of the AI / ML model 308, retraining of the AI / ML model 308, monitoring of the AI / ML model 308, monitoring of functions performed by the UE 306-1, reasoning performed using the AI / ML model 308, and the like. Each task involving the AI / ML model 308 may require different types of data. Therefore, the UE 306-1 may need to collect certain data. In this regard, in operation, the UE 306-1 may send a data collection request 416, which may be received by the network device 302. The data collection request 416 may include an identifier 424 of a task for which data is to be collected (referred to as a task identifier 424), which may involve the AI / ML model 308 and may be performed using the same.

[0117] At block 904, the network device may determine a measurement configuration corresponding to the identifier of the task. For example, upon receiving the data collection request 416, the configuration engine 304 of the network device 302 may process it and accordingly determine a measurement configuration 418 corresponding to the task identifier 424. It may be noted that the measurement configuration 418 may be viewed as defining the manner in which the UE may perform any communication or data exchange, based on which the UE 306-1 will perform the data collection process.

[0118] At block 906, the network device may send a response message to the UE. The response message may include a measurement identifier corresponding to the measurement configuration. For example, once the measurement configuration 418 is determined, the configuration engine 304 may generate a response message 420 based on the identified measurement configuration 418. In one example, the response message 420 may include an identifier (referred to as a measurement configuration identifier 430 or identifier 430) of the identified measurement configuration 418. The response message 420 may be sent to the UE 306-1 within the access network.

[0119] Fig.10An example method 1000 to be implemented in a user equipment (UE) according to an implementation of the present subject matter is illustrated for implementing data collection for tasks involving an artificial intelligence / machine learning (AI / ML) model present in the UE. The method 1000 may be performed by a programmed and / or configured network device present within a network, wherein such a device includes, for example, Figure 3 and Figure 5-Figure 8 User device 306-1 is depicted in FIG.

[0120] At block 1002, a user equipment (UE) may send a data collection request. The data collection request may include an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected. The AI / ML model may be provided in the UE. For example, a plurality of user equipments (UEs) 306 may communicate with a network device 302 in an access network. Each UE 306 may include an artificial intelligence / machine learning (AI / ML) model 308. In order to utilize the AI / ML model 308 to implement various UE-related functions performed by a UE (e.g., UE 306-1), various tasks may need to be performed, such as training of the AI / ML model 308, retraining of the AI / ML model 308, monitoring of the AI / ML model 308, monitoring of functions performed by the UE 306-1, reasoning performed using the AI / ML model 308, and the like. Each task involving the AI / ML model 308 may require different types of data. Therefore, the UE 306-1 may need to collect certain data. In this regard, in operation, UE 306-1 may send a data collection request 416 to network device 302. Data collection request 416 may include an identifier 424 of a task for which data is to be collected (referred to as task identifier 424), which task may involve AI / ML model 308 and may be performed using the same.

[0121] At block 1004, the UE may receive a response message. The response message may include an identifier of a measurement configuration corresponding to the identifier of the task. For example, upon receiving the data collection request 416, the configuration engine 304 of the network device 302 may process it and accordingly determine a measurement configuration 418 corresponding to the task identifier 424. It may be noted that the measurement configuration 418 may be viewed as defining the manner in which the UE may perform any communication or data exchange, based on which the UE 306-1 will perform the data collection process. A response message 420 may be sent to the UE 306-1 within the access network.

[0122] At block 1006, the UE may collect data based on the measurement configuration of the task involving the AI / ML model. For example, the UE 306-1 may receive the response message 420. After receiving the response message 420, a radio communication link associated with the UE 306-1 may be configured. The radio communication link associated with the UE 306-1 may be configured based on the measurement configuration corresponding to the measurement configuration identifier 430 received in the response message 420. Once the radio communication link is configured, the collection engine 310 may collect data based on the configuration of the network device 302.

[0123] Fig.11 Depicted is an example method 1100 for implementing data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE) in accordance with implementations of the present subject matter.

