Techniques for beam management using model adaptation

Abstract

Methods, systems, and devices for wireless communication are described. A user equipment (UE) can be configured to receive, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting output from the machine learning model. The UE can perform a first set of measurements on a set of reference signals received from the network entity and predict a first set of beam quality metrics associated with a set of received beams at the UE based on inputting the first set of measurements into the machine learning model. The UE can then send a control message indicating the first set of measurements to the network entity based on the first set of beam quality metrics satisfying the one or more thresholds.

Description

Technical Field

[0001] The following relates to wireless communication, including techniques for beam management using model adaptation. Background Art

[0002] Wireless communication systems are widely deployed to provide various types of communication content, such as voice, video, packet data, messaging, broadcasting, and so on. These systems may be capable of supporting communication with multiple users by sharing available system resources (e.g., time, frequency, and power). Examples of such multi-access systems include fourth-generation (4G) systems (such as Long-Term Evolution (LTE) systems, LTE-Advanced (LTE-A) systems, or LTE-A Pro systems) and fifth-generation (5G) systems (which may be referred to as New Radio (NR) systems). These systems may employ techniques such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal FDMA (OFDMA), or Discrete Fourier Transform Spread Orthogonal Frequency Division Multiplexing (DFT-S-OFDM). A wireless multi-access communication system may include one or more base stations, each of which supports wireless communication for communication devices, which may be referred to as User Equipment (UE).

[0003] Some wireless devices (e.g., UE) may be configured to perform measurements on reference signals from the network and input these measurements into a machine learning algorithm to predict the relative quality of the receive (Rx) beam to be applied to future communications. The machine learning model may not be able to adapt as the network and environment change and may result in inaccurate beam predictions. Summary of the Invention

[0004] The described techniques relate to improved methods, systems, devices, and apparatuses that support techniques for beam management using model adaptation. Generally, aspects of the present disclosure relate to techniques for efficiently training a machine learning model that is used to perform beam stealing. Specifically, aspects of the present disclosure relate to signaling and configuration that enable a User Equipment (UE) to determine when to report beam measurements to be used for offline model training and to determine when the output of the machine learning model is available for online model training. For example, the network may configure the UE with a machine learning model to be used for beam prediction at the UE. The network may configure the UE with a first set of thresholds and / or a second set of thresholds, where the first set of thresholds is used to determine which beam measurements are to be reported to the network for offline training, and the second set of thresholds is used to determine which beam measurements and model outputs can be used by the UE for online model training. In this example, the UE may input the measurements performed by the UE into the machine learning model to predict a beam quality metric and report these measurements to the network for offline training or use these measurements to perform online training based on whether the predicted beam quality metric meets the corresponding thresholds.

[0005] Describes a method. The method may include: receiving, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; performing a first set of measurements on a set of reference signals received from the network entity; predicting a first set of beam quality metrics associated with a set of receive (Rx) beams at the UE based on inputting the first set of measurements into the machine learning model; and sending, to the network entity, a control message indicating the first set of measurements based on the first set of beam quality metrics meeting the one or more thresholds.

[0006] Describes a device. The device may include: a processor; a memory coupled to the processor; and instructions stored in the memory. The instructions are executable by the processor to cause the device to: receive, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; perform a first set of measurements on a set of reference signals received from the network entity; predict a first set of beam quality metrics associated with a set of Rx beams at the UE based on inputting the first set of measurements into the machine learning model; and send, to the network entity, a control message indicating the first set of measurements based on the first set of beam quality metrics meeting the one or more thresholds.

[0007] Describes another device. The device may include: means for receiving, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; means for performing a first set of measurements on a set of reference signals received from the network entity; means for predicting a first set of beam quality metrics associated with a set of Rx beams at the UE based on inputting the first set of measurements into the machine learning model; and means for sending, to the network entity, a control message indicating the first set of measurements based on the first set of beam quality metrics meeting the one or more thresholds.

[0008] Describes a non-transitory computer-readable medium storing code. The code may include instructions executable by a processor to perform the following operations: receive control signaling from a network entity indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; perform a first set of measurements on a set of reference signals received from the network entity; predict a first set of beam quality metrics associated with a set of Rx beams at the UE based on inputting the first set of measurements into the machine learning model; and send a control message indicating the first set of measurements to the network entity based on the first set of beam quality metrics satisfying the one or more thresholds.

[0009] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: receive additional control signaling from the network entity indicating an updated version of the machine learning model, where the additional control signaling may be received based on sending the control message.

[0010] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the first set of beam quality metrics includes a set of probability metrics indicating the relative probability that a corresponding beam in the set of Rx beams meets a threshold beam performance, and the first set of beam quality metrics satisfies the one or more thresholds based on each beam quality metric in the first set of beam quality metrics being less than the one or more thresholds.

[0011] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: perform a second set of measurements on a second set of reference signals received from the network entity; and predict a second set of beam quality metrics associated with the set of Rx beams based on the second set of measurements, where the first set of beam quality metrics satisfies the one or more thresholds based on one or more differences between the first set of beam quality metrics and the second set of beam quality metrics being greater than or equal to the one or more thresholds.

[0012] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the first set of beam quality metrics and the second set of beam quality metrics include predicted reference signal received power (RSRP) measurements associated with the set of Rx beams, and the one or more thresholds include a threshold RSRP metric.

[0013] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the second set of beam quality metrics includes RSRP metrics that may be predicted using one or more mathematical operations different from the machine learning model.

[0014] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the one or more differences between the first set of beam quality metrics and the second set of beam quality metrics include a minimum mean square error (MMSE) metric.

[0015] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: performing a second set of measurements on a second set of reference signals received from the network entity; predicting a second set of beam quality metrics associated with the set of Rx beams based on inputting the second set of measurements into the machine learning model; and training the machine learning model using the second set of measurements based on at least one beam quality metric of the second set of beam quality metrics satisfying the one or more additional thresholds.

[0016] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: generating a pseudo-label associated with the at least one beam quality metric of the second set of beam quality metrics that satisfies the one or more additional thresholds; and inputting the pseudo-label, the at least one beam quality metric of the second set of beam quality metrics, and at least one measurement of the second set of measurements corresponding to the at least one beam quality metric into the machine learning model, wherein the training may be based on the input.

[0017] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the second set of beam quality metrics includes a set of probability metrics indicating the relative probability that a corresponding beam in the set of Rx beams meets a threshold beam performance, and based on the at least one beam quality metric being greater than or equal to the one or more additional thresholds, the at least one beam quality metric of the second set of beam quality metrics satisfies the one or more additional thresholds.

[0018] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: predicting a first set of beam indices corresponding to the first set of beam quality metrics, the first set of beam indices indicating predicted Rx beams associated with a set of multiple time instances; and selectively modifying the beam indices in the first set of beam indices based on one or more differences between the beam indices and one or more additional beam indices corresponding to one or more of the multiple time instances within a consistency window.

[0019] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: receiving an indication of the consistency window via the control signaling, wherein selectively modifying the beam indices may be based on receiving the control signaling.

[0020] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, based on the at least one predicted beam quality metric satisfying an outlier threshold, at least one predicted beam quality metric of the set of multiple predicted beam quality metrics is selectively modified during the second time interval.

[0021] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: selecting an Rx beam from the set of Rx beams based on the first set of beam quality metrics; and based on the selection, receiving one or more messages from the network entity using the Rx beam.

[0022] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: sending, to the network entity, capability signaling indicating one or more capabilities associated with the UE, wherein receiving the control signaling may be based on the capability signaling.

[0023] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: performing a second set of measurements on a second set of reference signals received from the network entity; determining a second set of beam quality metrics associated with the set of Rx beams, a set of additional Rx beams, or both, based on inputting the second set of measurements into the machine learning model; and avoiding reporting the second set of measurements to the network entity based on at least one beam quality metric of the second set of beam quality metrics failing to meet the one or more thresholds.

[0024] A method is described. The method may include: sending, to a UE, control signaling indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; sending, based on the control signaling, a set of reference signals to the UE; and receiving, from the UE, a control message indicating a first set of measurements associated with the set of reference signals, wherein the control message is received based on a first set of beam quality metrics associated with the first set of measurements satisfying the one or more thresholds, wherein the first set of beam quality metrics includes the output of the machine learning model.

[0025] A device is described. The device may include: a processor; a memory coupled to the processor; and instructions stored in the memory. The instructions are executable by the processor to cause the device to: send control signaling to a UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; send a set of reference signals to the UE based on the control signaling; and receive a control message from the UE indicating a first set of measurements associated with the set of reference signals, wherein the control message is received based on a first set of beam quality metrics associated with the first set of measurements meeting the one or more thresholds, wherein the first set of beam quality metrics includes the output of the machine learning model.

[0026] Another device is described. The device may include: means for sending control signaling to a UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; means for sending a set of reference signals to the UE based on the control signaling; and means for receiving a control message from the UE indicating a first set of measurements associated with the set of reference signals, wherein the control message is received based on a first set of beam quality metrics associated with the first set of measurements meeting the one or more thresholds, wherein the first set of beam quality metrics includes the output of the machine learning model.

[0027] A non-transitory computer-readable medium storing code is described. The code may include instructions executable by a processor to: send control signaling to a UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; send a set of reference signals to the UE based on the control signaling; and receive a control message from the UE indicating a first set of measurements associated with the set of reference signals, wherein the control message is received based on a first set of beam quality metrics associated with the first set of measurements meeting the one or more thresholds, wherein the first set of beam quality metrics includes the output of the machine learning model.

[0028] Some examples of the methods, devices, and non-transitory computer-readable media described herein may further include operations, features, means, or instructions for performing the following actions: training the machine learning model based on the first set of measurements, the first set of beam quality metrics, or both; and sending additional control signaling to the UE indicating an updated version of the machine learning model based on training the machine learning model.

[0029] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the first set of beam quality metrics includes a set of probability metrics indicating the relative probability that a corresponding beam in the set of Rx beams meets a threshold beam performance, and the first set of beam quality metrics meets the one or more thresholds based on each beam quality metric in the first set of beam quality metrics being less than the one or more thresholds.

[0030] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the first set of beam quality metrics meets the one or more thresholds based on one or more differences between the first set of beam quality metrics and a second set of beam quality metrics being greater than or equal to the one or more thresholds.

[0031] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the first set of beam quality metrics and the second set of beam quality metrics include predicted RSRP measurements associated with the set of Rx beams, and the one or more thresholds include a threshold RSRP metric.

[0032] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the second set of beam quality metrics includes RSRP metrics that can be predicted using one or more mathematical operations different from the machine learning model.

[0033] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the one or more differences between the first set of beam quality metrics and the second set of beam quality metrics include an MMSE metric.

[0034] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, the control signaling further indicates one or more additional thresholds for training the machine learning model at the UE.

[0035] Some examples of the methods, apparatuses, and non-transitory computer-readable media described herein may further include operations, features, components, or instructions for performing the following actions: receiving, from the UE, capability signaling indicating one or more capabilities associated with the UE, wherein transmitting the control signaling may be based on the capability signaling. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Examples of wireless communication systems that illustrate techniques supporting beam management using model adaptation in accordance with one or more aspects of the present disclosure are shown.

[0037] Figure 2 Examples of wireless communication systems that illustrate techniques supporting beam management using model adaptation in accordance with one or more aspects of the present disclosure are shown.

[0038] Figure 3 An example of a machine learning model configuration that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated.

[0039] Figure 4 An example of a process flow that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated.

[0040] Figure 5 An example of a process flow that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated.

[0041] Figure 6 and Figure 7 A block diagram of a device that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated.

[0042] Figure 8 A block diagram of a communication manager that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated.

[0043] Figure 9 A diagram of a system that includes a device that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated.

[0044] Figures 10 to 13 A flowchart of a method that shows techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated. Detailed Description

[0045] Some wireless devices (e.g., user equipment (UE)) may be configured to perform measurements on reference signals from a network and input those measurements into a machine learning algorithm to predict the relative quality of receive (Rx) beams to be applied to future communications. Machine learning models may not be able to adapt as the network and environment change and may result in inaccurate beam predictions. In such cases, the model may need to be further trained, which includes offline training and online training. However, both of these training methods have their respective drawbacks.

[0046] For example, using offline training, the UE can report measurements to the network so that the network can further "offline" train the model (e.g., when the model is not in use), where the network then reconfigures the UE with the newly trained model. In such cases, offline training may require a large amount of signaling overhead to report measurements to the network, and offline training may result in increased beam prediction latency. In comparison, using online training, the UE can select an Rx beam prediction from a machine learning model and use the Rx beam prediction as an input to further "online" train the model (e.g., while the UE continues to use the model). However, online training techniques may not be effective when the model does not accurately perform beam prediction, such as in cases where network and environmental conditions change rapidly.

[0047] Accordingly, aspects of the present disclosure relate to techniques for enabling a wireless device to efficiently train a machine learning model for performing beam stealing. Specifically, aspects of the present disclosure relate to signaling and configuration that enable the UE to determine when to report beam measurements to be used for offline model training and to determine when the output of a machine learning model is available for online model training. For example, the network can configure the UE with a machine learning model to be used for beam prediction at the UE. The network can configure the UE with a first set of thresholds and / or a second set of thresholds, where the first set of thresholds is used to determine which beam measurements to report to the network for offline training, and the second set of thresholds is used to determine which beam measurements and model outputs can be used by the UE for online model training. In this example, the UE can input measurements performed by the UE into the machine learning model to predict a beam quality metric and report these measurements to the network for offline training or use these measurements to perform online training based on whether the predicted beam quality metric meets the corresponding threshold.

[0048] In some cases, beam measurements / model outputs exhibiting relatively low confidence values or accuracies can meet the first set of thresholds and can thus be reported to the network. For example, a model output / prediction with a small confidence value (e.g., a relatively low probability of accuracy) can meet the first set of thresholds and can thus be reported to the network for offline training. Additionally, beam predictions that significantly deviate from predictions performed using non-machine learning techniques can also meet the first set of thresholds and can thus be reported to the network for offline training. In comparison, beam predictions exhibiting a high confidence level (e.g., a relatively high probability of accuracy) can meet the second set of thresholds and can thus be used for online model training at the UE.

[0049] Aspects of the present disclosure are first described in the context of a wireless communication system. Additional aspects of the present disclosure are described in the context of an example machine learning model configuration and an example process flow. Aspects of the present disclosure are further illustrated and described by and with reference to apparatus diagrams, system diagrams, and flowcharts related to techniques for beam management using model adaptation.

[0050] Figure 1 Illustrates an example of a wireless communication system 100 that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure. The wireless communication system 100 may include one or more network entities 105, one or more UEs 115, and a core network 130. In some examples, the wireless communication system 100 may be a Long Term Evolution (LTE) network, an LTE-Advanced (LTE-A) network, an LTE-A Pro network, a New Radio (NR) network, or a network operating according to other systems and radio technologies, including future systems and radio technologies not explicitly mentioned herein.

[0051] The network entities 105 may be dispersed throughout a geographical area to form the wireless communication system 100 and may include devices in different forms or having different capabilities. In various examples, the network entities 105 may be referred to as network elements, mobility elements, radio access network (RAN) nodes, or network equipment, among other designations. In some examples, the network entities 105 and the UEs 115 may communicate wirelessly via one or more communication links 125 (e.g., radio frequency (RF) access links). For example, the network entities 105 may support a coverage area 110 (e.g., a geographical coverage area) within which the UEs 115 and the network entities 105 may establish one or more communication links 125. The coverage area 110 may be an example of a geographical area within which the network entities 105 and the UEs 115 may support signal communication according to one or more radio access technologies (RATs).

[0052] The UEs 115 may be dispersed throughout the coverage area 110 of the wireless communication system 100, and each UE 115 may be stationary or mobile or both stationary and mobile at different times. The UEs 115 may be devices in different forms or having different capabilities. Figure 1 Some example UEs 115 are illustrated. The UEs 115 described herein may be capable of supporting communication with various types of devices, such as Figure 1 other UEs 115 or network entities 105 as shown.

[0053] As described herein, a node of the wireless communication system 100 (which may be referred to as a network node or a wireless node) can be a network entity 105 (e.g., any network entity described herein), a UE 115 (e.g., any UE described herein), a network controller, a device, an apparatus, a computing system, one or more components, or another suitable processing entity configured to perform any of the techniques described herein. For example, the node can be a UE 115. As another example, the node can be a network entity 105. As yet another example, a first node can be configured to communicate with a second node or a third node. In one aspect of this example, the first node can be a UE 115, the second node can be a network entity 105, and the third node can be a UE 115. In another aspect of this example, the first node can be a UE 115, the second node can be a network entity 105, and the third node can be a network entity 105. In other aspects of this example, the first node, the second node, and the third node can be different from these examples. Similarly, references to UE 115, network entity 105, device, apparatus, computing system, etc. can include the disclosure of UE 115, network entity 105, device, apparatus, computing system, etc. as nodes. For example, the disclosure that a UE 115 is configured to receive information from a network entity 105 also discloses that a first node is configured to receive information from a second node.

[0054] In some examples, the network entity 105 can communicate with the core network 130 or with each other or both. For example, the network entity 105 can communicate with the core network 130 via one or more backhaul communication links 120 (e.g., according to S1, N2, N3, or other interface protocols). In some examples, the network entity 105 can communicate with each other directly (e.g., directly between the network entities 105) or indirectly (e.g., via the core network 130) via the backhaul communication links 120 (e.g., according to X2, Xn, or other interface protocols). In some examples, the network entity 105 can communicate with each other via a midhaul communication link 162 (e.g., according to a midhaul interface protocol) or a fronthaul communication link 168 (e.g., according to a fronthaul interface protocol) or any combination thereof. The backhaul communication link 120, the midhaul communication link 162, or the fronthaul communication link 168 can be or include one or more wired links (e.g., electrical links, optical fiber links), one or more wireless links (e.g., radio links, wireless optical links), etc. or various combinations thereof. The UE 115 can communicate with the core network 130 via a communication link 155.

[0055] One or more of the network entities 105 described herein may include or may be referred to as a base station 140 (e.g., transceiver base station, radio base station, NR base station, access point, radio transceiver, Node B, evolved Node B (eNB), next-generation Node B or gigabit Node B (either of which may be referred to as a gNB), 5G NB, next-generation eNB (ng-eNB), home Node B, home evolved Node B or other suitable terms). In some examples, the network entity 105 (e.g., base station 140) may be implemented in an integrated (e.g., monolithic, stand-alone) base station architecture that may be configured to utilize a protocol stack physically or logically integrated within a single network entity 105 (e.g., a single RAN node, such as base station 140).

