Beam correspondence between artificial intelligence model training and inferring

Abstract

Methods, systems, and devices for wireless communication are described. A user equipment (UE) may support beam prediction based on artificial intelligence (AI) or machine learning (ML). The trained AI model uses the measurements of the first set of beams to predict measurements of the second set of beams. In some examples, the set of AI models may be standardized or predefined, and the UE and the network entity may conform to mutually supporting AI models from the set of AI models. Thus, the UE and the network entity may agree with the input and output of the AI model for beam prediction. In some examples, the UE and the network entity may perform a training process of the AI model, and may ensure a correspondence between inputs and outputs of the AI model for the training process and inputs and outputs of the AI model for the inference process.

Description

Technical Field

[0001] The following discussion pertains to wireless communication, including beam correspondence between artificial intelligence (AI) model training and inference. Background Technology

[0002] Wireless communication systems are widely deployed to provide various types of communication content, such as voice, video, packet data, message sending and receiving, broadcasting, and so on. These systems can support communication with multiple users by sharing available system resources (e.g., time, frequency, and power). Examples of such multiple access systems include fourth-generation (4G) systems (such as Long Term Evolution (LTE) systems, LTE-A 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 technologies such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal FDMA (OFDMA), or Discrete Fourier Transform Extended Orthogonal Frequency Division Multiplexing (DFT-S-OFDM). A wireless multiple access communication system may include one or more base stations, each supporting wireless communication of communication devices, which may be referred to as User Equipment (UE). Summary of the Invention

[0003] The described technology relates to improved methods, systems, devices, and apparatuses for supporting beam correspondence between training and inference of artificial intelligence (AI) models. User equipment (UE) can support beam prediction based on AI or machine learning (ML). An AI model can be trained using measurements of a first beam set (e.g., set B beams) to predict measurements of a future beam set (e.g., set A beams). The trained AI model uses actual measurements of a third beam set (e.g., set B beams) to predict measurements of a fourth beam set (e.g., set A beams) (e.g., referred to as beam inference), which the UE can report in a beam measurement report. In some examples, the AI ​​model set can be standardized or predefined, and the UE and network entities can agree on mutually supporting AI models from the AI ​​model set. Therefore, the UE and network entities can agree on the inputs and outputs of the AI ​​model used for beam prediction. In some examples, the UE and network entities can perform the training process of the AI ​​model and can ensure the correspondence between the inputs and outputs of the AI ​​model used for the training process and the inputs and outputs of the AI ​​model used for the inference process.

[0004] A method for wireless communication by a UE is described. The method may include: receiving a first message from a network entity indicating a first subset of AI model identifiers supported by the network entity from a predefined set of AI model identifiers associated with beam prediction; sending a second message to the network entity indicating a second subset of AI model identifiers supported by the UE from the predefined set of AI model identifiers; receiving control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of an AI model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; receiving a set of reference signals via the beam measurement resource set; and sending a report indicating a set of predicted values ​​for the beam set generated using the AI ​​model and based on the set of reference signals.

[0005] A UE for wireless communication is described. The UE may include: one or more memories storing processor-executable code; and one or more processors coupled to the one or more memories. The one or more processors may be able to operate individually or jointly to execute code such that the UE: receives a first message from a network entity indicating a first subset of AI model identifiers supported by the network entity from a predefined set of AI model identifiers associated with beam prediction; sends a second message to the network entity indicating a second subset of AI model identifiers supported by the UE from the predefined set of AI model identifiers; receives control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; receives a set of reference signals via the beam measurement resource set; and sends a report indicating a set of predicted values ​​for the beam set generated using the AI ​​model and based on the set of reference signals.

[0006] Another UE for wireless communication is described. The UE may include: components for receiving a first message from a network entity, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction; components for sending a second message to the network entity, the second message indicating a second subset of AI model identifiers supported by the UE from the predefined set of multiple AI model identifiers; components for receiving control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of an AI model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; components for receiving a set of reference signals via the beam measurement resource set; and components for sending a report indicating a set of predicted values ​​for the beam set generated using the AI ​​model and based on the reference signal set.

[0007] A non-transitory computer-readable medium storing code for wireless communication is described. The code may include instructions executable by a processor to: receive a first message from a network entity indicating a first subset of AI model identifiers supported by the network entity from a predefined set of AI model identifiers associated with beam prediction; send a second message to the network entity indicating a second subset of AI model identifiers supported by a UE from the predefined set of AI model identifiers; receive control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; receive a set of reference signals via the beam measurement resource set; and send a report indicating a set of predicted values ​​for the beam set generated using the AI ​​model and based on the set of reference signals.

[0008] In some examples of the methods described herein, UEs, and non-transitory computer-readable media, the first message may be one of system information, radio resource control messages (RRCs), media access control (MAC) control elements (MAC-CEs), or downlink control information (DCIs).

[0009] In some examples of the methods described herein, UEs, and non-transitory computer-readable media, the second message may be one of an RRC message, a MAC-CE, or an uplink control information (UCI).

[0010] In some examples of the methods, UEs, and non-transitory computer-readable media described herein, in order to indicate the beam measurement resource set, control signaling indicates the Channel State Information Reference Signal (CSI-RS) index set or the Synchronization Signal Block (SSB) index set, and the AI ​​model identifier can indicate the mapping of the CSI-RS index set or the SSB index set to the input of the AI ​​model.

[0011] In the methods described herein, in some examples of UEs and non-transitory computer-readable media, AI model identifiers can indicate the corresponding beamwidth and corresponding beam direction of each beam in the beam measurement resource set and each beam in the beam set.

[0012] In some examples of the methods, UEs, and nontransitory computer-readable media described herein, AI model identifiers can indicate quasi-co-location relationships between beams in a beam measurement resource set and beams in a beam set.

[0013] In the methods described herein, and in some examples of UEs and non-transitory computer-readable media, AI model identifiers can indicate the relative timing between the set of beam measurement resources and the set of predicted values.

[0014] In the methods described herein, and in some examples of UEs and non-transitory computer-readable media, the report also indicates a pre-decoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink departure angle, or a departure direction, based on a set of predicted values.

[0015] In the methods described herein, and in some examples of UEs and non-transitory computer-readable media, control signaling also schedules channel state information reports, and the reports can be included within the channel state information reports.

[0016] The methods described herein, some examples of UEs and non-transitory computer-readable media may also include operations, features, components or instructions for: performing a training process of an AI model with a network entity using a training beam measurement resource set to predict a training measurement result set of the training beam set, wherein the beam measurement resource set corresponds to the training beam measurement resource set, and wherein the beam set corresponds to the training beam set.

[0017] In some examples of the methods, UEs, and nontransitory computer-readable media described herein, the training beam measurement resource set includes a first CSI-RS set or a first SSB set with a first index value, the beam measurement resource set includes a second CSI-RS set or a second SSB set with a first index value, and the same mapping from the first index value to the input value of the AI ​​model can be used to predict the training measurement result set and generate a predicted value set.

[0018] The methods described herein, examples of UEs, and non-transitory computer-readable media may also include operations, features, components, or instructions for: receiving from a network entity a verification beam measurement resource set and a corresponding reference truth prediction target for a second beam set for an AI model; receiving a second reference signal set via the verification beam measurement resource set; and transmitting a second report indicating that a second set of predictions for the second beam set generated using the AI ​​model based on the reference signal set satisfies an accuracy metric regarding the corresponding reference truth prediction target, wherein the reception of control signaling may be based on the transmission of the second report.

[0019] A method for wireless communication by a network entity is described. The method may include: sending a first message to a UE indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction; receiving a second message from the UE indicating a second subset of AI model identifiers supported by the UE from the predefined set of multiple AI model identifiers; sending control signaling to the UE indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; sending a set of reference signals via the beam measurement resource set; and receiving a report from the UE indicating a set of predicted values ​​for the beam set based on the AI ​​model identifiers and the set of reference signals.

[0020] A network entity for wireless communication is described. The network entity may include: one or more memories storing processor-executable code; and one or more processors coupled to the one or more memories. The one or more processors may be able to operate individually or jointly to execute code such that the network entity: sends a first message to a UE indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction; receives a second message from the UE indicating a second subset of AI model identifiers supported by the UE from the predefined set of multiple AI model identifiers; sends control signaling to the UE indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; transmits a set of reference signals via the beam measurement resource set; and receives a report from the UE indicating a set of predicted values ​​for the beam set based on the AI ​​model identifiers and the set of reference signals.

[0021] Another network entity for wireless communication is described. This network entity may include: components for sending a first message to a UE, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction; components for receiving a second message from the UE, the second message indicating a second subset of AI model identifiers supported by the UE from the predefined set of multiple AI model identifiers; components for sending control signaling to the UE, the control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; components for sending a set of reference signals via the beam measurement resource set; and components for receiving a report from the UE, the report indicating a set of predicted values ​​for the beam set based on the AI ​​model identifiers and the reference signal set.

[0022] A non-transitory computer-readable medium storing code for wireless communication is described. The code may include instructions executable by a processor to: send a first message to a UE indicating a first subset of AI model identifiers supported by a network entity from a predefined set of AI model identifiers associated with beam prediction; receive a second message from the UE indicating a second subset of AI model identifiers supported by the UE from the predefined set of AI model identifiers; send control signaling to the UE indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set; send a set of reference signals via the beam measurement resource set; and receive a report from the UE indicating a set of predicted values ​​for the beam set based on the AI ​​model identifiers and the set of reference signals.

[0023] In some examples of the methods, network entities, and non-transitory computer-readable media described herein, the first message may be one of system information, RRC message, MAC-CE, or DCI.

[0024] In some examples of the methods, network entities, and non-transitory computer-readable media described herein, the second message may be one of an RRC message, a MAC-CE message, or a UCI message.

[0025] In some examples of the methods, network entities, and nontransitory computer-readable media described herein, in order to indicate a beam measurement resource set, control signaling indicates a CSI-RS index set or an SSB index set, and an AI model identifier can indicate a mapping of the CSI-RS index set or SSB index set to the input of the AI ​​model associated with the AI ​​model identifier.

[0026] In some examples of the methods, network entities, and non-transitory computer-readable media described herein, AI model identifiers can indicate the corresponding beamwidth and corresponding beam direction of each beam in the beam measurement resource set and each beam in the beam set.

[0027] In some examples of the methods, network entities, and nontransitory computer-readable media described herein, AI model identifiers can indicate quasi-co-location relationships between beams in a beam measurement resource set and beams in a beam set.

[0028] In some examples of the methods, network entities, and nontransitory computer-readable media described herein, AI model identifiers can indicate the relative timing between the set of beam measurement resources and the set of predicted values.

[0029] In some examples of the methods, network entities, and non-transitory computer-readable media described herein, the report also indicates a pre-decoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink departure angle, or a departure direction, based on a set of predicted values.

[0030] In some examples of the methods, network entities, and non-transitory computer-readable media described herein, control signaling also schedules channel state information reports, and the reports can be included within the channel state information reports.

[0031] Some examples of the methods, network entities, and non-transitory computer-readable media described herein may also include operations, features, components, or instructions for: performing a training process of an AI model with a UE using a training beam measurement resource set to predict a training measurement result set of the training beam set, wherein the beam measurement resource set corresponds to the training beam measurement resource set, and wherein the beam set corresponds to the training beam set.

[0032] In some examples of the methods, network entities, and nontransitory computer-readable media described herein, the training beam measurement resource set includes a first CSI-RS set or a first SSB set with a first index value, the beam measurement resource set includes a second CSI-RS set or a second SSB set with a first index value, and the same mapping from the first index value to the input value of the AI ​​model can be used to predict the training measurement result set and generate a predicted value set.

