Artificial intelligence radio function model management in communication network

Abstract

The wireless user equipment transmits learning model information corresponding to a learning model that facilitates the radio function to the network node. In response, the node sends a model management indication configuration corresponding to the learning model information to the user equipment. The user equipment monitors a learning model parameter metric indicative of learning model performance. The monitored metrics may be used to determine control actions or operations to perform based on analysis of the metrics relative to model performance metrics. A model management indication indicating the determined control action may be sent to the user device. Examples of control operations may include deactivating a learning model of the current operation or retraining the learning model.

Description

[0001] Related Applications

[0002] This application claims priority to U.S. Non - Provisional Patent Application No. 18 / 072,365, filed on November 30, 2022, and titled "Artificial intelligence radio function model management in a communication network", the entire content of which is incorporated herein by reference. Background Art

[0003] The term "New Radio" (NR) associated with the fifth - generation mobile wireless communication system ("5G") refers to the technical aspects used in the radio access network ("RAN"), which includes several quality of service classes (QoS), including ultra - reliable and low - latency communication ("URLLC"), enhanced mobile broadband ("eMBB"), and massive machine - type communication ("mMTC"). The URLLC QoS class is associated with strict latency requirements (e.g., low latency or low signal / message delay) and high reliability of radio performance, while conventional eMBB use cases may be associated with high - capacity wireless communication, which may allow less strict latency requirements (e.g., higher latency than URLLC) and less reliable radio performance compared to URLLC. The performance requirements of mMTC may be lower than those of eMBB use cases. Some use - case applications involving mobile devices or mobile user equipment (such as smart phones, wireless tablets, smart watches, etc.) may impose on varying given RAN resource loads or demands. Summary of the Invention

[0004] A brief summary of the disclosed subject matter is presented below in order to provide a basic understanding of some embodiments. This summary is not an exhaustive overview of all embodiments. Its purpose is neither to identify key or critical elements of all embodiments nor to delineate the scope of any or all embodiments. Its sole purpose is to present some concepts of the present disclosure in a concise form as a prelude to the more detailed description that follows.

[0005] In an example embodiment, a method may include: receiving, by a user equipment including a processor, a model management indication configuration corresponding to a first radio function of the user equipment from a radio access network node. The user equipment may monitor model performance parameters to obtain monitored model performance metrics, where the model performance parameters may be indicated for monitoring in a first radio function model configuration corresponding to a first radio function learning model for implementing a radio function on the user equipment. The monitored model performance metrics may be analyzed relative to a model performance metric standard to obtain an analyzed monitored performance metric. The user equipment may transmit a model management indication to the radio access network node based on the analyzed monitored performance metric being determined to not meet the model performance metric standard, e.g., a threshold corresponding to learning model parameters of the monitored metric. The example method may further include receiving, by the user equipment, a control instruction corresponding to an operation of the radio function learning model based on the model management indication, and implementing at least one control operation according to the control instruction.

[0006] The at least one control operation may include deactivating the first radio function learning model and activating a configured default radio function to perform the radio function, where the at least one control operation is determined by the radio access network node.

[0007] The first radio function model configuration may include at least one configured radio function model parameter value, and where the at least one control operation includes retraining the first radio function learning model using the at least one configured radio function model parameter value to obtain a retrained first radio function learning model. The at least one configured radio function model parameter value may be part of a retraining data set.

[0008] The first radio function model configuration may include at least one configured radio function model parameter value, and the method may further include: receiving an updated first radio function model configuration that includes at least one configured updated radio function model parameter value; replacing the at least one configured radio function model parameter value with the at least one configured updated radio function model parameter value; and implementing the radio function using the at least one configured updated radio function model parameter value (which may be part of a hold-out data set) with the first radio function learning model.

[0009] The model management indication may include a control operation request for requesting an updated first radio function model configuration, and the method may further include: in response to the model management indication, receiving the updated first radio function model configuration from a radio access network node; in response to receiving the updated first radio function model configuration, deactivating the use of the first radio function learning model according to the first radio function model configuration; and activating the use of the first radio function learning model according to the updated first radio function model configuration. In another embodiment, the control operation may include instructions for deactivating the use of the first radio function learning model according to the first radio function model configuration and activating the implementation of the radio function according to a previously configured default model, where the previously configured default model may be a configured probability learning model or it may be a configured deterministic model or function.

[0010] At least one control operation may include: retraining the first radio function learning model using at least one configured radio function model parameter value to obtain a retrained first radio function learning model. At least one configured radio function model parameter value may correspond to a data set corresponding to an index. The index may be transmitted in the model management indication.

[0011] The model performance parameters include at least one of the following: mean squared error value, root mean squared error value, normalized mean squared error value, mean absolute error value, R-squared value, generalized cosine similarity value, squared generalized cosine similarity value, accuracy value, number of true negative values, number of true positive values, number of false negative values, number of false positive values, precision value, recall value, or F1 score value.

[0012] In an embodiment, the example method may further include: transmitting, by a user equipment, an indication of radio function learning model information including learning model information corresponding to a first radio function to a radio access network node, where the indication of radio function learning model information may include a first indication, and where the learning model information includes at least one of the following: at least one learning model type indication, at least one radio function indication indicating at least one corresponding radio function; a list of at least one metric parameter to be estimated or reported corresponding to at least one learning model corresponding to at least one learning model type indication, a value representing the number of learning models stored by the user equipment for implementing at least one corresponding radio function, or a second indication of at least one learning model category corresponding to at least one data set to be used by the radio access network node to determine an analyzed monitoring performance metric.

[0013] The first radio function learning model may be one of a plurality of different radio function learning models capable of implementing a radio function. The model performance parameters include mathematical functions, such as the examples shown in Table 1.

[0014] The control instruction may include an instruction for performing an operation to deactivate the use of the first radio function learning model, wherein the control instruction includes an instruction for activating a second radio function learning model in a different radio function learning model to implement the radio function, wherein the first radio function model configuration defines the first radio function learning model and the second radio function model configuration defines the second radio function learning model, and wherein the first radio function model configuration is different from the second radio function model configuration.

[0015] In another exemplary embodiment, a method may include a radio access network node including a processor receiving learning model information corresponding to a current radio function learning model of a user equipment. In this embodiment, the method may further include: the radio access network node generating a model management indication configuration based on the learning model information; transmitting the model management indication configuration to the user equipment; receiving a performance parameter metric corresponding to a learning model performance metric parameter associated with the performance of the current radio function learning model; and transmitting a control instruction corresponding to the operation of the current radio function learning model by the user equipment based on the performance parameter metric.

[0016] The performance parameter metric corresponds to a learning model performance metric parameter. The learning model performance metric parameter may include at least one of the following: mean squared error value, root mean squared error value, normalized mean squared error value, mean absolute error value, R-squared value, generalized cosine similarity value, squared generalized cosine similarity value, accuracy value, number of true negative values, number of true positive values, number of false negative values, number of false positive values, precision value, recall value, or F1 score value.

[0017] The learning model information may include at least one of the following: at least one learning model type indication, at least one radio function indication indicating a corresponding at least one radio function, at least one metric to be estimated or reported corresponding to at least one learning model corresponding to at least one learning model type indication, the number of learning models that the user equipment can store for at least one radio function, or an indication of at least one learning model category corresponding to at least one data set to be used by the radio access network node to determine the performance parameter metric.

[0018] The model management indication may include a data size based at least on a performance parameter metric. Thus, depending on the metric being monitored and reported, the amount of data (e.g., number of bytes) used to transmit the metric may vary. For example, a quantized indication using the monitored metric may result in a smaller number of bytes or a configured uniform number of bytes being used for the transmitted data size of the indicated metric. Thus, the performance parameter metric may be indicated by a configured quantized performance parameter metric value or a quantization index. The quantization value or index may correspond to the range into which the monitored metric value falls. In an embodiment, the model management indication configuration includes a quantized performance parameter metric value. In an embodiment, the performance parameter metric may be a preamble corresponding to a user equipment.

[0019] The control instruction includes an instruction to deactivate the current radio function learning model and activate the default radio function. The control instruction includes an instruction to train the current radio function learning model to obtain an updated radio function learning model. The control instruction may include a recommended configured training data set indication indicating a configured training data set to be used for training the current radio function learning model that may be configured at the user equipment.

[0020] Example method embodiments may further include: analyzing a performance parameter metric relative to a performance parameter metric criterion to obtain an analyzed performance parameter metric, determining that the analyzed performance parameter metric fails to meet the performance parameter metric criterion, and determining a transmission control instruction based on determining that the performance parameter metric does not meet the performance parameter metric criterion.

[0021] In an embodiment, the model management indication includes a request from the user equipment for a transmission control instruction.

[0022] In another embodiment, a radio access network node operatively communicating with a plurality of user equipment may be configured to receive a second radio function learning model information indication from a second user equipment, the second radio function learning model information indication including learning model information corresponding to a second radio function learning model of the second user equipment. The radio access network node may generate a second model management indication configuration corresponding to the second radio function learning model, transmit the second model management indication configuration to the second user equipment, and receive a second performance parameter metric associated with the performance of the second radio function learning model from the second user equipment. The second performance parameter metric may correspond to a second learning model performance metric parameter; and the radio access network node may transmit a second control instruction corresponding to the operation of the second radio function learning model by the second user equipment based on the second performance parameter metric, wherein the first radio function learning model corresponds to a first learning model type, wherein the second radio function learning model corresponds to a second learning model type, wherein the first learning model type and the second learning model type are of the same type, wherein the first learning model type operates according to a first learning model configuration and the second learning model type operates according to a second learning model configuration, and wherein the first learning model configuration is different from the second learning model configuration. The second control instruction corresponding to the operation of the second radio function learning model by the second user equipment may be based on an analysis of the second performance parameter metric by the user equipment relative to a standard.

[0023] In an embodiment, the analysis may be performed by the radio access network node. In an embodiment, the analysis may be based on current or historical performance data of the learning model at the user equipment, which may include a number of data metrics. The learning model performance data may be combined with network operating conditions, context information (such as environment or weather-related information). The data for performance analysis may be directly generated by the network device or may be provided by a third-party application. Thus, based on the fact that the radio access network node has a greater perspective on the network aspect than the user equipment, the data for analysis, such as determining the standard for comparison with the learning model parameter metric, may facilitate the user equipment in determining its ability to promote the learning model performance for radio functions. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A wireless communication system environment is shown.

[0025] Figure 2 An example environment with radio functions implemented in conjunction with corresponding learning models is shown.

[0026] Figure 3A 、 Figure 3B 、 Figure 3C and Figure 3DShows a learning model metric report embodiment.

[0027] Figure 4 Shows a timing diagram of an example embodiment method for configuring a user equipment to manage a radio function learning model.

[0028] Figure 5 Shows an example embodiment of a radio function learning model retraining configuration.

[0029] Figure 6 Shows a timing diagram of an example embodiment method for configuring a user equipment to retrain a radio function learning model.

[0030] Figure 7 Shows a flowchart of an example method for managing a learning model that facilitates radio functions at a user equipment.

[0031] Figure 8 Shows a block diagram of an example method.

[0032] Figure 9 Shows a block diagram of an example user equipment.

[0033] Figure 10 Shows a block diagram of an example non-transitory machine-readable medium.

[0034] Figure 11 Shows a block diagram of an example method.

[0035] Figure 12 Shows a block diagram of an example user equipment.

[0036] Figure 13 Shows a block diagram of an example non-transitory machine-readable medium.

[0037] Figure 14 Shows an example computer environment.

[0038] Figure 15 Shows a block diagram of an example wireless UE. Detailed Description

[0039] First, those skilled in the art will readily understand that the embodiments of the present application have broad utility and applications. Many methods, embodiments, and adaptations of the present application, as well as many variations, modifications, and equivalent arrangements, will be apparent or reasonably suggested from the substance or scope of the various embodiments of the present application, in addition to those described herein.

[0040] Accordingly, although the present application has been described in detail with respect to various embodiments, it should be understood that the present disclosure merely illustrates one or more concepts expressed by the respective example embodiments and is made only for the purpose of providing a complete and implementable disclosure. The following disclosure is not intended to or should be construed as limiting the present application or otherwise excluding any such other embodiments, adaptations, variations, modifications, and equivalent arrangements, and the presently described embodiments are limited only by the appended claims and their equivalents.

[0041] As used in the present disclosure, in some embodiments, the terms "component", "system", etc. are intended to refer to or include a computer-related entity or an entity related to an operating device having one or more specific functions, where the entity can be hardware, a combination of hardware and software, software, or software in execution. By way of example, a component can be, but is not limited to, a process operating on a processor, a processor, an object, an executable, an execution thread, computer-executable instructions, a program, and / or a computer. By way of illustration and not limitation, both an application operating on a server and the server can be components.