[0056] In some examples, the network entity 105 may be implemented in a disaggregated architecture (e.g., disaggregated base station architecture, disaggregated RAN architecture) that may be configured to utilize a protocol stack physically or logically distributed between two or more network entities 105 (such as an integrated access backhaul (IAB) network, an open RAN (O-RAN) (e.g., a network configuration sponsored by the O-RAN Alliance) or a virtualized RAN (vRAN) (e.g., cloud RAN (C-RAN)). For example, the network entity 105 may include one or more of the following: a central unit (CU) 160, a distributed unit (DU) 165, a radio unit (RU) 170, a RAN intelligent controller (RIC) 175 (e.g., near-real-time RIC (near RT RIC), non-real-time RIC (non RT RIC)), a service management and orchestration (SMO) 180 system or any combination thereof. The RU 170 may also be referred to as a radio head, an intelligent radio head, a remote radio head (RRH), a remote radio unit (RRU) or a transmit receive point (TRP). One or more components of the network entity 105 in a disaggregated RAN architecture may be co-located, or one or more components of the network entity 105 may be located in distributed locations (e.g., separate physical locations). In some examples, one or more network entities 105 of the disaggregated RAN architecture may be implemented as virtual units (e.g., virtual CU (VCU), virtual DU (VDU), virtual RU (VRU)).

[0057] The functional split between the CU 160, DU 165, and RU 170 is flexible and can support different functions, depending on which functions are performed at the CU 160, DU 165, or RU 170 (e.g., network layer functions, protocol layer functions, baseband functions, RF functions, and any combination thereof). For example, a functional split of the protocol stack can be employed between the CU 160 and DU 165 such that the CU 160 can support one or more layers of the protocol stack and the DU 165 can support one or more different layers of the protocol stack. In some examples, the CU 160 can host higher protocol layer (e.g., layer 3 (L3), layer 2 (L2)) functions and signaling (e.g., radio resource control (RRC), service data adaptation protocol (SDAP), packet data convergence protocol (PDCP)). The CU 160 can be connected to one or more DU 165s or RU 170s, and one or more DU 165s or RU 170s can host lower protocol layers, such as layer 1 (L1) (e.g., physical (PHY) layer) or L2 (e.g., radio link control (RLC) layer, media access control (MAC) layer) functions and signaling, and can each be at least partially controlled by the CU 160. Additionally or alternatively, a functional split of the protocol stack can be employed between the DU 165 and RU 170 such that the DU 165 can support one or more layers of the protocol stack and the RU 170 can support one or more different layers of the protocol stack. The DU 165 can support one or more different cells (e.g., via one or more RU 170s). In some cases, the functional split between the CU 160 and DU 165 or between the DU 165 and RU 170 can be within a protocol layer (e.g., some functions of a protocol layer can be performed by one of the CU 160, DU 165, or RU 170, while other functions of that protocol layer are performed by a different one of the CU 160, DU 165, or RU 170). The CU 160 can be further functionally split into a CU control plane (CU-CP) and a CU user plane (CU-UP) function. The CU 160 can be connected to one or more DU 165s via an intermediate transport communication link 162 (e.g., F1, F1-c, F1-u), and the DU 165 can be connected to one or more RU 170s via a fronthaul communication link 168 (e.g., open fronthaul (FH) interface). In some examples, the intermediate transport communication link 162 or the fronthaul communication link 168 can be implemented according to the interfaces (e.g., channels) between the layers of the protocol stack, which are supported by the respective network entities 105 communicating via such communication links.

[0058] In some wireless communication systems (e.g., wireless communication system 100), the infrastructure and spectrum resources for radio access may support wireless backhaul link capabilities to supplement wired backhaul connections, thereby providing an IAB network architecture (e.g., to a core network 130). In some cases, in an IAB network, one or more network entities 105 (e.g., IAB nodes 104) may be partially controlled by each other. One or more IAB nodes 104 may be referred to as donor entities or IAB donors. One or more DUs 165 or one or more RUs 170 may be partially controlled by one or more CUs 160 associated with a donor network entity 105 (e.g., donor base station 140). One or more donor network entities 105 (e.g., IAB donors) may communicate with one or more additional network entities 105 (e.g., IAB nodes 104) via the supported access and backhaul links (e.g., backhaul communication link 120). An IAB node 104 may include an IAB mobile terminal (IAB-MT) controlled (e.g., scheduled) by a DU 165 of a coupled IAB donor. The IAB-MT may include a separate set of antennas for relaying communications with a UE 115, or may share the same antennas (e.g., of an RU 170 of the IAB node 104) used for access via the DU 165 of the IAB node 104 (e.g., referred to as a virtual IAB-MT (vIAB-MT)). In some examples, the IAB node 104 may include a DU 165 that supports communication links with additional entities (e.g., IAB nodes 104, UEs 115) within a relay chain or configuration (e.g., downstream) of the access network. In such cases, one or more components of a split RAN architecture (e.g., one or more IAB nodes 104 or components of an IAB node 104) may be configured to operate according to the techniques described herein.

[0059] For example, the access network (AN) or RAN may include communication between an access node (e.g., an IAB donor), an IAB node 104, and one or more UEs 115. The IAB donor may facilitate a connection between the core network 130 and the AN (e.g., via a wired or wireless connection to the core network 130). That is, the IAB donor may refer to a RAN node having a wired or wireless connection to the core network 130. The IAB donor may include a CU 160 and at least one DU 165 (e.g., and an RU 170), in which case the CU 160 may communicate with the core network 130 via an interface (e.g., a backhaul link). The IAB donor and the IAB node 104 may communicate via an F1 interface according to a protocol that defines signaling messages (e.g., the F1 AP protocol). Additionally or alternatively, the CU 160 may communicate with the core network via an interface (which may be an example of a part of the backhaul link) and may communicate with other CUs 160 (e.g., CUs 160 associated with alternative IAB donors) via an Xn-C interface (which may be an example of a part of the backhaul link).

[0060] The IAB node 104 may refer to a RAN node that provides IAB functionality (e.g., access for UEs 115, wireless self-backhaul capabilities, etc.). The DU 165 may act as a distributed scheduling node towards sub-nodes associated with the IAB node 104, and the IAB-MT may act as a scheduled node towards a parent node associated with the IAB node 104. That is, the IAB donor may be referred to as a parent node that communicates with one or more sub-nodes (e.g., the IAB donor may relay transmissions for UEs through one or more other IAB nodes 104). Additionally or alternatively, depending on the relay chain or configuration of the AN, the IAB node 104 may also be referred to as a parent node or a sub-node of other IAB nodes 104. Thus, the IAB-MT entity of the IAB node 104 may provide a Uu interface for a sub-IAB node 104 to receive signaling from a parent IAB node 104, and a DU interface (e.g., the DU 165) may provide a Uu interface for a parent IAB node 104 to signal to a sub-IAB node 104 or a UE 115.

[0061] For example, the IAB node 104 may be referred to as a parent node supporting communication for sub-IAB nodes or as a sub-node associated with the IAB donor or both. The IAB donor may include a CU 160 having a wired or wireless connection (e.g., a fronthaul communication link 120) to the core network 130 and may act as the parent node of the IAB node 104. For example, the DU 165 of the IAB donor may relay transmissions to the UE 115 via the IAB node 104, or may signal transmissions directly to the UE 115, or both. The CU 160 of the IAB donor may signal communication link establishment to the IAB node 104 via the F1 interface, and the IAB node 104 may schedule transmissions (e.g., transmissions relayed from the IAB donor to the UE 115) via the DU 165. That is, data may be relayed to and from the IAB node 104 via signaling over the NR Uu interface to the MT of the IAB node 104. Communication with the IAB node 104 may be scheduled by the DU 165 of the IAB donor, and communication with the IAB node 104 may be scheduled by the DU 165 of the IAB node 104.

[0062] In the context where the techniques described herein are applied to a split RAN architecture, one or more components of the split RAN architecture may be configured to support the techniques for beam management using model adaptation as described herein. For example, some operations described as being performed by the UE 115 or the network entity 105 (e.g., the base station 140) may additionally or alternatively be performed by one or more components of the split RAN architecture (e.g., the IAB node 104, the DU 165, the CU 160, the RU 170, the RIC 175, the SMO 180).

[0063] The UE 115 may include or may be referred to as a mobile device, a wireless device, a remote device, a handheld device, or a subscriber device or some other suitable term, where "device" may also be referred to as a unit, a station, a terminal, or a client, etc. The UE 115 may also include or may be referred to as a personal electronic device, such as a cellular phone, a personal digital assistant (PDA), a tablet computer, a laptop computer, or a personal computer. In some examples, the UE 115 may include or may be referred to as a wireless local loop (WLL) station, an Internet of Things (IoT) device, an Internet of Everything (IoE) device, or a machine type communication (MTC) device, etc., which may be implemented in various objects such as appliances or vehicles, meters, etc.

[0064] The UE 115 described herein may be capable of communicating with various types of devices such as other UEs 115 that may sometimes act as relays, as well as network entities 105 and network equipment including macro eNBs or gNBs, small cell eNBs or gNBs, or relay base stations, etc., asFigure 1 as shown

[0065] UE 115 and network entity 105 may wirelessly communicate with each other via one or more communication links 125 (e.g., access links) using resources associated with one or more carriers. The term "carrier" may refer to a set of RF spectrum resources having a defined physical layer structure for supporting communication link 125. For example, a carrier for communication link 125 may include a portion (e.g., bandwidth part (BWP)) of an RF spectrum band operating according to one or more physical layer channels for 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 carrier operation, user data, or other signaling. Wireless communication system 100 may support communication with UE 115 using carrier aggregation or multi-carrier operation. According to a carrier aggregation configuration, UE 115 may be configured to have multiple downlink component carriers and one or more uplink component carriers. Carrier aggregation may be used for both frequency division duplex (FDD) and time division duplex (TDD) component carriers. Communication between network entity 105 and other devices may refer to communication between these devices and any part (e.g., entity, sub-entity) of network entity 105. For example, the terms "transmit", "receive", or "communicate" when referring to network entity 105 may refer to any part of network entity 105 of the RAN (e.g., base station 140, CU 160, DU 165, RU 170) communicating with another device (e.g., directly or via one or more other network entities 105).

[0066] In some examples, such as in a carrier aggregation configuration, a carrier may also have acquisition signaling or control signaling for coordinating the operation of other carriers. A carrier may be associated with a frequency channel (e.g., evolved universal mobile telecommunications system terrestrial radio access (E-UTRA) absolute RF channel number (EARFCN)) and may be identified according to a channel raster for discovery by UE 115. A carrier may operate in stand-alone mode, in which case initial acquisition and connection may be performed by UE 115 via the carrier, or a carrier may operate in non-stand-alone mode, in which case the connection is anchored using a different carrier (e.g., different carriers of the same or different radio access technologies).

[0067] The communication link 125 shown in the wireless communication system 100 may include a downlink transmission (e.g., forward link transmission) from the network entity 105 to the UE 115, an uplink transmission (e.g., reverse link transmission) from the UE 115 to the network entity 105, or other transmission configurations such as both. A carrier may carry downlink communication or uplink communication (e.g., in FDD mode), or may be configured to carry both downlink communication and uplink communication (e.g., in TDD mode).

[0068] A carrier may be associated with a specific bandwidth of the RF 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 set of bandwidths of carriers of a particular radio access technology (e.g., 1.4 megahertz (MHz), 3 MHz, 5 MHz, 10 MHz, 15 MHz, 20 MHz, 40 MHz, or 80 MHz). Devices of the wireless communication system 100 (e.g., the network entity 105, the UE 115, or both) may have a hardware configuration that supports communication using a particular carrier bandwidth, or may be configurable to support communication using one of a set of carrier bandwidths. In some examples, the wireless communication system 100 may include a network entity 105 or a UE 115 that supports concurrent communication using carriers associated with multiple carrier bandwidths. In some examples, each served UE 115 may be configured to operate using a portion (e.g., subband, BWP) or all of the carrier bandwidth.

[0069] The signal waveform transmitted via a carrier may include multiple subcarriers (e.g., using a multi-carrier modulation (MCM) technique such as orthogonal frequency division multiplexing (OFDM) or discrete Fourier transform spread OFDM (DFT-S-OFDM)). In a system employing an MCM technique, a resource element may refer to the resource of one symbol period (e.g., the duration of one modulation symbol) and one subcarrier, in which case the symbol period and the subcarrier spacing may be inversely related. The number of bits carried by each resource element may depend on the modulation scheme (e.g., the order of the modulation scheme, the decoding rate of the modulation scheme, or both), such that a relatively high number of resource elements (e.g., during the transmission duration) and a relatively high order of the modulation scheme may correspond to a relatively high rate of communication. Wireless communication resources may refer to a combination of RF spectrum resources, time resources, and space resources (e.g., spatial layer or beam), and the use of multiple space resources may increase the data rate or data integrity for communication with the UE 115.

[0070] One or more parameter sets for a carrier may be supported, and the parameter sets may include subcarrier spacing (Δf) and cyclic prefix. The carrier may be divided into one or more BWPs having the same or different parameter sets. In some examples, UE 115 may be configured with multiple BWPs. In some examples, a single BWP of a carrier may be active at a given time, and the communication of UE 115 may be restricted to one or more active BWPs.

[0071] A time interval for network entity 105 or UE 115 may be expressed as a multiple of a basic time unit, and the basic time unit may refer to, for example, the sampling period T s = 1 / (Δf max ·N f ) seconds, where Δf max may represent the supported subcarrier spacing, and N f may represent the supported discrete Fourier transform (DFT) size. The time intervals of communication resources may be organized according to radio frames each having a specified duration (e.g., 10 milliseconds (ms)). Each radio frame may be identified by a system frame number (SFN) (e.g., ranging from 0 to 1023).

[0072] Each frame may include a plurality of consecutively numbered subframes or time slots, and each subframe or time slot may have the same duration. In some examples, a frame may be divided (e.g., in the time domain) into subframes, and each subframe may be further divided into a certain number of time slots. Alternatively, each frame may include a variable number of time slots, and the number of time slots may depend on the subcarrier spacing. Each time slot may include a certain number of symbol periods (e.g., depending on the length of the cyclic prefix appended to each symbol period). In some wireless communication systems 100, a time slot may be further divided into a plurality of mini-slots associated with one or more symbols. Excluding the cyclic prefix, each symbol period may be associated with one or more (e.g., N f ones) sampling periods. The duration of a symbol period may depend on the subcarrier spacing or the operating frequency band.

[0073] A subframe, time slot, mini-slot, or symbol may be the smallest scheduling unit (e.g., in the time domain) of wireless communication system 100 and may be referred to as a transmission time interval (TTI). In some examples, the TTI duration (e.g., the number of symbol periods in a TTI) may be variable. Additionally or alternatively, the smallest scheduling unit of wireless communication system 100 may be dynamically selected (e.g., in a burst of shortened TTIs (sTTIs)).

[0074] According to various techniques, carriers can be used to multiplex physical channels for communication. For example, one or more of time-division multiplexing (TDM) techniques, frequency-division multiplexing (FDM) techniques, or hybrid TDM-FDM techniques can be used to multiplex physical control channels and physical data channels for signaling via a downlink carrier. The control region of a physical control channel (e.g., a control resource set (CORESET)) can be defined by a set of symbol periods and can extend across the system bandwidth of a carrier or a subset of that system bandwidth. One or more control regions (e.g., CORESETs) can be configured for a set of UEs 115. For example, one or more of the UEs 115 can monitor or search a control region for control information according to one or more search space sets, and each search space set can include one or more control channel candidates in one or more aggregation levels arranged in a cascaded manner. The aggregation level of a control channel candidate can refer to the amount of control channel resources (e.g., control channel elements (CCEs)) associated with the coded information for a control information format with a given payload size. The search space set can include a common search space set configured to transmit control information to multiple UEs 115 and a UE-specific search space set for transmitting control information to a specific UE 115.

[0075] In some examples, the network entity 105 (e.g., the base station 140, the RU 170) can be movable and thus provide communication coverage for a moving coverage area 110. In some examples, different coverage areas 110 associated with different techniques can overlap, but the different coverage areas 110 can be supported by the same network entity 105. In some other examples, overlapping coverage areas 110 associated with different techniques can be supported by different network entities 105. The wireless communication system 100 can include, for example, a heterogeneous network in which different types of network entities 105 use the same or different radio access technologies to provide coverage for various coverage areas 110.

[0076] 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). The UE 115 can be designed to support ultra-reliable, low-latency, or critical functions. Ultra-reliable communication can include private communication or group communication and can be supported by one or more services (such as push-to-talk, video, or data). Support for ultra-reliable, low-latency functions can include prioritization of services, and such services can be used for public safety or general commercial applications. The terms ultra-reliable, low-latency, and ultra-reliable low-latency can be used interchangeably herein.

[0077] In some examples, the UE 115 may be configured to communicate directly with other UEs 115 via a device-to-device (D2D) communication link 135 (e.g., according to a peer-to-peer (P2P), D2D, or sidelink protocol). In some examples, one or more UEs 115 performing D2D communication in a group may be within the coverage area 110 of a network entity 105 (e.g., base station 140, RU 170), and this network entity may support aspects of such D2D communication configured (e.g., scheduled) by the network entity 105. In some examples, one or more UEs 115 in such a group may be outside the coverage area 110 of the network entity 105, or may otherwise be unable or not configured to receive transmissions from the network entity 105. In some examples, a group of UEs 115 communicating via D2D communication may support a one-to-many (1:M) system, where each UE 115 transmits to each of the other UEs 115 in the group. In some examples, the network entity 105 may facilitate the scheduling of resources for D2D communication. In some other examples, D2D communication may be performed between UEs 115 without involving the network entity 105.

[0078] In some systems, the D2D communication link 135 may be an example of a communication channel (such as a sidelink communication channel) between vehicles (e.g., UE 115). In some examples, vehicles may communicate using vehicle-to-everything (V2X) communication, vehicle-to-vehicle (V2V) communication, or some combination of these. Vehicles may signal 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 may communicate with roadside infrastructure (such as roadside units), or communicate with the network via one or more network nodes (e.g., network entity 105, base station 140, RU 170) using vehicle-to-network (V2N) communication, or both.