[0033] Some examples of the methods, network entities, and non-transitory computer-readable media described herein may also include operations, features, components, or instructions for: sending to the UE a verification beam measurement resource set and a corresponding reference truth prediction target for a second beam set for an AI model; sending a second reference signal set via the verification beam measurement resource set; and receiving from the UE a second report indicating that a second set of predictions for the second beam set generated at the UE using the AI ​​model based on the reference signal set satisfies an accuracy metric regarding the corresponding reference truth prediction target, wherein the transmission of control signaling may be based on the reception of the second report. Attached Figure Description

[0034] Figure 1 An example of a wireless communication system supporting beam correspondence between training and inference of an artificial intelligence (AI) model, according to one or more aspects of this disclosure, is shown.

[0035] Figure 2 An example of a wireless communication system supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0036] Figure 3 An example of an AI model training and inference graph supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0037] Figure 4 An example of a wireless communication system supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0038] Figure 5 An example of a wireless communication system supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0039] Figure 6 An example of a beam correspondence diagram supporting the beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0040] Figure 7 An example of a beam correspondence diagram supporting the beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0041] Figure 8 An example of a quasi-co-location (QCL) relationship consistency graph supporting the beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown.

[0042] Figure 9 An example of an AI model ID registration scenario supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0043] Figure 10 An example of a wireless communication system supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0044] Figure 11 An example of a process flow supporting the beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0045] Figure 12 An example of an AI model training and inference graph supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0046] Figure 13 An example of an AI model training graph supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0047] Figure 14An example of an AI model inference graph supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0048] Figure 15 Examples of AI model training and inference scenarios supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, are shown.

[0049] Figure 16 and Figure 17 A block diagram of an apparatus supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0050] Figure 18 A block diagram of a communication manager supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0051] Figure 19 A diagram of a system including a device supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0052] Figure 20 and Figure 21 A block diagram of an apparatus supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0053] Figure 22 A block diagram of a communication manager supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0054] Figure 23 A diagram of a system including a device supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown.

[0055] Figure 24 and Figure 25 A flowchart illustrating a method for supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. Detailed Implementation

[0056] User equipment (UE) can support beam prediction based on artificial intelligence (AI) or machine learning (ML). Such a UE can collect data measurements (e.g., Reference Signal Received Power (RSRP), Signal-to-Interference Plus-Noise Ratio (SINR), Channel Impulse Response (CIR)) based on SSB or CSI-RS measurements via Synchronous System Block (SSB) beams or Channel State Information (CSI) Reference Signal-RS beams. The AI ​​model can be trained using measurements of a first beam set (e.g., set B beams) to predict measurements of future beam sets (e.g., set A beams). The trained AI model uses actual measurements of a third beam set (e.g., set B beams) to predict measurements for a fourth beam set (e.g., set A beams) (e.g., this process can be referred to as beam inference), which the UE 115 can report in a beam measurement report. The mapping of beam measurements to the inputs of the AI ​​model affects the output of the AI ​​model (e.g., the predicted measurements). Therefore, for accurate use of the AI ​​model, the UE and network entities (e.g., gNB) should agree on the mapping and ordering of Channel Measurement Resources (CMR) or Interference Measurement Resources (IMR) to AI model inputs. However, existing CMR or IMR frameworks do not allow the UE to identify the order of AI model inputs because network entities can flexibly map beamforming codebooks to CMR or IMR. The beamforming codebook information used by network entities for CMR or IMR can be proprietary, and network entity vendors may be unwilling to signal such information.

[0057] In some examples, the set of trained AI models can be standardized or predefined. For example, these standards can define the mapping or ordering of beam measurements (e.g., which SSB or CSI-RS index) to the AI ​​model inputs of the standardized AI models. The UE and network entity can negotiate mutually supported AI models from the list of standardized AI models. For example, the network entity can indicate a subset of AI model IDs from the set of standardized AI model IDs, and the UE can indicate which AI model IDs in the indicated subset it supports. The network entity can schedule a set of beam measurement resources and can indicate the AI ​​model IDs mutually supported by the network entity and the UE. The UE can use the indicated AI models and beam measurement resources to generate and transmit beam prediction reports. By using mutually supported AI models from the list of trained AI models, the UE and network entity can agree on the inputs and outputs of the AI ​​models, which enables accurate beam prediction.

[0058] In some examples, network entities can identify the AI ​​model to the UE during the beam inference process, allowing the UE to ensure that the actual beam used for the beam inference process (e.g., SSB or CSI-RS) corresponds to the actual beam used to train the AI ​​model (e.g., SSB or CSI-RS) (e.g., a one-to-one association), and that the measurement of the predicted beam used for the beam inference process (e.g., the predicted measurement of the SSB or CSI-RS) corresponds to the predicted beam used to train the AI ​​model (e.g., the predicted measurement of the SSB or CSI-RS) (e.g., a one-to-one association). In such examples, the UE and network entities can first perform a training process using a training input set and a desired output set, or the AI ​​model. By ensuring the correspondence between the AI ​​model's inputs and outputs between the training and inference processes, the UE and network entities can use the AI ​​model to achieve accurate beam prediction.

[0059] Consistency between the input and output of an AI model can reduce latency and overhead, as well as improve beam selection accuracy for beam management purposes (e.g., beam prediction in the temporal and / or spatial domains). Consistency between the input and output of an AI model can also enable lifecycle management of the AI ​​model, including model training, deployment, inference, monitoring, and updates.

[0060] First, aspects of this disclosure are described within the context of a wireless communication system. These aspects are further illustrated and described with reference to AI model training and inference graphs, beam correspondence graphs, quasi-colocation (QCL) consistency graphs, AI model ID registration scenarios, process flows, AI model training and inference graphs, AI model training graphs, AI model inference graphs, and AI model training and inference scenarios. These aspects are further illustrated and described with reference to apparatus diagrams, system diagrams, and flowcharts relating to the beam correspondence between AI model training and inference.

[0061] Figure 1 An example of a wireless communication system 100 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. 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 Advanced LTE (LTE-A) network, an LTE-A Pro network, a New Radio (NR) network, or a network operating under other systems and radio technologies, including future systems and radio technologies not explicitly mentioned herein.

[0062] Network entity 105 may be distributed across a geographical area to form wireless communication system 100, and may include devices employing different forms or having different capabilities. In various examples, network entity 105 may be referred to as a network element, mobility element, radio access network (RAN) node, or network equipment, etc. In some examples, network entity 105 and UE 115 may wirelessly communicate via one or more communication links 125 (e.g., radio frequency (RF) access links). For example, network entity 105 may support coverage area 110 (e.g., a geographical coverage area) within which UE 115 and network entity 105 may establish one or more communication links 125. Coverage area 110 may be an example of a geographical area within which network entity 105 and UE 115 may support the transmission of signals according to one or more radio access technologies (RATs).

[0063] UE 115 can be distributed throughout the coverage area 110 of wireless communication system 100, and each UE 115 can be stationary or mobile, or stationary and mobile at different times. UE 115 can be devices in different forms or with different capabilities. Figure 1 Some example UE 115s are illustrated herein. The UE 115 described herein can be able to support various types of devices (such as, e.g., ...). Figure 1 It communicates with other UEs (115 or network entity 105) as shown.

[0064] As described herein, nodes of the wireless communication system 100 (which may be referred to as network nodes or wireless nodes) may be network entity 105 (e.g., any network entity described herein), UE 115 (e.g., any UE described herein), network controller, apparatus, device, computing system, one or more components, or another suitable processing entity configured to perform any of the techniques described herein. For example, a node may be UE 115. Alternatively, a node may be network entity 105. Furthermore, a first node may be configured to communicate with a second or third node. In one aspect of this example, the first node may be UE 115, the second node may be network entity 105, and the third node may be UE 115. In another aspect of this example, the first node may be UE 115, the second node may be network entity 105, and the third node may be network entity 105. In other aspects of this example, the first node, the second node, and the third node may be different from these examples. Similarly, references to UE 115, network entity 105, device, equipment, computing system, etc., may include disclosures of UE 115, network entity 105, device, equipment, computing system, etc., as nodes. For example, a disclosure that UE 115 is configured to receive information from network entity 105 also discloses that a first node is configured to receive information from a second node.

[0065] In some examples, network entity 105 may communicate with core network 130, communicate with each other, or both. For example, network entity 105 may communicate with 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, network entities 105 may communicate with each other directly (e.g., directly between network entities 105) or indirectly (e.g., via core network 130) via backhaul communication links 120 (e.g., according to X2, Xn, or other interface protocols). In some examples, network entities 105 may communicate with each other via midhaul communication link 162 (e.g., according to midhaul interface protocol) or fronthaul communication link 168 (e.g., according to fronthaul interface protocol) or any combination thereof. Backhaul communication link 120, midhaul communication link 162, or fronthaul communication link 168 may be or include one or more wired links (e.g., electrical links, fiber optic links), one or more wireless links (e.g., radio links, wireless optical links), etc., or various combinations thereof. UE 115 can communicate with core network 130 via communication link 155.

[0066] One or more network entities in network entity 105 described herein may include or be referred to as base station 140 (e.g., transceiver base station, radio base station, NR base station, access point, radio transceiver, node B, eNodeB (eNB), next-generation node B or gigabit node B (any of which may be referred to as gNB), 5G NB, next-generation eNB (ng-eNB), home node B, home evolution node B, or other suitable terms). In some examples, network entity 105 (e.g., base station 140) may be implemented in an aggregated (e.g., monolithic, standalone) base station architecture that may be configured to utilize a protocol stack that is physically or logically integrated within a single network entity 105 (e.g., a single RAN node, such as base station 140).

[0067] In some examples, network entity 105 may be implemented in a decomposed architecture (e.g., a decomposed base station architecture, a decomposed RAN architecture) that can be configured to utilize protocol stacks physically or logically distributed across 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., a cloud RAN (C-RAN)). For example, 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., a near real-time RIC, a non-real-time RIC), a service management and orchestration (SMO) 180 system, or any combination thereof. 170 may also be referred to as a radio headend, intelligent radio headend, remote radio headend (RRH), remote radio unit (RRU), or transmit / receive point (TRP). One or more components of network entity 105 in a decomposed RAN architecture may be co-located, or one or more components of network entity 105 may be located in distributed locations (e.g., separate physical locations). In some examples, one or more network entities 105 in a decomposed RAN architecture may be implemented as virtual units (e.g., virtual CU (VCU), virtual DU (VDU), virtual RU (VRU)).

[0068] The functional splitting among CU 160, DU 165, and RU 170 is flexible and can support different functionalities depending on which functions (e.g., network layer functions, protocol layer functions, baseband functions, RF functions, and any combination thereof) are performed at CU 160, DU 165, or RU 170. For example, a protocol stack functional splitting can be used between CU 160 and DU 165, allowing CU 160 to support one or more layers of the protocol stack, and DU 165 to support one or more different layers of the protocol stack. In some examples, CU 160 can host higher protocol layer (e.g., Layer 3 (L3), Layer 2 (L2)) functionalities and signaling (e.g., Radio Resource Control (RRC), Serving Data Adaptation Protocol (SDAP), Packet Data Convergence Protocol (PDCP)). CU 160 can connect to one or more DU 165 or RU 170, and one or more DU 165 or RU 170 can host lower protocol layers, such as Layer 1 (L1) (e.g., Physical (PHY) layer) or L2 (e.g., Radio Link Control (RLC) layer, Medium Access Control (MAC) layer) functionality and signaling, and each can be at least partially controlled by CU 160. Additionally or alternatively, a protocol stack functional split can be employed between DU 165 and RU 170, such that DU 165 can support one or more layers of the protocol stack, and RU 170 can support one or more different layers of the protocol stack. DU 165 can support one or more different cells (e.g., via one or more RU 170). In some cases, functional decomposition between CU 160 and DU 165, or between DU 165 and RU 170, can be performed within the protocol layer (e.g., some functions of the protocol layer can be performed by one of CU 160, DU 165, or RU 170, while other functions of the protocol layer can be performed by different of CU 160, DU 165, or RU 170). CU 160 can be further functionally decomposed into CU control plane (CU-CP) functions and CU user plane (CU-UP) functions. CU 160 can be connected to one or more DU 165 via midhaul communication link 162 (e.g., F1, F1-c, F1-u), and DU 165 can be connected to one or more RU 170 via fronthaul communication link 168 (e.g., open fronthaul (FH) interface). In some examples, the midhaul communication link 162 or the fronthaul communication link 168 may be implemented based on the interfaces (e.g., channels) between the layers of the protocol stack, which are supported by the corresponding network entities 105 communicating via such communication links.