[0124] At block 1102, a user equipment (UE) may send a data collection request with a task identifier to a network device. For example, a plurality of user equipments (UEs) 306 may communicate with a network device 302 in an access network. Each UE 306 may include an artificial intelligence / machine learning (AI / ML) model 308. In order to utilize the AI / ML model 308 to implement various UE-related functions performed by a UE 306 (e.g., UE 306-1), various tasks may need to be performed, such as training of the AI / ML model 308, retraining of the AI / ML model 308, monitoring of the AI / ML model 308, monitoring of functions performed by the UE 306-1, reasoning performed using the AI / ML model 308, and the like. Each task involving the AI / ML model 308 may require different types of data. Therefore, the UE 306-1 may need to collect certain data. In this regard, in operation, the UE 306-1 may send a data collection request 416 to the network device 302. The data collection request 416 may include an identifier 424 of a task for which data is to be collected (referred to as task identifier 424), which task may involve and may be performed using the AI / ML model 308. The data collection request 416 may include additional information in addition to the identifier 424. For example, the data collection request may include a first set of attributes (referred to as first attributes 426), a type of request (referred to as request type 428), or other information.

[0125] At block 1104, based on the task identifier, the network device may determine a measurement configuration. For example, the network device 302 may receive a data collection request 416 from the UE 306-1. After receiving the data collection request 416, the configuration engine 304 of the network device 302 may process it and determine a measurement configuration 418 corresponding to the task identifier 424 accordingly.

[0126] At block 1106, a first configuration may be determined. For example, a radio communication link associated with UE 306-1 may be configured using different configuration settings. Configuration engine 304 may determine the first configuration. The first configuration may correspond to an earlier configured radio communication link associated with UE 306-1.

[0127] At block 1108, a determination may be made to confirm whether the first configuration can be used for configuration. For example, based on the determined measurement configuration 418, the configuration engine 304 may evaluate the first configuration to determine whether the first configuration can be used to configure the radio communication link associated with the UE 306-1 for data collection. After the evaluation, if the configuration engine 304 determines that the first configuration can be used to configure the radio communication link associated with the UE 306-1 for data collection ("yes" path from block 1108), the configuration engine 304 may use the first configuration as the measurement configuration 418 (block 1110).

[0128] However, after evaluating the first configuration, if the configuration engine 304 determines that the first configuration cannot be used to configure the radio communication link associated with the UE 306-1 for data collection ("No" path from block 1108), the configuration engine 304 may determine a second configuration (block 1112). Thereafter, the configuration engine 304 may use the second configuration as the measurement configuration 418 (block 1114).

[0129] In another example, the radio communication link associated with the UE 306-1 may be previously configured. In this case, when determining the measurement configuration 418, the configuration engine 304 may also consider measurements of the previously configured radio communication link associated with the UE 306-1.

[0130] At block 1116, based on the measurement configuration, a response message may be generated. For example, once the measurement configuration 418 is determined, the configuration engine 304 may generate a response message 420 based on the identified measurement configuration 418. In one example, the response message 420 may include an identifier of the identified measurement configuration 418 (referred to as a measurement configuration identifier 430 or identifier 430).

[0131] A response message may be sent to the user equipment (UE) at block 1118. For example, thereafter, a response message 420 may be sent to the UE 306-1 within the access network.

[0132] At block 1120, a radio communication link associated with the UE may be configured. For example, UE 306-1 may then receive response message 420. After receiving response message 420, a radio communication link associated with UE 306-1 may be configured. In one example, the configuration of the radio communication link may be performed through RRC reconfiguration.

[0133] At block 1122, the user equipment (UE) may send a confirmation message to the network device. For example, after receiving the response message 420, the UE 306-1 may send a confirmation message to the network device 302. Such a confirmation message may indicate the successful or unsuccessful receipt of the measurement configuration in the received response message 420.

[0134] At block 1124, a user equipment (UE) may collect data for a task involving an artificial intelligence / machine learning (AI / ML) model. For example, once a radio communication link is configured, the collection engine 310 may collect data based on the configuration performed by the network device 302. Data is to be collected for a task involving an AI / ML model 308. Examples of such tasks include, but are not limited to, training of an AI / ML model 308, retraining of an AI / ML model 308, monitoring of an AI / ML model 308, monitoring of a function performed by the UE 306-1, reasoning using an AI / ML model 308, or a combination thereof.