[0079] 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 for managing access and mobility (e.g., a mobility management entity (MME), an access and mobility management function (AMF)) and at least one user plane entity for routing packets or interconnecting to an external network (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 a network entity 105 (e.g., a base station 140) associated with the core network 130. User IP packets may be passed through the user plane entity, which may provide IP address allocation and other functions. The user plane entity may be connected to the IP services 150 of one or more network operators. The IP services 150 may include access to the Internet, an intranet, an IP multimedia subsystem (IMS), or packet switched streaming services.

[0080] The wireless communication system 100 may operate using one or more frequency bands that may be in the range of 300 megahertz (MHz) to 300 gigahertz (GHz). Generally, the region from 300 MHz to 3 GHz is known as the ultra-high frequency (UHF) region or the decimeter band because, in terms of length, the wavelength range is from approximately one decimeter to one meter. UHF waves may be blocked or redirected by buildings and environmental features (which may be referred to as clusters), but these waves may be sufficient to penetrate structures so that macrocells can serve UEs 115 located indoors. Compared to communications using smaller frequencies and longer waves in the high frequency (HF) or very high frequency (VHF) portions of the spectrum below 300 MHz, communications using UHF waves may be associated with smaller antennas and shorter ranges (e.g., less than 100 kilometers).

[0081] The wireless communication system 100 may utilize both licensed RF spectrum bands and unlicensed RF spectrum bands. For example, the wireless communication system 100 may use an unlicensed band (such as the 5 GHz industrial, scientific, and medical (ISM) band) to employ licensed-assisted access (LAA), LTE unlicensed (LTE-U) radio access technology, or NR technology. When operating using an unlicensed RF spectrum band, devices such as the network entity 105 and the UE 115 may employ carrier sensing for collision detection and avoidance. In some examples, operation using an unlicensed band may be combined with operation using a component carrier of a licensed band based on a carrier aggregation configuration (e.g., LAA). Operation using unlicensed spectrum may include downlink transmission, uplink transmission, peer-to-peer (P2P) transmission, or device-to-device (D2D) transmission, etc.

[0082] The network entity 105 (e.g., base station 140, RU 170) or the UE 115 may be equipped with multiple antennas, which may be used to employ techniques such as transmit diversity, receive diversity, multiple-input multiple-output (MIMO) communication, or beamforming. The antennas of the network entity 105 or the UE 115 may be located within one or more antenna arrays or antenna panels, which may support MIMO operation or transmit or receive beamforming. For example, one or more base station antennas or antenna arrays may be co-located at an antenna assembly (such as an antenna tower). In some examples, the antennas or antenna arrays associated with the network entity 105 may be located at different geographical locations. The network entity 105 may include an antenna array having a set of antenna ports in multiple rows and columns that the network entity 105 may use for beamforming to support communication with the UE 115. Similarly, the UE 115 may include one or more antenna arrays, which may support various MIMO or beamforming operations. Additionally or alternatively, the antenna panel may support RF beamforming for signals transmitted via the antenna ports.

[0083] The network entity 105 or the UE 115 may use MIMO communication to take advantage of multipath signal propagation and improve spectral efficiency by transmitting or receiving multiple signals via different spatial layers. Such techniques may be referred to as spatial multiplexing. The multiple signals may be transmitted, for example, by a transmitting device via different antennas or different combinations of antennas. Similarly, the multiple signals may be received by a receiving device via different antennas or different combinations of antennas. Each of the multiple signals may be referred to as a separate spatial stream and may carry information associated with the same data stream (e.g., the same codeword) or different data streams (e.g., different codewords). Different spatial layers may be associated with different antenna ports for channel measurement and reporting. MIMO techniques include: single-user MIMO (SU-MIMO), for which multiple spatial layers are transmitted to the same receiving device; and multi-user MIMO (MU-MIMO), for which multiple spatial layers are transmitted to multiple devices.

[0084] Beamforming (which may 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., network entity 105, UE 115) to shape or steer an antenna beam (e.g., Tx beam, Rx beam) along a spatial path between the transmitting device and the receiving device. Beamforming can be achieved by combining signals communicated via the antenna elements of an antenna array such that some signals propagating along a particular direction relative to the antenna array experience constructive interference while other signals experience destructive interference. Adjustment of the signals communicated 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 that device. The adjustment associated with each of these 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).

[0085] Network entity 105 or UE 115 can use beam scanning techniques as part of beamforming operations. For example, network entity 105 (e.g., base station 140, RU 170) can use multiple antennas or antenna arrays (e.g., antenna panels) to perform beamforming operations for directional communication with UE 115. Some signals (e.g., synchronization signals, reference signals, beam selection signals, or other control signals) can be sent by network entity 105 multiple times in different directions. For example, network entity 105 can send signals according to different sets of beamforming weights associated with different transmission directions. Transmission along different beam directions can be used to identify (e.g., by the transmitting device such as network entity 105, or by the receiving device such as UE 115) the beam direction for later transmission or reception by network entity 105.

[0086] Some signals (such as data signals associated with a particular receiving device) can be sent by a transmitting device (e.g., transmitting network entity 105, transmitting UE 115) along a single beam direction (e.g., a direction associated with a receiving device such as receiving network entity 105 or receiving UE 115). In some examples, the beam direction associated with transmission along a single beam direction can be determined based on signals transmitted along one or more beam directions. For example, UE 115 can receive one or more of the signals sent by network entity 105 in different directions and can report to network entity 105 an indication of the signal that UE 115 receives with the highest signal quality or other acceptable signal quality.

[0087] In some examples, transmissions performed by a device (e.g., by network entity 105 or UE 115) may be carried out using multiple beam directions, and the device may use a combination of digital precoding or beamforming to generate a combined beam for transmission (e.g., from network entity 105 to UE 115). UE 115 may report feedback indicating precoding weights for one or more beam directions, and the feedback may correspond to a configured set of beams across the system bandwidth or one or more sub-bands. Network entity 105 may transmit reference signals (e.g., cell-specific reference signal (CRS), channel state information reference signal (CSI-RS)), which may or may not be precoded. UE 115 may provide feedback for beam selection, which may 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 network entity 105 (e.g., base station 140, RU 170) in one or more directions, UE 115 may use similar techniques for transmitting signals multiple times in different directions (e.g., for identifying beam directions used by UE 115 for subsequent transmissions or receptions), or for transmitting signals in a single direction (e.g., for transmitting data to a receiving device).

[0088] A receiving device (e.g., UE 115) may perform reception operations according to multiple reception configurations (e.g., directional listening) when receiving various signals (such as synchronization signals, reference signals, beam selection signals, or other control signals) from a receiving device (e.g., network entity 105). For example, the receiving device may perform reception according to multiple reception directions by: receiving via different antenna sub-arrays, processing the received signals according to different antenna sub-arrays, receiving according to different sets of reception beamforming weights (e.g., different directional listening weight sets) applied to signals received at multiple antenna elements of an antenna array, or processing the received signals according to different sets of reception beamforming weights applied to signals received at multiple antenna elements of an antenna array. Any of these operations may be referred to as "listening" according to different reception configurations or reception directions. In some examples, the receiving device may use a single reception configuration to receive along a single beam direction (e.g., when receiving a data signal). The single reception configuration may be aligned along a beam direction determined based on listening according to different reception configuration directions (e.g., a beam direction determined to have the highest signal strength, highest signal-to-noise ratio (SNR), or other acceptable signal quality based on listening according to multiple beam directions).

[0089] The wireless communication system 100 can be a packet-based network operating according to a layered protocol stack. In the user plane, the communication at the bearer or PDCP layer can be IP-based. The RLC layer can perform packet segmentation and reassembly for conveyance via logical channels. The MAC layer can perform priority handling and multiplexing of logical channels into transport channels. The MAC layer can also implement error detection techniques, error correction techniques, or both to support retransmission to improve link efficiency. In the control plane, the RRC layer can provide the establishment, configuration, and maintenance of the RRC connection that supports the radio bearers for user plane data between the UE 115 and the network entity 105 or the core network 130. The PHY layer can map the transport channels to physical channels.

[0090] The UE 115 and the network entity 105 can support the 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 via a communication link (e.g., communication link 125, D2D communication link 135). HARQ can 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)). HARQ can improve the throughput at the MAC layer under poor radio conditions (e.g., low signal-to-noise ratio conditions). In some examples, the device can support same-slot HARQ feedback, in which case the device can provide HARQ feedback for data received via previous symbols in a particular slot during that slot. In some other examples, the device can provide HARQ feedback in a subsequent slot or according to some other time interval.

[0091] In some aspects, the corresponding devices of the wireless communication system 100 can support techniques that enable a wireless device (e.g., UE 115) to efficiently train a machine learning model for performing beam stealing. Specifically, the wireless communication system 100 can support signaling and configuration that enable the UE 115 to determine when to report beam measurements that will be used for offline model training and to determine when the output of the machine learning model is available for online model training.

[0092] For example, a network entity 105 of a wireless communication system 100 may configure a UE 115 with a machine learning model to be used for beam prediction at the UE 115. The network entity 105 may configure the UE 115 with a first set of thresholds and / or a second set of thresholds, where the first set of thresholds is used to determine which beam measurements are to be reported to the network for offline training, and the second set of thresholds is used to determine which beam measurements and model outputs can be used by the UE 115 for online model training. The UE 115 may input measurements performed by the UE 115 into the machine learning model to predict a beam quality metric associated with an Rx beam at the UE 115. In this example, the UE 115 may be configured to report these measurements to the network entity 105 for offline training or use these measurements to perform online training based on whether the predicted beam quality metric meets a corresponding threshold.

[0093] The techniques described herein enable efficient training of a machine learning model for beam prediction while reducing the control signaling overhead for performing model training. Specifically, by configuring the UE 115 with thresholds for reporting inputs / outputs of a machine model to the network for offline training, the techniques described herein enable the UE 115 to report model inputs / outputs only when offline training is likely to be expected or needed. Thus, the techniques described herein can reduce the control signaling overhead associated with offline training, reduce the frequency of offline training, and reduce the latency of beam prediction performed using the machine learning model. Additionally, by configuring the UE 115 with thresholds for reporting inputs / outputs of a machine model to the network for offline training, the techniques described herein enable the UE 115 to perform online training using model outputs with high confidence values, which are expected to improve the efficiency and reliability of the machine learning model for performing future beam predictions.

[0094] Figure 2 An example of a wireless communication system 200 supporting techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated. In some examples, aspects of the wireless communication system 200 may implement aspects of the wireless communication system 100, or may be implemented by these aspects. Specifically, the wireless communication system 200 may support techniques for training a machine learning model for beam prediction, as previously described herein.

[0095] For example, the wireless communication system 200 may include a UE 115-a and a network entity 105-a (which may be examples of wireless devices as described herein). In some aspects, the UE 115-a and the network entity 105-a may communicate with each other using a communication link 205, which may be an example of an NR or LTE link, a sidelink (e.g., a PC5 link), etc. between the corresponding devices. In some cases, the communication link 205 may include an example of an access link (e.g., a Uu link), which may include a bi-directional link that enables both uplink communication and downlink communication. For example, the UE 115-a may use the communication link 205 to send an uplink signal, such as an uplink control signal or an uplink data signal, to one or more components of the network entity 105-a, and one or more components of the network entity 105-a may use the communication link 205 to send a downlink signal, such as a downlink control signal or a downlink data signal, to the UE 115-a.

[0096] As previously mentioned herein, some wireless devices (e.g., UE 115) may be configured to receive reference signals from the network using different Rx beams and perform measurements on the received reference signals to identify the relative quality of the corresponding Rx beams. In such cases, the UE 115 may send a measurement report (e.g., a CSI report) indicating the measurements to the network so that the network may schedule communication at the UE 115 using the Rx beam that is best for the UE 115 (e.g., an Rx beam that exhibits a threshold quality).

[0097] The wireless communication system may be configured to implement various types of machine learning models to perform beam prediction, including neural network functionality (NNF), neural network models, etc. In the context of NNF, each NNF may be associated with a function Y = F(X), where a normalized input X is input into the corresponding NNF to generate a normalized output Y. Each NNF within the wireless communication system may be identified by a normalized NNF identifier (ID), where some wireless communications may also support a non-normalized NNF ID associated with an NNF for a proprietary extension. The NNF may be associated with a mandatory information element for inter-vendor interoperability and / or an optional information element for flexible implementation. In some cases, a single NNF may be supported by multiple models (e.g., vendor-specific implementations).

[0098] A neural network model can be defined as a model structure and a set of parameters, where the neural network model can be defined by an operator, a vendor, a third party, or any combination thereof. The model structure of the neural network model can be identified by a model ID, which can include a default parameter set. Each model ID can be unique within a wireless communication system, where each model ID can be associated with one or more corresponding NNFs. The set of parameters for the neural network model can include the weights of the neural network model (e.g., the weights for the corresponding inputs of the model) and other configuration parameters. The parameter set can be location and / or configuration specific.

[0099] In some cases, the UE 115 can utilize past measurements to predict beam measurements (e.g., future Rx beam quality) at some point in the future for the same or a different beam set, and can report the predicted / extrapolated beam measurements to the network. In other words, a wireless device of a wireless communication system can be configured to train an algorithm (e.g., a machine learning model) to predict future RSRP measurements for beam set #2 based on past RSRP measurements of beam set #1, where beam set #1 and beam set #2 can include the same beams, overlapping beams, or completely different beam sets. In some cases, the algorithm / machine learning model can include a recurrent neural network or a traditional algorithm.

[0100] In some cases, a wireless device (e.g., network entity 105-a, UE 115-a) implementing a machine learning model can be configured to measure reference signals, such as SSBs (e.g., a subset of SSBs), and input these measurements into the machine learning model to predict future measurements (e.g., SSB measurements). For example, the UE 115-a can be configured to perform measurements on the received SSBs and input these measurements into the machine learning model / algorithm in order to predict some refined CSI-RS beams for unicast PDSCH / PDCCH. Additionally or alternatively, the output of the machine learning model can include beam IDs (e.g., the beam ID of the beam predicted to exhibit the highest performance) expected to exhibit a certain threshold performance at some point in the future.

[0101] The use of a machine learning model for performing beam prediction can reduce the reference signal overhead within the network, because the prediction of future beams and / or beam metrics can reduce the frequency at which the corresponding devices send / receive reference signals for beam / channel tracking. Additionally, using a machine learning model to perform beam prediction can reduce uplink feedback, because this can reduce the frequency at which the UE115-a sends channel estimation feedback. Therefore, by reducing the frequency of channel estimation feedback, the machine learning model for beam prediction can also reduce the power consumption at the UE 115-a.

[0102] A machine learning model / algorithm for beam prediction can be trained and maintained by a network (e.g., network entity 105-a), where the machine learning model is run / executed by the network or distributed to and run / executed by a UE 115 within the network. In the case where the machine learning model is run / executed by the UE 115, the network can train and configure the model and distribute the trained model to the corresponding UE 115.

[0103] For example, according to a first specific implementation, a machine learning module / model for beam prediction can be trained and run by network entity 105-a. In such a case, UE 115-a can send a reference signal (e.g., SRS) to network entity 105-a, where network entity 105-a performs measurements on the received reference signal and inputs these measurements into the machine learning model to perform beam prediction (e.g., predicting future RSRP, future beam ID, etc.). Network entity 105-a can then send the prediction results (e.g., predicted RSRP, predicted beam ID) to UE 115-a and can schedule communication between network entity 105-a and UE 115-a based on the prediction. UE 115-a can perform measurements and / or generate reports (e.g., CSI report, beam report) provided to network entity 105-a such that network entity 105-a can utilize the measurements / generated reports to further train / update the machine learning model (e.g., by comparing the predicted RSRP with the actual RSRP measurements performed by UE 115-a). This first specific implementation where the machine learning model is trained and run by network entity 105-a may be useful in cases where UE 115-a exhibits relatively low power and / or processing capabilities.

[0104] In comparison, according to a second specific implementation, a machine learning module / model for beam prediction can be initially trained by network entity 105-a but run / executed by UE 115-a. For example, network entity 105-a can configure UE 115-a with the machine learning model, where UE 115-a is configured to run the configured machine learning model using measurements performed on a reference signal received from network entity 105-a. In this example, UE 115-a can report the prediction output (e.g., predicted RSRP, predicted beam ID) of the machine learning model to network entity 105-a, such as based on certain trigger events / conditions for reporting the model output. The second specific implementation where the machine learning model is run / executed by UE 115-a can take advantage of the fact that network entity 105-a typically sends reference signals more frequently than UE 115-a. Additionally, compared to the first specific implementation, the second specific implementation can result in less signaling overhead but may require higher computational capabilities and power consumption at UE 115-a.

[0105] Compared with traditional algorithms for beam determination / prediction, the use of machine learning-based beam prediction has been shown to yield significant performance gains. However, such machine learning solutions for beam prediction can depend on the data distribution of the input provided to the machine learning model. For example, as the environment changes (e.g., as UE 115-a moves within the network or as the channel conditions between UE 115-a and network entity 105-a change), the configured machine learning model may not adapt to the change and may thus fail to make accurate predictions for beam management. In such cases, the pre-configured machine learning model may not be able to make accurate predictions due to the changing input resulting from the changing environment (e.g., the model may make inaccurate or unreliable predictions due to the changing environment / input).

[0106] In cases where the machine learning model cannot perform accurate predictions, the model may need to be further trained, which can include offline training and / or online training. However, both of these training methods have their respective drawbacks.

[0107] In the context of offline training, the wireless device training the model can collect new data (e.g., perform new measurements), manually label the data, and use the new labeled data to further train / fine-tune the model. For example, UE 115-a can report measurements to the network, where the network further fine-tunes the model and reconfigures UE 115-a with the updated model to be used for subsequent beam prediction. However, offline training may increase the beam prediction latency because UE 115-a may not be able to use the outdated model while the model is being offline trained by the network. Additionally, by the time UE 115-a is reconfigured with the updated model, the updated / fine-tuned model may already be outdated and unable to perform accurate predictions.

[0108] In comparison, online model training / adaptation can be based on pseudo-labels, where UE 115-a can select the data it is convinced of and set the prediction as the pseudo-label. In other words, UE 115-a can perform online model training by obtaining the output / prediction from the machine learning model, labeling the output / prediction as accurate or precise, and inputting the labeled output / prediction back into the model to further train the model. Online model training can enable the model to more efficiently adapt to environmental changes and reduce the latency of model training. However, online training techniques may not be effective in cases where the model does not accurately perform beam prediction, such as in cases where the network and environmental conditions change rapidly. In other words, not all outputs from the model are valid for model fine-tuning.