[0069] In a wireless communication system (e.g., wireless communication system 100), the infrastructure and spectrum resources for radio access can support wireless backhaul link capabilities to supplement wired backhaul connections, thereby providing an IAB network architecture (e.g., to core network 130). In some cases, in an IAB network, one or more network entities 105 (e.g., IAB node 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 DU 165s or one or more RU 170s may be partially controlled by one or more CU 160s associated with 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 node 104) via supported access and backhaul links (e.g., backhaul communication link 120). 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 communication with UE 115, or may share the same antennas (e.g., those of RU 170) for access to IAB node 104 via DU 165 of IAB node 104. (e.g., referred to as a virtual IAB-MT (vIAB-MT)). In some examples, IAB node 104 may include a DU 165 that supports communication links with additional entities (e.g., IAB node 104, UE 115) within a relay chain or configuration (e.g., downstream) of the access network. In such cases, one or more components of the decomposed RAN architecture (e.g., one or more IAB nodes 104 or components of IAB node 104) may be configured to operate according to the techniques described herein.

[0070] For example, the access network (AN) or RAN may include communication between an access node (e.g., an IAB donor), IAB node 104, and one or more UEs 115. The IAB donor may facilitate connectivity between the core network 130 and the AN (e.g., via a wired or wireless connection to the core network 130). That is, an 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 RU 170), where the CU 160 may communicate with the core network 130 via an interface (e.g., a backhaul link). The IAB donor and IAB node 104 may communicate via an F1 interface according to a protocol defining the 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 part of a backhaul link) and may communicate with other CU 160s (e.g., CU 160 associated with an alternative IAB donor) via an Xn-C interface (which may be part of a backhaul link).

[0071] IAB node 104 may refer to a RAN node that provides IAB functionality (e.g., access for UE 115, radio self-backhaul capability, etc.). DU 165 may act as a distributed scheduling node toward child nodes associated with IAB node 104, and IAB-MT may act as a scheduled node toward a parent node associated with IAB node 104. That is, an IAB donor may be referred to as a parent node communicating with one or more child nodes (e.g., an IAB donor may relay for UE transmissions via one or more other IAB nodes 104). Additionally or alternatively, depending on the AN's relay chain or configuration, IAB node 104 may also be referred to as a parent or child node of other IAB nodes 104. Therefore, the IAB-MT entity of IAB node 104 can provide a Uu interface for child IAB node 104 to receive signaling from parent IAB node 104, and the DU interface (e.g., DU 165) can provide a Uu interface for parent IAB node 104 to send signaling notifications to child IAB node 104 or UE 115.

[0072] For example, IAB node 104 may be referred to as a parent node supporting communication to child IAB nodes, or as a child IAB node associated with an IAB donor, or both. An IAB donor may include a CU 160 having a wired or wireless connection to core network 130 (e.g., backhaul communication link 120) and may act as a parent node of IAB node 104. For example, the IAB donor's DU 165 may relay transmissions to UE 115 via IAB node 104, or may signal transmissions directly to UE 115, or both. The IAB donor's CU 160 may signal the establishment of a communication link to IAB node 104 via an F1 interface, and IAB node 104 may schedule transmissions via DU 165 (e.g., transmissions relayed from the IAB donor to UE 115). That is, data may be relayed to and from IAB node 104 via signaling through the NR Uu interface of the MT to IAB node 104. Communication with IAB node 104 can be scheduled by DU 165 of the IAB donor, and communication with IAB node 104 can be scheduled by DU 165 of IAB node 104.

[0073] In the context of applying the techniques described herein to a decomposed RAN architecture, one or more components of the decomposed RAN architecture can be configured to support beam correspondence between AI model training and inference as described herein. For example, some operations described as being performed by UE 115 or network entity 105 (e.g., base station 140) may additionally or alternatively be performed by one or more components of the decomposed RAN architecture (e.g., IAB node 104, DU 165, CU 160, RU 170, RIC 175, SMO 180).

[0074] UE 115 may include or be referred to as a mobile device, wireless device, remote device, handheld device, or subscriber device, or any other suitable term, wherein "device" may also be referred to as a cell, station, terminal, or client, etc. UE 115 may also include or be referred to as a personal electronic device, such as a cellular phone, personal digital assistant (PDA), tablet computer, laptop computer, or personal computer. In some examples, UE 115 may include or be referred to as a wireless local loop (WLL) station, Internet of Things (IoT) device, Internet of Everything (IoE) device, or machine-type communication (MTC) device, etc., which may be implemented in various objects such as appliances or vehicles, meters, etc.

[0075] The UE 115 described herein can communicate with various types of devices, such as other UEs 115 that sometimes act as relays, network entities 105, and network equipment including macro eNBs or gNBs, small cell eNBs or gNBs, or relay base stations, etc. Figure 1 As shown.

[0076] UE 115 and network entity 105 can 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" can 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 of the RF spectrum band (e.g., a bandwidth portion (BWP)) 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 coordinating carrier operation, user data, or other signaling. Wireless communication system 100 can support communication with UE 115 using carrier aggregation or multi-carrier operation. Depending on the carrier aggregation configuration, UE 115 may be configured with multiple downlink component carriers and one or more uplink component carriers. Carrier aggregation can be used in conjunction with both frequency division duplex (FDD) component carriers and time division duplex (TDD) component carriers. Communication between network entity 105 and other devices can refer to communication between these devices and any part of network entity 105 (e.g., entity, sub-entity). For example, the terms “send,” “receive,” or “communicate” when referring to network entity 105 can refer to any part of the RAN’s network entity 105 (e.g., base station 140, CU160, DU 165, RU 170) communicating with another device (e.g., directly or via one or more other network entities 105).

[0077] In some examples, such as in carrier aggregation configurations, a carrier may also have acquisition signaling or control signaling to coordinate the operation of other carriers. A carrier may be associated with a frequency channel (e.g., an Evolved Universal Mobile Telecommunications System Terrestrial Radio Access (E-UTRA) Absolute RF Channel Number (EARFCN)) and may be identified according to a channel grating used for discovery by UE 115. A carrier may operate in standalone mode, in which case initial acquisition and connection can be made by UE 115 via that carrier, or the carrier may operate in non-standalone mode, in which case different carriers (e.g., the same or different radio access technologies) are used to anchor the connection.

[0078] The communication link 125 shown in the wireless communication system 100 may include downlink transmission (e.g., forward link transmission) from network entity 105 to UE 115, uplink transmission (e.g., return link transmission) from UE 115 to network entity 105, or both, as well as other transmission configurations. A carrier may carry downlink communication or uplink communication (e.g., in FDD mode), or may be configured to carry both downlink and uplink communication (e.g., in TDD mode).

[0079] 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 carrier or the “system bandwidth” of the wireless communication system 100. For example, the carrier bandwidth may be one bandwidth in a set of bandwidths for a particular radio access technology (e.g., 1.4 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., network entity 105, UE 115, or both) may have hardware configurations that support communication using a specific carrier bandwidth, or may be configured to support communication using one carrier bandwidth in a set of carrier bandwidths. In some examples, the wireless communication system 100 may include network entity 105 or 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.

[0080] The signal waveform transmitted via a carrier may include multiple subcarriers (e.g., using multi-carrier modulation (MCM) techniques, such as orthogonal frequency division multiplexing (OFDM) or discrete Fourier transform extended OFDM (DFT-S-OFDM)). In a system employing MCM, a resource element may refer to a resource of one symbol period (e.g., the duration of one modulation symbol) and one subcarrier, in which case the symbol period and 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., in the transmission duration) and a relatively high modulation scheme order correspond to a relatively high communication rate. Wireless communication resources may refer to a combination of RF spectrum resources, temporal resources, and spatial resources (e.g., spatial layers or beams), and the use of multiple spatial resources may increase the data rate or data integrity used for communication with UE 115.

[0081] It can support one or more sets of parameters for a carrier, and the set of parameters may include subcarrier spacing ( (and cyclic prefix). A carrier can be divided into one or more BWPs with the same or different sets of parameters. In some examples, UE 115 can be configured with multiple BWPs. In some examples, a single BWP of a carrier can be active at a given time, and the communication of UE 115 can be constrained to one or more active BWPs.

[0082] The time interval for network entity 105 or UE 115 can be expressed as a multiple of a basic time unit, such as the sampling period. seconds, of which It can represent the supported subcarrier spacing, and This can represent the supported Discrete Fourier Transform (DFT) size. The time interval of the communication resources can be organized according to radio frames, each with a specified duration (e.g., 10 milliseconds (ms)). Each radio frame can be identified by a System Frame Number (SFN) (e.g., ranging from 0 to 1023).

[0083] Each frame may include multiple consecutively numbered subframes or time slots, and each subframe or time slot may have the same duration. In some examples, a frame may (e.g., in the time domain) be divided into subframes, and each subframe may be further divided into a 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 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, time slots may be further divided into multiple micro-time slots associated with one or more symbols. Excluding the cyclic prefix, each symbol period may be associated with one or more (e.g., The duration of a symbol period is associated with a (number) sampling period. The duration of a symbol period can depend on the subcarrier spacing or the operating frequency band.

[0084] A subframe, time slot, micro-time slot, or symbol can be the smallest scheduling unit of the wireless communication system 100 (e.g., in the time domain) and can be referred to as a transmission time interval (TTI). In some examples, the duration of the TTI (e.g., the number of symbol periods in the TTI) can be variable. Additionally or alternatively, the smallest scheduling unit of the wireless communication system 100 can be dynamically selected (e.g., in a burst of shortened TTIs (sTTIs)).

[0085] Depending on the technology, carriers can be used to multiplex physical channels for communication. One or more of Time Division Multiplexing (TDM), Frequency Division Multiplexing (FDM), or hybrid TDM-FDM techniques can be used, for example, to multiplex physical control channels and physical data channels for signaling via a downlink carrier. The control region (e.g., control resource set (CORESET)) of the physical control channel can be defined by a set of symbol periods and can extend across the system bandwidth of the carrier or a subset of that bandwidth. One or more control regions (e.g., CORESET) can be configured for a set of UEs 115. For example, one or more UEs in UE 115 can monitor or search for control regions to obtain control information based on 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 concatenated manner. The aggregation level of control channel candidates can refer to the amount of control channel resources (e.g., control channel elements (CCEs)) associated with coded information for a control information format having a given payload size. The search space set may include: a common search space set configured to transmit control information to multiple UEs 115, and a UE-specific search space set used to transmit control information to a specific UE 115.