[0042] One or more components can reside in a process and / or an execution thread, and a component can be located on one computer and / or distributed between two or more computers. Further, these components can execute from various computer-readable media on which various data structures are stored. The components can communicate via local and / or remote processes, such as in accordance with a signal having one or more data packets (e.g., data from one component interacts with another component in a local system, a distributed system, and / or across a network such as the Internet via the signal). As another example, a component can be a device having a specific function provided by mechanical components operated by circuitry or electronic circuitry, which is operated by a software application or firmware application executed by a processor, where the processor can be located inside or outside the device and executes at least a portion of the software or firmware application. As yet another example, a component can be a device having a specific function provided by electronic components without mechanical components, and these electronic components can include a processor therein to execute software or firmware that at least partially imparts the function to the electronic components. Although the various components have been shown as separate components, it will be appreciated that multiple components can be implemented as a single component, or a single component can be implemented as multiple components, without departing from the example embodiments.

[0043] As used herein, the term "facilitate" is in the context of a system, device, or component "facilitating" one or more actions or operations, which is related to the nature of a complex computing environment where multiple components and / or multiple devices may be involved in some computing operations. Non-limiting examples of operations that may or may not involve multiple components and / or multiple devices include: transmitting or receiving data, establishing a connection between devices, determining intermediate results towards obtaining a result, etc. In this regard, a computing device or component can facilitate an operation by playing any part in completing the operation. Thus, when the operations of a component are described herein, it should be understood that in cases where these operations are described as being facilitated by that component, these operations can optionally be completed in cooperation with one or more other computing devices or components (such as but not limited to sensors, antennas, audio and / or video output devices, other devices, etc.).

[0044] In addition, various embodiments can be implemented as using standard programming and / or engineering techniques to produce software, firmware, hardware, or any combination thereof to control a computer to implement the methods, apparatuses, or articles of manufacture of the disclosed subject matter. As used herein, the term "article of manufacture" is intended to cover a computer program accessible from any computer-readable (or machine-readable) device or computer-readable (or machine-readable) storage / communication medium. For example, computer-readable storage media can include but are not limited to magnetic storage devices (e.g., hard disks, floppy disks, magnetic strips), optical disks (e.g., compact disc (CD), digital versatile disc (DVD)), smart cards, and flash memory devices (e.g., cards, sticks, key drives). Of course, those skilled in the art will recognize that many modifications can be made to this configuration without departing from the scope or spirit of the various embodiments.

[0045] Artificial intelligence ("AI") and machine learning ("ML") models can facilitate performance and operational functions and improvements in 5G implementations, such as, for example, network automation, optimizing signaling overhead, energy savings at the device, and traffic-capacity maximization. AI / ML capabilities can be implemented and structured in many different forms and in various vendor-proprietary designs. A 5G radio network may only sense the AI / ML capabilities and capacities of user equipment, including radio functions performed or enhanced by learning models. To facilitate consistent behavior of user equipment devices, the radio access network node ("RAN") of the network to which a user equipment device may be attached or registered manages or controls the real-time AI / ML model performance for various radio functions at different user equipment devices, even if the RAN does not know the (multiple) actual AI / ML model implementations at each user equipment. Such management or control can facilitate minimizing the propagation of learning model errors corresponding to underperforming AI / ML models at the device at multiple execution instances of the corresponding radio function execution, which may degrade the performance of the radio function.

[0046] As disclosed herein, several embodiments facilitate the dynamic management of various AI / ML models deployed at different user equipment devices. The network RAN can dynamically control the activation, deactivation, and triggering of model retraining (which may be specific to radio functions) of a learning model based on the monitoring and analysis of defined real-time performance metrics corresponding to the learning model. This thus allows for the detection of underperforming AI / ML models in near real-time. It will be appreciated that in some embodiments disclosed herein, even if the learning model may be implementing a specific radio function, the metrics monitored or analyzed may be learning model metrics and not necessarily radio function metrics (e.g., mathematical / statistical metrics are not necessarily radio function metrics such as, for example, signal strength).

[0047] Conventional rule-based models can be implemented in user equipment to perform various radio (“RF”) functions or signal processing functions such as beamforming, channel estimation, demodulation, and decoding, and can be based on well-established system models. As long as such models closely follow the actual behavior of the radio network system in which the user equipment is operating, these models can achieve satisfactory performance. However, the performance of traditional conventional models may provide sub-optimal performance. AI / ML-based models generally outperform their conventional counterparts; unlike conventional rule-based models, AI / ML-based models may be data-based rather than based on the rules of pre-determined conventional models. Thus, the output or result of a conventional rule-based model can be considered “deterministic” as the input is applied to static rules that result in a “determined” output, while the output or result of an AI / ML model can be considered probabilistic as the learning model generally infers a likely output based on coefficients, factors, functions, or other variables that may have been derived based on previous inputs to the model.

[0048] Although AI / ML-based models trained with data from actual real-world operations may outperform traditional rule-based models, in cases where the radio system / environment may have undergone changes that may not have been experienced or "seen" during the training of the learning model, the learning model may not be robust enough and thus provide less than ideal results. Therefore, in such cases where the learning model is "unknown", the learning model may infer a less than ideal output compared to a static rule-based model. Such problem situations may be caused by, for example, specific network / user equipment conditions or configurations, or by the architecture of the AI / ML model, or a combination thereof. Therefore, it is desirable to implement a process such that the network RAN can monitor the AI / ML learning model performance of the user equipment via UE feedback / signaling and react to metrics corresponding to such monitoring, which may enable a fallback mechanism or initiate, trigger, or otherwise cause retraining of a poorly performing learning model (e.g., as determined according to the monitored learning model metrics) at the user equipment.

[0049] Unlike conventional rule-based technical solutions, the implementation of an AI / ML learning model for radio functions is inherently probabilistic (e.g., it will not determine a deterministic output for a given input), and the output will depend on the number and quality of samples seen during the training of a given model architecture.

[0050] For example, for the implementation of an AI / ML learning model for radio functions at a user equipment, the user equipment or the gNB / RAN may predict the modulation and coding scheme ("MCS") and use a given number of channel state information reporting instances for it. However, channel conditions or interference conditions that did not exist during training may systematically lead to incorrect MCS selection, which may correspondingly result in violation of the minimum device performance target. Additionally, due to the probabilistic nature of the AI / ML learning model, variable conditions make user equipment performance testing problematic. Therefore, a network RAN that can perceive the AI / ML learning model performance close to real time can facilitate dynamically pre-empting, managing, or calibrating the AI / ML learning model. Such control allows the network to activate or deactivate a detected poorly performing AI / ML learning model, or assist the user equipment device by triggering adaptive model retraining or dataset distribution to support / accelerate the retraining of the learning model, and correspondingly help the user equipment to recover a satisfactory performance of the AI / ML model performance.

[0051] The embodiments disclosed herein can facilitate the RAN (e.g., via a device that is part of the RAN) to dynamically manage and control the performance of various AI / ML learning models that may be facilitating different radio functions at multiple user equipment devices. Uncontrolled AI / ML learning model performance can cause or allow degradation of radio functions and may thus result in violation of the minimum performance requirements of a given user equipment. Embodiments for the RAN to dynamically manage different AI / ML proprietary learning models at various user equipment facilitate the detection and potentially the recovery of underperforming models via additional auxiliary inference, retraining signaling, or data sets.

[0052] The embodiments disclosed herein can include the dynamic management of AI / ML models at different user equipment devices by a network RAN node and can include the definition and compilation of model-specific performance metrics, dynamic reporting of AI / ML learning model performance metrics, and adaptive inference retraining assistance, which can be adjusted according to a given radio function facilitated by the corresponding AI / ML learning model.

[0053] Turning now to the drawings, Figure 1 An example of a wireless communication system 100 that supports blind decoding of PDCCH candidates or search spaces in accordance with aspects of the present disclosure is shown. The wireless communication system 100 can include one or more base stations 105, one or more UEs 115, and a core network 130. In some examples, the wireless communication system 100 can be a Long Term Evolution (LTE) network, an LTE-Advanced (LTE-A) network, an LTE-A Pro network, or a New Radio (NR) network. In some examples, the wireless communication system 100 can support enhanced broadband communication, ultra-reliable (e.g., mission-critical) communication, low latency communication, communication with low-cost and low-complexity devices, or any combination thereof. As shown in the figure, examples of UEs 115 can include smart phones, cars or other vehicles, or drones or other aircraft. Another example of a UE can be a virtual reality apparatus 117, such as smart glasses, virtual reality headsets, augmented reality headsets, and other similar devices that can provide images, videos, audio, tactile, taste, or smell to a wearer. A UE (such as VR apparatus 117) can transmit or receive wireless signals with a RAN base station 105 via a long-range wireless link 125, or the UE / VR apparatus can receive or transmit wireless signals via a short-range wireless link 137, which can include a wireless link with the UE device 115, such as a Bluetooth link, a Wi-Fi link, etc. A UE (such as apparatus 117) can communicate via multiple wireless links simultaneously, such as by communicating via link 125 with the base station 105 and via a short-range wireless link. The VR apparatus 117 can also communicate with a wireless UE via a cable or other wired connection. The RAN or its components can be referred to byFigure 12 Implemented by one or more of the described computer components.

[0054] Continue Figure 1 Continuing the discussion, base stations 105 can be dispersed throughout a geographic area to form a wireless communication system 100, and can be devices in different forms or with different capabilities. Base stations 105 and UEs 115 can communicate wirelessly via one or more communication links 125. Each base station 105 can provide a coverage area 110, and a UE 115 and a base station 105 can establish one or more communication links 125 over the coverage area. Coverage area 110 can be an example of a geographic area over which a base station 105 and a UE 115 can support communication of signals according to one or more radio access technologies.

[0055] UEs 115 can be dispersed within the coverage area 110 of the wireless communication system 100, and each UE 115 can be stationary, mobile, or both at different times. UEs 115 can be devices in different forms or with different capabilities. Some example UEs 115 are shown in Figure 1 . The UEs 115 described herein can communicate with various types of devices, such as other UEs 115, base stations 105, or network devices (e.g., core network nodes, relay devices, integrated access and backhaul (IAB) nodes, or other network devices), as Figure 1 shown in

[0056] Base stations 105 can communicate with the core network 130, or with each other, or both. For example, a base station 105 can interface with the core network 130 via one or more backhaul links 120 (e.g., via S1, N2, N3, or other interfaces). Base stations 105 can communicate with each other directly (e.g., directly between base stations 105) or indirectly (e.g., via the core network 130) or both via backhaul links 120 (e.g., via X2, Xn, or other interfaces). In some examples, backhaul link 120 can include one or more wireless links.

[0057] One or more of the base stations 105 described herein can include or can be referred to by those of ordinary skill in the art as a base transceiver station, radio base station, access point, wireless transceiver, NodeB, eNodeB (eNB), next-generation NodeB, or gigabit NodeB (any of which can be referred to as a bNodeB or gNB), home NodeB, home eNodeB, or other suitable terms.

[0058] The UE 115 may include or may be referred to as a mobile device, wireless device, remote device, handheld device, subscriber device, or other suitable terms, where "device" may also be referred to as a unit, station, terminal, or client, among other examples. The UE 115 may also include or may be referred to as a personal electronic device, such as a cellular phone, personal digital assistant (PDA), tablet computer, laptop computer, personal computer, or router. In some examples, the UE 115 may include or may be referred to as a wireless local loop (WLL) station, Internet of Things (IoT) device, Internet of Everything (IOE) device, or machine type communication (MTC) device, among other examples, which may be implemented in various objects such as appliances, vehicles, or smart meters, among other examples.

[0059] The UE 115 may be capable of communicating with various types of devices, such as other UE 115s that may sometimes act as relays, as well as base station 105 and network devices (including macro base stations (eNBs) or gNBs, small cell base stations (eNBs) or gNBs, or relay base stations, among other examples, as Figure 1 shown therein.

[0060] The UE 115 and the base station 105 may communicate wirelessly with each other via one or more communication links 125 on one or more carriers. The term "carrier" may refer to a set of radio spectrum resources having a defined physical layer structure for supporting the communication link 125. For example, a carrier for the communication link 125 may include a portion of a radio spectrum band (e.g., a bandwidth part (BWP)) that operates according to one or more physical layer channels of a given radio access technology (e.g., LTE, LTE-A, LTE-A Pro, NR). Each physical layer channel may carry acquisition signaling (e.g., synchronization signals, system information), control signaling for coordinating the operation of the carrier, user data, or other signaling. The wireless communication system 100 may support communication with the UE 115 using carrier aggregation or multi-carrier operation. The UE 115 may be configured with multiple downlink component carriers and one or more uplink component carriers according to a carrier aggregation configuration. Carrier aggregation may be used with both frequency division duplex (FDD) and time division duplex (TDD) component carriers.