[0109] Such deficiencies in offline model training and online model training can reduce the efficacy of machine learning-based beam prediction techniques and, as a result, prevent the widespread use of machine learning models for beam prediction.

[0110] Accordingly, aspects of the present disclosure relate to techniques for enabling a wireless device to efficiently train a machine learning model for performing beam stealing. Specifically, aspects of the present disclosure relate to signaling and configuration that enable UE 115 to determine when to report beam measurements to be used for offline model training and to determine when the output of a machine learning model is available for online model training. In this regard, aspects of the present disclosure can enable online model adaptation with lower latency and can enable UE 115-a to select outputs from a model that can be used to efficiently train and fine-tune the model at a limited bandwidth cost.

[0111] For example, referring to Figure 2 the wireless communication system 200 shown, UE 115-a may send capability signaling 225 to network entity 105-a. The capability signaling 225 may indicate one or more capabilities associated with the ability of UE 115-a to perform future beam prediction, such as the supported machine learning models 215, processing capabilities, memory capabilities, etc. For example, the capability signaling 225 may indicate one or more machine learning models 215 that can be used to perform beam prediction supported by UE 115-a.

[0112] UE 115-a may receive control signaling 230 from network entity 105-a that indicates the machine learning model 215 to be used for beam prediction at UE 115-a. For example, the control signaling 230 may indicate the machine learning model 215 supported by UE 115-a, as indicated via the capability signaling 225. In this regard, network entity 105-a may indicate what model inputs 210 will be used for the machine learning model 215 and what model outputs 220 UE 115-a will use the machine learning model 215 to predict for beam prediction.

[0113] In some aspects, the control signaling 230 may additionally indicate one or more thresholds for reporting the output from the machine learning model 215. For example, as will be further described herein, the control signaling 230 may indicate thresholds β and / or δ that UE 115-a uses to determine whether / when it is expected that UE 115-a will report the output from the machine learning model 215 to the network for offline training. Additionally or alternatively, the control signaling 230 may indicate a threshold τ that UE 115-a uses to determine whether / when UE 115-a can utilize the input / output of the machine learning model 215 for online model training at UE 115-a.

[0114] In some aspects, network entity 105-a may send a set of reference signals 235 (e.g., Synchronization Signal Block (SSB)) to UE 115-a. Network entity 105-a may send reference signals 235 that will be used by UE 115-a to estimate the channel condition and perform beam prediction.

[0115] In some aspects, UE 115-a may use one or more Rx beams at UE 115-a to receive reference signals 235. For example, UE 115-a may scan a set of Rx beams (e.g., Rx beam 1, Rx beam 2, Rx beam 3, etc.) to receive the corresponding reference signals 235 in order to evaluate the relative performance or efficiency of the corresponding Rx beams for receiving communications from network entity 105-a.

[0116] In some aspects, UE 115-a may perform a set of measurements on a set of reference signals 235. For example, in a case where UE 115-a receives reference signals 235 having a set of Rx beams including a first Rx beam, a second Rx beam, and a third Rx beam, UE 115-a may perform measurements for each of the first Rx beam, the second Rx beam, and the third Rx beam. Additionally or alternatively, UE 115-a may perform measurements using the same Rx beam at different time points. For example, network entity 105-a may send reference signals 235 within a first time interval (T1), a second time interval (T2), and a third time interval (T3). In this example, UE 115-a may use the same Rx beam (or a set of Rx beams) for each of the corresponding time intervals to receive reference signals 235, and thus may perform measurements to evaluate the relative performance of the Rx beam over time.

[0117] The first set of measurements may include but are not limited to RSRP, RSRQ, SNR, SINR, CQI, RSSI, or any combination thereof. In some cases, the measurements performed by UE 115-a may be based on what model inputs 210 the machine learning model 215 will use for beam prediction.

[0118] UE 115-a may predict a first set of beam quality metrics associated with the Rx beams at UE 115-a based on inputting the first set of measurements into the machine learning model 215. In this regard, the measurements may include model inputs 210, and the beam quality metrics may include model outputs 220. The first set of beam quality metrics (e.g., model outputs 220) may include predicted metrics associated with the Rx beams within one or more future time intervals. In some cases, the Rx beams associated with the output / prediction of the machine learning model 215 may be the same as or different from the Rx beams used to receive reference signals 235.

[0119] The model output 220 (e.g., predicted beam quality metric) can include any output known in the art, including predicted measurements associated with the Rx beam (e.g., predicted RSRP, RSRQ, SNR, SINR, CQI, RSSI), predicted probability or confidence values (e.g., a probability metric corresponding to the probability that the corresponding Rx beam will exhibit a certain threshold level of performance or quality at a future time), or both.

[0120] For example, according to a first specific implementation, the model output 220 can include a predicted beam index and a corresponding confidential probability value (e.g., a probability metric). This specific implementation can be referred to Figure 3 for further illustration and description.

[0121] Figure 3 An example of a machine learning model configuration 300 that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure is illustrated. In some examples, aspects of the machine learning model configuration 300 can implement aspects of the wireless communication system 100, the wireless communication system 200, or both, or can be implemented by these aspects.

[0122] Specifically, the machine learning model configuration 300 can illustrate Figure 2 the example of the machine learning model 215 shown and described. For example, the machine learning model configuration 300 illustrates a model input 305 and a model output 315, which can include Figure 2 the examples of the model input 210 and the model output 220 shown. The machine learning model configuration 300 also illustrates a set of convolutional layers 310 of a machine learning model (such as Figure 2 the machine learning model 215 shown).

[0123] In some specific implementations, the UE 115 can input the model input 305 into the convolutional layers 310 of the machine learning model, where the machine learning model is configured to generate a model output 315 associated with the predicted beam quality metric. As Figure 3 shown, the model output 315 can include a confidential probability vector 320 and a beam prediction vector 325.

[0124] The confidential probability vector 320 may include probability metrics that indicate the relative probability that a corresponding Rx beam (represented by a beam index) will exhibit a certain threshold level of performance or quality within a certain future time interval. In some cases, the sum of the probability metrics of the confidential probability vector may add up to 1. The beam prediction vector 325 may be based on the confidential probability vector 320 and may indicate a prediction as to which Rx beam is to be applied to communication within a certain future time interval. Thus, the beam prediction vector 325 may include a single "1" value indicating the Rx beam predicted to have the highest quality / performance, with the remaining values in the beam prediction vector 325 for the remaining Rx beams being set to "0".

[0125] For example, based on the model input 305 provided to the model, Figure 3 the model shown may predict that Rx beam index 2 is associated with the highest probability metric. In other words, the machine learning model may predict that Rx beam index 2 has the highest probability of exhibiting a certain threshold level of quality / performance among Rx beam indices 0–4. Thus, in Figure 3 this example shown, the machine learning model may thus predict that Rx beam index 2 is to be applied to receive a downlink signal during a certain future time interval.

[0126] In some aspects, the UE 115 may be configured to compare the probability metrics of the confidential probability vector 320 with a threshold β to determine whether the UE 115 should report the measurement / model input 305 and / or the model output 315 to the network for offline training. In some cases, it may be expected that the UE 115 reports the model input 305 (e.g., measurement) and / or the corresponding model output 315 (e.g., the confidential probability vector 320, the beam prediction vector 325) to the network only when the confidential probability is less than or equal to the threshold β. In some aspects, the threshold β may be configured or indicated by the network, predefined by the network and / or at the UE 115, or both.

[0127] In some cases, the confidential probability may include the highest probability metric of the confidential probability vector 320, the median or average probability metric of the confidential probability vector 320, etc. Additionally, the network may configure the machine learning model to output a confidential probability to identify whether it is valuable. Such a machine learning model may be used to perform beam prediction, output a confidential probability value, or both.

[0128] For example, the network may configure the threshold β for reporting the model output 315 to the network as β = 0.6. Thus, any prediction with a confidential probability lower than β should be recorded for offline optimization. In Figure 3In the example shown, the UE 115 may input measurements that result in a highest probability metric (e.g., confidentiality probability) of 0.5. In this example, the UE 115 may determine that the confidentiality probability (e.g., the highest probability metric) is less than a threshold (e.g., 0.5 < β), and may thus report the model input 305 and / or the model output 315 to the network (e.g., via Figure 2 the control message 240 shown) so that the network can utilize the model input 305 and / or the model output 315 for offline training.

[0129] As another example, the network may again configure the threshold β for reporting the model output 315 to the network as β = 0.6. Thus, any prediction with a confidentiality probability below β should be recorded for offline optimization. In this example, the UE 115 may input measurements that result in a highest probability metric (e.g., confidentiality probability) of 0.8. In this example, the UE 115 may determine that the confidentiality probability (e.g., the highest probability metric) is greater than the threshold (e.g., 0.8 > β), and may thus avoid reporting the model input 305 and / or the model output 315 to the network. In other words, since the model predicts at least one Rx beam with a relatively high confidence value / probability metric, the UE 115 may determine that offline training is not expected.

[0130] Reference will again be made to Figure 2 . In an additional or alternative embodiment, the model output 220 predicted by the machine learning model 215 may include predicted measurements (e.g., predicted RSRP, RSRQ, SNR, SINR, CQI, RSSI) associated with an Rx beam at some future time interval.

[0131] In some embodiments, to monitor the state / accuracy of the machine learning model 215, the UE 115-a may be configured to compare the model output 220 (e.g., predicted RSRP measurement) with a prediction performed by a non-machine learning-based technique or algorithm. In such cases, the UE 115-a may be configured to report the model input 220 and / or the model output 220 to the network for offline training only if the difference between the machine learning-based model output 210 and the estimate from the monitoring reference signal 235 is greater than or equal to a threshold δ. In some aspects, the threshold δ may be configured or indicated by the network, predefined by the network and / or at the UE 115, or both.

[0132] In such cases, the monitoring reference signal 235 may be configured with a sparse pattern to reduce resource utilization, and the network may instruct the UE 115-a to perform beam prediction using both the machine learning model 215 and a traditional algorithm (e.g., mathematical operations different from the machine learning model 215). Additionally, the monitoring reference signal 235 may include a subset of the reference signal 235 for beam management and / or a newly configured reference signal 235.

[0133] In other words, in a case where the first set of beam quality metrics includes predicted measurements associated with a set of Rx beams, UE 115-b may compare the predicted measurements (e.g., model output 220) with actual measurements predicted using a mathematical operation different from the machine learning model (e.g., a traditional algorithm), and may compare the difference between the machine learning predicted measurements (e.g., model output 220) with a threshold metric (δ).

[0134] For example, the first set of beam quality metrics (e.g., model output 220) may include predicted RSRP measurements associated with a set of Rx beams predicted by machine learning model 215. In this example, UE 115-a may use non-machine learning-based techniques (e.g., traditional algorithms) to perform a second set of measurements. UE 115-a may then determine the difference (e.g., median minimum mean square error (MMSE) metric) between a first set of RSRP measurements (RSRP1) predicted by machine learning model 215 and a second set of RSRP measurements (RSRP2) predicted by a non-machine learning-based algorithm. For example, in beam management, one predicted RSRP value for each Rx beam may be generated based on previous RSRP measurements, and a reference signal 235 may be used to estimate the true RSRP based on a traditional algorithm.

[0135] In this example, if one or more differences between the first set of RSRP measurements (RSRP1) predicted by the model and the second set of RSRP measurements (RSRP2) predicted using non-machine learning operations are greater than or equal to a threshold RSRP metric (δ) (e.g., the threshold RSRP metric δ satisfied in the case of |RSRP1 - RSRP2| ≥ δ), then the first set of beam quality metrics (e.g., the first set of RSRP measurements) predicted by machine learning model 215 may satisfy the threshold RSRP metric (δ). For example, if the MMSE is greater than the threshold δ (where RSRP predicted is the RSRP measurement predicted using machine learning model 215, and where RSRP estimated is the RSRP measurement predicted using other non-machine learning-based mathematical operations), then UE 115-a may determine the MMSE based on MMSE = mean{(RSRP predicted - RSRP estimated ) 2} and may report the input / output (e.g., via control message 240).

[0136] In such cases where the threshold δ is satisfied, this may indicate that the model output 220 of machine learning model 215 significantly deviates from what would be predicted using a non-machine learning-based algorithm, and the input and / or output should be reported to the network for offline training.

[0137] In some aspects, the control signaling 230 may additionally or alternatively indicate a threshold for determining whether / when the UE 115-a should use the model output 220 to perform online model training at the UE 115-a. As previously mentioned herein, online model training / adaptation may utilize recorded new data (e.g., the model output 220) to fine-tune the machine learning model 215 (as opposed to offline training techniques used by the network to reconfigure the model). Such online training techniques may be used to adapt the machine learning model 215 to a new data distribution (e.g., new model inputs 210 resulting from changing channel conditions) using pseudo-labels.

[0138] Thus, in the context of beam management, model outputs 220 with a high confidence probability (e.g., a high probability metric indicated by the confidence probability vector 320 in Figure 3 ) may be used for pseudo-labels and may be re-input into the machine learning model 215 to fine-tune and perform online model training.

[0139] For example, reference will again be made to Figure 3 . Based on the model input 305 provided to the model, Figure 3 the model shown may generate a model output 315 including a confidence probability vector 320 that includes probability metrics associated with respective Rx beams. Additionally, the network may configure or indicate a threshold τ = 0.8 for online model training. In this example, the UE 115 may determine that any probability metric greater than the threshold τ = 0.8 meets the threshold and may thus be used for online model training (e.g., re-input back into the model for online training). In other words, based on the confidence probability vector 320, probability metrics with a high probability that meet the threshold τ = 0.8 may be set as pseudo-labels, where the corresponding model input 305 and the generated pseudo-labels may be recorded and re-input into the machine learning model for online training to fine-tune the model.

[0140] As described herein, both the thresholds β and τ may be associated with the probability metrics of the confidence probability vector 320. However, the threshold β for reporting the output of the machine learning model is different from the threshold τ for online model training. Specifically, τ is greater than β. For example, the network may set the thresholds to β = 0.5 and τ = 0.9. In such cases, the UE 115-a may report beam quality metrics (e.g., probability metrics) for offline training in cases where none of the values in the confidence probability vector 320 are greater than β = 0.5, and may utilize beam quality metrics (e.g., probability metrics) for online training for the corresponding beam quality metrics that are greater than τ = 0.9.

[0141] Reference will again be made to Figure 2。In some cases, it can be expected that the model output 220 predicted by the machine learning model 215 is coherent in the time domain, frequency domain, and / or spatial domain. Therefore, when applied to beam management and prediction, it can be expected that the machine learning prediction is consistent within a given time window. In other words, when predicting which Rx beams should be used within a short time interval, UE 115-a can expect that the machine learning model 215 will predict that the same Rx beams will exhibit the best performance within the short time interval, rather than predicting that five different Rx beams will exhibit the best performance within different subsets of the short time interval.

[0142] In other words, in some aspects, the wireless device can expect time domain consistency with the model output 220 of the machine learning model 215. The term "time domain consistency" means that the predictions within a given time window should be consistent. The duration of the consistency window can depend on different factors, such as the moving speed of UE 115-a. For example, a slower moving speed can result in a slower change in the channel environment, and thus the model output 220 should change more slowly and less significantly over time (e.g., a longer consistency window where the model output 220 is expected to be consistent / coherent). In comparison, a faster moving speed can result in a faster change in the channel environment, and thus the model output 220 should change more quickly and more significantly over time (e.g., a shorter consistency window where the model output 220 is expected to be consistent / coherent).

[0143] Therefore, in some aspects, the network entity 105-a can indicate (e.g., via the control signaling 230) the consistency window, in which UE 115-a is configured to selectively modify the prediction of the machine learning model 215 such that the model output 220 is consistent / coherent within the consistency window. In other words, within the window, the beam prediction (and / or other model output 220) should be the same with a high probability. The length of the consistency window can be defined or indicated by the network, where. In addition, in some cases, the length of the consistency window can be adapted to network conditions or different deployment scenarios (e.g., a shorter length of the consistency window in an indoor or high traffic scenario with rapidly changing network conditions).

[0144] For example, UE 115-a can utilize the machine learning model 215 for a consistency window that includes ten different time intervals (e.g., having time intervals {RS t0 ,RS t1 ,...,RS t9}(consistency window) for beam prediction. The machine learning model can make ten predictions corresponding to the Rx beams used in ten corresponding time intervals, as indicated by the beam indices [1, 2, 1, 1, 0, 1, 1, 2, 1, 1]. In this example, since it is expected that the model output 220 (e.g., Rx beam index prediction) is coherent / consistent within the consistency window, UE 115-a can selectively adjust the model output 220 to [1, 1, 1, 1, 1, 1, 1, 1, 1, 1]. In this example, the selectively adjusted sample / model output 220 (e.g., RS t1 , RS t4 , RS t7 )) can be used for model training (e.g., using {RS t1 , beam index 1}, {RS t4 , beam index 1}, {RS t7 , beam index 1} for additional model fine-tuning / training).

[0145] In some cases, the consistency window can be defined in the time domain, frequency domain, and / or spatial domain. In other words, it can be expected that the model output 220 is coherent / consistent with respect to the time domain, frequency domain, and / or spatial domain (e.g., for a given time interval at a given spatial position, it should be expected that the machine learning model 215 predicts the same Rx beam to be applied to a certain number of consecutive frequency bands).

[0146] In the case of the model output 220 including predicted measurements such as predicted RSRP measurements, it can be expected that the predicted RSRP measurements vary smoothly between adjacent model outputs 220 within the consistency window. The system (network or standard) can define a filtering method to handle the predictions within the consistency window. The filtering methods for selectively modifying the model output 220 within the consistency window can include but are not limited to averaging operations, infinite impulse response (IIR) filters (e.g., 1-tap IIR filters), etc.

[0147] For example, the model output 220 may include predicted RSRP values for one Rx beam at different times: RSRP (dBm) = [10, 11, 13, 40, 14, 11, 12,...]. In this example, UE 115-a may filter the RSRP value "40" based on the assumption that the predicted RSRP measurements are expected to vary smoothly between adjacent model outputs 220 within the consistency window. For example, UE 115-a may perform an averaging operation based on adjacent values (e.g., selectively modify the "40" value to a "12.25" value by the following operation: (11 + 13 + 14 + 11) / 4 = 12.25, such that the model output is adjusted from [10, 11, 13, 40, 14, 11, 12,...] to [10, 11, 13, 12.25, 14, 11, 12,...]. In this example, the model output 220 for t3 (e.g., {RS t3 , 12.25}) may be recorded and used for online and / or offline model training / adaptation.