[0135] Fig.12 A computing environment 1200 is illustrated that implements a non-transitory computer-readable medium 1202 according to an implementation of the present subject matter, which is used to implement data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE). In one example, the computing environment 1200 may include the network device 302 described above. The computing environment 1200 includes a processing resource 1204 that is communicatively coupled to the non-transitory computer-readable medium 1202 via a communication link 1206. In one example, the processing resource 1204 may be a processor or controller within the network device 302 that retrieves and executes computer-readable instructions from the non-transitory computer-readable medium 1202.

[0136] The non-transitory computer-readable medium 1202 may be, for example, an internal memory device or an external memory device. In one example, the communication link 1206 may be a direct communication link, such as any memory read / write interface. In another example, the communication link 1206 may be an indirect communication link, such as a network interface. In this case, the processing resource 1204 may access the non-transitory computer-readable medium 1202 via a network 1208. The network 1208 may be a single network or a combination of multiple networks, and may use a variety of different communication protocols.

[0137] In one example, the non-transitory computer-readable medium 1202 includes executable instructions 1210 for serving the network device 302. For example, the non-transitory computer-readable medium 1202 may include executable instructions 1210 for implementing functions performed by the network device 302. The instructions 1210, when executed, may cause the network device 302 to receive a data collection request 416 from a user equipment (UE) 306-1. The received data collection request 416 may include an identifier of a task for which data is to be collected (referred to as a task identifier 424), the task involving an artificial intelligence / machine learning (AI / ML) model 308 provided in the UE 306-1. Examples of such tasks involving the AI / ML model 308 may include, but are not limited to, training of the AI / ML model 308, retraining of the AI / ML model 308, monitoring of the AI / ML model 308, monitoring of functions performed by the UE 306-1, or reasoning using the AI / ML model 308. It may be noted that the following examples are merely indicative and not exhaustive. Other example tasks are possible without departing from the scope of the present subject matter.

[0138] Thereafter, instructions 1210 may also be executed to determine a measurement configuration 418 corresponding to the task identifier 424. For example, the measurement configuration 418 indicates the manner in which the UE 306-1 may perform any communication or enable the exchange of data.

[0139] Thereafter, instructions 1210 may be executed to send a response message 420 to UE 306-1. Response message 420 may include an identifier of the measurement configuration, referred to as measurement configuration identifier 430. Once UE 306-1 receives response message 420, the response message may cause a radio communication link associated with UE 306-1 to be reconfigured.

[0140] Fig.13 A computing environment 1300 is illustrated that implements a non-transitory computer-readable medium 1302 according to another implementation of the present subject matter, which is used to implement data collection for tasks involving artificial intelligence / machine learning (AI / ML) models present in a user equipment (UE). In one example, the computing environment 1300 may include the user equipment 306-1 described above. The computing environment 1300 includes a processing resource 1304 that is communicatively coupled to the non-transitory computer-readable medium 1302 via a communication link 1306. In one example, the processing resource 1304 can be a processor or controller within the user equipment (UE) 306-1 that retrieves and executes computer-readable instructions from the non-transitory computer-readable medium 1302.

[0141] In another example, the non-transitory computer readable medium 1302 , the communication link 1306 , and the network 1308 can be implemented in a manner similar to the non-transitory computer readable medium 1202 , the communication link 1206 , and the network 1208 , respectively.