[0061] In some examples (e.g., in a carrier aggregation configuration), a carrier may also have acquisition signaling or control signaling that coordinates the operation for other carriers. A carrier may be associated with a frequency channel (e.g., an evolved universal mobile telecommunications system terrestrial radio access (E-UTRA) absolute radio frequency channel number (EARFCN)), and may be positioned according to a channel raster for discovery by UE 115. A carrier may operate in an independent mode where initial acquisition and connection may be performed by UE 115 via the carrier, or a carrier may operate in a non-independent mode where a different carrier (e.g., of the same or different radio access technology) is used to anchor the connection.

[0062] The communication link 125 shown in the wireless communication system 100 may include an uplink transmission from UE 115 to the base station 105, or a downlink transmission from the base station 105 to UE 115. A carrier may carry downlink or uplink communication (e.g., in FDD mode), or may be configured to carry both downlink and uplink communication (e.g., in TDD mode).

[0063] A carrier may be associated with a specific bandwidth of the radio spectrum, and in some examples, the carrier bandwidth may be referred to as the "system bandwidth" of the carrier or the wireless communication system 100. For example, the carrier bandwidth may be one of a plurality of bandwidths determined for a carrier of a particular radio access technology (e.g., 1.4, 3, 5, 10, 15, 20, 40, or 80 megahertz (MHz)). Devices of the wireless communication system 100 (e.g., base station 105, UE 115, or both) may have a hardware configuration that supports communication on a specific carrier bandwidth, or may be configured to support communication on one of a set of carrier bandwidths. In some examples, the wireless communication system 100 may include a base station 105 or UE 115 that supports simultaneous communication via carriers associated with multiple carrier bandwidths. In some examples, each served UE 115 may be configured to operate on a portion (e.g., a sub-band, BWP) or all of the carrier bandwidth.

[0064] The signal waveform transmitted through a carrier can be composed of multiple subcarriers (e.g., using multi-carrier modulation (MCM) techniques such as orthogonal frequency division multiplexing (OFDM) or discrete Fourier transform spread OFDM (DFT-S-OFDM)). In a system employing MCM techniques, a resource element can be composed of a symbol period (e.g., the duration of a modulated symbol) and a subcarrier, where the symbol period and the subcarrier spacing are inversely correlated. The number of bits carried by each resource element may depend on the modulation scheme (e.g., the order of the modulation scheme, the coding rate of the modulation scheme, or both). Therefore, the more resource elements the UE 115 receives and the higher the order of the modulation scheme, the higher the data rate of the UE can be. Wireless communication resources can refer to a combination of radio spectrum resources, time resources (e.g., search spaces), or space resources (e.g., spatial layers or beams), and using multiple spatial layers can further increase the data rate or data integrity for communicating with the UE 115.

[0065] One or more numerologies of a carrier can be supported, where a numerology can include a subcarrier spacing (Δf) and a cyclic prefix. A carrier can be divided into one or more BWPs with the same or different numerologies. In some examples, the UE 115 can be configured with multiple BWPs. In some examples, a single BWP for a carrier can be active at a given time, and the communication for the UE 115 can be restricted to one or more active BWPs.

[0066] The time interval of the base station 105 or the UE 115 can be represented by a multiple of a basic time unit, which can refer to, for example, T S = 1 / (Δf max ·N f ) seconds of sampling period, where Δf max can represent the maximum supported subcarrier spacing, and N f can represent the maximum supported discrete Fourier transform (DFT) size. The time interval of the communication resources can be organized according to each radio frame with a specified duration (e.g., 10 milliseconds (ms)). Each radio frame can be identified by a system frame number (SFN) (e.g., ranging from 0 to 1023).

[0067] Each frame may include a plurality of consecutively numbered sub - frames or time slots, and each sub - frame or time slot may have the same duration. In some examples, a frame may be divided (e.g., in the time domain) into sub - frames, and each sub - frame may be further divided into a plurality of time slots. Alternatively, each frame may include a variable number of time slots, and the number of time slots may depend on the sub - carrier spacing. Each time slot may include a plurality of symbol periods, e.g., depending on the length of the cyclic prefix added to each symbol period. In some wireless communication systems 100, a time slot may be further divided into a plurality of mini - time slots each containing one or more symbols. In addition to the cyclic prefix, each symbol period may contain one or more (e.g., N f

[0068]

[0069]

[0070] physical channels may be multiplexed on a carrier according to various techniques. The physical control channel and the physical data channel may be multiplexed on a downlink carrier, e.g., using one or more of time - division multiplexing (TDM) techniques, frequency - division multiplexing (FDM) techniques, or hybrid TDM - FDM techniques. The control region of the physical control channel (e.g., control resource set (CORESET)) may be defined by a plurality of symbol periods and may span the system bandwidth of the carrier or a subset of the system bandwidth. One or more control regions (e.g., CORESET) may be configured for a group of UEs 115. For example, one or more of the UEs 115 may monitor or search a control region or space for control information according to one or more search space sets, and each search space set may include one or more control channel candidates at one or more aggregation levels arranged in a cascaded manner. The aggregation level of a control channel candidate may refer to the number of control channel resources (e.g., control channel elements (CCE)) associated with the coded information of a control information format having a given payload size. The search space set may include a common search space set configured for transmitting control information to a plurality of UEs 115 and a UE - specific search space set for transmitting control information to a particular UE 115. Other search spaces and configurations for monitoring and decoding them are disclosed herein, which are novel and not conventional.

[0070] Base station 105 may provide communication coverage via one or more cells (e.g., macro cells, small cells, hotspots, or other types of cells, or any combination thereof). The term "cell" may refer to a logical communication entity for communicating with base station 105 (e.g., via a carrier), and may be associated with an identifier for differentiating adjacent cells (e.g., physical cell identifier (PCID), virtual cell identifier (VCID), or others). In some examples, a cell may also refer to a geographic coverage area 110 or a portion of the geographic coverage area 110 (e.g., a sector) on which the logical communication entity operates. The scope of such a cell may range from a smaller area (e.g., a structure, a subset of a structure) to a larger area, depending on various factors such as the capabilities of base station 105. For example, a cell may be or include a building, a subset of a building, or an external space located between or overlapping with geographic coverage areas 110, among other examples.

[0071] Macro cells typically cover a relatively large geographic area (e.g., with a radius of several kilometers) and may allow unrestricted access by UEs 115 having a service agreement with the network provider supporting the macro cell. Compared with macro cells, small cells may be associated with base stations 105 having lower power, and small cells may operate in the same or different frequency bands (e.g., licensed, unlicensed) as macro cells. Small cells may provide unrestricted access to UEs 115 having a service agreement with the network provider, or may provide restricted access to UEs 115 associated with the small cell (e.g., UEs 115 in a closed subscriber group (CSG), UEs 115 associated with users in a home or office). Base station 105 may support one or more cells and may also support communication on one or more cells using one or more component carriers.

[0072] In some examples, a carrier may support multiple cells, and different cells may be configured according to different protocol types (e.g., MTC, narrowband IoT (NB-IoT), enhanced mobile broadband (eMBB)), and different protocol types may provide access for different types of devices.

[0073] In some examples, base station 105 may be movable and thus may provide communication coverage for a mobile geographic coverage area 110. In some examples, different geographic coverage areas 110 associated with different technologies may overlap, but different geographic coverage areas 110 may be supported by the same base station 105. In other examples, overlapping geographic coverage areas 110 associated with different technologies may be supported by different base stations 105. The wireless communication system 100 may include, for example, a heterogeneous network in which different types of base stations 105 provide coverage for various geographic coverage areas 110 using the same or different radio access technologies.

[0074] The wireless communication system 100 may support synchronous or asynchronous operation. For synchronous operation, the base stations 105 may have similar frame times, and transmissions from different base stations 105 may be approximately aligned in time. For asynchronous operation, the base stations 105 may have different frame times, and in some examples, transmissions from different base stations 105 may not be aligned in time. The techniques described herein may be used for synchronous or asynchronous operation.

[0075] Some UEs 115 (such as MTC or IoT devices) may be low-cost or low-complexity devices and may provide automated communication between machines (e.g., via machine-to-machine (M2M) communication). M2M communication or MTC may refer to data communication technologies that allow devices to communicate with each other or with the base station 105 without human intervention. In some examples, M2M communication or MTC may include communication from devices such as integrated sensors or meters that measure or capture information and relay such information to a central server or application that utilizes the information or presents the information to a human who interacts with the application. Some UEs 115 may be designed to collect information or implement automated behavior of machines or other devices. Examples of applications of MTC devices include smart metering, inventory monitoring, water level monitoring, device monitoring, healthcare monitoring, wildlife monitoring, weather and geological event monitoring, fleet management and tracking, remote security sensing, physical access control, and transaction-based business billing.

[0076] Some UEs 115 may be configured to operate in power-saving modes, such as half-duplex communication (e.g., a mode that supports one-way communication via transmission or reception but does not support simultaneous transmission and reception). In some examples, half-duplex communication may be performed at a lower peak rate. Other energy-saving techniques for UEs 115 include entering a power-saving deep sleep mode when not actively communicating, operating on a limited bandwidth (e.g., according to narrowband communication), or a combination of these techniques. For example, some UEs 115 may be configured to operate using a narrowband protocol type associated with a defined portion or range (e.g., a set of subcarriers or resource blocks (RBs)) within a carrier, within a carrier guard band, or outside a carrier.

[0077] The wireless communication system 100 can be configured to support ultra-reliable communication or low-latency communication, or various combinations thereof. For example, the wireless communication system 100 can be configured to support ultra-reliable low-latency communication (URLLC) or mission-critical communication. The UE 115 can be designed to support ultra-reliability, low latency, or critical functions (e.g., mission-critical functions). Ultra-reliable communication can include private communication or group communication, and can be supported by one or more mission-critical services, such as mission-critical push-to-talk (MCPTT), mission-critical video (MCVideo), or mission-critical data (MCData). Support for mission-critical functions can include prioritization of services, and mission-critical services can be used for public safety or general commercial applications. The terms ultra-reliable, low-latency, mission-critical, and ultra-reliable low-latency are used interchangeably herein.

[0078] In some examples, the UE 115 can also communicate directly with other UEs 115 via a device-to-device (D2D) communication link 135 (e.g., using peer-to-peer (P2P) or D2D protocols). The communication link 135 can include a sidelink communication link. One or more UEs 115 utilizing D2D communication can be within the geographical coverage 110 of the base station 105. Other UEs 115 in such a group can be outside the geographical coverage 110 of the base station 105, or otherwise unable to receive transmissions from the base station 105. In some examples, a group of UEs 115 communicating via D2D communication can utilize a one-to-many (1:M) system, where a UE transmits to each other UE in the group. In some examples, the base station 105 facilitates the scheduling of resources for D2D communication. In other cases, D2D communication is performed between UEs 115 without the participation of the base station 105.

[0079] In some systems, the D2D communication link 135 can be an example of a communication channel (such as a sidelink communication channel) between vehicles (e.g., UE 115). In some examples, vehicles can communicate using vehicle-to-everything (V2X) communication, vehicle-to-vehicle (V2V) communication, or some combination of these communication methods. Vehicles can transmit information related to traffic conditions, signal scheduling, weather, safety, emergencies, or any other information related to the V2X system. In some examples, vehicles in a V2X system can communicate with roadside infrastructure (such as roadside units), or communicate with the network via one or more RAN network nodes (e.g., base station 105) using vehicle-to-network (V2N) communication, or communicate with both.

[0080] The core network 130 may provide user authentication, access authorization, tracking, Internet Protocol (IP) connectivity, and other access, routing, or mobility functions. The core network 130 may be an evolved packet core (EPC) or a 5G core (5GC), which may include at least one control plane entity that manages access and mobility (e.g., a mobility management entity (MME), an access and mobility management function (AMF)) and at least one user plane entity that routes packets or interconnects to external networks (e.g., a serving gateway (S-GW), a packet data network (PDN) gateway (P-GW), or a user plane function (UPF)). The control plane entity may manage non-access stratum (NAS) functions, such as mobility, authentication, and bearer management of the UE 115 served by the base station 105 associated with the core network 130. User IP packets may be transmitted through the user plane entity, which may provide IP address allocation and other functions. The user plane entity may be connected to IP services 150 for one or more network operators. The IP services 150 may include access to the Internet, an intranet(s), an IP multimedia subsystem (IMS), or packet-switched streaming services.

[0081] Some network devices, such as the base station 105, may include subcomponents, such as an access network entity 140, which may be an example of an access node controller (ANC). Each access network entity 140 may communicate with the UE 115 through one or more other access network transmission entities 145, which may be referred to as radio heads, intelligent radio heads, or transmission / reception points (TRPs). Each access network transmission entity 145 may include one or more antenna panels. In some configurations, the various functions of each access network entity 140 or the base station 105 may be distributed across various network devices (e.g., radio heads and ANCs) or combined into a single network device (e.g., the base station 105).