[0148] The techniques described herein can enable efficient training of machine learning models for beam prediction while reducing the control signaling overhead for performing model training. Specifically, by configuring UE 115-a with thresholds for reporting the input / output of the machine model to the network for offline training, the techniques described herein can enable UE 115-a to report the model input / output only when offline training is likely to be expected or needed. Thus, the techniques described herein can reduce the control signaling overhead associated with offline training, reduce the frequency of offline training, and reduce the latency of beam prediction performed using the machine learning model. Additionally, by configuring UE 115-a with thresholds for reporting the input / output of the machine model to the network for offline training, the techniques described herein can enable UE 115-a to perform online training using model outputs with high confidence values, which are expected to improve the efficiency and reliability of the machine learning model for performing future beam predictions.

[0149] Figure 4 Illustrates an example of a process flow 400 that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure. In some examples, aspects of the process flow 400 may implement aspects of the wireless communication system 100, the wireless communication system 200, the machine learning model configuration 300, or any combination thereof, or may be implemented by these aspects. For example, the process flow 400 illustrates techniques for determining whether UE 115-b will report measurements for offline model training, as previously described herein.

[0150] The process flow 400 includes UE 115-b and network entity 105-b, which may be examples of wireless devices as described herein. For example, Figure 4 the illustrated UE 115-b and network entity 105-b may include examples of UE 115-a and network entity 105-a as Figure 2 respectively illustrated.

[0151] In some examples, the operations illustrated in process flow 400 may be performed by hardware (e.g., including circuits, processing blocks, logic components, and other components), code executed by a processor (e.g., software or firmware), or any combination thereof. Alternative examples may be implemented, where some steps are performed in an order different from the described order or not performed at all. In some cases, the steps may include additional features not mentioned below, or other steps may be added.

[0152] At 405, UE 115-b may send capability signaling to network entity 105-b. The capability signaling may indicate one or more capabilities associated with the ability of UE 115-b to perform future beam prediction, such as supported machine learning models, processing capabilities, memory capabilities, etc. For example, the capability signaling may indicate one or more machine learning models available for performing beam prediction supported by UE 115-b.

[0153] At 410, UE 115-b may receive control signaling from network entity 105-b that indicates the machine learning model to be used for beam prediction at UE 115-b. For example, the control signaling may indicate the machine learning model supported by UE 115-b, as indicated by the capability signaling at 405. In this regard, network entity 105-b may indicate what input will be used for the machine learning model and what output UE 115-b will use the machine learning model to predict for beam prediction.

[0154] In some aspects, the control signaling may additionally indicate one or more thresholds for reporting the output from the machine learning model. For example, as described herein, the control signaling may indicate thresholds β and / or δ that UE 115-b uses to determine whether / when it is expected that UE 115-b reports the output from the machine learning model to the network for offline training.

[0155] At 415, network entity 105-b may send a set of reference signals (e.g., SSB) to UE 115-b. Network entity 105-b may send the reference signals that will be used by UE 115-b to estimate the channel conditions and perform beam prediction. In this regard, network entity 105-b may send the reference signals at 415 based on receiving the capability signaling at 405, sending the control signaling at 410, or both.

[0156] In some aspects, UE 115-b may receive reference signals using one or more Rx beams at UE 115-b. For example, UE 115-b may scan a set of Rx beams (e.g., Rx beam 1, Rx beam 2, Rx beam 3, etc.) to receive corresponding reference signals in order to evaluate the relative performance or efficiency of the corresponding Rx beams for receiving communications from network entity 105-b.

[0157] At 420, UE 115-b may perform a set of measurements on a set of reference signals received at 415. For example, in a case where UE 115-b receives reference signals with a set of Rx beams including a first Rx beam, a second Rx beam, and a third Rx beam, UE 115-b may perform measurements for each of the first Rx beam, the second Rx beam, and the third Rx beam. Additionally or alternatively, UE 115-b may perform measurements using the same Rx beam at different time points. For example, network entity 105-b may send reference signals within a first time interval (T1), a second time interval (T2), and a third time interval (T3). In this example, UE 115-b may use the same Rx beam (or a set of Rx beams) for each of the corresponding time intervals to receive reference signals and thus may perform measurements to evaluate the relative performance of the Rx beam over time.

[0158] The first set of measurements may include but are not limited to RSRP, RSRQ, SNR, SINR, CQI, RSSI, or any combination thereof. In some cases, the measurements performed by UE 115-b may be based on what inputs the machine learning model will use for beam prediction. In this regard, UE 115-b may perform a set of measurements at 420 based on sending capability signaling at 405, receiving control signaling at 410, receiving reference signals at 415, or any combination thereof.

[0159] At 435, UE 115-b may predict a first set of beam quality metrics associated with the Rx beams at UE 115-b based on inputting the first set of measurements into a machine learning model. The first set of beam quality metrics may include predicted metrics associated with the Rx beams within one or more future time intervals. In some cases, the Rx beam associated with the output / prediction of the machine learning model may be the same as or different from the Rx beam used to receive the reference signals at 415.

[0160] As previously described herein, the first set of beam quality metrics may include various different types of beam quality metrics. For example, as referred to Figure 3As shown and described, the first set of beam quality metrics may include a set of probability metrics (e.g., confidential probability vector 320) indicating the relative probability that a corresponding beam in a set of Rx beams meets a threshold beam performance. In an additional or alternative embodiment, the set of beam quality metrics may include predicted RSRP values, RSRQ values, SNR values, SINR values, CQI values, RSSI values, or any combination thereof.

[0161] UE 115-b may additionally be configured to use a machine learning model to predict a set of beam indices corresponding to the first set of beam quality metrics. For example, as shown and described with reference to Figure 3 UE 115-b may use a machine learning model to predict a set of beam indices (e.g., beam prediction vector 325) corresponding to the first set of beam quality metrics (e.g., confidential probability vector 320).

[0162] At 430, UE 115-b may determine whether the first set of beam quality metrics (e.g., RSRP values, probability metrics) meets one or more thresholds for reporting model output. In this regard, UE 115-b may perform the determination at 430 based on sending capability signaling at 405, receiving control signaling at 410, receiving reference signals at 415, performing measurements at 420, predicting the first set of beam quality metrics at 425, or any combination thereof.

[0163] The thresholds (e.g., β and / or δ) for performing the determination at 430 may be based on the type of beam quality metrics predicted at 425. For example, in the case where the first set of beam quality metrics includes predicted probability metrics (e.g., confidential probability vector 320), UE 115-b may compare the predicted probability metrics with a confidence / probability threshold β. In this example, if all probability metrics (e.g., all values of the confidential probability vector 320) are less than or equal to the threshold β (e.g., a threshold satisfied in the case where all probability metrics ≤ β), then the first set of beam quality metrics may meet the threshold β. In such cases where the threshold β is met, this may indicate that the output of the machine learning model is associated with a relatively low confidence value, and the input and / or output should be reported to the network for offline training.

[0164] As another example, in the case where the first set of beam quality metrics includes predicted RSRP measurements associated with a set of Rx beams, UE 115-b may compare the predicted RSRP measurements with actual RSRP measurements predicted using a different mathematical operation (e.g., a traditional algorithm) from the machine learning model, and may compare the difference between the predicted RSRP measurements with a threshold RSRP metric (δ).

[0165] For example, the first set of beam quality metrics may include predicted RSRP measurements associated with a set of Rx beams predicted by a machine learning model. In this example, UE 115-b may use non-machine learning-based techniques (e.g., traditional algorithms) to perform a second set of measurements. UE 115-b may then determine the difference (e.g., MMSE metric) between the first set of RSRP measurements (RSRP1) predicted by the machine learning model and the second set of RSRP measurements (RSRP2) predicted by the non-machine learning-based algorithm. In this example, if one or more differences between the first set of RSRP measurements (RSRP1) and the second set of RSRP measurements (RSRP2) are greater than or equal to a threshold RSRP metric (δ) (e.g., a threshold RSRP metric δ satisfied in the case of |RSRP1 - RSRP2| ≥ δ), then the first set of beam quality metrics (e.g., the first set of RSRP measurements) predicted by the model may satisfy the threshold RSRP metric (δ). In such cases where the threshold δ is satisfied, this may indicate that the output of the machine learning model significantly deviates from what would be predicted using non-machine learning-based algorithms, and the input and / or output should be reported to the network for offline training.

[0166] In the case where the first set of beam quality metrics predicted at 425 fails to meet one or more thresholds (e.g., β, δ) at 430 (e.g., step 430 = "no"), process flow 400 may proceed to 450. In other words, if the first set of beam quality metrics fails to meet the threshold for reporting the output of the machine learning model, UE 115-b may avoid reporting the measurements (input) and / or output from the model.

[0167] In comparison, in the case where the first set of beam quality metrics predicted at 425 meets one or more thresholds (e.g., β, δ) at 430 (e.g., step 430 = "yes"), process flow 400 may proceed to 435.

[0168] At 435, UE 115-b may send a control message to network entity 105-b, where the control message indicates the first set of measurements input to the machine learning model. In this regard, in the case where the threshold is met at 430, UE 115-b may report the measurements input to the machine learning model so that the network can utilize the measurements to perform offline training for the machine learning model. In some cases, the control message may additionally indicate the first set of beam quality metrics output / predicted by the machine learning model.

[0169] At 440, network entity 105-b may utilize the information reported at 435 to further train and update the machine learning model (e.g., offline training). For example, network entity 105-b may input the measurements and / or the first set of beam quality metrics into the machine learning model to perform offline training.

[0170] At 445, UE 115-b may receive control signaling from network entity 105-b indicating an updated version of a machine learning model. In this regard, UE 115-b may receive an updated version of the machine learning model, which will be used for future beam prediction. UE 115-b may receive the control signaling at 445 based on sending measurements and / or model outputs to network entity 105-b at 435.

[0171] At 450, UE 115-b may select one or more Rx beams at UE 115-b to be used for receiving communications from network entity 105-b. For example, as Figure 3 shown and described, UE 115-b may select an Rx beam associated with the highest probability metric / confidence value, the highest RSRP value, etc. In the case where UE 115-b receives the updated machine learning model at 445, UE 115-b may be configured to utilize the updated machine learning model to predict additional beam quality metrics and may select one or more Rx beams at 450 based on the additional beam quality metrics predicted via the updated machine learning model.

[0172] At 455, UE 115-b may receive one or more messages (e.g., PDSCH message, PDCCH message) from network entity 105-b. Specifically, UE 115-b may use the one or more Rx beams selected at 450 to receive the one or more messages at 455.

[0173] Figure 5 Illustrates an example of process flow 500 supporting techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure. In some examples, aspects of process flow 500 may implement aspects of wireless communication system 100, wireless communication system 200, machine learning model configuration 300, process flow 400, or any combination thereof, or may be implemented by these aspects. For example, process flow 500 illustrates techniques for determining which model outputs will be used for online model training, as previously described herein.

[0174] Process flow 500 includes UE 115-c and network entity 105-c, which may be examples of wireless devices as described herein. For example, Figure 5 the illustrated UE 115-c and network entity 105-c may include examples of UE 115-a and network entity 105-a as Figure 2 respectively shown.

[0175] In some examples, the operations illustrated in process flow 500 may be performed by hardware (e.g., including circuitry, processing blocks, logic components, and other components), code executed by a processor (e.g., software or firmware), or any combination thereof. Alternative examples may be implemented, where some steps are performed in an order different from that described or not performed at all. In some cases, steps may include additional features not mentioned below, or other steps may be added.

[0176] At 505, UE 115-c may send capability signaling to network entity 105-c. The capability signaling may indicate one or more capabilities associated with the ability of UE 115-c to perform future beam prediction, such as supported machine learning models, processing capabilities, memory capabilities, etc. For example, the capability signaling may indicate one or more machine learning models available for performing beam prediction supported by UE 115-c.

[0177] At 510, UE 115-c may receive control signaling from network entity 105-c that indicates the machine learning model to be used for beam prediction at UE 115-c. For example, the control signaling may indicate a machine learning model supported by UE 115-c, as indicated via the capability signaling at 505. In this regard, network entity 105-c may indicate what inputs will be used for the machine learning model and what outputs UE 115-c will use the machine learning model to predict for beam prediction.

[0178] In some aspects, the control signaling may additionally indicate one or more thresholds for performing online model training at UE 115-c. For example, as described herein, the control signaling may indicate a threshold τ that UE 115-c uses to determine whether / when UE 115-c may utilize the inputs / outputs of the machine learning model to perform online model training at UE 115-c.

[0179] At 515, network entity 105-c may send a set of reference signals (e.g., SSB) to UE 115-c. Network entity 105-c may send the reference signals that will be used by UE 115-c to estimate the channel conditions and perform beam prediction. In this regard, network entity 105-c may send the reference signals at 515 based on receiving the capability signaling at 505, sending the control signaling at 510, or both.

[0180] In some aspects, UE 115-c may receive reference signals using one or more Rx beams at UE 115-c. For example, UE 115-c may scan a set of Rx beams (e.g., Rx beam 1, Rx beam 2, Rx beam 3, etc.) to receive corresponding reference signals in order to evaluate the relative performance or efficiency of the corresponding Rx beams for receiving communications from network entity 105-c.

[0181] At 520, UE 115-c may perform a set of measurements on a set of reference signals received at 515. For example, in a case where UE 115-c receives reference signals having a set of Rx beams including a first Rx beam, a second Rx beam, and a third Rx beam, UE 115-c may perform measurements for each of the first Rx beam, the second Rx beam, and the third Rx beam. Additionally or alternatively, UE 115-c may perform measurements using the same Rx beam at different time points. For example, network entity 105-c may send reference signals within a first time interval (T1), a second time interval (T2), and a third time interval (T3). In this example, UE 115-c may use the same Rx beam (or a set of Rx beams) to receive reference signals for each of the corresponding time intervals, and thus may perform measurements to evaluate the relative performance of the Rx beam over time.

[0182] The first set of measurements may include, but is not limited to, RSRP, RSRQ, SNR, SINR, CQI, RSSI, or any combination thereof. In some cases, the measurements performed by UE 115-c may be based on what inputs the machine learning model will use for beam prediction. In this regard, UE 115-c may perform a set of measurements at 520 based on sending capability signaling at 505, receiving control signaling at 510, receiving reference signals at 515, or any combination thereof.

[0183] At 435, UE 115-c may predict a first set of beam quality metrics associated with the Rx beam at UE 115-c based on inputting the first set of measurements into a machine learning model. The first set of beam quality metrics may include predicted metrics associated with the Rx beam within one or more future time intervals. In some cases, the Rx beam associated with the output / prediction of the machine learning model may be the same as or different from the Rx beam used to receive the reference signals at 515.

[0184] As previously described herein, the first set of beam quality metrics may include various different types of beam quality metrics. For example, as referred to Figure 3As shown and described, the first set of beam quality metrics may include a set of probability metrics (e.g., the confidential probability vector 320) indicating the relative probability that a corresponding beam in a set of Rx beams meets a threshold beam performance. In additional or alternative embodiments, the set of beam quality metrics may include predicted RSRP values, RSRQ values, SNR values, SINR values, CQI values, RSSI values, or any combination thereof.

[0185] UE 115-c may additionally be configured to use a machine learning model to predict a set of beam indices corresponding to the first set of beam quality metrics. For example, as referred to Figure 3 As shown and described, UE 115-c may use a machine learning model to predict a set of beam indices (e.g., the beam prediction vector 325) corresponding to the first set of beam quality metrics (e.g., the confidential probability vector 320).

[0186] At 530, UE 115-c may determine whether the first set of beam quality metrics (e.g., probability metrics) meets one or more thresholds for reporting model output. In this regard, UE 115-c may perform the determination at 530 based on sending capability signaling at 505, receiving control signaling at 510, receiving reference signals at 515, performing measurements at 520, predicting the first set of beam quality metrics at 525, or any combination thereof.

[0187] For example, in the case where the first set of beam quality metrics includes predicted probability metrics (e.g., the confidential probability vector 320), UE 115-c may compare the predicted probability metrics with a confidence / probability threshold τ. In this example, the beam quality metrics in the first set that are greater than or equal to the threshold τ may meet the threshold (e.g., the threshold τ satisfied in the case where the probability metric ≥ τ). In such cases where the threshold τ is met, this may indicate that the corresponding output of the machine learning model is associated with a relatively high confidence value and may thus be used to further train the model using online training.

[0188] In some cases, UE 115-c may generate pseudo-labels for the beam quality metrics in the first set that meet the threshold τ. For example, the prediction / output with the highest probability metric / confidence value (e.g., the highest confidential probability vector 320) may be set as the pseudo-label.

[0189] In the case where the first set of beam quality metrics predicted at 325 fails to meet one or more thresholds (e.g., τ) at 530 (e.g., step 530 = "no"), the process flow 500 may proceed to 540. In other words, if the first set of beam quality metrics fails to meet the threshold for online training, UE 115-c may avoid using the output and corresponding measurements to perform online training.

[0190] In comparison, when at least one beam quality metric among the first set of beam quality metrics predicted at 525 meets one or more thresholds (e.g., τ) at 530 (e.g., step 530 = "Yes"), the process flow 500 may proceed to 435.

[0191] At 535, UE 115-c may perform online training for the machine learning model by inputting the beam quality metrics that meet the thresholds into the machine learning model. Additionally, UE 115-b may input the corresponding measurements and / or the generated pseudo-labels input into the model into the machine learning model to perform online training.

[0192] At 540, UE 115-c may select, at UE 115-c, one or more Rx beams to be used to receive communications from network entity 105-c. For example, as Figure 3 shown and described, UE 115-c may select the Rx beam associated with the highest probability metric / confidence value, the highest RSRP value, etc. In the case where UE 115-c performs online training at 535 to update the machine learning model, UE 115-c may be configured to use the updated machine learning model to predict additional beam quality metrics, and may select one or more Rx beams at 540 based on the additional beam quality metrics predicted via the updated machine learning model.

[0193] At 545, UE 115-c may receive one or more messages (e.g., PDSCH message, PDCCH message) from network entity 105-c. Specifically, UE 115-b may use the one or more Rx beams selected at 540 to receive one or more messages at 545.