[0086] Network entity 105 may provide communication coverage via one or more cells (e.g., macro cells, small cells, hotspots, or other types of cells, or any combination thereof). The term "cell" may refer to a logical communication entity used to communicate with network entity 105 (e.g., using a carrier) and may be associated with an identifier used to distinguish adjacent cells (e.g., Physical Cell Identifier (PCID), Virtual Cell Identifier (VCID), or other cell identifier). In some examples, a cell may also refer to a coverage area 110 or a portion of coverage area 110 (e.g., a sector) in which a logical communication entity operates. Depending on various factors such as the capabilities of network entity 105, the extent of such cells may range from smaller areas (e.g., structures, subsets of structures) to larger areas. For example, a cell may be or may include buildings, subsets of buildings, or external space between or overlapping coverage areas 110, etc.

[0087] Macro cells typically cover a relatively large geographical area (e.g., a radius of several kilometers) and allow unrestricted access to UE 115 that has a service subscription with a network provider supporting the macro cell. In contrast, small cells may be associated with a lower-power network entity 105 (e.g., a lower-power base station 140) and may operate using the same or different (e.g., licensed, unlicensed) frequency bands as macro cells. Small cells may provide unrestricted access to UE 115 that has a service subscription with a network provider, or restricted access to UE 115 associated with a small cell (e.g., UE 115 in a closed subscriber group (CSG), or UE 115 associated with a user in a home or office). Network entity 105 may support one or more cells and may also use one or more component carriers to support communication via one or more cells.

[0088] In some examples, a carrier can support multiple cells and can be configured with different cells based on different protocol types that can provide access for different types of devices (e.g., MTC, Narrowband IoT (NB-IoT), Enhanced Mobile Broadband (eMBB)).

[0089] In some examples, network entity 105 (e.g., base station 140, RU 170) may be mobile, and thus provide communication coverage to mobile coverage areas 110. In some examples, different coverage areas 110 associated with different technologies may overlap, but the different coverage areas 110 may be supported by the same network entity 105. In some other examples, overlapping coverage areas 110 associated with different technologies may be supported by different network entities 105. The wireless communication system 100 may 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.

[0090] Wireless communication system 100 may be configured to support ultra-reliable communication or low-latency communication, or various combinations thereof. For example, wireless communication system 100 may be configured to support ultra-reliable low-latency communication (URLLC). UE 115 may be designed to support ultra-reliable, low-latency, or critical functions. Ultra-reliable communication may include private or group communication and may be supported by one or more services, such as push-to-talk, video, or data. Support for ultra-reliable, low-latency functions may include prioritizing services, and such services may be used for public safety or general business applications. The terms “ultra-reliable,” “low-latency,” and “ultra-reliable low-latency” are used interchangeably herein.

[0091] In some examples, UE 115 may be configured to support direct communication with other UE 115s via device-to-device (D2D) communication link 135 (e.g., according to peer-to-peer (P2P), D2D, or sidelink protocols). In some examples, one or more UE 115s performing D2D communication in a group may be within the coverage area 110 of network entity 105 (e.g., base station 140, RU 170), which may support aspects of such D2D communication configured (e.g., scheduled by network entity 105). In some examples, one or more UE 115s in such a group may be outside the coverage area 110 of network entity 105, or may otherwise be unable or not configured to receive transmissions from network entity 105. In some examples, the group of UE 115s communicating via D2D communication may support a one-to-many (1:M) system, where each UE 115 transmits to each of the other UE 115s in the group. In some examples, network entity 105 may facilitate the scheduling of resources used for D2D communication. In other examples, D2D communication may be performed between UEs 115 without involving network entity 105.

[0092] Core network 130 provides user authentication, access authorization, tracking, Internet Protocol (IP) connectivity, and other access, routing, or mobility functions. Core network 130 can be an evolved packet core (EPC) or a 5G core (5GC), which may include at least one control plane entity (e.g., a mobility management entity (MME), access and mobility management function (AMF)) for managing access and mobility, and at least one user plane entity (e.g., a serving gateway (S-GW), packet data network (PDN) gateway (P-GW), or user plane function (UPF)) for routing packets or interconnecting to external networks. The control plane entity manages non-access stratum (NAS) functions, such as mobility, authentication, and bearer management of UE 115 served by network entity 105 (e.g., base station 140) associated with core network 130. User IP packets can be transferred through user plane entities, which provide IP address allocation and other functions. User plane entities can connect to one or more network operator IP services 150. IP services 150 may include access to the Internet, intranets, IP Multimedia Subsystem (IMS), or packet-switched streaming services.

[0093] Wireless communication system 100 can operate using one or more frequency bands in the range of 300 MHz to 300 GHz. Generally, the area from 300 MHz to 3 GHz is referred to as the Ultra High Frequency (UHF) band or decimeter band because the wavelength range is approximately one decimeter to one meter in length. UHF waves may be blocked or redirected by buildings and environmental features (which may be referred to as clusters), but these waves are sufficient to penetrate structures so that macrocells can provide service to UE 115 located indoors. Compared to communication using smaller frequencies and longer wavelengths in the lower frequency (HF) or very high frequency (VHF) portions of the spectrum below 300 MHz, communication using UHF waves can be associated with smaller antennas and shorter ranges (e.g., less than 100 km).

[0094] Wireless communication system 100 may utilize both licensed and unlicensed RF spectrum bands. For example, wireless communication system 100 may use unlicensed bands (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 with unlicensed RF spectrum, devices such as network entity 105 and UE 115 may employ carrier sensing for collision detection and avoidance. In some examples, operation using unlicensed bands may be combined with component carriers operating with licensed bands based on carrier aggregation configurations (e.g., LAA). Operation using unlicensed spectrum may include downlink transmission, uplink transmission, P2P transmission, or D2D transmission, etc.

[0095] Network entity 105 (e.g., base station 140, RU 170) or UE 115 may be equipped with multiple antennas that can be used to employ techniques such as transmit diversity, receive diversity, multiple-input multiple-output (MIMO) communication, or beamforming. The antennas of network entity 105 or UE 115 may be located within one or more antenna arrays or antenna panels, which can 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 network entity 105 may be located at different geographical locations. Network entity 105 may include an antenna array having a collection of multiple rows and columns of antenna ports that network entity 105 can use to support beamforming for communication with UE 115. Similarly, UE 115 may include one or more antenna arrays that can support various MIMO or beamforming operations. Additionally or alternatively, the antenna panel may support RF beamforming for signals transmitted via the antenna ports.

[0096] Network entity 105 or UE 115 can use MIMO communication to leverage multipath signal propagation and improve spectral efficiency by transmitting or receiving multiple signals via different spatial layers. This technique is known as spatial multiplexing. The multiple signals can be transmitted, for example, by a transmitting device via different antennas or different combinations of antennas. Similarly, the multiple signals can be received by a receiving device via different antennas or different combinations of antennas. Each of the multiple signals can be referred to as a separate spatial stream and can carry information associated with the same data stream (e.g., the same codeword) or different data streams (e.g., different codewords). Different spatial layers can be associated with different antenna ports used for channel measurement and reporting. MIMO techniques include: single-user MIMO (SU-MIMO), where multiple spatial layers are transmitted to the same receiving device; and multi-user MIMO (MU-MIMO), where multiple spatial layers are transmitted to multiple devices.

[0097] Beamforming (also known as spatial filtering, directional transmission, or directional reception) is a signal processing technique that can be used at a transmitting or receiving device (e.g., network entity 105, UE 115) to shape or guide an antenna beam (e.g., a transmit beam, a receive beam) along a spatial path between the transmitting and receiving devices. Beamforming can be achieved by combining signals transmitted via antenna elements of an antenna array such that some signals propagating along a specific orientation relative to the antenna array experience constructive interference, while other signals experience destructive interference. Adjustments to the signals transmitted via the antenna elements may include applying amplitude shifts, phase shifts, or both to the signals carried via the antenna elements associated with the device. The adjustments associated with each of these antenna elements may be defined by a beamforming weight set associated with a specific orientation (e.g., relative to the antenna array of the transmitting or receiving device or relative to some other orientation).

[0098] Network entity 105 or UE 115 may use beam scanning technology as part of beamforming operations. For example, network entity 105 (e.g., base station 140, RU 170) may 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) may be transmitted multiple times by network entity 105 along different directions. For example, network entity 105 may transmit signals according to different beamforming weight sets associated with different transmission directions. Transmission along different beam directions may be used to identify (e.g., by a transmitting device (such as network entity 105) or by a receiving device (such as UE 115)) the beam direction for later transmission or reception by network entity 105.

[0099] Some signals (such as data signals associated with a specific receiving device) may be transmitted by a transmitting device (e.g., transmitting network entity 105, transmitting UE 115) along a single beam direction (e.g., the direction associated with the 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 may be determined based on the signals transmitted along one or more beam directions. For example, UE 115 may receive one or more signals transmitted by network entity 105 along different directions and may report to network entity 105 an indication of signals received by UE 115 with the highest signal quality or other acceptable signal quality.

[0100] In some examples, transmissions performed by a device (e.g., network entity 105 or UE 115) may be performed using multiple beam directions, and the device may use a combination of digital pre-decoding or beamforming to generate a combined beam for transmission (e.g., from network entity 105 to UE 115). UE 115 may report feedback indicating pre-decoding weights for one or more beam directions, and this feedback may correspond to a beam set configured across system bandwidth or one or more sub-bands. Network entity 105 may transmit reference signals that may be pre-decoded or undecoded (e.g., cell-specific reference signals (CRS), CSI-RS). UE 115 may provide feedback on beam selection, which may be a pre-decoded matrix indicator (PMI) or codebook-based feedback (e.g., multi-panel codebook, linear combination codebook, port selection codebook). Although these techniques are described with reference to signals transmitted by network entity 105 (e.g., base station 140, RU 170) along one or more directions, UE 115 may use similar techniques to transmit signals multiple times along different directions (e.g., to identify the beam direction used by UE 115 for subsequent transmission or reception), or to transmit signals along a single direction (e.g., to transmit data to a receiving device).

[0101] 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 transmitting device (e.g., network entity 105). For example, the receiving device may perform reception according to multiple reception directions by: receiving via different antenna subarrays; processing the received signal according to different antenna subarrays; 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 the antenna array; or processing the received signal according to different sets of reception beamforming weights applied to signals received at multiple antenna elements of the 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 a data signal is received). A single receiver configuration can be aligned along a beam direction determined based on listening according to different receiver 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).

[0102] The wireless communication system 100 can be a packet-based network operating according to a layered protocol stack. In the user plane, communication at the bearer or PDCP layer can be IP-based. The RLC layer performs packet segmentation and reassembly for transmission via logical channels. The MAC layer performs priority handling and multiplexing of logical channels to transport channels. The MAC layer can also use error detection, error correction, or both to support retransmissions to improve link efficiency. In the control plane, the RRC layer provides the establishment, configuration, and maintenance of RRC connections between the UE 115 and network entity 105 or core network 130 that support user plane data radio bearers. The PHY layer maps transport channels to physical channels.

[0103] UE 115 can support AI- or ML-based beam prediction (e.g., referred to as AI-based beam prediction). Such a UE can collect data measurements (e.g., RSRP, SINR, CIR) based on SSB or CSI-RS measurements via SSB or CSI-RS beams. The AI ​​model can be trained using measurements of a first beam set (e.g., set B beams) to predict measurements of future beam sets (e.g., set A beams). The trained AI model uses actual measurements of a third beam set (e.g., set B beams) to predict measurements of a fourth beam set (e.g., set A beams) (e.g., referred to as a beam inference process), which UE 115 can report in a beam measurement report. For example, beam measurement reports can be used for purposes such as initial access, beam management in RRC connection mode, beam fault recovery (BFR), and radio link fault recovery response. In some examples, the AI ​​model can support spatial domain downlink beam prediction for set A beams based on measurements of set B beams. In some examples, the AI ​​model can support temporal downlink beam prediction for set A based on historical measurements of set B beams. In some examples, set A and set B beams can be in the same frequency range. In some examples, set B beams can be a subgroup of set A beams. In some examples, set B beams can be different from set A beams (e.g., set A beams can be narrower than set B beams). As described herein, the codebook construction of set A and set B beams can be vendor-specific to different gNB (e.g., network entity 105) providers.