[0142] In one example, the non-transitory computer-readable medium 1302 includes executable instructions 1310 for serving a user equipment (UE) 306-1. For example, the non-transitory computer-readable medium 1302 may include executable instructions 1310 for implementing functions performed by the UE 306-1. The instructions 1310, when executed, may cause the UE 306-1 to send a data collection request 416. The data collection request 416 may include an identifier of a task for which data is to be collected (referred to as a task identifier 424), the task involving an artificial intelligence / machine learning (AI / ML) model 308 provided in the UE 306-1. Examples of such tasks involving the AI / ML model 308 may include, but are not limited to, training of the AI / ML model 308, retraining of the AI / ML model 308, monitoring of the AI / ML model 308, monitoring of functions performed by the UE 306-1, or reasoning using the AI / ML model 308. It may be noted that the following examples are merely indicative and not exhaustive. Other example tasks are also possible without departing from the scope of the present subject matter.

[0143] Thereafter, the instructions 1310 may also be executed to receive a response message 420. The response message 420 may include an identifier of the measurement configuration (referred to as a measurement configuration identifier 430).

[0144] Thereafter, instructions 1310 may be executed to collect data based on a measurement configuration of tasks involving the AI / ML model 308 .

[0145] Although examples of the present disclosure have been described in language specific to structural features and / or methods, it should be understood that the appended claims are not necessarily limited to the specific features or methods described. Rather, the specific features and methods are disclosed and interpreted as examples of the present disclosure.

Claims

1. A network device, comprising: at least one processor; as well as at least one memory storing instructions, which, when executed by the at least one processor, cause the network device to at least: Receiving a data collection request from a user equipment (UE), wherein the received data collection request includes: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected, wherein the AI / ML model is provided in the UE; determining a measurement configuration corresponding to the identifier of the task; and A response message is sent to the UE, wherein the response message includes a measurement identifier corresponding to the measurement configuration.

2. The network device of claim 1 , wherein the instructions, when executed by the at least one processor, further cause the network device to: A radio communication link associated with the UE is reconfigured based on the measurement configuration corresponding to the measurement identifier.

3. The network device according to any one of claims 1 to 2, wherein the data collection request further comprises at least one of the following: a first set of attributes for the data collection corresponding to the data collection request; or A type of request, wherein the type indicates one of initiation or termination of the data collection.

4. The network device of claim 3, wherein the first set of attributes comprises at least one of: a time instance at which data collection is initiated, a time instance at which data collection is terminated, a periodicity of data collection, a sensitivity of data requirements, or characteristics of a geographic area for performing the data collection. 5 . The network device according to claim 3 , wherein the response message further comprises a list of a second set of attributes, wherein the second set of attributes is determined using the first set of attributes.

6. The network device according to any one of claims 1 to 5, wherein the instructions, when executed by the at least one processor, further cause the network device to: retrieving a mapping between a plurality of tasks and a plurality of measurement configurations; and The measurement configuration is determined based on the mapping and the identifier of the task obtained from the data collection request.

7. The network device according to any one of claims 1 to 6, wherein the instructions, when executed by the at least one processor, further cause the network device to: determining a first configuration corresponding to an earlier configured radio communications link associated with the UE; evaluating the first configuration based on the determined measurement configuration to determine whether the first configuration can be used to configure the radio communication link associated with the UE for the data collection; as well as Upon confirming that the first configuration can be used to configure the radio communication link associated with the UE for the data collection, the first configuration is used as the measurement configuration.

8. The network device of claim 7, wherein the instructions, when executed by the at least one processor, further cause the network device to: based on the evaluation of the first configuration, confirming that the first configuration cannot be used to configure the radio communication link associated with the UE for the data collection; determining a second configuration; as well as The second configuration is used as the measurement configuration for configuring the radio communication link associated with the UE for the data collection.

9. The network device according to any one of claims 1 to 8, wherein the instructions, when executed by the at least one processor, further cause the network device to: processing the data collection request received from the UE, wherein the radio communication link associated with the UE is previously configured; determining the measurement configuration based on previously configured measurements of the UE; as well as The response message is sent to the UE, wherein the response message includes: the measurement identifier corresponding to the measurement configuration to configure the previously configured radio communication link associated with the UE.

10. The network device according to any one of claims 1 to 9, wherein the instructions, when executed by the at least one processor, further cause the network device to: processing a data collection termination request received from the UE, wherein the termination request comprises: a log related to the data collection performed in accordance with the data collection request; as well as Upon determining that the termination request meets a predefined condition, a control instruction is sent to the UE, wherein the control instruction is executable by the UE to reconfigure a radio communication link associated with the UE to terminate the data collection by the UE.