[0194] Figure 6 FIG. 600 illustrates a block diagram of a device 605 that supports techniques for using model adaptation for beam management in accordance with one or more aspects of the present disclosure. Device 605 may be an example of aspects of network entity 105 as described herein. Device 605 may include a receiver 610, a transmitter 615, and a communication manager 620. Device 605 may also include a processor. Each of these components may communicate with each other (e.g., via one or more buses).

[0195] The receiver 610 may provide components for obtaining (e.g., receiving, determining, identifying) information associated with various channels (e.g., control channels, data channels, information channels, channels associated with a protocol stack) such as user data, control information, or any combination thereof (e.g., I / Q samples, symbols, packets, protocol data units, service data units). The information may be delivered to other components of the device 605. In some examples, the receiver 610 may support obtaining information by receiving signals via one or more antennas. Additionally or alternatively, the receiver 610 may support obtaining information by receiving signals via one or more wired (e.g., electrical, optical fiber) interfaces, wireless interfaces, or any combination thereof.

[0196] The transmitter 615 may provide components for outputting (e.g., transmitting, providing, conveying, delivering) information generated by other components of the device 605. For example, the transmitter 615 may output information associated with various channels (e.g., control channels, data channels, information channels, channels associated with a protocol stack) such as user data, control information, or any combination thereof (e.g., I / Q samples, symbols, packets, protocol data units, service data units). In some examples, the transmitter 615 may support outputting information by transmitting signals via one or more antennas. Additionally or alternatively, the transmitter 615 may support outputting information by transmitting signals via one or more wired (e.g., electrical, optical fiber) interfaces, wireless interfaces, or any combination thereof. In some examples, the transmitter 615 and the receiver 610 may be co-located in a transceiver, which may include a modem or be coupled to a modem.

[0197] The communication manager 620, the receiver 610, the transmitter 615, or various combinations or various components thereof may be examples of components for performing aspects of the techniques for beam management using model adaptation as described herein. For example, the communication manager 620, the receiver 610, the transmitter 615, or various combinations or components thereof may support methods for performing one or more of the functions described herein.

[0198] In some examples, the communication manager 620, the receiver 610, the transmitter 615, or various combinations or components thereof may be implemented in hardware (e.g., in a communication management circuit). The hardware may include a processor, a DSP, a CPU, an ASIC, an FPGA, or other programmable logic device, a microcontroller, discrete gate or transistor logic, discrete hardware components, or any combination thereof configured to or otherwise supporting components for performing the functions described in this disclosure. In some examples, a processor and a memory coupled to the processor may be configured to perform one or more of the functions described herein (e.g., by the processor executing instructions stored in the memory).

[0199] Additionally or alternatively, in some examples, the communication manager 620, the receiver 610, the transmitter 615, or various combinations or components thereof may be implemented in code executed by a processor (e.g., implemented as communication management software or firmware). If implemented in code executed by a processor, the functions of the communication manager 620, the receiver 610, the transmitter 615, or various combinations or components thereof may be performed by a general-purpose processor, a DSP, a CPU, an ASIC, an FPGA, a microcontroller, or any combination of these or other programmable logic devices (e.g., components configured or otherwise supporting the functions described in this disclosure).

[0200] In some examples, the communication manager 620 may be configured to perform various operations (e.g., receive, obtain, monitor, output, transmit) using or otherwise in cooperation with the receiver 610, the transmitter 615, or both. For example, the communication manager 620 may receive information from the receiver 610, convey information to the transmitter 615, or integrate in combination with the receiver 610, the transmitter 615, or both to obtain information, output information, or perform various other operations as described herein.

[0201] For example, the communication manager 620 may be configured to or otherwise support components for receiving control signaling from a network entity indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The communication manager 620 may be configured to or otherwise support components for performing a first set of measurements on a set of reference signals received from a network entity. The communication manager 620 may be configured to or otherwise support components for predicting a first set of beam quality metrics associated with a set of Rx beams at the UE based on inputting the first set of measurements into a machine learning model. The communication manager 620 may be configured to or otherwise support components for sending a control message indicating the first set of measurements to the network entity based on the first set of beam quality metrics meeting one or more thresholds.

[0202] For example, the communication manager 620 may be configured to or otherwise support components for sending control signaling to a UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting outputs from the machine learning model. The communication manager 620 may be configured to or otherwise support components for sending a set of reference signals to the UE based on the control signaling. The communication manager 620 may be configured to or otherwise support components for receiving, from the UE, a control message indicating a first set of measurements associated with the set of reference signals, where the control message is received based on a first set of beam quality metrics associated with the first set of measurements satisfying one or more thresholds, and where the first set of beam quality metrics includes the output of the machine learning model.

[0203] By including or configuring a communication manager 620 according to the examples described herein, a device 605 (e.g., a control receiver 610, a transmitter 615, the communication manager 620, or a processor combined with or otherwise coupled to them) may support techniques that enable efficient training of a machine learning model for beam prediction while reducing control signaling overhead for performing model training. Specifically, by configuring the UE 115 with thresholds for reporting the input / output of a machine model to the network for offline training, the techniques described herein may enable the UE 115 to report model input / output only when offline training is likely to be expected or needed. Thus, the techniques described herein may reduce control signaling overhead associated with offline training, reduce the frequency of offline training, and reduce the latency of beam prediction performed using the machine learning model. Additionally, by configuring the UE 115 with thresholds for reporting the input / output of a machine model to the network for offline training, the techniques described herein may enable the UE 115 to perform online training using model outputs with high confidence values, which are expected to improve the efficiency and reliability of the machine learning model for performing future beam prediction.

[0204] Figure 7 Block diagram 700 illustrates a device 705 supporting techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure. The device 705 may be an example of aspects of the device 605 or the network entity 105 described herein. The device 705 may include a receiver 710, a transmitter 715, and a communication manager 720. The device 705 may also include a processor. Each of these components may communicate with each other (e.g., via one or more buses).

[0205] The receiver 710 may provide components for obtaining (e.g., receiving, determining, identifying) information associated with various channels (e.g., control channels, data channels, information channels, channels associated with a protocol stack) such as user data, control information, or any combination thereof (e.g., I / Q samples, symbols, packets, protocol data units, service data units). The information may be delivered to other components of the device 705. In some examples, the receiver 710 may support obtaining information by receiving signals via one or more antennas. Additionally or alternatively, the receiver 710 may support obtaining information by receiving signals via one or more wired (e.g., electrical, optical fiber) interfaces, wireless interfaces, or any combination thereof.

[0206] The transmitter 715 may provide components for outputting (e.g., transmitting, providing, conveying, delivering) information generated by other components of the device 705. For example, the transmitter 715 may output information associated with various channels (e.g., control channels, data channels, information channels, channels associated with a protocol stack) such as user data, control information, or any combination thereof (e.g., I / Q samples, symbols, packets, protocol data units, service data units). In some examples, the transmitter 715 may support outputting information by transmitting signals via one or more antennas. Additionally or alternatively, the transmitter 715 may support outputting information by transmitting signals via one or more wired (e.g., electrical, optical fiber) interfaces, wireless interfaces, or any combination thereof. In some examples, the transmitter 715 and the receiver 710 may be co-located in a transceiver, which may include a modem or be coupled to a modem.

[0207] The device 705 or its various components may be examples of components for performing aspects of the techniques for model-adaptive beam management as described herein. For example, the communication manager 720 may include a control signaling reception manager 725, a measurement manager 730, a machine learning model manager 735, a control message transmission manager 740, a control signaling transmission manager 745, a reference signal transmission manager 750, a control message reception manager 755, or any combination thereof. The communication manager 720 may be an example of aspects of the communication manager 620 as described herein. In some examples, the communication manager 720 or its various components may be configured to perform various operations (e.g., receive, obtain, monitor, output, transmit) using or otherwise in cooperation with the receiver 710, the transmitter 715, or both. For example, the communication manager 720 may receive information from the receiver 710, convey information to the transmitter 715, or integrate in combination with the receiver 710, the transmitter 715, or both to obtain information, output information, or perform various other operations as described herein.

[0208] The control signaling reception manager 725 may be configured to or otherwise support components for receiving control signaling from a network entity indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The measurement manager 730 may be configured to or otherwise support components for performing a first set of measurements on a set of reference signals received from a network entity. The machine learning model manager 735 may be configured to or otherwise support components for predicting a first set of beam quality metrics associated with a set of Rx beams at the UE based on inputting the first set of measurements into the machine learning model. The control message transmission manager 740 may be configured to or otherwise support components for transmitting a control message indicating the first set of measurements to a network entity based on the first set of beam quality metrics meeting one or more thresholds.

[0209] The control signaling transmission manager 745 may be configured to or otherwise support components for transmitting control signaling to the UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The reference signal transmission manager 750 may be configured to or otherwise support components for transmitting a set of reference signals to the UE based on the control signaling. The control message reception manager 755 may be configured to or otherwise support components for receiving from the UE a control message indicating a first set of measurements associated with a set of reference signals, where the control message is received based on a first set of beam quality metrics associated with the first set of measurements meeting one or more thresholds, and where the first set of beam quality metrics includes an output of the machine learning model.

[0210] Figure 8Block diagram 800 illustrates a communication manager 820 that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure. The communication manager 820 may be an example of the communication manager 620, the communication manager 720, or aspects of both as described herein. The communication manager 820 or its various components may be examples of components for performing various aspects of the techniques for beam management using model adaptation as described herein. For example, the communication manager 820 may include a control signaling reception manager 825, a measurement manager 830, a machine learning model manager 835, a control message transmission manager 840, a control signaling transmission manager 845, a reference signal transmission manager 850, a control message reception manager 855, a beam index manager 860, a beam quality metric manager 865, a downlink reception manager 870, a capability signaling transmission manager 875, a capability signaling reception manager 880, a pseudo-label manager 885, or any combination thereof. Each of these components may communicate directly or indirectly with each other (e.g., via one or more buses), which communication may include communication within protocol layers of a protocol stack, communication associated with logical channels of the protocol stack (e.g., between protocol layers of the protocol stack, within devices, components, or virtualized components associated with network entity 105, between devices, components, or virtualized components associated with network entity 105), or any combination thereof.

[0211] The control signaling reception manager 825 may be configured to or otherwise support components for receiving control signaling from a network entity indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The measurement manager 830 may be configured to or otherwise support components for performing a first set of measurements on a set of reference signals received from a network entity. The machine learning model manager 835 may be configured to or otherwise support components for predicting a first set of beam quality metrics associated with a set of Rx beams at a UE based on inputting the first set of measurements into a machine learning model. The control message transmission manager 840 may be configured to or otherwise support components for transmitting a control message indicating the first set of measurements to a network entity based on the first set of beam quality metrics meeting one or more thresholds.

[0212] In some examples, the control signaling reception manager 825 may be configured to or otherwise support components for receiving additional control signaling from a network entity indicating an updated version of a machine learning model, where the additional control signaling is received based on transmitting the control message.

[0213] In some examples, the first set of beam quality metrics includes a set of probability metrics indicating the relative probability that a corresponding beam in a set of Rx beams meets a threshold beam performance. In some examples, the first set of beam quality metrics meets one or more thresholds based on each beam quality metric in the first set of beam quality metrics being less than one or more thresholds.

[0214] In some examples, the measurement manager 830 may be configured to or otherwise support components for performing a second set of measurements on a second set of reference signals received from a network entity. In some examples, the machine learning model manager 835 may be configured to or otherwise support components for predicting a second set of beam quality metrics associated with a set of Rx beams based on the second set of measurements, where the first set of beam quality metrics meets one or more thresholds based on one or more differences between the first set of beam quality metrics and the second set of beam quality metrics being greater than or equal to one or more thresholds.

[0215] In some examples, the first set of beam quality metrics and the second set of beam quality metrics include predicted RSRP measurements associated with a set of Rx beams. In some examples, one or more thresholds include a threshold RSRP metric. In some examples, the second set of beam quality metrics includes RSRP metrics predicted using one or more mathematical operations different from the machine learning model. In some examples, one or more differences between the first set of beam quality metrics and the second set of beam quality metrics include an MMSE metric.

[0216] In some examples, the measurement manager 830 may be configured to or otherwise support components for performing a second set of measurements on a second set of reference signals received from a network entity. In some examples, the machine learning model manager 835 may be configured to or otherwise support components for predicting a second set of beam quality metrics associated with a set of Rx beams based on inputting the second set of measurements into a machine learning model. In some examples, the machine learning model manager 835 may be configured to or otherwise support components for training the machine learning model using the second set of measurements based on at least one beam quality metric in the second set of beam quality metrics meeting one or more additional thresholds.

[0217] In some examples, the pseudo-label manager 885 may be configured to or otherwise support components for generating pseudo-labels associated with at least one beam quality metric in the second set of beam quality metrics that meets one or more additional thresholds. In some examples, the machine learning model manager 835 may be configured to or otherwise support components for inputting the pseudo-labels, at least one beam quality metric in the second set of beam quality metrics, and at least one measurement in the second set of measurements corresponding to the at least one beam quality metric into the machine learning model, where the training is based on the input.

[0218] In some examples, the second set of beam quality metrics includes a set of probability metrics indicating the relative probability that a corresponding beam in a set of Rx beams meets a threshold beam performance. In some examples, based on at least one beam quality metric being greater than or equal to one or more additional thresholds, at least one beam quality metric in the second set of beam quality metrics meets one or more additional thresholds.

[0219] In some examples, the machine learning model manager 835 may be configured to or otherwise support components for predicting a first set of beam indices corresponding to the first set of beam quality metrics, the first set of beam indices indicating predicted Rx beams associated with a set of multiple time instances. In some examples, the beam index manager 860 may be configured to or otherwise support components for selectively modifying beam indices in the first set of beam indices based on one or more differences between a beam index and one or more additional beam indices corresponding to one or more of the multiple time instances within a coherence window.

[0220] In some examples, the control signaling reception manager 825 may be configured to or otherwise support components for receiving an indication of a coherence window via control signaling, wherein selectively modifying the beam index is based on receiving the control signaling.

[0221] In some examples, the beam quality metric manager 865 may be configured to or otherwise support components for selectively modifying at least one predicted beam quality metric in a set of multiple predicted beam quality metrics during a second time interval based on at least one predicted beam quality metric meeting an outlier threshold.

[0222] In some examples, the downlink reception manager 870 may be configured to or otherwise support components for selecting an Rx beam from a set of Rx beams based on the first set of beam quality metrics. In some examples, the downlink reception manager 870 may be configured to or otherwise support components for receiving one or more messages from a network entity using the Rx beam based on the selection.

[0223] In some examples, the capability signaling transmission manager 875 may be configured to or otherwise support components for transmitting capability signaling indicating one or more capabilities associated with the UE to a network entity, wherein receiving the control signaling is based on the capability signaling.

[0224] In some examples, the measurement manager 830 may be configured to or otherwise support components for performing a second set of measurements on a second set of reference signals received from a network entity. In some examples, the machine learning model manager 835 may be configured to or otherwise support components for determining a second set of beam quality metrics associated with a set of Rx beams, a set of additional Rx beams, or both, based on inputting the second set of measurements into a machine learning model. In some examples, the control message sending manager 840 may be configured to or otherwise support components for avoiding reporting the second set of measurements to the network entity based on at least one beam quality metric of the second set of beam quality metrics failing to meet one or more thresholds.

[0225] The control signaling sending manager 845 may be configured to or otherwise support components for sending control signaling to the UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The reference signal sending manager 850 may be configured to or otherwise support components for sending a set of reference signals to the UE based on the control signaling. The control message receiving manager 855 may be configured to or otherwise support components for receiving, from the UE, a control message indicating a first set of measurements associated with a set of reference signals, where the control message is received based on a first set of beam quality metrics associated with the first set of measurements meeting one or more thresholds, and where the first set of beam quality metrics includes an output of the machine learning model.

[0226] In some examples, the machine learning model manager 835 may be configured to or otherwise support components for training a machine learning model based on the first set of measurements, the first set of beam quality metrics, or both. In some examples, the control signaling sending manager 845 may be configured to or otherwise support components for sending additional control signaling to the UE indicating an updated version of the machine learning model based on training the machine learning model.

[0227] In some examples, the first set of beam quality metrics includes a set of probability metrics indicating a relative probability that a corresponding beam in a set of Rx beams meets a threshold beam performance. In some examples, the first set of beam quality metrics meets one or more thresholds based on each beam quality metric of the first set of beam quality metrics being less than one or more thresholds.

[0228] In some examples, the first set of beam quality metrics meets one or more thresholds based on one or more differences between the first set of beam quality metrics and the second set of beam quality metrics being greater than or equal to one or more thresholds.

[0229] In some examples, the first set of beam quality metrics and the second set of beam quality metrics include predicted RSRP measurements associated with a set of Rx beams. In some examples, one or more thresholds include a threshold RSRP metric. In some examples, the second set of beam quality metrics includes RSRP metrics predicted using one or more mathematical operations different from the machine learning model. In some examples, one or more differences between the first set of beam quality metrics and the second set of beam quality metrics include MMSE metrics.

[0230] In some examples, the control signaling further indicates one or more additional thresholds for training the machine learning model at the UE.

[0231] In some examples, the capability signaling reception manager 880 can be configured to or otherwise support components for receiving, from a UE, capability signaling indicating one or more capabilities associated with the UE, wherein the transmission of the control signaling is based on the capability signaling.

[0232] Figure 9 FIG. illustrates a system 900 including a device 905 that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure. The device 905 can be an example of, or include components of, the device 605, the device 705, or the network entity 105 as described herein. The device 905 can communicate with one or more network entities 105, one or more UEs 115, or any combination thereof, and the communication can include communication via one or more wired interfaces, via one or more wireless interfaces, or any combination thereof. The device 905 can include components that support output and obtaining of communication, such as a communication manager 920, a transceiver 910, an antenna 915, a memory 925, code 930, and a processor 935. These components can be electronically communicated or otherwise (e.g., operatively, communicatively, functionally, electronically, electrically) coupled via one or more buses (e.g., bus 940).