[0104] The mapping from beamforming measurements to AI model inputs affects the AI ​​model's output (e.g., predicted measurements). Therefore, for accurate use of the AI ​​model, UE 115 and network entity 105 (e.g., gNB) should agree on the mapping and ordering of CMR or IMR to AI model inputs. However, existing CMR or IMR frameworks do not allow UE 115 to identify the order of AI model inputs because network entity 105 can flexibly map beamforming codebooks to CMR or IMR. Beamforming codebook information used by network entity 105 for CMR or IMR can be proprietary, and network entity vendors may be reluctant to signal such information. Furthermore, additional AI-based beamforming prediction reporting considerations may include mechanisms related to reporting, the content of the report (e.g., information associated with or configured for the reported data samples, such as timestamps, signal-to-noise ratio (SNR), and data quality), reduction of reporting overhead, and UE complexity and power consumption.

[0105] In some examples, the set of trained AI models can be standardized or predefined. These standards can define the mapping or ordering of beam measurements (e.g., which SSB or CSI-RS index) to the AI ​​model inputs and / or AI model outputs of the standardized AI models. UE 115 and network entity 105 can negotiate mutually supported AI models from the list of standardized AI models. For example, network entity 105 can indicate a subset of AI model IDs from the set of standardized AI model IDs, and UE 115 can indicate which AI model IDs in the indicated subset it supports. Network entity 105 can schedule a set of beam measurement resources and can indicate the AI ​​model IDs mutually supported by network entity 105 and UE 115. UE 115 can use the indicated AI models and beam measurement resources to generate and send beam prediction reports. By using mutually supported AI models from the list of trained AI models, UE 115 and network entity 105 can agree on the inputs and outputs of the AI ​​models, which enables accurate beam prediction.

[0106] In some examples, network entity 105 may identify the AI ​​model to UE 115 during the beam inference process, enabling the UE to ensure that the actual beam used for the beam inference process (e.g., SSB or CSI-RS) corresponds to the actual beam used to train the AI ​​model (e.g., SSB or CSI-RS) (e.g., a one-to-one association), and that the measurement of the predicted beam used for the beam inference process (e.g., the predicted measurement of SSB or CSI-RS) corresponds to the predicted beam used to train the AI ​​model (e.g., the predicted measurement of SSB or CSI-RS) (e.g., a one-to-one association). In such examples, UE 115 and network entity 105 may first perform the training process using a training input set and a desired output set or the AI ​​model.

[0107] Figure 2 An example of a wireless communication system 200 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. Wireless communication system 200 may implement, or be implemented by, aspects of wireless communication system 100. For example, wireless communication system 200 may include UE 115-a, which may be an example of UE 115 as described herein. Wireless communication system 200 may include network entity 105-a, which may be an example of network entity 105 as described herein.

[0108] UE 115-a can communicate with network entity 105-a using communication link 125-a. Communication link 125-a can be an example of an NR or LTE link between UE 115-a and network entity 105-a. Communication link 125-a can include a bidirectional link that enables both uplink and downlink communication. For example, UE 115-a can use communication link 125-a to send uplink signals 205 (e.g., uplink transmissions), such as uplink control signals or uplink data signals, to network entity 105-a, and network entity 105-a can use communication link 125-a to send downlink signals 210 (e.g., downlink transmissions), such as downlink control signals or downlink data signals, to UE 115-b. In some examples, network entity 105-a can perform beamforming to send downlink signals 210 to UE 115-a via one or more beams 215.

[0109] As described herein, UE 115-a can use an AI model to perform beam prediction to predict future measurements of beam 215. For example, network entity 105-a can send control signaling 230 that schedules a set of beam measurement resources for reference signal set 235 (e.g., SSB or CSI-RS). Network entity 105-a can transmit reference signal set 235 via a first beam set 215 (e.g., set B beam). UE 115-a can perform measurements on reference signal set 235 and can use an AI model to predict measurements for a second beam set 215 (e.g., set A beam). The UE can send a report 240 indicating the predicted measurements for the second beam set 215. In some examples, report 240 can be a CSI report and can also indicate measurements for reference signal set 235.

[0110] As described herein, in some examples, a training process can be used to train the AI ​​model used at UE 115-a. During UE-side data collection for model training, UE 115-a can identify the order of the corresponding set A beams and set B beams, and can map the order of the corresponding set A beams and set B beams to the input and output features of the AI ​​model. During AI model inference (e.g., during online beam prediction to generate the predicted measurement results to be indicated in report 240), UE 115-a and network entity 105-a should agree on the order of the corresponding set A beams and set B beams mapped to the input and output features of the AI ​​model, such that UE 115-a knows which set B beam's measured L1-RSRP to input into which AI model input feature, and that UE 115-a can determine the predicted L1-RSRP of a given set A beam based on the given AI model output features. However, because network entity 105-a has the flexibility to change the mapping between beamforming codebook and CMR / IMR, the CMR / IMR indexing framework may not be able to satisfy a consistent mapping between set B beams to AI model input features and set A beams to AI model output features.

[0111] A direct approach to consistently mapping set B beams to AI model input features and set A beams to AI model output features would be for network entity 105-a to indicate beamforming codebook information (e.g., beam adjacency information or beam pointing direction information) regarding CMR or IMR. However, the network entity's vendor may be unwilling to disclose such beamforming codebook information, which may be proprietary. Therefore, alternative solutions can be used to ensure beam ID labeling consistency or correspondence for the lifecycle management of AI models used for beam prediction.

[0112] In some examples, the set of trained AI models can be standardized or predefined. These standards can define the mapping or ordering of beam measurements (e.g., which SSB or CSI-RS index) to the AI ​​model inputs and / or AI outputs of the standardized AI models. UE 115-a and network entity 105-a can negotiate mutually supported AI models from the list of standardized AI models. For example, network entity 105 can send a first message 220 indicating a subset of AI model IDs from the set of standardized AI model IDs, and UE 115-a can send a second message 225 indicating which AI model IDs in the indicated subset are supported by UE 115. Network entity 105 can indicate the AI ​​model IDs mutually supported by network entity 105-a and UE 115-a in control signaling 230. UE 115-a can use the indicated AI models and beam measurement resources to generate beam prediction values ​​indicated in report 240.

[0113] In some examples, network entity 105-a may identify the AI ​​model to UE 115-a in control signaling 230, such that UE 115-a can ensure that the actual beam used for the beam inference process (e.g., SSB or CSI-RS) corresponds to the actual beam used to train the AI ​​model (e.g., SSB or CSI-RS) (e.g., a one-to-one association), and the measurement of the predicted beam used for the beam inference process (e.g., the predicted measurement of SSB or CSI-RS) corresponds to the predicted beam used to train the AI ​​model (e.g., the predicted measurement of SSB or CSI-RS) (e.g., a one-to-one association). In such examples, UE 115-a and network entity 105-a may first perform the training process using a training input set and a expected output set or the AI ​​model.

[0114] Figure 3 An example of an AI model training and inference diagram 300 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The AI ​​model training and inference diagram 300 may be implemented by, or can be implemented by, aspects of wireless communication system 100 or wireless communication system 200. For example, AI model 310 may be operated by UE 115 as described herein.

[0115] In training procedure 305, UE 115 collects measurement results associated with the first and second sets of SSB / CSI-RS resources to train an AI model for prediction in inference procedure 315. The channel characteristics of the fourth set of SSB / CSI-RS / virtual resources (i.e., set A beams) are predicted based on measurements of the third set of SSB / CSI-RS resources (i.e., set B beams). The AI ​​model 310 trained by UE 115 in training procedure 305 can be requested later by network entity 105 for use in inference procedure 315. The measured results of the first or second set of SSB / CSI-RS or the predicted channel characteristics of the fourth set of SSB / CSI-RS / virtual resources may include at least one of L1-RSRP, L1-SINR, the top K resources in terms of L1-RSRP / L1-SINR, Channel Quality Index (CQI), rank, or CIR. UE 115 and network entity 105 may identify the same AI model ID for AI model 310 during data collection for the training process and data collection for online prediction for the inference process 315, such that the following resource associations are determined to correspond to the identified AI model ID: (1) a one-to-one association between the third set of SSB / CSI-RS / virtual resources considered for model inference and the corresponding SSB / CSI-RS resources in the first set of SSB / CSI-RS resources used for data collection; and (2) a one-to-one association between the fourth set of SSB / CSI-RS resources considered for model inference and the corresponding SSB / CSI-RS resources in the second set of SSB / CSI-RS resources used for data collection.

[0116] In some examples, this correspondence can be maintained across network entities. For instance, the training process 305 can be performed using a first network entity 105, and the inference process can be performed using a second, different network entity 105 (e.g., a different serving cell, component carrier (CC), bandwidth portion (BWP), DU, or CU). For example, if the AI ​​model ID is identified by the subsequent inference process 315 regarding the online beam prediction at the second network entity 105 trained using the first network entity 105: (1) the third set of SSB / CSI-RS / virtual resources considered for model inference associated with the first network entity 105 (e.g., the first serving cell / CC / BWP / DU / CU) is associated one-to-one with the corresponding SSB / CSI-RS resources in the third set of SSB / CSI-RS resources associated with the second network entity (e.g., the second serving cell / CC / BWP / DU / CU); and (2) the fourth set of SSB / CSI-RS / virtual resources considered for model inference associated with the first network entity 105 (e.g., the first serving cell / CC / BWP / DU / CU) is associated one-to-one with the corresponding SSB / CSI-RS resources in the second set of SSB / CSI-RS resources associated with the second network entity (e.g., the second serving cell / CC / BWP / DU / CU).

[0117] In some examples, UE 115 can anticipate consistency in the number of resources between the training process 305 and the inference process. For example, UE 115 can anticipate that the first and third resource groups may include the same number of resources, and that the second and fourth resource groups may include the same number of resources. In some examples, UE 115 can anticipate a one-to-one association between the first and third resource groups. For example, during the inference process 315, UE 115 can use measurements associated with the nth resource in the third resource group to determine the input features of the AI ​​model trained during the data collection phase of the training process 305 based on the nth resource in the first resource group. In some examples, UE 115 can anticipate a one-to-one association between the second and fourth resource groups. For example, during the inference process 315, UE 115 can use the output features of the AI ​​model trained during the data collection phase of the training process 305 based on the nth resource in the second resource group to determine the predicted channel characteristics associated with the nth resource in the fourth resource group. The one-to-one association between the first and third resource groups and the second and fourth resource groups can be applied to the relationship between the first and second resource groups of the first network entity 105 (e.g., the first serving cell / CC / BWP / DU / CU) and the second and fourth resource groups of the second network entity 105 (e.g., the second serving cell / CC / BWP / DU / CU).