[0082] The wireless communication system 100 may operate using one or more frequency bands, typically in the range of 300 megahertz (MHz) to 300 gigahertz (GHz). Generally, the region from 300 MHz to 3 GHz is referred to as the ultra-high frequency (UHF) region or the decimeter band because the length of the wavelength range is from approximately 1 decimeter to 1 meter. UHF waves may be blocked or redirected by buildings and environmental features, but these waves can penetrate buildings sufficiently to enable macrocells to serve UEs 115 located indoors. Compared to transmissions at smaller frequencies and longer wavelengths using the high frequency (HF) or very high frequency (VHF) portions of the spectrum below 300 MHz, UHF wave transmissions may be associated with smaller antennas and shorter ranges (e.g., less than 100 kilometers).

[0083] The wireless communication system 100 may also operate in the super high frequency (SHF) region using a spectrum band from 3 GHz to 30 GHz (also known as the centimeter band) or in the extremely high frequency (EHF) region of the spectrum (e.g., from 30 GHz to 300 GHz) (also known as the millimeter band). In some examples, the wireless communication system 100 may support millimeter wave (mmW) communication between the UE 115 and the base station 105, and the EHF antennas of the corresponding devices may be smaller and closer spaced than UHF antennas. In some examples, this may facilitate the use of antenna arrays within the device. However, the propagation of EHF transmissions may be subject to even greater atmospheric attenuation and shorter ranges than SHF or UHF transmissions. The techniques disclosed herein may be employed across transmissions using one or more different frequency regions, and the designated use of the spectrum bands across these frequency regions may vary by country or regulatory body.

[0084] The wireless communication system 100 may utilize both licensed and unlicensed radio spectrum bands. For example, the wireless communication system 100 may employ licensed assisted access (LAA), LTE-unlicensed (LTE-U) radio access technology, or NR technology in an unlicensed spectrum band such as the 5 GHz industrial, scientific, and medical (ISM) spectrum band. When operating in an unlicensed radio spectrum band, devices such as the base station 105 and the UE 115 may employ carrier sensing for collision detection and avoidance. In some examples, operation in an unlicensed spectrum band may be based on a carrier aggregation configuration that combines component carriers operating in a licensed spectrum band (e.g., LAA). Operation in the unlicensed spectrum may include downlink transmissions, uplink transmissions, peer-to-peer (P2P) transmissions, or device-to-device (D2D) transmissions, among other examples.

[0085] The base station 105 or the UE 115 may be equipped with multiple antennas that may be used to employ techniques such as transmit diversity, receive diversity, multiple-input multiple-output (MIMO) communication, or beamforming. The antennas of the base station 105 or the UE 115 may be located within one or more antenna arrays or antenna panels that may support MIMO operation or transmission or receive beamforming. For example, one or more base station antennas or antenna arrays may be co-located on an antenna assembly such as an antenna tower. In some examples, the antennas or antenna arrays associated with the base station 105 may be located in different geographical locations. The base station 105 may have an antenna array having multiple rows and columns of antenna ports that the base station 105 may use to support beamforming of communications with the UE 115. Similarly, the UE 115 may have one or more antenna arrays that may support various MIMO or beamforming operations. Additionally or alternatively, an antenna panel may support radio beamforming of signals transmitted via the antenna ports.

[0086] The base station 105 or the UE 115 can use MIMO communication to utilize multipath signal propagation and improve spectral efficiency by transmitting or receiving multiple signals via different spatial layers. This technique can be referred to as spatial multiplexing. For example, a transmitting device can transmit multiple signals via different antennas or different combinations of antennas. Similarly, a receiving device can also receive multiple signals via different antennas or different combinations of antennas. Each of the multiple signals can be referred to as a separate spatial stream and can carry bits associated with the same data stream (e.g., the same codeword) or different data streams (e.g., different codewords). Different spatial layers can be associated with different antenna ports for channel measurement and reporting. MIMO techniques include: single-user MIMO (SU-MIMO), where multiple spatial layers are transmitted to the same receiving device; and multi-user MIMO (MU-MIMO), where multiple spatial layers are transmitted to multiple devices.

[0087] Beamforming, which can also be referred to as spatial filtering, directional transmission, or directional reception, is a signal processing technique that can be used at a transmitting device or a receiving device (e.g., the base station 105, the UE 115) to shape or manipulate an antenna beam (e.g., a transmission beam, a reception beam) along a spatial path between the transmitting device and the receiving device. Beamforming can be achieved by combining signals transmitted via the antenna elements of an antenna array such that some signals propagating in a particular orientation relative to the antenna array experience constructive interference while other signals experience destructive interference. The adjustment of the signals transmitted via the antenna elements can include the transmitting device or the receiving device applying an amplitude offset, a phase offset, or both to the signals carried via the antenna elements associated with the device. The adjustment associated with each of the antenna elements can be defined by a set of beamforming weights associated with a particular orientation (e.g., relative to the antenna array of the transmitting device or the receiving device, or relative to some other orientation).

[0088] The base station 105 or the UE 115 can use beam scanning techniques as part of beamforming operations. For example, the base station 105 can use multiple antennas or antenna arrays (e.g., antenna panels) to perform beamforming operations for directional communication with the UE 115. Some signals (e.g., synchronization signals, reference signals, beam selection signals, or other control signals) can be transmitted by the base station 105 multiple times in different directions. For example, the base station 105 can transmit signals according to different sets of beamforming weights associated with different transmission directions. Transmissions in different beam directions can be used (e.g., by the transmitting device such as the base station 105 or the receiving device such as the UE 115) to identify the beam direction for later transmission or reception by the base station 105.

[0089] Some signals, such as data signals associated with a particular receiving device, can be transmitted by base station 105 in a single beam direction (e.g., the direction associated with a receiving device such as UE 115). In some examples, the beam direction associated with transmission along a single beam direction can be determined based on signals transmitted in one or more beam directions. For example, UE 115 can receive one or more signals transmitted by base station 105 in different directions and can report to the base station an indication of the signal received by UE 115 having the highest signal quality or other acceptable signal quality.

[0090] In some examples, transmissions made by a device (e.g., by base station 105 or UE 115) can be performed using multiple beam directions, and the device can use a combination of digital precoding or radio beamforming to generate a combined beam for transmission (e.g., from base station 105 to UE 115). UE 115 can report feedback indicating precoding weights for one or more beam directions, and the feedback can correspond to the configured number of beams across the system bandwidth or one or more subbands. Base station 105 can transmit reference signals (e.g., cell-specific reference signal (CRS), channel state information reference signal (CSI-RS)), which can be precoded or non-precoded. UE 115 can provide feedback for beam selection, which can be a precoding matrix indicator (PMI) or codebook-based feedback (e.g., multi-panel type codebook, linear combination type codebook, port selection type codebook). Although these techniques are described with reference to signals transmitted by base station 105 in one or more directions, UE 115 can employ similar techniques for transmitting signals multiple times in different directions (e.g., for identifying beam directions for subsequent transmission or reception by UE 115), or for transmitting signals in a single direction (e.g., for transmitting data to a receiving device).

[0091] A receiving device (e.g., UE 115) may attempt multiple receive configurations (e.g., directional listening) when receiving various signals (such as synchronization signals, reference signals, beam selection signals, or other control signals) from a base station 105. For example, the receiving device may attempt multiple receive directions by: receiving via different antenna sub-arrays; processing received signals according to different antenna sub-arrays; receiving according to different receive beamforming weight sets (e.g., different directional listening weight sets) applied to signals received at multiple antenna elements of an antenna array; or processing according to different receive beamforming weight sets applied to signals received at multiple antenna elements of an antenna array. Any of the above ways may be referred to as "listening" according to different receive configurations or receive directions. In some examples, the receiving device may use a single receive configuration to receive along a single beam direction (e.g., when receiving a data signal). This single receive configuration may be aligned with the beam direction determined based on listening in different receive configuration directions (e.g., based on determining the beam direction with the highest signal strength, highest signal-to-noise ratio (SNR), or other acceptable signal quality according to listening in multiple beam directions).

[0092] The wireless communication system 100 may be a packet-based network operating according to a hierarchical protocol stack. In the user plane, communication at the bearer or packet data convergence protocol (PDCP) layer may be based on IP. The radio link control (RLC) layer may perform packet segmentation and reassembly for communication over logical channels. The media access control (MAC) layer may perform priority handling and multiplex logical channels into transport channels. The MAC layer may also use error detection techniques, error correction techniques, or both to support retransmission at the MAC layer to improve link efficiency. In the control plane, the radio resource control (RRC) protocol layer may provide the establishment, configuration, and maintenance of an RRC connection between the UE 115 and the base station 105 or the core network 130, thus supporting radio bearers for user plane data. At the physical layer, transport channels may be mapped to physical channels.

[0093] UE 115 and base station 105 may support retransmission of data to increase the likelihood that the data is successfully received. Hybrid Automatic Repeat reQuest (HARQ) feedback is a technique for increasing the likelihood of correctly receiving data over communication link 125. HARQ may include a combination of error detection (e.g., using Cyclic Redundancy Check (CRC)), Forward Error Correction (FEC), and retransmission (e.g., Automatic Repeat reQuest (ARQ)). Under poor radio conditions (e.g., low signal-to-noise ratio conditions), HARQ can improve throughput at the MAC layer. In some examples, a device may support same-slot HARQ feedback, where the device may provide HARQ feedback in a particular slot for data received in the previous symbol in that slot. In other cases, the device may provide HARQ feedback in a subsequent slot or according to some other time interval.

[0094] Now turning to Figure 2 , the figure shows a system 200 that includes a RAN node 105 communicating with a user equipment 115 via a wireless link 125. UE 115 may perform various radio functions 205A - 205n facilitated by corresponding machine learning models 215A - 215n, respectively. UE 115 may transmit an indication of radio function learning model information 207 to RAN 105. RAN105 may transmit a machine learning model management indication configuration 210 corresponding to or based on the learning model information 207 to UE 115. During wireless operation of UE 115 and communication with RAN 105, the UE may transmit parameter metric reports 220A - 220n, which may include one or more learning model parameter metrics corresponding to 215A - 215n, respectively. In one embodiment, reports 220A - 220n may include one or more control action requests, e.g., requests to deactivate or retrain one or more of models 215A - 215n. UE 115 may determine a control action request based on monitoring metrics corresponding to the operation of models 215A - 215n, or RAN 115 may determine a control action based on the monitoring metrics transmitted in one or more of reports 220A - 220n.

[0095] AI / ML model inference performance monitoring and dynamic reporting of AI / ML model performance indications.

[0096] The network RAN can monitor the inference performance of the UE's learning model by transmitting a test data set to the UE. The RAN can also provide the UE with one or more optional minimum performance requirements associated with the test data set. The test data set can be transmitted periodically according to a network-configured period, or the test data set can be dynamically transmitted to the UE based on a trigger (e.g., a change in the serving node / RAN providing wireless network services to the UE). The network RAN can provide the UE with the (multiple) test data sets, which can represent the current conditions experienced by the UE. The RAN can generally "learn" or determine these conditions based on information related to the UE channel and radio environment determined via various UE reports or sensing information that can be provided from the UE to the RAN. The attributes of the test data set, the size or format of the data set, can be configured for each UE or each group of related UEs.

[0097] The AI / ML learning model deployed at the UE device 115 (such as Figure 2 the model 215 shown in) can be implementation-specific (e.g., a vendor-proprietary learning model). (Examples of vendors that can provide proprietary learning models include user equipment manufacturers or providers of user equipment applications, network equipment providers or providers of network equipment applications, or mobile network operators or providers of mobile network operator applications.) The network RAN can determine the overall performance of the learning model deployed at the UE to facilitate minimum device performance requirements. As disclosed herein, a dynamic reporting process can facilitate the user equipment compiling and reporting indications that can be configured or pre-configured, reflecting or indicating the model performance of the corresponding learning model.

[0098] A particular user equipment can employ several different AI / ML learning model implementations to run, execute, or otherwise facilitate different radio functions. Different learning model parameter metrics can indicate the performance of different learning models. The embodiments disclosed herein can enable the user equipment to compile and report one or more different learning model performance indication parameter metrics or indications for each learning model. Different learning model metrics can be associated with different corresponding filtering or time resolution configurations. Thus, such customized metric reporting for a given learning model can facilitate the optimized tracking and reporting of each active learning model of each user equipment device 115 that can be served by the RAN105, as Figure 1 or Figure 2As shown. Thus, the network RAN 105 can obtain and use the real-time performance of each learning model active at the UE 115 to facilitate the best performance of the learning model and the inferences it may generate. Additionally, several report variants can be customized to accommodate various AI / ML learning model implementations and purposes. For example, exact absolute, exact relative, quantization, or temporal (e.g., historical) metric reports are disclosed and described herein. The network node RAN 105 can dynamically trade off the AI / ML learning model reporting overhead against the accuracy of obtaining AI / ML model performance metrics.