[0233] The transceiver 910 may support bidirectional communication via a wired link, a wireless link, or both as described herein. In some examples, the transceiver 910 may include a wired transceiver and may communicate bidirectionally with another wired transceiver. Additionally or alternatively, in some examples, the transceiver 910 may include a wireless transceiver and may communicate bidirectionally with another wireless transceiver. In some examples, the device 905 may include one or more antennas 915, which may be capable of (e.g., concurrently) sending or receiving wireless transmissions. The transceiver 910 may also include a modem for: modulating a signal; providing the modulated signal for transmission (e.g., via one or more antennas 915, via a wired transmitter); receiving the modulated signal (e.g., from one or more antennas 915, from a wired receiver); and demodulating the signal. In some implementations, the transceiver 910 may include one or more interfaces, such as one or more interfaces coupled to one or more antennas 915 configured to support various receiving or obtaining operations, or one or more interfaces coupled to one or more antennas 915 configured to support various sending or outputting operations, or a combination thereof. In some implementations, the transceiver 910 may include or be configured to be coupled to one or more processors or memory components, which are operable to perform or support operations based on received or obtained information or signals, or to generate information or other signals for transmission or other output, or any combination thereof. In some implementations, the transceiver 910, or the transceiver 910 and one or more antennas 915, or the transceiver 910 and one or more antennas 915 and one or more processors or memory components (e.g., processor 935 or memory 925 or both) may be included in a chip or chip assembly installed in the device 905. In some examples, the transceiver may be operable to support communications via one or more communication links (eg, communication link 125, backhaul communication link 120, midhaul communication link 162, fronthaul communication link 168).

[0234] The memory 925 may include RAM and ROM. The memory 925 may store computer-readable, computer-executable code 930 including instructions that, when executed by the processor 935, cause the device 905 to perform various functions described herein. The code 930 may be stored in a non-transitory computer-readable medium (such as system memory or another type of memory). In some cases, the code 930 may not be directly executable by the processor 935, but may (e.g., when compiled and executed) cause the computer to perform the functions described herein. In some cases, the memory 925 may also include a BIOS, etc., which may control basic hardware or software operations, such as interaction with peripheral components or devices.

[0235] The processor 935 may include intelligent hardware devices (e.g., general-purpose processors, DSPs, ASICs, CPUs, FPGAs, microcontrollers, programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof). In some cases, the processor 935 may be configured to operate a memory array using a memory controller. In some other cases, the memory controller may be integrated into the processor 935. The processor 935 may be configured to execute computer-readable instructions stored in a memory (e.g., memory 925) to cause the device 905 to perform various functions (e.g., functions or tasks supporting techniques for beam management using model adaptation). For example, the device 905 or components of the device 905 may include the processor 935 and the memory 925 coupled to the processor 935, and the processor 935 and the memory 925 are configured to perform the various functions described herein. The processor 935 may be an example of a cloud computing platform (e.g., one or more physical nodes and supporting software such as an operating system, virtual machine, or container instance) that may host functions for performing the functions of the device 905 (e.g., by executing code 930). The processor 935 may be any one or more suitable processors capable of executing scripts or instructions of one or more software programs stored in the device 905 (such as within the memory 925). In some specific implementations, the processor 935 may be a component of a processing system. A processing system generally may refer to a system or series of machines or components that receive inputs and process those inputs to produce a set of outputs (which may be delivered to other systems or components of, for example, the device 905). For example, the processing system of the device 905 may refer to a system including various other components or sub-components of the device 905 (such as the processor 935, or the transceiver 910, or the communication manager 920, or a combination of other components or components of the device 905). The processing system of the device 905 may interface with other components of the device 905 and may process information (such as inputs or signals) received from other components or output information to other components. For example, a chip or modem of the device 905 may include a processing system and one or more interfaces for outputting information or for obtaining information or both. One or more interfaces may be implemented as or otherwise include a first interface configured to output information and a second interface configured to obtain information or the same interface configured to output information and obtain information, etc. In some specific implementations, one or more interfaces may refer to an interface between the processing system of a chip or modem and a transmitter such that the device 905 may transmit information output from the chip or modem. Additionally or alternatively, in some specific implementations, one or more interfaces may refer to an interface between the processing system of a chip or modem and a receiver such that the device 905 may obtain information or signal inputs, and the information may be delivered to the processing system.One of ordinary skill in the art will readily recognize that the first interface may also receive information or signal inputs, and the second interface may also output information or signal outputs.

[0236] In some examples, bus 940 may support communication within a protocol layer of a protocol stack (e.g., within the protocol layer). In some examples, bus 940 may support communication associated with a logical channel of a protocol stack (e.g., between protocol layers of the protocol stack), which may include communication performed within components of device 905, or communication performed between different components of device 905 that may be co-located or located at different locations (e.g., where device 905 may refer to a system in which one or more of communication manager 920, transceiver 910, memory 925, code 930, and processor 935 may be located in one component or divided between different components).

[0237] In some examples, communication manager 920 may manage (e.g., via one or more wired or wireless backhaul links) aspects of communication with core network 130. For example, communication manager 920 may manage the delivery of data communication for client devices such as one or more UEs 115. In some examples, communication manager 920 may manage communication with other network entities 105, and may include a controller or scheduler for coordinating with other network entities 105 to control communication with UEs 115. In some examples, communication manager 920 may support the X2 interface within LTE / LTE-A radio communication network technologies to provide communication between network entities 105.

[0238] For example, communication manager 920 may be configured to or otherwise support components for receiving control signaling from a network entity indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. Communication manager 920 may be configured to or otherwise support components for performing a first set of measurements on a set of reference signals received from a network entity. Communication manager 920 may be configured to or otherwise support components for predicting a first set of beam quality metrics associated with a set of Rx beams at a UE based on inputting the first set of measurements into a machine learning model. Communication manager 920 may be configured to or otherwise support components for sending a control message indicating the first set of measurements to a network entity based on the first set of beam quality metrics meeting one or more thresholds.

[0239] For example, communication manager 920 may be configured to or otherwise support components for sending control signaling to a UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. Communication manager 920 may be configured to or otherwise support components for sending a set of reference signals to the UE based on the control signaling. Communication manager 920 may be configured to or otherwise support components for receiving, from the UE, a control message indicating a first set of measurements associated with the set of reference signals, where the control message is received based on a first set of beam quality metrics associated with the first set of measurements satisfying one or more thresholds, where the first set of beam quality metrics includes an output of the machine learning model.

[0240] By including or configuring communication manager 920 according to examples as described herein, device 905 may support techniques that enable efficient training of a machine learning model for beam prediction while reducing control signaling overhead for performing model training. Specifically, by configuring UE 115 with thresholds for reporting the input / output of a machine model to a network for offline training, the techniques described herein may enable UE 115 to report model input / output only in cases where offline training may be expected or needed. Thus, the techniques described herein may reduce control signaling overhead associated with offline training, reduce the frequency of offline training, and reduce the latency of beam prediction performed using the machine learning model. Additionally, by configuring UE 115 with thresholds for reporting the input / output of a machine model to a network for offline training, the techniques described herein may enable UE 115 to perform online training using model outputs with high confidence values, which are expected to improve the efficiency and reliability of the machine learning model for performing future beam predictions.

[0241] In some examples, communication manager 920 may be configured to use or otherwise cooperate with transceiver 910, one or more antennas 915 (e.g., where applicable), or any combination thereof to perform various operations (e.g., receive, obtain, monitor, output, transmit). Although communication manager 920 is illustrated as a separate component, in some examples, one or more of the functions described with reference to communication manager 920 may be supported or performed by transceiver 910, processor 935, memory 925, code 930, or any combination thereof. For example, code 930 may include instructions executable by processor 935 to cause device 905 to perform aspects of the techniques described herein for model-adaptive beam management, or processor 935 and memory 925 may otherwise be configured to perform or support such operations.

[0242] Figure 10Illustrates a flowchart of method 1000 that illustrates techniques supporting beam management using model adaptation in accordance with one or more aspects of the present disclosure. Operations of method 1000 may be implemented by a network entity or its components as described herein. For example, operations of method 1000 may be performed by a network entity as described with reference to Figures 1 to 9 as described. In some examples, the network entity may execute an instruction set to control functional elements of the network entity to perform the described functions. Additionally or alternatively, the network entity may use dedicated hardware to perform aspects of the described functions.

[0243] At 1005, the method may include: receiving, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The operation of 1005 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1005 may be performed by a control signaling reception manager 825 as described with reference to Figure 8 as described.

[0244] At 1010, the method may include: performing a first set of measurements on a set of reference signals received from a network entity. The operation of 1010 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1010 may be performed by a measurement manager 830 as described with reference to Figure 8 as described.

[0245] At 1015, the method may include: predicting a first set of beam quality metrics associated with a set of Rx beams at a UE based on inputting the first set of measurements into a machine learning model. The operation of 1015 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1015 may be performed by a machine learning model manager 835 as described with reference to Figure 8 as described.

[0246] At 1020, the method may include: sending, to a network entity, a control message indicating the first set of measurements based on the first set of beam quality metrics satisfying one or more thresholds. The operation of 1020 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1020 may be performed by a control message transmission manager 840 as described with reference to Figure 8 as described.

[0247] Figure 11 Illustrates a flowchart of method 1100 that illustrates techniques supporting beam management using model adaptation in accordance with one or more aspects of the present disclosure. Operations of method 1100 may be implemented by a network entity or its components as described herein. For example, operations of method 1100 may be performed by a network entity as described with reference to Figures 1 to 9performed by the described network entity. In some examples, the network entity may execute an instruction set to control functional elements of the network entity to perform the described functions. Additionally or alternatively, the network entity may use dedicated hardware to perform aspects of the described functions.

[0248] At 1105, the method may include: receiving, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The operation of 1105 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1105 may be performed by the control signaling reception manager 825 as described with reference to Figure 8 the described control signaling reception manager 825.

[0249] At 1110, the method may include: performing a first set of measurements on a set of reference signals received from a network entity. The operation of 1110 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1110 may be performed by the measurement manager 830 as described with reference to Figure 8 the described measurement manager 830.

[0250] At 1115, the method may include: predicting a first set of beam quality metrics associated with a set of Rx beams at a UE based on inputting the first set of measurements into a machine learning model. The operation of 1115 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1115 may be performed by the machine learning model manager 835 as described with reference to Figure 8 the described machine learning model manager 835.

[0251] At 1120, the method may include: sending, to the network entity, a control message indicating the first set of measurements based on the first set of beam quality metrics satisfying one or more thresholds. The operation of 1120 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1120 may be performed by the control message sending manager 840 as described with reference to Figure 8 the described control message sending manager 840.

[0252] At 1125, the method may include: receiving, from the network entity, additional control signaling indicating an updated version of the machine learning model, wherein the additional control signaling is received based on sending the control message. The operation of 1125 may be performed in accordance with examples disclosed herein. In some examples, aspects of the operation of 1125 may be performed by the control signaling reception manager 825 as described with reference to Figure 8 the described control signaling reception manager 825.

[0253] Figure 12FIG. 1200 is a flow chart illustrating a method 1200 that supports techniques for beam management using model adaptation in accordance with one or more aspects of the present disclosure. Operations of method 1200 may be implemented by a network entity or components thereof as described herein. For example, operations of method 1200 may be performed by a network entity as described with reference to Figures 1 to 9 The described network entity performs. In some examples, the network entity may execute an instruction set to control functional elements of the network entity to perform the described functions. Additionally or alternatively, the network entity may use dedicated hardware to perform aspects of the described functions.

[0254] At 1205, the method may include: receiving, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at a UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The operation of 1205 may be performed in accordance with examples as disclosed herein. In some examples, aspects of the operation of 1205 may be performed by a control signaling reception manager 825 as described with reference to Figure 8 The described performs.

[0255] At 1210, the method may include: performing a first set of measurements on a set of reference signals received from a network entity. The operation of 1210 may be performed in accordance with examples as disclosed herein. In some examples, aspects of the operation of 1210 may be performed by a measurement manager 830 as described with reference to Figure 8 The described performs.

[0256] At 1215, the method may include: predicting a first set of beam quality metrics associated with a set of Rx beams at a UE based on inputting the first set of measurements into a machine learning model. The operation of 1215 may be performed in accordance with examples as disclosed herein. In some examples, aspects of the operation of 1215 may be performed by a machine learning model manager 835 as described with reference to Figure 8 The described performs.

[0257] At 1220, the method may include: sending, to a network entity, a control message indicating the first set of measurements based on the first set of beam quality metrics satisfying one or more thresholds. The operation of 1220 may be performed in accordance with examples as disclosed herein. In some examples, aspects of the operation of 1220 may be performed by a control message transmission manager 840 as described with reference to Figure 8 The described performs.

[0258] At 1225, the method may include: selecting an Rx beam from a set of Rx beams based on the first set of beam quality metrics. The operation of 1225 may be performed in accordance with examples as disclosed herein. In some examples, aspects of the operation of 1225 may be performed by a downlink reception manager 870 as described with reference to Figure 8 The described performs.

[0259] At 1230, the method may include: receiving, based on a selection, one or more messages from a network entity using an Rx beam. The operations at 1230 may be performed according to examples disclosed herein. In some examples, aspects of the operations at 1230 may be performed by a downlink reception manager 870 as described with reference to Figure 8 what is described.

[0260] Figure 13 FIG. illustrates a flowchart of a method 1300 that illustrates techniques supporting beam management using model adaptation in accordance with one or more aspects of the present disclosure. The operations of method 1300 may be implemented by a network entity or its components as described herein. For example, the operations of method 1300 may be performed by a network entity as described with reference to Figures 1 to 9 what is described. In some examples, the network entity may execute an instruction set to control functional elements of the network entity to perform the described functions. Additionally or alternatively, the network entity may use dedicated hardware to perform aspects of the described functions.

[0261] At 1305, the method may include: sending control signaling to a UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model. The operations at 1305 may be performed according to examples disclosed herein. In some examples, aspects of the operations at 1305 may be performed by a control signaling transmission manager 845 as described with reference to Figure 8 what is described.

[0262] At 1310, the method may include: sending a set of reference signals to the UE based on the control signaling. The operations at 1310 may be performed according to examples disclosed herein. In some examples, aspects of the operations at 1310 may be performed by a reference signal transmission manager 850 as described with reference to Figure 8 what is described.

[0263] At 1315, the method may include: receiving from the UE a control message indicating a first set of measurements associated with a set of reference signals, wherein the control message is received based on a first set of beam quality metrics associated with the first set of measurements satisfying one or more thresholds, wherein the first set of beam quality metrics includes an output of a machine learning model. The operations at 1315 may be performed according to examples disclosed herein. In some examples, aspects of the operations at 1315 may be performed by a control message reception manager 855 as described with reference to Figure 8 what is described.

[0264] An overview of aspects of the present disclosure is provided below:

[0265] Aspect 1: A method for wireless communication at a UE, the method comprising: receiving, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; performing a first set of measurements on a set of reference signals received from the network entity; predicting, at least in part, a first set of beam quality metrics associated with a set of Rx beams at the UE based on inputting the first set of measurements into the machine learning model; and sending, at least in part, a control message indicating the first set of measurements to the network entity based on the first set of beam quality metrics meeting the one or more thresholds.

[0266] Aspect 2: The method according to aspect 1, the method further comprising: receiving, from the network entity, additional control signaling indicating an updated version of the machine learning model, wherein the additional control signaling is received at least in part based on sending the control message.

[0267] Aspect 3: The method according to any one of aspects 1 to 2, wherein the first set of beam quality metrics includes a set of probability metrics indicating a relative probability that a corresponding beam in the set of Rx beams meets a threshold beam performance, and the first set of beam quality metrics meets the one or more thresholds based at least in part on each beam quality metric in the first set of beam quality metrics being less than the one or more thresholds.

[0268] Aspect 4: The method according to any one of aspects 1 to 3, the method further comprising: performing a second set of measurements on a second set of reference signals received from the network entity; and predicting, at least in part, a second set of beam quality metrics associated with the set of Rx beams based on the second set of measurements, wherein the first set of beam quality metrics meets the one or more thresholds based at least in part on one or more differences between the first set of beam quality metrics and the second set of beam quality metrics being greater than or equal to the one or more thresholds.

[0269] Aspect 5: The method according to aspect 4, wherein the first set of beam quality metrics and the second set of beam quality metrics include predicted RSRP measurements associated with the set of Rx beams, and the one or more thresholds include a threshold RSRP metric.

[0270] Aspect 6: The method according to any one of aspects 4 to 5, wherein the second set of beam quality metrics includes RSRP metrics predicted using one or more mathematical operations different from the machine learning model.

[0271] Aspect 7: The method according to any one of Aspects 4 to 6, wherein the one or more differences between the first set of beam quality metrics and the second set of beam quality metrics include an MMSE metric.

[0272] Aspect 8: The method according to any one of Aspects 1 to 7, wherein the control signaling further indicates one or more additional thresholds for training the machine learning model at the UE, and the method further includes: performing a second set of measurements on a second set of reference signals received from the network entity; predicting a second set of beam quality metrics associated with the set of Rx beams at least in part based on inputting the second set of measurements into the machine learning model; and training the machine learning model using the second set of measurements at least in part based on at least one beam quality metric in the second set of beam quality metrics satisfying the one or more additional thresholds.

[0273] Aspect 9: The method according to Aspect 8, the method further includes: generating a pseudo-label associated with the at least one beam quality metric in the second set of beam quality metrics that satisfies the one or more additional thresholds; and inputting the pseudo-label, the at least one beam quality metric in the second set of beam quality metrics, and at least one measurement in the second set of measurements corresponding to the at least one beam quality metric into the machine learning model, wherein the training is at least in part based on the input.

[0274] Aspect 10: The method according to any one of Aspects 8 to 9, wherein the second set of beam quality metrics includes a set of probability metrics indicating a relative probability that a corresponding beam in the set of Rx beams meets a threshold beam performance, and at least one beam quality metric in the second set of beam quality metrics satisfies the one or more additional thresholds at least in part based on the at least one beam quality metric being greater than or equal to the one or more additional thresholds.

[0275] Aspect 11: The method according to any one of Aspects 1 to 10, the method further includes: predicting a first set of beam indices corresponding to the first set of beam quality metrics, the first set of beam indices indicating predicted Rx beams associated with a plurality of time instances; and selectively modifying a beam index in the first set of beam indices at least in part based on one or more differences between the beam index and one or more additional beam indices corresponding to one or more of the plurality of time instances within a coherence window.

[0276] Aspect 12: The method according to Aspect 11, the method further includes: receiving an indication of the coherence window via the control signaling, wherein selectively modifying the beam index is at least in part based on receiving the control signaling.

[0277] Aspect 13: The method according to any one of Aspects 1 to 12, wherein the first set of measurements includes a plurality of measurements performed on a plurality of reference signals received by the UE using an Rx beam from the set of Rx beams within a first time interval, and wherein the first set of beam quality metrics includes a plurality of predicted beam quality metrics associated with the Rx beam within a second time interval, the method further comprising: selectively modifying at least one of the plurality of predicted beam quality metrics within the second time interval at least in part based on the at least one predicted beam quality metric satisfying an outlier threshold.