[0118] In some examples, UE 115 can determine the resource ID order so that the "m-th" resource and the "n-th" resource can be consistently identified during both the training process 305 and the inference process 315. In a first example, the resource ID order can be identified within a resource set of a single RRC configuration. For example, the corresponding resource within a single set of resources can be identified based on a resource set of a single RRC configuration (e.g., SSB resource set, CSI-RS resource set, prediction resource set, virtual resource set). Each resource can be associated with a given relative resource ID order within the corresponding resource set. The first example can be applied to situations where the data collection for both the training process 305 and the inference process 315 is associated with a CSI measurement or reporting process. In a second example, the order can be determined at network nodes other than network entity 105 (e.g., Figure 4 The resource ID order is identified with the assistance of network node 405. For example, during AI model ID registration for the training process or during AI model ID identification for the inference process 315, resources considered for a given resource group can be reported from network entity 105 to network node 405 first. Network node 405 can determine the ID order of the resources reported by network entity 105 and can signal the ID order to both UE 115 and network entity 105 (so that network node 405 can ensure the ordering consistency between training process 305 and inference process 315). The second example can be applied to situations where the resources involved cannot be directly grouped into resource sets of a single RRC configuration (e.g., set B beam may include multiple CSI-RS resource sets or may include both CSI-RS and SSB resources).

[0119] In some examples, AI model 310 can be based on a logical or physical model. For example, the AI ​​model identified by a model ID can be logical, and how the logical model maps to the physical AI model can be implementation-specific. A logical AI model can refer to the AI ​​model that is identified and assigned a model ID, and a physical AI model can refer to the actual implementation of such a model.

[0120] In some examples, measurements or predicted targets generated by the AI ​​model at UE 115 can be applied to associations corresponding to the first and third groups of resources, as well as the second and fourth groups of resources (e.g., Predecoding Matrix Indicator (PMI), Channel Quality Indicator (CQI), CIR, Rank Indicator (RI), Angle of Arrival (AoA), Downlink Angle of Arrival (DoA), Downlink Angle of Departure (AoD), Direction of Departure (DoD), Location or Position, Orientation or Blocking). For example, one-to-one associations can be at the resource level, resource port level, or resource set level. For example, for CSI compression, the input to AI model 310 can be the CIR estimated from CSI-RS ports. Therefore, the first and second groups of resources can each include a single resource, and resources can include multiple ports (e.g., 32 or 64 CSI-RS ports). Furthermore, for some use cases, one-to-one associations can be applied only to the input or only to the output. For example, for CSI compression, the AI ​​model input can be the CIR estimated from the CSI-RS port, and the AI ​​model output can be the compressed CSI based on this CIR (where the label is the PMI calculated based on the CIR). Therefore, in this case, the one-to-one association can be applied only to the input resources of the AI ​​model.

[0121] Figure 4 An example of a wireless communication system 400 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The wireless communication system 400 may implement, or may be implemented by, aspects of wireless communication system 100, wireless communication system 200, or AI model training and inference diagram 300. For example, the wireless communication system 400 may include a UE 115-b, which may be an example of UE 115 as described herein. The wireless communication system 400 may include a network entity 105-b, which may be an example of network entity 105 as described herein. UE 115-b and network entity 105-b may communicate using communication link 125-b, which may be an example of communication link 125 as described herein. In some examples, network entity 105-b may communicate with a server 410 of the supplier of network entity 105-b via communication link 430, which may be an example of communication link 120.

[0122] The wireless communication system 400 may include a network node 405 that can assist in signaling for maintaining the consistency of resource ID order between AI model training and AI model inference. For example, network node 405 may be dedicated to managing AI / ML model IDs associated with different AI / ML functionalities or features. Network node 405 may communicate with UE 115-b via communication link 420, which may be an example of communication link 125 or communication link 135 as described herein. Network node 405 may communicate with network entity 105-b via communication link 415, which may be an example of communication link 125 or communication link 120 as described herein.

[0123] When initiating data collection for the inference process using a trained AI model, network entity 105-b, UE 115-b, network entity 105-b, and network node 405 can agree on the AI ​​model ID used for training the AI ​​model. For example, AI model features or functionality can be temporal and spatial L1-RSRP predictions on set A beams for each of the set B beams within 40+80+120ms later than the last measurement timing of set B beams, with 8 set B beam inputs and 32 set A beam outputs.

[0124] For example, network entity 105-b may indicate or suggest supported AI model IDs, or it may register AI model IDs with network node 405. Network node 405 may be responsible for AI model lifecycle management and may manage the existing model IDs of network entity 105-b, as well as the associated AI model features and functionality. Network node 405 may forward AI model IDs to be used in the inference process to UE 115-b (e.g., these may be AI model IDs suggested by network entity 105-b).

[0125] Data collection for the inference process can be initiated by upper-layer commands transmitted from network node 405 to both network entity 105-b and UE 115-b, which may indicate AI models and / or AI model features or functionalities. Network entity 105-b may suggest or register AI model IDs associated with the indicated AI model features or functionalities to network node 405 (e.g., via upper-layer protocols through communication link 415). For example, server 410 of a vendor of network entity 105-b may include a database recording which AI models and AI model features or functionalities are supported by network entity 105-b associated with that vendor. Network node 405 may forward the suggested AI model IDs registered by network entity 105-b to UE 115-b via communication link 125, or network entity 105-b may also indicate the model IDs to UE 115-b via communication link 420-b.

[0126] Different vendors can register different AI model IDs for the same functionality or feature. For example, a network entity vendor can run server 410 outside the network to coordinate AI model IDs, making it possible to identify which AI model IDs are registered by the vendor and which are registered by other vendors on the 3GPP network. In this case, during data collection, network entity 105-b can register or re-register only the vendor-coordinated AI model IDs of network entity 105-b, and network entity 105-b may not interact with model IDs not associated with its vendors (e.g., based on a database of vendor-backed registered AI model IDs stored in server 410).

[0127] Figure 5 An example of a wireless communication system 500 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The wireless communication system 500 may implement, or can be implemented by, aspects of wireless communication system 100, AI model training and inference diagram 300, wireless communication system 200, or wireless communication system 400. For example, the wireless communication system 500 may include UE 115-c, which may be an example of UE 115 as described herein. The wireless communication system 500 may include network entity 105-c, which may be an example of network entity 105 as described herein. The wireless communication system 500 may include network node 405, which may be an example of network node 405 as described herein. The wireless communication system 500 may include server 410-a, which may be an example of server 410 as described herein. In some examples, network entity 105-c may communicate with server 410-a via communication link 430-a, which may be an example of communication link 120.

[0128] In some examples, such as in the first scenario 505, when initiating the AI ​​model inference process, the AI ​​model ID registered for the corresponding AI feature or function can be signaled from network node 405-a to network entity 105-c in the first message 515-a, and the AI ​​model ID can be signaled to UE 115-c in the second message 515-b. In some examples, such as in the second scenario 510, when initiating the AI ​​model inference process, the AI ​​model ID registered for the corresponding AI feature or function can be signaled directly from network entity 105-c to UE 115-c in message 520.

[0129] Network entity 105-c can ensure a one-to-one resource-level association between the training process of an AI model and the initiated inference process, for example, as referenced Figure 3 As described. In some examples, if network entity 105-c is scheduled to participate in the inference process and also in the UE-side data collection process of training the same AI model on the same AI features or functionality, then the guarantee of one-to-one resource-level association for network entity 105-c can be direct, for example, as described in the reference. Figures 6 to 8 As described. If the network entity 105-c scheduled to participate in the inference process is different from the network entity participating in the training process of the same AI model in terms of the same AI features or functionality, the additional standard pre-specification can be used to ensure one-to-one resource-level association.

[0130] If the AI ​​model ID is signaled to the UE 115-c from network node 405-a, as in the first scenario 505, the AI ​​model ID can be indicated in the command from network node 405-a that initiates the inference process. As described herein, the model ID can be based on an AI model ID suggested or recommended by network entity 105-c (e.g., based on server 410-a operated by the vendor, network entity 105-c may know the AI ​​model ID associated with the vendor of network entity 105-c, so that network entity 105-c can recommend supported AI model IDs to network node 405-a for network node 405-a or UE 115-c to choose from).

[0131] If the AI ​​model ID is signaled directly to the UE 115-c from network entity 105-c, as in scenario 510, the inference process can be initiated directly by network entity 105-c. Since network entity 105-c is aware of the AI ​​model ID associated with its vendor based on the vendor-operated server 410-a, it can indicate the supported AI model ID to the UE 115-c.

[0132] Figure 6 An example of a beam correspondence diagram 600 supporting the beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The beam correspondence diagram 600 may be implemented by, or can be implemented by, aspects of wireless communication system 100, wireless communication system 200, AI model training and inference diagram 300, wireless communication system 400, or wireless communication system 500.

[0133] Relative direction consistency diagram 605 illustrates a first beam set 615 including beams 615-a, 615-b, 615-c, 615-d, and 615-e, and a second beam set 620 including beams 620-a, 620-b, 620-c, 620-d, and 620-e. In some examples, relative beam direction metrics may be required to ensure a one-to-one resource-level association between the training and inference processes of the AI ​​model. For example, in the local coordinate system (LCS) at the center of the antenna panel based on network entity 105, (e.g., Figure 3 The relative pointing relationship between the k-th resource and the j-th resource within the first group of resources, and (for example, Figure 3 The relative line-of-sight pointing relationship between the k-th resource and the j-th resource within the third group of resources should be the same. For example, the first beam set 615 can be the first group of resources, and the second beam set 620 can be the third group of resources, and the relative line-of-sight pointing relationship between beams in the first beam set 615 can be the same as the relative line-of-sight pointing relationship between beams in the second beam set 620 (e.g., the line-of-sight pointing relationship between beams 615-a and 615-b can be the same as the line-of-sight pointing relationship between beams 620-a and 620-b). For example, if the pointing direction of the k-th resource is a negative offset relative to the j-th resource within the first group of resources in a certain LCS dimension, then the pointing direction of the k-th resource should also be relative to (e.g., Figure 3 The j-th resource within the third group of resources has the same negative offset in the same LCS dimension. When the same network entity 105 is used for training and inference, the same relative eye axis pointing relationship should also apply to the second and fourth groups of resources during training and inference (e.g., Figure 3 The second and fourth groups of resources, also known as the set A beam, are used in this context. The same relative line-of-sight pointing relationship should also apply to the first / second / third / fourth groups of resources for different network entities using the same trained AI model.

[0134] The relative beamwidth consistency diagram 610 illustrates a first beam set 625 including beams 625-a, 625-b, 625-c, 625-d, and 625-e, and a second beam set 630 including beams 630-a, 630-b, 630-c, 630-d, and 630-e. In some examples, relative beamwidth relationship metrics may be needed to ensure a one-to-one resource-level association between the training and inference processes of an AI model. For example, (e.g., Figure 3 The relative beamwidth relationship between the k-th and j-th resources within the first group of resources and (e.g., Figure 3The relative beamwidth relationship between the k-th and j-th resources within the third group of resources should be the same. For example, if the beamwidth of the k-th resource is narrower than the beamwidth of the j-th resource within the first group of resources, then the beamwidth of the k-th resource should also be narrower than the beamwidth of the j-th resource within the third group of resources. For example, if beam 625-e is narrower than beam 625-d in the first beam set 625 corresponding to the first group of resources, then beam 630-e should be narrower than beam 630-d in the second beam set 630 corresponding to the third group of resources. When the same network entity 105 is used for training and inference processes, the same relative beamwidth relationship should also apply to the second and fourth groups of resources in the training and inference processes (e.g., ...). Figure 3 The second and fourth groups of resources, also referred to as set A beams, are used in this context. The same relative beamwidth relationships should also apply to the first / second / third / fourth groups of resources for different network entities using the same trained AI model.

[0135] Figure 7 An example of a beam correspondence diagram 700 supporting the beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The beam correspondence diagram 700 may be implemented by or can be implemented by aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400 or wireless communication system 500.