[0099] For AI / ML learning model performance, various parameters and their corresponding metrics can be considered, analyzed, or evaluated based on the nature of the problem being solved and the corresponding learning model function (e.g., regression or classification), or the radio function that the learning model is performing or facilitating. For example, for radio functions such as channel estimation or channel state information (“CSI”) compression, a regression function can be used in a learning model with the following parameters or their corresponding metrics that may be evaluated: mean squared error (“MSE”); root mean squared error (“RMSE”); normalized mean squared error (“NMSE”); mean absolute error (“MAE”); R-squared; generalized cosine similarity (“GCS”); or squared generalized cosine similarity (“SGCS”). Table 1 shows an example function that defines the corresponding learning model parameters, and the corresponding metrics can be monitored and evaluated as listed above.

[0100]

[0101]

[0102] Table 1

[0103] For classification problems such as beam index prediction, accuracy parameter metrics can be analyzed to determine the performance of the learning model that facilitates beam index prediction. Other example learning model parameter metrics that can indicate the performance of a learning model solving a classification problem can include, but are not limited to: the absolute numbers of true negatives, true positives, false negatives, and false positives; precision and recall; or the F1 score. The F1 score can include an evaluation metric that is used to represent the performance of a machine learning model or classifier and provides combined information about the precision and recall of the learning model. A high F1 score metric generally indicates high values for both the recall and precision metrics.

[0104] The implementation of AI / ML learning models at different devices can be vendor - proprietary as described above and can be transparent to network nodes (e.g., the RAN serving a UE may not have access to the specific functionality and programming of a given learning model deployed in the UE that facilitates radio functions). To manage and facilitate the UE device to achieve performance goals, the network RAN can be made aware of the UE's capabilities and the overall AI / ML learning model performance. Thus, an active UE device can transmit device - specific AI / ML capability information when first connecting to the serving network RAN, including the following information elements ("IE"): the types of AI / ML - supported algorithms, including supervised learning, unsupervised learning, and reinforcement learning; a list of radio functions supported by AI / ML; a list of supported AI / ML model - specific metrics for estimation and reporting; the model library size for each radio function, e.g., the number of models that can be stored for each radio function; or an indication of model classification (small / medium / large), which can facilitate the network RAN to define or determine the data set to be used by the learning model. For example, for a large number of neurons (e.g., nodes of a learning model neural network), the determination of a commensurate number of information samples can be used to avoid overfitting of the learning model. The AI / ML capability IE can become part of the device - capability signaling based on subsequent radio resource control ("RRC") signaling or on dynamically scheduled uplink control information ("UCI") transmission.

[0105] Thus, the network RAN can configure a UE device with AI / ML learning model capabilities to monitor or estimate and report certain model - specific parameter metrics indicating the learning model performance, and report the monitored metrics as model management indications via an uplink channel. The learning model can be identified in the model management indication by a network - assigned model identifier. As Figure 3A shown, in one embodiment, the model management indication includes, at the UE device operating these models, separately associated with Figure 2Independent performance metrics 310, 315, and 320 corresponding to the active AI / ML models 215A, 215B, and 215n shown. Metrics 310, 315, and 320 may correspond to model identifiers 311, 316, and 321, respectively. Enabling multiple metric configurations and reports per model allows the device and network to coordinate the best AI / ML performance metrics that reflect the actual performance of the corresponding AI / ML models, since different models may be represented by various performance metrics with different computational complexities and reporting overheads. The RAN or UE may determine the metric configuration for reporting learning model metrics indicative of the performance of the learning model. Different metric reports may cause the size of the model management indication to vary over time, depending on the active model and the corresponding reported metric size (e.g., various performance metrics at a given time may have different sizes, either reported as exact values or reported as quantization levels or ranges corresponding to a value). Thus, the model management indication can be dynamically scheduled via the uplink data channel (PUSCH) to facilitate scheduling efficiency and flexibility.

[0106] Examples of analyzing the monitored metric values relative to a model performance metric standard may include: the standard is a threshold, and the user equipment compares the monitored metric with the threshold (e.g., comparing the mean squared error (“MSE”), or comparing any other monitored metric value from Table 1 with the threshold, and reporting to the radio access network if the MSE is greater than the threshold). Another example may include: the standard is a range of values, and the user equipment determines whether the monitored metric value is within a given range, and reporting to the radio access network if the monitored metric is outside the standard range.

[0107] In Figure 3A it, the dynamic learning model management metric report 305 may dynamically report learning model-specific metrics according to the learning model identifier corresponding to the respective learning model. Report 305 may include metric reports 310, 315, and 320 corresponding to the learning models 215A, 215B, and 215n shown in Figure 2 respectively.

[0108] To reduce the overhead of reporting AI / ML model management indications, quantization of model-specific performance metrics may be used, where the metric-specific quantization levels corresponding to ranges may be configured by the RAN. As Figure 3BAs shown, the UE device indicated by the transmission model management can select a quantized representation of the estimated model-specific metric instead of reporting the exact metric value, resulting in a reduction in the total size of the model management indication. The quantization indication can be in the following forms: a continuous quantization level indication (i.e., bitwise), or a sequence / leading code representation of the determined quantization metric levels. An example of quantization would be that the UE determines the actual metric value of a parameter falling within a certain range and can transmit an index value instead of the actual metric value to reduce the amount of data used to transmit the model management indication to the serving RAN.

[0109] In another embodiment, the network RAN can configure a learning model parameter metric standard for a device with AI / ML learning model capabilities, which can be referred to as a model performance metric standard, such as a threshold associated with an active AI / ML model. Thus, when the learning model of the UE device fails or violates the configured learning model-specific performance metric threshold, the UE can transmit a model-specific "FAIL" indication to the serving RAN. Therefore, the network RAN can determine whether to trigger a retraining data set based on the received model FAIL indication. In Figure 3B In, the example dynamic learning model management metric report 325 can dynamically report the quantized learning model metrics specific to the learning model according to the learning model identifier corresponding to the respective learning model. The report 325 can include metric reports 330, 335, and 340 corresponding to the learning models 215A, 215B, and 215n shown in Figure 2 respectively. The metric reports 330, 335, and 340 can correspond to the model identifiers 331, 336, and 341 respectively.

[0110] In another embodiment, the network RAN can configure the UE device to report historical model-specific performance metrics over time within a configured time period. This may help predict an AI / ML learning model (e.g., a time model) for specific radio conditions or actions at a future set of moments. Thus, as Figure 3C shown, the UE device can compile a model-specific model management indication report within the configured historical or time reporting period, including exact or quantized indications of the selected model-specific parameters. In another embodiment, the network RAN can configure the UE device for a filtered time report of specific model parameters metrics (e.g., indicated by the corresponding model ID). This can instruct the UE device to perform filtering on the selected model-specific metric samples within the configured reporting period according to the indicated layer 1 filter type and filtering coefficient, thus further reducing the network resource overhead used to report the model management indication (which could otherwise be used to transmit volume data).

[0111] In Figure 3CIn this example, the historical learning model management metric report 350 can report learning model metrics specific to a historical learning model based on a learning model identifier corresponding to the learning model. The report 350 can include metric reports 355, 360, and 365 corresponding to one of the learning models 215, x2, and x3 shown, for example, Figure 2 in. The metric reports 355, 360, and 365 can correspond to metric report identifiers 356, 361, and 365, respectively.

[0112] Figure 3D An embodiment of an example historical learning model management metric report 370 is shown that can report learning model metrics specific to a historical learning model based on a learning model identifier corresponding to the respective learning model. The report 370 can include metric reports 375, 380, and 385 corresponding to the learning models 215A, 215B, and 215n shown, for example, Figure 2 in. The metric reports 375, 380, and 385 can correspond to model identifiers 376, 381, and 386, respectively. Thus, each of the metrics 375, 380, and 385 can include a historical report for the respective models 376, 381, and 386. The reports 375, 380, and 385 can be, for example, information or data that can be included in the historical report 350 described in the reference Figure 3A . In other words, the reports 375, 380, and 385 can include the historical report 350, each historical report corresponding to a different model. Thus, the report 370 can include multiple reports 350, each report corresponding to a different learning model.

[0113] Now turning to Figure 4, the figure shows a timing diagram of an example method for configuring a user equipment 115 using a learning model metric report configuration from a network node / RAN 105. At action 405, a UE or wireless transmit and receive unit (WTRU) 115 (hereinafter sometimes referred to only as UE 115) transmits capability information to the serving cell RAN 105, which may include an indication of radio function learning model information, which may include learning model information corresponding to one or more radio function learning models of the user equipment. The capability information may include an indication of one or more radio functions that may be implemented, operated, executed, or otherwise facilitated by the UE 115. The capability information may include information indicating the type of AI / ML learning model (e.g., supervised, unsupervised, reinforcement learning, etc.) that may correspond to, support, or facilitate the radio function. In operation 410, the RAN 105 transmits one or more AI / ML learning model management configurations, and the UE / WTRU 115 receives one or more AI / ML learning model management configurations from the RAN 105. The learning model management configuration may include format information corresponding to an AI / ML model management request that may be made by the UE 115.

[0114] The format information transmitted and received at operation 410 may include: information for a current indication of the learning model metric used by the UE 115 to generate each active learning model, which may have a model identifier associated with it assigned by the RAN 105; information for an indication of the generation of aggregated metric information corresponding to multiple or all learning models, which may currently be active based on the model identifier associated with the model assigned by the RAN 105; information corresponding to the number of determined or configured learning models or the historical metrics of the model for each active learning model; and the corresponding RAN-assigned model identifier. The format information may include learning model-specific performance metrics reported according to the model identifier or reporting conditions of the model management request, including the reporting period or learning model degradation criteria, which may be referred to as model performance metric criteria associated with the configured metrics specific to the model, e.g., thresholds, for triggering the transmission of a model management request from the UE 115 to the RAN 105 based on non-satisfaction of the model performance metric criteria.

[0115] At action 415, in the case where the period for transmitting an AI / ML learning model management request has expired or the associated reporting conditions corresponding to the learning model have been met (e.g., the monitored metrics do not meet the criteria), the UE / WTRU 115 may transmit a learning model management request for the AI / ML learning model that may be actively facilitating the radio function at the UE model based on the configured AI model management request format, thereby indicating a request to deactivate one or more active AI models.

[0116] RAN 105 receives the learning model management request transmitted by UE 115 at action 415, and the UE may have an active AI / ML learning model that actively promotes radio functions. At action 420, RAN node 105 may transmit a model deactivation request associated with one or more indicated model identifiers of one or more learning models operating on UE 115. The model deactivation request may instruct UE 115 to deactivate one or more learning models that are currently promoting radio functions, such that the radio functions revert to operating according to a previous learning model configuration, a default learning model configuration, or a default deterministic radio function model. At action 425, RAN node 105 may abort or refresh an active inference assistance signaling process that may correspond to the model deactivated at UE 115, based on the learning model identifier associated with the learning model deactivated at UE 115.

[0117] In an embodiment, RAN 105 may transmit, at action 420, a request for UE 115 to retransmit the learning model management request transmitted at action 415 to verify if there is a problem at the UE.

[0118] AI / ML model retraining.

[0119] If based on, for example, reference Figure 2 , Figure 3, or Figure 4The described embodiments detect poor performance of AI / ML learning models, and the serving RAN can use additional training data sets to assist the served UE device in calibrating and restoring the degraded inference performance of one or more models with poor performance. The additional training data sets can be specific to the radio functions facilitated by the AI / ML models with poor performance. Accordingly, a list of training data codebooks is defined, where, for example, each codebook or multiple codebooks is associated with a specific radio function (such as power control, scheduling, or beam management). The training codebooks can include training data sets specifically designed for training or calibrating learning models that perform specific radio functions. Thus, the training codebooks can indicate to the UE device various data sets, training / retraining durations, time resolutions (e.g., the time scale or instance between retraining data transmissions in terms of OFDM symbols, mini-slots, slots, frames, total number of frames, etc.), or resource sets for dynamic or semi-static configuration of the models. The RAN node can configure the available training codebooks for various radio functions for the device. When a model with poor performance of one or more active radio functions is detected or determined at the user equipment device or network node (based on metrics transmitted by the user equipment), a request indication indicating a request for inference retraining or model restoration can be transmitted together with a recommended inference codebook determined based on the real-time performance of the AI / ML model (based on the monitored metrics) and the tolerance / requirements of the corresponding radio function. The network RAN can dynamically or semi-statically schedule the transmission of the selected training data sets with the corresponding time durations, resource sizes, and time resolutions. Multiple available data sets for each AI / ML-facilitated radio function can be transmitted to facilitate AI / ML learning models that perform differently for the same radio function, thus having training data sets customized with respect to their corresponding real-time AI / ML performance.