[0278] Aspect 14: The method according to any one of Aspects 1 to 13, the method further comprising: selecting an Rx beam from the set of Rx beams at least in part based on the first set of beam quality metrics; and receiving one or more messages from the network entity using the Rx beam at least in part based on the selection.

[0279] Aspect 15: The method according to any one of Aspects 1 to 14, the method further comprising: sending capability signaling indicating one or more capabilities associated with the UE to the network entity, wherein receiving the control signaling is at least in part based on the capability signaling.

[0280] Aspect 16: The method according to any one of Aspects 1 to 15, the method further comprising: performing a second set of measurements on a second set of reference signals received from the network entity; determining a second set of beam quality metrics associated with the set of Rx beams, a set of additional Rx beams, or both at least in part based on inputting the second set of measurements into the machine learning model; and avoiding reporting the second set of measurements to the network entity at least in part based on at least one beam quality metric of the second set of beam quality metrics failing to meet the one or more thresholds.

[0281] Aspect 17: A method for wireless communication at a network entity, the method comprising: sending control signaling to a UE indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; sending a set of reference signals to the UE at least in part based on the control signaling; and receiving a control message from the UE indicating a first set of measurements associated with the set of reference signals, wherein the control message is received at least in part based on a first set of beam quality metrics associated with the first set of measurements satisfying the one or more thresholds, wherein the first set of beam quality metrics includes the output of the machine learning model.

[0282] Aspect 18: The method according to aspect 17, the method further comprising: training the machine learning model at least in part based on the first set of measurements, the first set of beam quality metrics, or both; and sending additional control signaling to the UE indicating an updated version of the machine learning model at least in part based on training the machine learning model.

[0283] Aspect 19: The method according to any one of aspects 17 to 18, wherein the first set of beam quality metrics comprises a set of probability metrics indicating a relative probability that a corresponding beam in the set of Rx beams meets a threshold beam performance, and the first set of beam quality metrics meets the one or more thresholds at least in part based on each beam quality metric in the first set of beam quality metrics being less than the one or more thresholds.

[0284] Aspect 20: The method according to any one of aspects 17 to 19, wherein the first set of beam quality metrics meets the one or more thresholds at least in part based on one or more differences between the first set of beam quality metrics and a second set of beam quality metrics being greater than or equal to the one or more thresholds.

[0285] Aspect 21: The method according to aspect 20, wherein the first set of beam quality metrics and the second set of beam quality metrics comprise predicted RSRP measurements associated with the set of Rx beams, and the one or more thresholds comprise a threshold RSRP metric.

[0286] Aspect 22: The method according to any one of aspects 20 to 21, wherein the second set of beam quality metrics comprises RSRP metrics predicted using one or more mathematical operations different from the machine learning model.

[0287] Aspect 23: The method according to any one of aspects 20 to 22, wherein the one or more differences between the first set of beam quality metrics and the second set of beam quality metrics comprise an MMSE metric.

[0288] Aspect 24: The method according to any one of aspects 17 to 23, wherein the control signaling further indicates one or more additional thresholds for training the machine learning model at the UE.

[0289] Aspect 25: The method according to any one of aspects 17 to 24, the method further comprising: receiving, from the UE, capability signaling indicating one or more capabilities associated with the UE, wherein sending the control signaling is at least in part based on the capability signaling.

[0290] Aspect 26: An apparatus, the apparatus comprising: a processor; a memory coupled to the processor; and instructions stored in the memory and executable by the processor to cause the apparatus to perform the method according to any one of Aspects 1 to 16.

[0291] Aspect 27: An apparatus, the apparatus comprising at least one component for performing the method according to any one of Aspects 1 to 16.

[0292] Aspect 28: A non-transitory computer-readable medium storing code including instructions executable by a processor to perform the method according to any one of Aspects 1 to 16.

[0293] Aspect 29: An apparatus, the apparatus comprising: a processor; a memory coupled to the processor; and instructions stored in the memory and executable by the processor to cause the apparatus to perform the method according to any one of Aspects 17 to 25.

[0294] Aspect 30: An apparatus, the apparatus comprising at least one component for performing the method according to any one of Aspects 17 to 25.

[0295] Aspect 31: A non-transitory computer-readable medium storing code including instructions executable by a processor to perform the method according to any one of Aspects 17 to 25.

[0296] It should be noted that the methods described herein depict possible specific implementations, and the operations and steps may be rearranged or otherwise modified and other specific implementations are possible. Additionally, aspects from two or more methods may be combined.

[0297] Although aspects of LTE, LTE-A, LTE-A Pro, or NR systems may be described for example purposes and the LTE, LTE-A, LTE-A Pro, or NR terminology may be used in most of the description, the techniques described herein are also applicable to networks other than LTE, LTE-A, LTE-A Pro, or NR networks. For example, the described techniques may be applicable to various other wireless communication systems such as Ultra Mobile Broadband (UMB), Institute of Electrical and Electronics Engineers (IEEE) 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802.20, Flash-OFDM, and other systems and radio technologies not explicitly mentioned herein.

[0298] The information and signals described herein can be represented using any of a variety of different technologies and methods. For example, data, instructions, commands, information, signals, bits, symbols, and chips mentioned throughout the specification can be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or optical particles, or any combination thereof.

[0299] The various illustrative blocks and components described in connection with the disclosure herein can be implemented or performed using a general-purpose processor, DSP, ASIC, CPU, FPGA, or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but in the alternative, the processor may be any processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices (e.g., a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration).

[0300] The functions described herein can be implemented using hardware, software executed by a processor, firmware, or any combination thereof. When implemented using software executed by a processor, the functions can be stored as one or more instructions or code on a computer-readable medium or transmitted using one or more instructions or code on a computer-readable medium. Other examples and specific implementations are within the scope of the present disclosure and the appended claims. For example, due to the nature of software, the functions described herein can be implemented using software, hardware, firmware, hardwiring, or any combination of these items executed by a processor. The features implementing the functions may also be physically located in different places, including being distributed such that different parts of the functions are implemented in different physical locations.

[0301] A computer-readable medium includes both a non-transitory computer storage medium and a communication medium, where the communication medium includes any medium that facilitates transfer of a computer program from one location to another. The non-transitory storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer. By way of example and not limitation, the non-transitory computer-readable medium can include RAM, ROM, electrically erasable programmable ROM (EEPROM), flash memory, compact disc (CD) ROM or other optical disc storage, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to carry or store desired program code components in the form of instructions or data structures and that can be accessed by a general-purpose or special-purpose computer or a general-purpose or special-purpose processor. Additionally, any connection is properly termed a computer-readable medium. For example, if software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of computer-readable medium. As used herein, disk and disc include CD, laser disc, optical disc, digital versatile disc (DVD), floppy disk, and Blu-ray disc. A disk can magnetically reproduce data, and a disc can optically reproduce data using a laser. Combinations of the above are also included within the scope of computer-readable medium.

[0302] As used herein (including in the claims), the "or" used in a list of items (e.g., a list of items accompanied by phrases such as "at least one of" or "one or more of") indicates an inclusive listing such that, for example, the listing of at least one of A, B, or C means A or B or C or AB or AC or BC or ABC (i.e., A and B and C). Additionally, as used herein, the phrase "based on" should not be construed as a reference to a closed set of conditions. For example, an example step described as "based on condition A" can be based on both condition A and condition B without departing from the scope of the present disclosure. In other words, as used herein, the phrase "based on" should be interpreted in the same manner as the phrase "at least partially based on".

[0303] The term "determine" encompasses a variety of actions, and thus, "determine" can include operations such as computing, calculating, processing, deriving, researching, looking up (such as looking up in a table, database, or other data structure), and ascertaining. Additionally, "determine" can include receiving (e.g., receiving information), accessing (e.g., accessing data stored in a memory), etc. Additionally, "determine" can include parsing, obtaining, selecting, choosing, establishing, and other such similar actions.

[0304] In the drawings, like components or features may have the same reference numerals. Additionally, various components of the same type may be distinguished by following the reference numeral with a dash and a second label used to differentiate among like components. If only the first reference numeral is used in the specification, the description may apply to any one of the like components having the same first reference numeral, regardless of the second reference numeral or any other subsequent reference numerals.

[0305] The description set forth herein in conjunction with the drawings describes exemplary configurations and does not represent all examples that may be implemented or that are within the scope of the claims. The term "exemplary" as used herein means "serving as an example, instance, or illustration" and does not mean "preferred" or "advantageous over other examples". The detailed description includes specific details for providing an understanding of the described techniques. However, the techniques may be practiced without these specific details. In some instances, well-known structures and devices are shown in block diagram form to avoid obscuring the concepts of the described examples.

[0306] The present description is provided to enable a person of ordinary skill in the art to make or use the present disclosure. Various modifications to the present disclosure will be readily apparent to those of ordinary skill in the art, and the general principles defined herein may be applied to other variations without departing from the scope of the present disclosure. Thus, the present disclosure is not limited to the examples and designs described herein and is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An apparatus for wireless communication at a user equipment (UE), the apparatus comprising: a processor; and a memory coupled to the processor, the memory storing instructions executable by the processor to cause the apparatus to: receive, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; perform a first set of measurements on a set of reference signals received from the network entity; predict, at least in part based on inputting the first set of measurements into the machine learning model, a first set of beam quality metrics associated with a set of receive beams at the UE; and send, at least in part based on the first set of beam quality metrics satisfying the one or more thresholds, a control message to the network entity indicating the first set of measurements.

2. The apparatus according to claim 1, wherein the instructions are further executable by the processor to cause the apparatus to: receive, from the network entity, additional control signaling indicating an updated version of the machine learning model, wherein the additional control signaling is received at least in part based on sending the control message.

3. The apparatus according to claim 1, wherein the first set of beam quality metrics includes a set of probability metrics indicating a relative probability that a corresponding beam in the set of receive beams meets a threshold beam performance, and wherein the first set of beam quality metrics satisfies the one or more thresholds at least in part based on each beam quality metric in the first set of beam quality metrics being less than the one or more thresholds.

4. The apparatus according to claim 1, wherein the instructions are further executable by the processor to cause the apparatus to: perform a second set of measurements on a second set of reference signals received from the network entity; and predict, at least in part based on the second set of measurements, a second set of beam quality metrics associated with the set of receive beams, wherein the first set of beam quality metrics satisfies the one or more thresholds at least in part based on one or more differences between the first set of beam quality metrics and the second set of beam quality metrics being greater than or equal to the one or more thresholds.

5. The apparatus according to claim 4, wherein the first set of beam quality metrics and the second set of beam quality metrics include predicted reference signal received power measurements associated with the set of receive beams, and wherein the one or more thresholds include a threshold reference signal received power metric.

6. The apparatus according to claim 4, wherein the second set of beam quality metrics includes reference signal received power metrics predicted using one or more mathematical operations different from the machine learning model.

7. The apparatus according to claim 4, wherein the one or more differences between the first set of beam quality metrics and the second set of beam quality metrics include a least mean square error metric.

8. The apparatus according to claim 1, wherein the control signaling further indicates one or more additional thresholds for training the machine learning model at the UE, and wherein the instructions are further executable by the processor to cause the apparatus to: perform a second set of measurements on a second set of reference signals received from the network entity; predict a second set of beam quality metrics associated with the set of received beams, at least in part by inputting the second set of measurements into the machine learning model; and train the machine learning model using the second set of measurements, at least in part based on at least one beam quality metric in the second set of beam quality metrics satisfying the one or more additional thresholds.

9. The apparatus according to claim 8, wherein the instructions are further executable by the processor to cause the apparatus to: generate a pseudo-label associated with the at least one beam quality metric in the second set of beam quality metrics that satisfies the one or more additional thresholds; and input the pseudo-label, the at least one beam quality metric in the second set of beam quality metrics, and at least one measurement in the second set of measurements corresponding to the at least one beam quality metric into the machine learning model, wherein the training is at least in part based on the input.

10. The apparatus according to claim 8, wherein the second set of beam quality metrics includes a set of probability metrics indicating a relative probability that a corresponding beam in the set of received beams meets a threshold beam performance, and wherein at least one beam quality metric in the second set of beam quality metrics satisfies the one or more additional thresholds, at least in part based on the at least one beam quality metric being greater than or equal to the one or more additional thresholds.

11. The apparatus according to claim 1, wherein the instructions are further executable by the processor to cause the apparatus to: predict a first set of beam indices corresponding to the first set of beam quality metrics, the first set of beam indices indicating predicted received beams associated with a plurality of time instances; and selectively modify the beam indices in the first set of beam indices, at least in part based on one or more differences between the beam indices and one or more additional beam indices corresponding to one or more of the plurality of time instances within a coherence window.

12. The apparatus according to claim 11, wherein the instructions are further executable by the processor to cause the apparatus to: receive an indication of the coherence window via the control signaling, wherein selectively modifying the beam indices is at least in part based on receiving the control signaling.

13. The apparatus according to claim 1, wherein the first set of measurements includes a plurality of measurements performed on a plurality of reference signals received by the UE in a first time interval using a received beam in the set of received beams, and wherein the first set of beam quality metrics includes a plurality of predicted beam quality metrics associated with the received beam in a second time interval, and wherein the instructions are further executable by the processor to cause the apparatus to: Selectively modify at least one predicted beam quality metric among the plurality of predicted beam quality metrics during the second time interval, at least in part based on the at least one predicted beam quality metric satisfying an outlier threshold.

14. The apparatus according to claim 1, wherein the instructions are further executable by the processor to cause the apparatus to: Select a receive beam from the set of receive beams at least in part based on the first set of beam quality metrics; and Receive one or more messages from the network entity using the receive beam, at least in part based on the selection.

15. The apparatus according to claim 1, wherein the instructions are further executable by the processor to cause the apparatus to: Send capability signaling indicating one or more capabilities associated with the UE to the network entity, wherein receiving the control signaling is at least in part based on the capability signaling.

16. The apparatus according to claim 1, wherein the instructions are further executable by the processor to cause the apparatus to: Perform a second set of measurements on a second set of reference signals received from the network entity; Determine a second set of beam quality metrics associated with the set of receive beams, a set of additional receive beams, or both, at least in part based on inputting the second set of measurements into the machine learning model; and And Avoid reporting the second set of measurements to the network entity, at least in part based on at least one beam quality metric in the second set of beam quality metrics failing to meet the one or more thresholds.

17. An apparatus for wireless communication at a network entity, the apparatus comprising: A processor; And A memory coupled to the processor, the memory storing instructions that are executable by the processor to cause the apparatus to: Send control signaling indicating a machine learning model to be used for beam prediction at a user equipment (UE), the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; Send a set of reference signals to the UE, at least in part based on the control signaling; and And Receive a control message from the UE indicating a first set of measurements associated with the set of reference signals, wherein the control message is received at least in part based on a first set of beam quality metrics associated with the first set of measurements satisfying the one or more thresholds, wherein the first set of beam quality metrics includes an output of the machine learning model.

18. The apparatus according to claim 17, wherein the instructions are further executable by the processor to cause the apparatus to: Train the machine learning model at least in part based on the first set of measurements, the first set of beam quality metrics, or both; and Send additional control signaling indicating an updated version of the machine learning model to the UE, at least in part based on training the machine learning model.

19. The apparatus according to claim 17, wherein the first set of beam quality metrics includes a set of probability metrics indicating the relative probability that a corresponding beam in a set of received beams meets a threshold beam performance, and wherein the first set of beam quality metrics meets the one or more thresholds based at least in part on each beam quality metric in the first set of beam quality metrics being less than the one or more thresholds.

20. The apparatus according to claim 17, wherein the first set of beam quality metrics meets the one or more thresholds based at least in part on one or more differences between the first set of beam quality metrics and a second set of beam quality metrics being greater than or equal to the one or more thresholds.

21. The apparatus according to claim 20, wherein the first set of beam quality metrics and the second set of beam quality metrics include predicted reference signal received power measurements associated with a set of received beams, and wherein the one or more thresholds include a threshold reference signal received power metric.

22. The apparatus according to claim 20, wherein the second set of beam quality metrics includes reference signal received power metrics predicted using one or more mathematical operations different from the machine learning model.

23. The apparatus according to claim 20, wherein the one or more differences between the first set of beam quality metrics and the second set of beam quality metrics include a least mean square error metric.

24. The apparatus according to claim 17, wherein the control signaling further indicates one or more additional thresholds for training the machine learning model at the UE.

25. The apparatus according to claim 17, wherein the instructions can be further executed by the processor to cause the apparatus to: receive, from the UE, capability signaling indicating one or more capabilities associated with the UE, wherein the control signaling is sent at least in part based on the capability signaling.

26. A method for wireless communication at a user equipment (UE), the method comprising: receiving, from a network entity, control signaling indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; performing a first set of measurements on a set of reference signals received from the network entity; predicting, at least in part based on inputting the first set of measurements into the machine learning model, a first set of beam quality metrics associated with a set of received beams at the UE; and sending, to the network entity, a control message indicating the first set of measurements based at least in part on the first set of beam quality metrics meeting the one or more thresholds.

27. The method according to claim 26, the method further comprising: receiving, from the network entity, additional control signaling indicating an updated version of the machine learning model, wherein the additional control signaling is received at least in part based on sending the control message.

28. The method according to claim 26, wherein the first set of beam quality metrics includes a set of probability metrics indicating a relative probability that a corresponding beam in the set of received beams meets a threshold beam performance, and wherein the first set of beam quality metrics meets the one or more thresholds based at least in part on each beam quality metric in the first set of beam quality metrics being less than the one or more thresholds.

29. A method for wireless communication at a network entity, the method comprising: sending control signaling to a user equipment (UE) indicating a machine learning model to be used for beam prediction at the UE, the control signaling further indicating one or more thresholds for reporting an output from the machine learning model; sending a set of reference signals to the UE based at least in part on the control signaling; and receiving, from the UE, a control message indicating a first set of measurements associated with the set of reference signals, wherein the control message is received based at least in part on a first set of beam quality metrics associated with the first set of measurements meeting the one or more thresholds, and wherein the first set of beam quality metrics includes the output of the machine learning model.

30. The method according to claim 29, the method further comprising: training the machine learning model based at least in part on the first set of measurements, the first set of beam quality metrics, or both; and sending additional control signaling to the UE indicating an updated version of the machine learning model based at least in part on training the machine learning model.

Images