[0136] Absolute directionality consistency diagram 705 illustrates the first beam set 715, the second beam set 720, and the third beam set 725. In some examples, absolute beam directionality metrics may be required to ensure a one-to-one resource-level association between the training and inference processes of the AI ​​model. For example, in the LCS at the center of the antenna panel based on network entity 105 (e.g., Figure 3 The relative pointing relationship between the k-th resource and the j-th resource within the first group of resources, and (for example, Figure 3The relative line-of-sight pointing relationship between the k-th and j-th resources within the third resource group should be within the standard predefined range. For example, if the pointing direction of the k-th resource is a negative offset of X degrees relative to the j-th resource in the first resource group in a certain LCS dimension, then the pointing direction of the k-th resource should also be a negative offset of X+Y degrees relative to the j-th resource in the third resource group in the same LCS dimension, where the value of Y (x > 0 and x ≪ |y|) should be standard predefined. For example, the relative directional offset between beams 715-a and 715-d can be X degrees, where the first beam set 715 is the first resource group, and the second beam set 720 or the third beam set 725 is the third resource group. Therefore, the relative directional offsets between beams 720-a and 720-d, and between beams 725-a and 725-d, should be within X+Y degrees. When the same network entity 105 is used in the training and inference processes, the same absolute eye axis pointing relationship should also apply to the second and fourth sets of resources in the training and inference processes (e.g., Figure 3 The second and fourth groups of resources, also known as the set A beam, are used in this context. The same absolute line-of-sight pointing relationship should also apply to the first / second / third / fourth groups of resources for different network entities using the same trained AI model.

[0137] Absolute beamwidth consistency diagram 710 illustrates the first beam set 730, the second beam set 735, and the third beam set 740. In some examples, absolute beamwidth relationship metrics may be needed to ensure a one-to-one resource-level correlation between the training and inference processes of an AI model. For example, (e.g., Figure 3 The relative beamwidth relationship between the k-th and j-th resources within the first group of resources and (e.g., Figure 3 The relative beamwidth relationship between the k-th and j-th resources within the third group of resources should not exceed ±1 degree, where 1 > 0 is a predefined value in the standard. For example, the relative beamwidth difference between beams 730-a and 730-d can be Z degrees, where the first beam set 730 is the first group of resources, and the second beam set 720 or the third beam set 725 is the third group of resources. Therefore, the relative beamwidth difference between beams 735-a and 735-d, and between beams 740-a and 740-d, should be within Z degrees. When the same network entity 105 is used for training and inference processes, the same absolute beamwidth relationship should also apply to the second and fourth groups of resources in the training and inference processes (e.g., ...). Figure 3 The second and fourth groups of resources, also referred to as set A beams, are used in this context. The same absolute beamwidth relationship should also apply to the first / second / third / fourth groups of resources for different network entities using the same trained AI model.

[0138] Figure 8 An example of a beam QCL relationship consistency map 800 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The beam QCL relationship consistency map 800 may be implemented by or can be implemented by aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400 or wireless communication system 500.

[0139] In some examples, to ensure a one-to-one resource-level association between the AI ​​model training process 805 and the inference process 810 using the AI ​​model, a resource-level D-type QCL consistency metric may be required, for example, based on a standard predefined one. QCL type D refers to spatial reception parameters shared between two signals or channels. For example... Figure 8 As shown, the first set of resources refers to the set of beams B in training process 805 (e.g., the 8 wide beams corresponding to the SSB resource set), and the second set of resources refers to the set of beams A in training process 805 (e.g., the 32 narrow beams corresponding to the CSI-RS resource set). Figure 8 As shown, the third set of resources refers to the set of B beams of the inference process 810 (e.g., the 8 wide beams corresponding to the SSB resource set), and the fourth set of resources refers to the set of A beams of the inference process 810 (e.g., the 32 narrow beams corresponding to the CSI-RS resource set).

[0140] For example, based on the QCL consistency metric, if the j-th resource in the first group of resources in training process 805 has a D-type QCL relationship with the k-th resource in the second group of resources in training process 805, then it can also be expected that the j-th resource in the third group of resources in inference process 810 has a D-type QCL relationship with the k-th resource in the fourth group of resources in inference process 810. Similarly, if the j-th resource in the first group of resources in training process 805 has a D-type QCL relationship with the k-th resource in the second group of resources in training process 805 of the first network entity 105 (e.g., the first serving cell / CC / BWP / DU / CU), then... kIf a resource has a D-type QCL relationship, then it can also be expected that the j-th resource in the first group of resources has a D-type QCL relationship with the k-th resource in the second group of resources in the training process 805 of the second network entity 105 (e.g., the second serving cell / CC / BWP / DU / CU). Similarly, if the j-th resource in the first group of resources in training process 805 has a D-type QCL relationship with the k-th resource in the second group of resources in the training process 805 of the first network entity 105 (e.g., the first serving cell / CC / BWP / DU / CU), then it can also be expected that the j-th resource in the third group of resources has a D-type QCL relationship with the k-th resource in the fourth group of resources in the inference process 810 of the second network entity 105 (e.g., the second serving cell / CC / BWP / DU / CU).

[0141] In some examples, to ensure a one-to-one resource-level association between the training process 805 of the AI ​​model and the inference process 810 using the AI ​​model, the time metric can be a standard predefined one. In some examples, the time metric can be defined such that if the nth resource in {first, second} resource groups is associated with periodicity A, then the nth resource in {third, fourth} resource groups should be associated with the same periodicity A. In some examples, the time metric can be defined such that if the nth AI model output feature is trained only on a finite set of future time-domain opportunities for the nth resource in the second resource group, then UE 115 should expect the nth AI model output feature to be used to determine the predicted channel characteristic associated with the nth resource in the fourth resource group, which is also associated only with the same finite set of future time-domain opportunities. In some examples, future time-domain opportunities can be defined based on a finite number of future time-domain offsets compared to the last symbol of the most recently measured resource in the first resource group during training process 805. In some examples, future time-domain timing can be defined based on a finite number of future time-domain offsets compared to the last symbol of the latest measured resource in the second set of resources during inference process 810.

[0142] Figure 9 An example of an AI model ID registration scenario 900 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The AI ​​model ID registration scenario 900 may be implemented by or can be implemented by aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400, or wireless communication system 500.

[0143] In some examples, AI model IDs can be predefined or standardized to ensure resource ID consistency between training and inference processes. For example, various logical or physical AI model IDs can be predefined or pre-listed in standards accessible to UE 115 and network entity 105 (e.g., 3GPP standards). In some examples, the AI ​​features or functionalities supported by the listed AI model IDs can be predefined or standardized. For example, for a beam prediction AI feature or functionality (or for a sub-feature or functionality under a beam prediction feature or functionality), different network entity vendors can "pre-register" multiple model IDs to be predefined or pre-listed in standards (e.g., in 3GPP standards). For example, such as... Figure 9 As shown, network entity vendor #1 can pre-register AI model IDs #1–50, network entity vendor #2 can pre-register AI model IDs #51–100, network entity vendor #3 can pre-register AI model IDs #101–150, network entity vendor #4 can pre-register AI model IDs #151–200, and so on. It is optional to specify in the standard which gNB vendor contributes to which model ID. (See reference...) Figure 2 As described, network entity 105 can signal to UE 115 which AI model IDs it supports, and UE 115 can signal to network entity 105 which AI model IDs it supports, so that network entity 105 and UE 115 can identify the AI ​​model IDs that both network entity 105 and UE 115 support.

[0144] The performance details of the AI ​​model associated with the registered AI model ID can be defined based on RAN4 requirements (e.g., top beam prediction accuracy, L1-RSRP prediction accuracy, etc.). See reference... Figure 3 – Figure 8As described, the resource-level one-to-one association and / or relative / absolute resource characteristic consistency between the first and second resource groups used for the training process and the third and fourth resource groups used for the inference process, or between inferences processed for different network entities, can be predefined in the standard (e.g., the definition of AI model ID in the standard). Where the standard does not specify resource-level one-to-one association and / or relative / absolute resource characteristic consistency, offline or interoperability device testing (IoDT) can be performed between the network entity and the UE vendor to ensure that the performance requirements specified by RAN4 for a given AI model are met (e.g., top beam prediction accuracy, L1-RSRP prediction accuracy, etc.). In some examples, UE 115 can signal whether it supports using only a pre-registered AI model ID for the inference process (e.g., referred to as Solution 1), whether it supports training an AI model and then using the trained AI model for the inference process while maintaining the resource-level one-to-one association (e.g., referred to as Solution 2), or whether UE 115 supports both Solution 1 and Solution 2.

[0145] about Figure 2 In some examples, the first message 220 may be sent via system information, RRC, Media Access Control (MAC) control element (MAC-CE), or Downlink Control Information (DCI). For example, network entity 105-a may broadcast a first message indicating the supported AI model ID in Residual Minimal System Information (RMSI) or other System Information (OSI). Alternatively, network entity 105-a may signal the first message 220 indicating the supported model ID via a UE-specific RRC, which may be based on a subset of model IDs already reported by the UE (e.g., in the second message 225) and / or may be based on a serving cell or BWP-specific configuration. Furthermore, network entity 105-a may signal the first message 220 indicating the supported model ID via CSI reporting settings, where the CSI reporting settings will be used for beam prediction result feedback, which may be based on a subset of model IDs already reported by the UE (e.g., in the second message 225). For example, network entity 105-a can signal a first message 220 indicating the supported model IDs by activating the MAC-CE of semi-persistent (SP) CSI reporting, where the CSI reporting settings will be used for beam prediction result feedback, which can be based on a subset of model IDs that have already been reported by the UE (e.g., in the second message 225). Alternatively, network entity 105-a can also signal a first message 220 indicating the supported model IDs by activating the DCI of aperiodic CSI reporting. CSI-AssociatedReportConfigInfoThe information element sends a first message 220 to signal the supported model IDs, where CSI reporting settings will be used for beam prediction result feedback, which can be based on a subset of model IDs that have been reported by the UE (e.g., in the second message 225). Network entity 105 can also use this signaling to indicate whether network entity 105 supports Solution 1 and / or Solution 2.

[0146] about Figure 2 In some examples, the second message 225 may be sent via UE capability report, RRC, MAC-CE, or uplink control information (UCI). For example, UE 115 may report the AI ​​model IDs supported by the UE during the initial access procedure via a UE capability report or via a UE-specific RRC message. This may be based on the AI ​​model IDs supported by the network entities indicated in the RMSI / OSI and / or may be based on different or the same capabilities reported for different serving cells, CCs, BWPs, DUs, CUs, or any combination thereof. In some examples, UE 115 may update the list of AI model IDs supported by the UE via MAC-CE or UCI, which may overwrite previously reported AI model IDs supported by the UE. This may be based on the AI ​​model IDs supported by the network entities indicated in the first message 220. UE 115 may also use this signaling to indicate whether UE 115 supports Solution 1 and / or Solution 2.

[0147] Figure 10 An example of a wireless communication system 1000 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The wireless communication system 1000 may implement, or can be implemented by, aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400, or wireless communication system 500 or AI model ID registration scenario 900. For example, the wireless communication system 1000 may include UE 115-d and UE 115-c, which may be an example of UE 115 as described herein. The wireless communication system 1000 may include network entity 105-d, which may be an example of network entity 105 as described herein. The wireless communication system 1000 may include network node 405-b, which may be an example of network node 405 as described herein. UE 115-d may communicate with a UE vendor (e.g., Figure 9 UE 115-e can be associated with different UE vendors (e.g., UE vendor #5). Figure 9 The UE vendor #7) is associated with network entity 105-d, and network entity 105-d can be associated with network entity vendor (e.g., Figure 9 Related to gNB supplier #1).