[0120] The performance of various AI / ML models that perform different radio functions may depend on several factors, including traffic channel conditions, traffic distribution, interference statistics, link adaptation, scheduling determination, etc. Such factors may change over time, and the rate of change depends on factors such as the traffic load level or traffic transmission type (e.g., short-term sporadic transmissions or large payload transmissions) of the RAN. Accordingly, the AI / ML learning models at the user equipment device may occasionally be improved to obtain or maintain the desired inference performance of the learning models. In an example, for an implementation with an AI / ML learning model that operates specifically at the user equipment, the network RAN can provide assistance signaling and data sets for calibrating and improving the corresponding inference performance in case of a detected AI / ML learning model with poor performance (e.g., the learning model metrics do not meet one or more learning model parameter criteria).

[0121] In Figure 5In the example learning model management retraining configuration 500 shown, the inference retraining data set 520 can be specific to each of one or more radio functions 515 that can be performed or facilitated by a corresponding AI / ML model to be improved at a user equipment. AI / ML learning models that can facilitate different radio functions can be configured or reconfigured using different inference retraining data sets. For example, an AI / ML learning model that facilitates power control at a user equipment device can use a data set that causes the user equipment to perform a number of uplink transmissions at a number of transmission power level values and power control parameters. However, a different learning model that facilitates, for example, beam management can use a data set that includes parameters for facilitating beam scanning, beam detection, and beam recovery processes. Thus, a user equipment with AI / ML capabilities can be configured with various AI / ML inference retraining data sets 520, where the retraining set can be associated with a specific radio function 515 and can be dynamically configured according to parameters 525, which can include data set length, resource size, time resolution, and coordinated network device actions. The retraining data set 520 can be delivered to the user equipment device using a system information block, RRC setup signaling, or via direct downlink control information (“DCI”). Additionally, for some radio functions, the network RAN can provide both samples and labels in the data set 520. For other radio functions, the RAN can first transmit a first data set 520 (e.g., samples), and then transmit a second data set 520 (e.g., labels) after the network processing information has been transmitted. For example, this segmented transmission of the data set 520 can be used if the network RAN does not have the exact user equipment device condition information available for use in generating the labels.

[0122] Accordingly, a user equipment with AI / ML capabilities can request inference retraining from the serving RAN, as well as an indication of: a recommended inference retraining data set or codebook 520 corresponding to the respective radio function 510 of a preconfigured inference retraining list, and the corresponding available inference data set / codebook associated with the radio function. The user equipment device can select an inference retraining data set in terms of data length, resource size, and timing resolution based on the active AI / ML learning model being executed by the user equipment and the current condition of the corresponding radio function, and include it in the request. In response, the network RAN can transmit the inference retraining data set to the requesting user equipment device based on the requested inference retraining data set. In one embodiment, the channel resources for transmitting the inference retraining data set can be dynamically scheduled according to the resource set size indicated by the user equipment device, e.g., using DCI signaling. In one embodiment, the transmission of the inference retraining data set can be semi-statically configured based on an inference retraining data set codebook, e.g., using RRC signaling and a reserved resource pool.

[0123] In another example embodiment, the inference training data set can be labeled or unlabeled to facilitate an AI / ML learning model leveraging supervised or unsupervised learning capabilities, respectively. Thus, the network RAN can transmit inference data samples along with the corresponding labels for each data sample. The data labels of the inference training data set can be dynamically configured according to the use case and the radio functions facilitated by the AI / ML learning model. The various labels can indicate different radio parameters or settings that can be adopted when transmitting the inference training data set.

[0124] Thus, the network RAN can transmit a set of inference training codebooks with an associated set of labels, or the network RAN can transmit a separate inference data set, or a set of labels for a previously transmitted corresponding inference data set. The latter case can be useful in certain AI / ML learning model deployments where the user equipment may have received inference data samples but may still be missing the associated set of labels. In such a case, the user equipment can transmit a request for the set of AI / ML learning model labels corresponding to the set of previously received inference samples to the serving cell RAN.

[0125] Now turning to Figure 6 , the figure shows a timing diagram of an example method embodiment 600 for configuring the user equipment 115 with a learning model management retraining configuration from the network node / RAN 105. At action 605, the user equipment 115 transmits learning model capabilities, similar to the transmission of learning model capabilities described in action 405 as shown in reference to Figure 4 . Continuing Figure 6 's description, based on the information transmitted by the UE 115 at action 605, at action 610, the RAN 105 transmits an AI / ML learning model management configuration to the UE 115, which can include a set, list, or codebook of available inference retraining data sets, such as Figure 5 's set 520 shown, where each data set is associated with one or more RF functions 515 and has a corresponding frequency resource size, time resolution, and time duration, such as represented by the information element 525 in Figure 5 . Continuing Figure 6In the description, at action 615, when the configured period for transmitting the AI model management request has expired or the reporting condition associated with triggering the transmission of the model management request is satisfied or met, UE 115 transmits an AI / ML learning model management request to UE 115 to obtain one or more AI / ML learning models active at the UE. The AI / ML model management request can be generated or transmitted according to a configured AI model management request format and can include an indication of a recommended inference retraining dataset. In an embodiment, the indication of the recommended inference retraining dataset can include a dataset index selected from a preconfigured list of datasets, indicating a request for immediate model retraining or inference retraining using the recommended dataset.

[0126] The AI / ML learning model management configuration transmitted at action 610 can include a format for the information used to generate or transmit the AI model management request, e.g., individual indications and current indications for each active model associated with a model identifier, summary indications and current indications for the active model identifier, or a predetermined or configured number of model history indications corresponding to each active model and the respective model identifier. The AI / ML learning model management configuration can include reporting conditions or criteria that, if satisfied, met, or otherwise triggered, can cause the UE 115 to generate or transmit a model management request. The reporting conditions can include one or more model parameter metrics, such as, for example, a reporting period or a learning model degradation criterion, which can be referred to as a model performance metric, e.g., a threshold, for triggering the transmission of a model management request when one or more criteria are not met.

[0127] The RAN node 105 receives the AI / ML learning model management request transmitted from the UE 115 at action 615 and, at action 620, the RAN 105 schedules and transmits a retraining dataset corresponding to the model identifier indicated in the request transmitted at action 615. The retraining dataset is transmitted from the RAN 105 to the UE 115 as part of the transmission of the control instruction. The UE 115 can perform retraining based on the received retraining dataset or other information transmitted at action 620.

[0128] Now turning to Figure 7, which shows a flowchart of an example method 700 for managing one or more learning models that can facilitate one or more corresponding radio functions. Method 700 begins at operation 705. At operation 710, the user equipment transmits information corresponding to one or more learning models used at the user equipment to the RAN node serving the user equipment to facilitate one or more corresponding radio functions. At operation 715, after receiving the learning model information transmitted at operation 710, the RAN generates a learning model management configuration based on the learning model information. At operation 720, the user equipment monitors metrics corresponding to learning model parameters. The parameters of the metrics monitored by the user equipment at operation 720 can include learning model parameters, such as statistical parameters as described in reference table 1, and not necessarily radio function metrics. Examples of radio function metrics can include signal strength, beam identifier, etc. In one embodiment, the user equipment may not determine control instructions, but may transmit the metrics monitored at operation 720 to the RAN at operation 740.

[0129] However, the user equipment may determine control instructions at operation 725. The user equipment may determine whether the metrics monitored at operation 720 meet the learning model performance metric criteria based on the metrics monitored at operation 720. If the monitored metrics reach or meet the criteria, such that for example the user equipment can potentially determine a recommended control instruction that can remedy the unsatisfactory performance by the learning model at the user equipment, then method 700 proceeds to operation 730. At operation 730, the user equipment may determine a control instruction request and transmit the control instruction request to the RAN at operation 735. After transmitting the control instruction request at operation 735, method 700 proceeds to operation 745.

[0130] At operation 745, the RAN may determine control instructions. If reaching operation 745 via operation 735, the RAN may evaluate the control instruction request transmitted from the user equipment at operation 735 and determine whether the implementation of the control operation will potentially mitigate the degraded performance of one or more learning models as determined by the user equipment at operation 730. For example, the RAN may know or have been informed of network conditions or other conditions that may not be relevant to the user equipment and that may have caused the control instruction determined by the user equipment at operation 730 to be beneficial to the user equipment, and the RAN may determine that the control instruction requested by the UE will be beneficial and may generate control instructions for the UE to implement the operation requested in the control operation request transmitted at operation 735.

[0131] Alternatively, the RAN may determine, at action 745, a control instruction that should implement a control instruction to reject the request transmitted at action 735. In other words, if the RAN determines at action 745 that the control action request generated at action 730 need not be executed because the request is unlikely to improve the performance of one or more learning models at the user equipment, the RAN may determine, at action 745, a control instruction to not change the operation of one or more learning models that the user equipment may be implementing.

[0132] If reaching action 745 via action 740, the RAN may determine the control instruction at action 745 based on an analysis of the metrics monitored at action 720 and transmitted at action 740 with respect to the learning model performance criteria.

[0133] After determining the control instruction at action 745, method 700 proceeds to step 750 and transmits the determined control instruction to the user equipment. At action 755, the user equipment may execute the control instruction, which, as described above, may include not performing any operation. Alternatively, the control instruction may include an instruction for the user equipment to deactivate one or more learning models corresponding to the metrics monitored at action 720. The control instruction may include an instruction to direct the user equipment to resume performing radio functions under or facilitated by the operation of a default learning model or a default deterministic model that does not include a learning model.

[0134] The RAN may also perform operations according to the control instruction. For example, the RAN may refresh, clear, or delete information, data, or configured values corresponding to one or more learning models at the user equipment, where the control instruction transmitted at action 750 includes a deletion instruction. The refresh performed at action 755 may include refreshing or deleting the inference assistance signaling process of the deactivated model according to the model identifier being executed by the UE. After performing the operation at action 755, method 700 proceeds to action 760 and ends.

[0135] The control instruction determined at action 745 may include an instruction to retrain the learning model whose metrics were monitored at action 720. The instruction to retrain the learning model may include a retraining data set, or an index that the user equipment may use to look up the retraining data set in a retraining codebook that may have been transmitted from the RAN to the user equipment (e.g., at action 715). If the control instruction determined at action 745 is for the user equipment to retrain one or more learning models, the user equipment retrains one or more learning models at action 755 based on the information (data set, codebook, or other data) transmitted by the RAN at action 750. After performing the operation at action 755, method 700 proceeds to action 760 and ends.

[0136] Now turn to Figure 8 , which shows the method 800 of an exemplary embodiment. At block 810, the method includes: receiving, by a user equipment including a processor, a model management indication configuration corresponding to a first radio function learning model of the user equipment from a radio access network node; at block 815, monitoring, by the user equipment, model performance parameters of a first radio function model configuration corresponding to the first radio function learning model that implements a radio function on the user equipment to obtain a monitored model performance metric; at block 820, analyzing the monitored model performance metric with respect to a model performance metric standard to obtain an analyzed monitored performance metric; at block 825, transmitting, by the user equipment, a model management indication to the radio access network node based on determining that the analyzed monitored performance metric does not meet the model performance metric standard according to the model management indication configuration; at block 830, receiving, by the user equipment, a control instruction corresponding to the operation of the radio function learning model based on the model management indication; and at block 835, performing at least one control operation according to the control instruction.

[0137] Now turn to Figure 9 , which shows the user equipment 900. At block 905, the user equipment includes a processor configured to: transmit an indication of radio function learning model information including learning model information corresponding to a radio function learning model to a radio access network node; at block 910, in response to the indication of the radio function learning model information, receive a model management indication configuration corresponding to the learning model information; at block 915, monitor model performance parameters of a first radio function model configuration that configures the radio function learning model to implement a radio function via the user equipment to obtain a monitored model performance metric; at block 920, transmit the monitored model performance metric to the radio access network node via a model management indication according to the model management indication configuration; at block 925, receive a control instruction from the radio access network node, where the control instruction is based on the monitored first model performance metric; at block 930, perform at least one control operation according to the control instruction; at block 935, where the at least one control operation includes: deactivating the radio function learning model configured with the first radio function model configuration; and at block 940, activating the radio function learning model configured with a second radio function model configuration to implement a radio function.

[0138] Now turn to Figure 10, which shows a non-transitory machine-readable medium 1000. At block 1005, the non-transitory machine-readable medium includes executable instructions that, when executed by a processor of a user equipment, facilitate the execution of operations, the operations including: monitoring model performance parameters corresponding to a first radio function model configuration of a radio function learning model that facilitates the execution of radio functions by the user equipment, the monitoring resulting in a monitored first model performance metric; at block 1010, analyzing the monitored first model performance metric with respect to a model performance metric criterion for the first radio function model configuration, the analysis resulting in an analyzed monitored performance metric; at block 1015, determining a control operation recommendation based on determining that the analyzed monitored performance metric does not meet the model performance metric criterion; and at block 1020, transmitting the control operation recommendation to a radio access network node according to a model management instruction configuration corresponding to the radio function learning model.