11. A user equipment (UE), comprising: at least one processor; as well as at least one memory storing instructions, which, when executed by the at least one processor, cause the user equipment (UE): sending a data collection request, wherein the data collection request includes: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected, and wherein the AI / ML model is provided in the UE; receiving, by the UE, a response message, wherein the response message includes: a measurement identifier of the measurement configuration corresponding to the identifier of the task; and Based on the measurement configuration, data is collected for the task involving the AI / ML model.

12. A user equipment (UE) according to claim 11, wherein the task comprises at least one of the following: training of the AI / ML model, retraining of the AI / ML model, monitoring of the AI / ML model, monitoring of functions performed by the UE, or reasoning using the AI / ML model.

13. The user equipment (UE) according to any one of claims 11 to 12, wherein the data collection request further comprises at least one of the following: a first set of attributes for the data collection corresponding to the data collection request; or A type of request, wherein the type indicates one of initiation or termination of the data collection.

14. A user equipment (UE) according to any one of claims 11 to 13, wherein the instructions, when executed by the at least one processor, further cause the user equipment to: Sending a termination request, wherein the termination request includes: a log related to said data collection performed in response to receiving said response message; and A control instruction received from a network device is executed in response to the termination request, wherein the control instruction, when executed, further reconfigures a radio communication link to terminate the data collection by the UE.

15. A method for communication, comprising: Receiving, by a network device, a data collection request from a user equipment (UE), wherein the received data collection request includes: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected, wherein the AI / ML model is provided in the UE; determining, by the network device, a measurement configuration corresponding to the identifier of the task; and The network device sends a response message to the UE, wherein the response message includes: a measurement identifier corresponding to the measurement configuration.

16. A method for communication, comprising: Sending, by a user equipment (UE), a data collection request, wherein the data collection request includes: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected, and wherein the AI / ML model is provided in the UE; receiving, by the UE, a response message, wherein the response message includes: an identifier of the measurement configuration corresponding to the identifier of the task; and The UE collects data for the task involving the AI / ML model based on the measurement configuration.

17. The method of claim 16, wherein the task comprises at least one of: training of the AI / ML model, retraining of the AI / ML model, monitoring of the AI / ML model, monitoring of functions performed by the UE, or reasoning using the AI / ML model.

18. The method according to any one of claims 16 to 17, wherein the data collection request comprises one of the following: a first set of attributes for the data collection corresponding to the data collection request; and A type of request, wherein the type indicates one of initiation or termination of the data collection.

19. The method of claim 18, wherein the first set of attributes comprises at least one of: a time instance at which data collection is initiated, a time instance at which data collection is terminated, a periodicity of data collection, a sensitivity of data requirements, or characteristics of a geographic area in which the data collection is performed.

20. A non-transitory computer readable medium comprising computer readable instructions which, when executed by a processor, cause a network device to: A data collection request is received from a user equipment (UE), wherein the received data collection request comprises: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected, wherein the AI / ML model is provided in the UE; determining a measurement configuration corresponding to the identifier of the task; as well as A response message is sent to the UE, wherein the response message includes: a measurement identifier corresponding to the measurement configuration.

21. The non-transitory computer-readable medium of claim 20, further comprising instructions that, when executed by the processor, cause a radio communication link associated with the UE to be reconfigured based on the measurement configuration corresponding to the measurement identifier.

22. A non-transitory computer-readable medium comprising computer-readable instructions which, when executed by a processor, cause a user equipment (UE): Sending a data collection request, wherein the data collection request includes: an identifier of a task involving an artificial intelligence / machine learning (AI / ML) model for which data is to be collected, and wherein the AI / ML model is provided in the UE; receiving a response message, wherein the response message comprises: a measurement identifier of a measurement configuration corresponding to the identifier of the task; as well as Based on the measurement configuration, data is collected for the task involving the AI / ML model.

Images