[0139] Now turn to Figure 11 , which shows an example embodiment method 1100. At block 1105, the method includes: receiving, by a radio access network node including a processor, learning model information corresponding to a current radio function learning model of a user equipment from the user equipment; at block 1110, generating, by the radio access network node, a model management instruction configuration based on the learning model information; at block 1115, transmitting, by the radio access network node, the model management instruction configuration to the user equipment; at block 1120, receiving, by the radio access network node, a performance parameter metric corresponding to a learning model performance metric parameter and associated with the performance of the current radio function learning model from the user equipment; at block 1125, transmitting, by the radio access network node, a control instruction corresponding to the operation of the current radio function learning model by the user equipment to the user equipment based on the performance parameter metric; and at block 1130, where the control instruction includes an instruction to deactivate the current radio function learning model and activate a default radio function.

[0140] Now turn to Figure 12, the figure shows an example system 1200, at block 1205, the system includes computer-executable components of a communication network node, the computer-executable components include a processor configured to: receive a first indication of first radio function learning model information including the first learning model information corresponding to the first radio function learning model of the first user equipment from the first user equipment; at block 1210, generate a first model management indication configuration corresponding to the first radio function learning model; at block 1215, transmit the first model management indication configuration to the first user equipment; at block 1220, receive a first performance parameter metric corresponding to the first learning model performance metric parameter associated with the performance of the first radio function learning model from the first user equipment; at block 1225, transmit a first control instruction corresponding to the operation of the first radio function learning model by the first user equipment to the first user equipment based on the first performance parameter metric; and at block 1230, wherein the first control instruction includes a first retraining instruction for retraining the first radio function learning model to obtain a retrained first radio function learning model, and wherein the processor is further configured to: transmit a first retraining data set for use by the first user equipment to retrain the first radio function learning model.

[0141] Now turning to Figure 13 , the figure shows a non-transitory machine-readable medium 1300, at block 1305, the non-transitory machine-readable medium includes executable instructions that, when executed by a processor of a radio access network node of a communication network, facilitate the execution of operations including: receiving a first radio function learning model information indication including learning model information corresponding to a first radio function learning model of a user equipment from the user equipment; at block 1310, generating a first model management indication configuration corresponding to the first radio function learning model; at block 1315, transmitting the first model management indication configuration to the user equipment; at block 1320, receiving a first model management indication from the user equipment according to the first model management indication configuration, the first model management indication including a first performance parameter metric corresponding to the first learning model performance metric parameter associated with the performance of the first radio function learning model; and at block 1325, transmitting a first control instruction corresponding to the operation of the first radio function learning model by the user equipment to the user equipment based on the first performance parameter metric.

[0142] To provide additional context for the various embodiments described herein, Figure 14The following discussion is intended to provide a brief general description of a suitable computing environment 1400 in which various embodiments of the embodiments described herein may be implemented. While the embodiments have been described above in the general context of computer-executable instructions that may operate on one or more computers, those skilled in the art will recognize that these embodiments may also be implemented in combination with other program modules and / or as a combination of hardware and software.

[0143] Generally, program modules include routines, programs, components, data structures, etc. that perform particular tasks or implement particular abstract data types. In addition, those skilled in the art will appreciate that these methods may also be practiced with other computer system configurations, including single-processor or multi-processor computer systems, minicomputers, mainframe computers, IoT devices, distributed computing systems, and personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, etc., each of which may be operably coupled to one or more associated devices.

[0144] The embodiments illustrated herein may also be practiced in a distributed computing environment where certain tasks are performed by remote processing devices linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

[0145] Computing devices generally include a variety of media, which may include computer-readable storage media, machine-readable storage media, and / or communication media, the difference in the use of these two terms is described below. Computer-readable storage media or machine-readable storage media can be any available storage media accessible by a computer and includes both volatile and non-volatile media, removable and non-removable media. For example, but not limited to, computer-readable storage media or machine-readable storage media can be implemented in combination with any method or technology for storing information such as computer-readable or machine-readable instructions, program modules, structured data, or unstructured data.

[0146] A computer-readable storage medium may include, but is not limited to, random access memory (RAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technologies, compact disc read only memory (CD-ROM), digital versatile disc (DVD), Blu-ray disc (BD) or other optical disc storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, solid state drives or other solid state storage devices, or other tangible and / or non-transitory media that can be used to store the desired information. In this regard, the terms "tangible" or "non-transitory" as applied to storage, memory or computer-readable media herein should be understood as a modifier that only excludes propagating transitory signals per se, without disclaiming the right to all standard storage, memory or computer-readable media that are not only propagating transitory signals per se.

[0147] A computer-readable storage medium can be accessed by one or more local or remote computing devices, such as via an access request, query, or other data retrieval protocol, to perform various operations with respect to the information stored by the medium.

[0148] A communication medium typically embodies computer-readable instructions, data structures, program modules, or other structured or unstructured data in a data signal (such as a modulated data signal, e.g., a carrier wave or other transmission mechanism), and includes any information delivery or transmission medium. The term "modulated data signal" or signal refers to a signal in which one or more of its characteristics are set or changed in such a manner as to encode information in the signal. By way of example, and not limitation, communication media include wired media (such as a wired network or direct-wired connection) and wireless media (such as acoustic waves, RF, infrared, and other wireless media).

[0149] Referring again to Figure 14 , an example environment 1400 for implementing various embodiments of the aspects described herein includes a computer 1402 that includes a processing unit 1404, a system memory 1406, and a system bus 1408. The system bus 1408 couples system components, including but not limited to the system memory 1406, to the processing unit 1404. The processing unit 1404 can be any of a variety of commercially available processors and can include cache memory. Dual microprocessors and other multi-processor architectures can also be employed as the processing unit 1404.

[0150] The system bus 1408 can be any of several types of bus structures, which can be further interconnected to a memory bus (with or without a memory controller), a peripheral bus, and a local bus using any of a variety of commercially available bus architectures. The system memory 1406 includes a ROM 1410 and a RAM 1412. The basic input / output system (BIOS) can be stored in non-volatile memory (such as ROM, erasable programmable read-only memory (EPROM), and electrically erasable programmable read-only memory (EEPROM)), and the BIOS contains basic routines that help transfer information between components within the computer 1402 (such as during startup). The RAM 1412 can also include high-speed RAM, such as static RAM for caching data.

[0151] The computer 1402 also includes an internal hard disk drive (HDD) 1414 (e.g., EIDE, SATA), one or more external storage devices 1416 (e.g., a magnetic floppy disk drive (FDD) 1416, a memory stick or flash drive reader, a memory card reader, etc.), and an optical disc drive 1420 (e.g., which can read or write CD-ROM discs, DVDs, BDs, etc.). Although the internal HDD 1414 is shown as being located within the computer 1402, the HDD 1414 can also be configured for external use in a suitable chassis (not shown). Additionally, although not shown in the environment 1400, a solid-state drive (SSD) can be used as an addition to or an alternative to the HDD 1414. The HDD 1414, the external storage device(s) 1416, and the optical disc drive 1420 can be connected to the system bus 1408 via an HDD interface 1424, an external storage interface 1426, and an optical disc drive interface 1428, respectively. The interface 1424 for external drive implementation can include at least one or both of a universal serial bus (USB) and Institute of Electrical and Electronics Engineers (IEEE) 1394 interface technologies. Other external drive connection technologies are also contemplated within the embodiments described herein.

[0152] The drives and their associated computer-readable storage media provide non-volatile storage of data, data structures, computer-executable instructions, and the like. For the computer 1402, the drives and storage media can store any data in a suitable digital format. Although the above description of computer-readable storage media refers to various types of storage devices, those skilled in the art should appreciate that other types of computer-readable storage media, whether currently existing or to be developed in the future, can be used in the exemplary operating environment, and further, any such storage media can contain computer-executable instructions for performing the methods described herein.

[0153] Multiple program modules can be stored in the drive and RAM 1412, including an operating system 1430, one or more application programs 1432, other program modules 1434, and program data 1436. All or part of the operating system, application programs, modules, and / or data can also be cached in RAM 1412. The systems and methods described herein can be implemented using a variety of commercially available operating systems or combinations of operating systems.

[0154] Computer 1402 can optionally include emulation technology. For example, a hypervisor (not shown) or other intermediary can emulate a hardware environment for the operating system 1430, and the emulated hardware can optionally be different from the Figure 14 hardware shown in. In such an embodiment, the operating system 1430 can include one VM among multiple virtual machines (VMs) hosted at the computer 1402. Additionally, the operating system 1430 can provide an operating-time environment for the application 1432, such as a JAVA operating-time environment or a.NET framework. The operating-time environment is a consistent execution environment that allows the application 1432 to operate on any operating system that includes the operating-time environment. Similarly, the operating system 1430 can support containers, and the application 1432 can be in the form of a container, which is a lightweight, independent, executable software package that includes, for example, the code of the application, the operating-time, system tools, system libraries, and settings.

[0155] Furthermore, the computer 1402 can include a security module, such as a Trusted Platform Module (TPM). For example, using the TPM, the boot component hashes the next boot component in time and waits for the result to match a security value before loading the next boot component. This process can be performed at any layer in the code execution stack of the computer 1402, for example, applied at the application execution layer or the operating system (OS) kernel layer, thus achieving security at any code execution layer.

[0156] A user may input commands and information into computer 1402 through one or more wired / wireless input devices (such as keyboard 1438, touch screen 1440, and pointing devices (such as mouse 1442)). Other input devices (not shown) may include a microphone, an infrared (IR) remote control, a radio (RF) remote control or other remote controls, a joystick, a virtual reality controller and / or a virtual reality headset, a gamepad, a stylus, an image input device (such as one or more cameras), a gesture sensor input device, a vision motion sensor input device, an emotion or face detection device, a biometric input device (such as a fingerprint or iris scanner), etc. These and other input devices are typically connected to processing unit 1404 through an input device interface 1444 that may be coupled to system bus 1408, but may also be connected through other interfaces, such as a parallel port, an IEEE 1394 serial port, a game port, a USB port, an IR interface, interfaces, etc.

[0157] Monitor 1446 or other types of display devices may also be connected to system bus 1408 via an interface (such as video adapter 1448). In addition to monitor 1446, a computer typically also includes other peripheral output devices (not shown), such as speakers, printers, etc.

[0158] Computer 1402 may operate in a networking environment with a logical connection to one or more remote computers (such as one or more remote computers 1450) via wired and / or wireless communication. The one or more remote computers 1450 may be workstations, server computers, routers, personal computers, portable computers, microprocessor-based entertainment appliances, peer devices, or other common network nodes, and typically include many or all of the elements described in relation to computer 1402, but for the sake of brevity, only memory / storage device 1452 is shown. The depicted logical connections include wired / wireless connections to a local area network (LAN) 1454 and / or a larger network (such as a wide area network (WAN) 1456). Such LAN and WAN networking environments are common in offices and companies and facilitate enterprise-level computer networks (such as intranets), all of which may be connected to a global communication network, such as the Internet.

[0159] When used in a LAN networking environment, computer 1402 may be connected to local network 1454 through a wired and / or wireless communication network interface or adapter 1458. Adapter 1458 may facilitate wired or wireless communication with LAN 1454, which may also include a wireless access point (AP) disposed thereon for communicating with adapter 1458 in a wireless mode.

[0160] When used in a WAN networking environment, computer 1402 may include a modem 1460 or may be connected to a communication server on WAN 1456 via other means (such as via the Internet) to establish communication through WAN 1456. The modem 1460, which may be an internal or external or wired or wireless device, may be connected to the system bus 1408 via the input device interface 1444. In a networking environment, program modules associated with the computer 1402 or portions thereof may be stored in the remote memory / storage device 1452. It will be appreciated that the network connections shown are examples and other means of establishing communication links between computers may be used.

[0161] When used in a LAN or WAN networking environment, in addition to or in place of the external storage device 1416 described above, computer 1402 may access a cloud storage system or other network-based storage system. Generally, the connection between the computer 1402 and the cloud storage system may be established through LAN 1454 or WAN 1456, for example, via adapter 1458 or modem 1460 respectively. After connecting the computer 1402 to the associated cloud storage system, the external storage interface 1426 may, with the help of adapter 1458 and / or modem 1460, manage the storage provided by the cloud storage system in the same way as other types of external storage. For example, the external storage interface 1426 may be configured to provide access to cloud storage sources as if these sources were physically connected to the computer 1402.

[0162] Computer 1402 may communicate with any wireless device or entity operably set up for wireless communication, such as printers, scanners, desktop and / or portable computers, portable data assistants, communication satellites, any device or location associated with a wirelessly detectable tag (e.g., self-service terminals, newsstands, store shelves, etc.) and telephones. This may include Wi-Fi and wireless technologies. Thus, the communication may be of a predefined structure like a conventional network or merely a point-to-point communication between at least two devices.

[0163] Go to Figure 15, which shows a block diagram of an example UE 1560. The UE 1560 can include a smart phone, a wireless tablet, a laptop computer with wireless capabilities, a wearable device, a machine device that can facilitate vehicle telematics, a tracking device, a remote sensing device, etc. The UE 1560 includes a first processor 1530, a second processor 1532, and a shared memory 1534. The UE 1560 includes a radio front-end circuit 1562, which may be referred to herein as a transceiver, but is understood to generally include transceiver circuitry, separate filters, and separate antennas for facilitating transmission and reception of signals over a wireless link (such as Figure 1 one or more of the wireless links 125, 135, and 137 shown in

[0164] Continuing Figure 15 the description, the UE 1560 may also include a SIM 1564 or a SIM profile, which may include information stored in a memory (memory 34 or a separate memory section) for facilitating wireless communication with Figure 1 the RAN 105 or the core network 130 shown in Figure 15 The SIM 1564 is shown as a single component having the shape of a conventional SIM card, but it will be appreciated that the SIM 1564 may represent multiple SIM cards, multiple SIM profiles, or multiple eSIMs, some or all of which may be implemented in hardware or software. It will be appreciated that the SIM profile may include information such as security credentials (e.g., encryption keys, values that can be used to generate encryption keys, or shared values shared between the SIM 1564 and another device, which may be Figure 1 a component of the RAN 105 or the core network 130 shown in

[0165] The SIM 1564 is shown as being coupled to both the first processor portion 1530 and the second processor portion 1532. An advantage that this implementation can provide is that the first processor portion 1530 can be without the need to request or receive from the SIM 1564 the information or data that the second processor 1532 may request, thereby eliminating the use of the first processor acting as a "middleman" when the second processor uses information from the SIM to perform its functions and execute applications. The first processor 1530 (which may be a modem processor or a baseband processor) is shown as being smaller than the processor 1532 (which may be a more complex application processor) to visually indicate the relative complexity levels (i.e., processing capabilities and performance) between the two processor portions and the corresponding relative levels of operating power consumption. Keeping the second processor portion 1532 in a sleep / inactive / low-power state when the UE 1560 does not require the second processor portion 1532 to execute applications and process application-related data provides the following advantage: reducing power consumption when the UE only needs to use the first processor portion 1530 in a listening mode to monitor the bearer management and mobility management / maintenance processes configured conventionally or to monitor the search space that the UE has been configured to monitor while the second processor portion remains inactive / sleeping.

[0166] The UE 1560 may also include sensors 1566, such as, for example, a temperature sensor, an accelerometer, a gyroscope, a barometer, a humidity sensor, etc., which may provide signals to the first processor 1530 or the second processor 1532. The output device 1568 may include, for example, one or more visual displays (e.g., a computer monitor, a VR device, etc.), acoustic transducers (such as a speaker or a microphone), a vibration component, etc. The output device 1568 may include software that interfaces with output devices external to the UE 1560 (e.g., a visual display, a speaker, a microphone, a tactile device, an olfactory or gustatory device, etc.).

[0167] The following glossary given in Table 2 may apply to the description of one or more embodiments disclosed herein.

[0168]

[0169]

[0170]

[0171] Table 2

[0172] The above description includes non - limiting examples of various embodiments. Of course, to describe the disclosed subject matter, it is not possible to describe every conceivable combination of components or methods, and one of ordinary skill in the art will recognize that further combinations and permutations of the various embodiments are possible. The disclosed subject matter is intended to cover all such changes, modifications, and variations that fall within the spirit and scope of the appended claims.

[0173] Regarding the various functions performed by the above - described components, devices, circuits, systems, etc., the terms used to describe such components (including references to "means") are intended to also include any (multiple) structures that perform the specified function of the described component (e.g., functional equivalents), even if not structurally equivalent to the disclosed structure. Additionally, although a particular feature of the disclosed subject matter may be disclosed only with respect to one of several implementations, such a feature may be combined with one or more other features of other implementations in accordance with the needs and advantages of any given or particular application.

[0174] As used herein, the term "exemplary" and / or "illustrative" or variations thereof are intended to mean serving as an example, instance, or illustration. To avoid doubt, the disclosed subject matter is not limited by such examples. Additionally, any aspect or design described herein as "exemplary" and / or "illustrative" should not necessarily be understood as being superior to or more advantageous than other aspects or designs, nor is it intended to exclude equivalent structures and techniques known to those of ordinary skill in the art. Further, when used in the detailed description or claims, the terms "comprising", "having", "including", and other similar words are intended to be inclusive - in a manner similar to the term "including" as an open - ended transitional word - and do not exclude any additional or other elements.

[0175] As used herein, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or". For example, the phrase "A or B" is intended to include instances of A, B, and both A and B. Additionally, as used in this application and the appended claims, the words "a" and "an" should generally be understood to mean "one or more" unless otherwise stated or clearly indicated from the context to be in the singular form.

[0176] As used herein, the term "set" does not include the empty set, i.e., a set that contains no elements. Thus, a "set" in this disclosure includes one or more elements or entities. Similarly, as used herein, the term "group" refers to a collection of one or more entities.

[0177] Unless otherwise clearly indicated by the context, the terms "first", "second", "third", etc. as used in the claims are for clarity only and do not otherwise indicate or imply any temporal order. For example, "first determination", "second determination", and "third determination" do not indicate or imply that the first determination should be made before the second determination, and vice versa, and so on.

[0178] The description of the illustrated embodiments of the present disclosure provided herein (including that described in the abstract) is not intended to be exhaustive or to limit the disclosed embodiments to the precise forms disclosed. While specific embodiments and examples are described herein for illustrative purposes, various modifications that are considered to be within the scope of such embodiments and examples can be made as would be recognized by those of ordinary skill in the art. In this regard, although the subject matter has been described herein in connection with various embodiments and the corresponding drawings, it should be understood that, where applicable, other similar embodiments can be used, or modifications and additions can be made to the described embodiments to perform the same, similar, alternative, or substitute functions as the disclosed subject matter without departing from the disclosed subject matter. Accordingly, the disclosed subject matter should not be limited to any single embodiment described herein, but should be construed in its breadth and scope in accordance with the appended claims.

Claims

1. A method, comprising: receiving, by a radio access network node including a processor, learning model information corresponding to a current radio function learning model of a user equipment from the user equipment; generating, by the radio access network node, a model management indication configuration based on the learning model information; transmitting, by the radio access network node, the model management indication configuration to the user equipment; receiving, by the radio access network node, from the user equipment a performance parameter metric corresponding to a learning model performance metric parameter and associated with the performance of the current radio function learning model; and transmitting, by the radio access network node, to the user equipment a control instruction corresponding to an operation of the current radio function learning model by the user equipment based on the performance parameter metric.

2. The method according to claim 1, wherein the performance parameter metric corresponds to a learning model performance metric parameter.

3. The method according to claim 2, wherein the learning model performance metric parameter includes at least one of the following: mean squared error value, root mean squared error value, normalized mean squared error value, mean absolute error value, R-squared value, generalized cosine similarity value, squared generalized cosine similarity value, accuracy value, number of true negative values, number of true positive values, number of false negative values, number of false positive values, precision value, recall value, or F1-score value.

4. The method according to claim 1, wherein the learning model information includes at least one of the following: at least one learning model type indication, at least one radio function indication indicating a corresponding at least one radio function, at least one metric to be estimated or reported corresponding to at least one learning model corresponding to the at least one learning model type indication, the number of learning models that the user equipment can store for the at least one radio function, or an indication of at least one learning model category corresponding to at least one data set to be used by the radio access network node to determine the performance parameter metric.

5. The method according to claim 1, wherein the model management indication includes a data size based at least on the performance parameter metric.

6. The method according to claim 1, wherein the performance parameter metric is a configured quantized performance parameter metric value.

7. The method according to claim 6, wherein the model management indication configuration includes the quantized performance parameter metric value.

8. The method according to claim 1, wherein the performance parameter metric is a preamble corresponding to the user equipment.

9. The method according to claim 1, wherein the control instruction includes an instruction to deactivate the current radio function learning model and an instruction to activate a default radio function.

10. The method according to claim 1, wherein the control instruction includes an instruction to train the current radio function learning model to obtain an updated radio function learning model.

11. The method according to claim 10, wherein the control instruction includes a recommended configured training dataset indication indicating a configured training dataset to be used to train the current radio function learning model.

12. The method according to claim 1, further comprising: analyzing the performance parameter metric with respect to a performance parameter metric standard to obtain an analyzed performance parameter metric; determining that the analyzed performance parameter metric fails to meet the performance parameter metric standard; and determining to transmit the control instruction based on the performance parameter metric being determined not to meet the performance parameter metric standard.

13. The method according to claim 1, wherein the model management indication includes a request from the user equipment to transmit the control instruction.

14. A system, comprising: a computer-executable component of a communication network node including a processor, the processor being configured to: receive a first indication of first radio function learning model information including first learning model information corresponding to a first radio function learning model of the first user equipment; generate a first model management indication configuration corresponding to the first radio function learning model; transmit the first model management indication configuration to the first user equipment; receive a first performance parameter metric associated with the performance of the first radio function learning model and corresponding to a first learning model performance metric parameter from the first user equipment; and transmit a first control instruction corresponding to the operation of the first radio function learning model by the first user equipment to the first user equipment based on the first performance parameter metric.

15. The system according to claim 14, wherein the first control instruction includes a first retraining instruction for retraining the first radio function learning model to obtain a retrained first radio function learning model, and wherein the processor is further configured to: transmit a first retraining dataset to be used by the first user equipment to retrain the first radio function learning model.

16. The system according to claim 14, wherein the processor is further configured to: analyze the first performance parameter metric with respect to a first performance parameter metric standard to obtain an analyzed first performance parameter metric; determine that the analyzed first performance parameter metric fails to meet the first performance parameter metric standard; and determine to transmit the first control instruction based on the analyzed first performance parameter metric being determined not to meet the first performance parameter metric standard.

17. The system according to claim 14, wherein the processor is further configured to: receive a second indication of second radio function learning model information including learning model information corresponding to a second radio function learning model of a second user equipment; generate a second model management indication configuration corresponding to the second radio function learning model; transmit the second model management indication configuration to the second user equipment; Receiving, from the second user equipment, a second performance parameter metric corresponding to a second learning model performance metric parameter and associated with the performance of the second radio function learning model; And Transmitting, based on the second performance parameter metric, a second control instruction corresponding to the operation of the second radio function learning model by the second user equipment, Wherein the first radio function learning model corresponds to a first learning model type, wherein the second radio function learning model corresponds to a second learning model type, wherein the first learning model type and the second learning model type are of the same type, wherein the first learning model type operates according to a first learning model configuration, and the second learning model type operates according to a second learning model configuration, and wherein the first learning model configuration is different from the second learning model configuration.

18. A non-transitory machine-readable medium comprising executable instructions that, when executed by a processor of a radio access network node of a communication network, facilitate the execution of operations including: Receiving, from a user equipment, a first radio function learning model information indication including learning model information corresponding to a first radio function learning model of the user equipment; Generating a first model management indication configuration corresponding to the first radio function learning model; Transmitting the first model management indication configuration to the user equipment; Receiving, according to the first model management indication configuration, a first model management indication from the user equipment and including a first performance parameter metric corresponding to a first learning model performance metric parameter and associated with the performance of the first radio function learning model; And Transmitting, based on the first performance parameter metric, a first control instruction corresponding to the operation of the first radio function learning model by the user equipment.

19. The non-transitory machine-readable medium according to claim 18, wherein the operations further include: Receiving, from the user equipment, a second radio function learning model information indication including learning model information corresponding to a second radio function learning model of the user equipment; Generating a second model management indication configuration corresponding to the second radio function learning model; Transmitting the second model management indication configuration to the user equipment; Receiving a first model management indication from the user equipment according to the first model management indication configuration; Receiving, from the user equipment, a second performance parameter metric corresponding to a second learning model performance metric parameter and associated with the performance of the second radio function learning model; And Transmitting, based on the second performance parameter metric, a second control instruction corresponding to the operation of the second radio function learning model by the user equipment, Wherein the first radio function learning model and the second radio function learning model respectively correspond to a first radio function and a second radio function, and wherein the first radio function is a radio function different from the second radio function.

20. The non-transitory machine-readable medium according to claim 19, wherein the first model management indication and the second model management indication include different data sizes.

Images