[0148] The first scenario 1005 of the wireless communication system 1000 shows a reference Figure 9 An example of Solution 1 described herein is provided, in which UE 115-d and network entity 105-d can perform IoDT training of an AI model that is predefined or registered by a standard. A second scenario 1010 of the wireless communication system 1000 illustrates an example of a modified Solution 2. For example, in the second scenario, predefined or pre-listed model IDs associated with each network entity vendor can be considered to have been pre-registered by network entity 105-d for Solution 2. In such examples, the predefined or pre-registered model IDs can be reserved and may not be allowed to be registered by a network entity of a given vendor unless the model ID is contributed by that given vendor (e.g., gNB vendor #1). Upper-layer security schemes can be used to implement registration security on a network entity vendor basis. Such AI models can use the methods of Solution 2 (e.g., reference...). Figure 3 – Figure 8 (as described) to train, and can meet the consistency requirements of Solution 2.

[0149] For example, based on a small number of network entities 105-d of gNB vendor #1 and UE 115-d of UE vendor #5 in a specific location (e.g., a city), network entity vendor #1 can train an AI model corresponding to AI model ID #1 for UE 115-d of UE vendor #5 via IoDT. The AI ​​model ID can be predefined as AI model ID #1. Then, according to Solution 2 as shown in Scenario 1010, once network entity 105-d is deployed, gNB vendor #1 can register AI model ID #1, and other network entity vendors may not be allowed to register AI model ID #1. UE 115-d belonging to UE vendor #5 can continue to collect data for training purposes through Solution 2 to improve the performance of the AI ​​model corresponding to AI model ID #1, and gNB vendor #1 can continue to ensure reference... Figure 3 – Figure 8 The described consistency requirements are maintained for Solution 2. UE 115 associated with other UE vendors (such as UE115-f associated with UE vendor #7) can then participate in data collection to train a model based on the registered AI model ID #1, and the performance of UE 115-f associated with UE vendor #7 using the AI ​​model associated with AI model ID #1 can be as good as that of UE 115-d associated with UE vendor #5, even though UE 115-f associated with UE vendor #7 did not participate in the IoDT data collection and training of the AI ​​model associated with AI model ID #1.

[0150] Figure 11 An example of a process flow 1100 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. Process flow 1100 may include UE 115-f, which may be an example of UE 115 as described herein. Process flow 1100 may include network entity 105-e, which may be an example of network entity 105 as described herein. In the following description of process flow 1100, operations between network entity 105-c and UE 115-f may be sent in a different order than the example order shown, or operations performed by network entity 105-e and UE 115-f may be performed in a different order or at different times. Some operations may also be omitted from process flow 1100, and other operations may be added to process flow 1100.

[0151] In some examples, at 1105, the vendor of network entity 105-e can pre-register a certain number of AI model IDs for beam prediction. For example, AI model IDs can be predefined / pre-listed in the 3GPP specification. Which vendor contributes to which model ID can be specified or agreed upon in the 3GPP standard. Each model ID can include multiple sub-IDs, where each sub-ID is associated with different numbers of set A beams and set B beams and / or different correspondences between set A beams and set B beams.

[0152] In some examples, at 1110, network entity 105-e (associated with the network entity vendor) may perform IoDT with UE 115-f (associated with the UE vendor) on an AI model ID pre-registered by the network entity vendor. In some examples, the network entity vendor may share a training dataset with the UE vendor, or the network entity vendor may coordinate over-the-air data collection with the UE vendor. Information that can be identified for each training data sample may include: (1) a given number of ordered (e.g., based on the SSB / CSI-RS resource order in a given resource set) set B beams, along with their L1-RSRP; (2) a given number of ordered set A beams, along with the baseline truth of the prediction target (e.g., L1-RSRP or the top K beams); and (3) the correspondence between the set A and set B beams involved (e.g., QCL / parent-child relationship, absolute / relative pointing direction, beamwidth), or such correspondence between set A beams or between set B beams. Item (3) may be optional. Such information, which can be identified by each training data sample of a given AI model ID, may be specified for the AI ​​model ID in the 3GPP standard. In some examples, if the AI ​​model ID includes multiple sub-model IDs, then shared data samples for the same sub-ID should be consistent with items (1), (2), and / or (3) above, or over-the-air data collection for the same sub-ID should be consistent with items (1), (2), and / or (3) above. In some examples, the number of set A beams and set B beams may need to be the same across different sub-IDs, while the correspondence between set A beams and set B beams may vary. The UE vendor may use the training dataset (e.g., shared offline by the network entity vendor or collected over-the-air via IoDT) to train at least one AI model for each sub-ID model of the AI ​​model ID.

[0153] At 1115, UE 115-f and network entity 105-e can perform verification tests to verify that the AI ​​model pre-registered by the network entity vendor meets RAN 4 requirements (e.g., top beam prediction accuracy, L1-RSRP prediction accuracy, etc.). During the verification tests, the network entity vendor can share inference datasets offline with UE 115-f, where each data sample includes the following (and should be provided in the same order as the datasets in the IoDT test at 1110): (1) L1-RSRP of set B beams involved in the model ID; (2) the baseline truth value of the prediction target with respect to set A beams involved in the model ID; and (3) the sub-ID associated with the data sample. Item (3) may be optional. UE 115-f can feed back the prediction results to network entity 105-e via cable or via the R19 air interface (e.g., where network entity 105-e may be a test rig). The feedback data should meet RAN 4 requirements, referencing the baseline ground truth of the predicted target (e.g., L1-RSRP or the top K beam accuracy metric). In some examples, the UE performing the IoDT training procedure at 1110 may differ from the UE performing the validation test at 1115.

[0154] In some examples, for instance, if the validation test for the registered AI model ID passes, UE 115-f and network entity 105-e can perform the inference process at 1120-1140 using the AI ​​model associated with the registered AI model ID. In some examples, the UE performing the inference process at 1120-1140 may be different from the UE performing the IoDT training process at 1110 and / or performing the validation test at 1115, and / or the network entity performing the inference process at 1120-1140 may be different from the network entity performing the IoDT training process at 1110 and / or performing the validation test at 1115.

[0155] At 1120, network entity 105-e can advertise the AI ​​model ID and / or sub-ID supported by network entity 105-e. For example, network entity 105-e can advertise the AI ​​model ID and / or sub-ID supported by network entity 105-e via broadcast through RMSI or OSI, through serving cell or BWP-specific RRC configuration, or through CSI reporting settings for UE115-f to report beam prediction results. In some examples, for instance, if network entity 105-e dynamically changes the pre-decoding of set A beams and / or set B beams, network entity 105-e can indicate the update of the supported AI model ID and / or sub-ID (e.g., update to an alternative sub-ID) via MAC-CE or DCI.

[0156] At 1125, UE 115-f can indicate the AI ​​model ID supported by UE 115-f (e.g., via UE capability report or RRC configuration). In some examples, UE 115-f can indicate an update to the supported AI model ID or sub-ID via MAC-CE or UCI.

[0157] At 1130, network entity 105-e may send control signaling indicating an AI model ID supported by both network entity 105-e and UE 115-f. The control signaling may indicate a beam measurement resource set. The control signaling may indicate a mapping of the beam measurement resource set to the inputs and / or outputs of the AI ​​model (and / or features or functionalities associated with the indicated sub-ID) associated with the AI ​​model ID.

[0158] At 1135, network entity 105-e can transmit a set of reference signals via a set of beam measurement resources (e.g., via set B beams).

[0159] At 1140, UE 115-f can send a report indicating the use of an AI model and a set of predicted values ​​for a set of beams (e.g., set A beams) generated based on a set of reference signals.

[0160] Network entity 105-e can guarantee the consistency of the following items (e.g., in control signaling at 1130): (1) configuring / indicating set A and set B beams to UE 115-f based on the same number of resources and order as identified during the IoDT procedure at 1110; (2) (optionally) the correspondence between or within the set A and set B beams involved (e.g., parent-child relationship, absolute / relative pointing direction, beamwidth). Regarding item (1), for example, if set B beam is an SSB or CSI-RS, and set B beam is configured by a CMR set associated with CSI report settings, then set B beam should be ordered in the resource set based on the same order as during the IoDT procedure at 1110. For example, regarding item (1), if set A beam is a virtual resource (e.g., not actually transmitted) and set A beam is configured by a “predicted” resource set associated with the CSI report settings, then set A beam should be ordered in the resource set based on the same order identified in the IoDT procedure at 1110.

[0161] Regarding the determination of the order of set B beams during model training (e.g., during the IoDT training process at 1110), for over-the-air data collection, UE 115-f can be configured with a set of SSB or CSI-RS resources comprising B SSB or CSI-RS resources (where each SSB / CSI-RS resource has its own serving cell-specific configuration resource ID). These B SSB or CSI-RS resources are associated with {1st, 2nd, ..., Bth} entries within the resource set, thus the B SSB or CSI-RS resources are ordered as {1st, 2nd, ..., Bth} SSB or CSI-RS resources within the SSB or CSI-RS resource set. UE 115-f can use channel characteristics measured from {1st, 2nd, ..., Bth} SSB or CSI-RS resources to determine input values ​​for {1st, 2nd, ..., Bth} AI model input features (or feature groups) associated with the corresponding AI model ID.

[0162] Regarding the determination of the order of the B-beam set during model training (e.g., during the IoDT training process at 1110), for offline data sharing, B L1-RSRPs can be provided to the UE 115-f, where each L1-RSRP should also be accompanied by a resource ID, such that the B L1-RSRPs are ordered as {1st, 2nd, ..., Bth} L1-RSRPs. The UE 115-f can use the channel characteristics based on the {1st, 2nd, ..., Bth} L1-RSRP identifiers to determine the input values ​​for the {1st, 2nd, ..., Bth} AI model input features (or feature groups) associated with the corresponding AI model ID. Reserved values ​​can be used to represent unmeasured or unreported L1-RSRPs.

[0163] As described herein, the order of the B beam sets should be consistent between training (e.g., at 1110) and inference (e.g., at 1120-1140) and between inferences involving different network entities. During model inference at 1120-1140, UE 115-f can be configured with a set of B SSB or CSI-RS resources (where each SSB or CSI-RS resource has its own serving cell-specific configuration resource ID). These B SSB / CSI-RS resources are associated with {1st, 2nd, ..., Bth} entries within the resource set, thus the B SSB / CSI-RS resources are ordered as {1st, 2nd, ..., Bth} SSB or CSI-RS resources within the SSB or CSI-RS resource set. UE 115-f can use channel characteristics measured from {1st, 2nd, ..., Bth} SSB / CSI-RS resources to determine the input values ​​of the {1st, 2nd, ..., Bth} AI model input features (or feature groups) associated with the corresponding AI model ID. Such SSB or CSI-RS resource sets can be configured as the CMR set set in the CSI report (e.g., in control signaling at 1130), where UE 115-f can provide beam prediction results regarding the model ID via the corresponding CSI report at 1135.

[0164] Regarding the determination of the order of set A beams during model training (e.g., during the IoDT training process at 1110), for airborne data collection, UE 115-f can be configured with a set of SSB or CSI-RS resources comprising A SSB or CSI-RS resources (where each SSB or CSI-RS resource has its own serving cell-specific configuration resource ID). These A SSB or CSI-RS resources are associated with {1st, 2nd, ..., Ath} entries within the resource set, thus the A SSB or CSI-RS are ordered as {1st, 2nd, ..., Ath} SSB or CSI-RS resources within the SSB or CSI-RS resource set. UE 115-f can use channel characteristics measured from the {1st, 2nd, ..., Ath} SSB or CSI-RS resources to determine the baseline truth label values ​​for the {1st, 2nd, ..., Ath} AI model output features (or feature groups) associated with the corresponding AI model ID.

[0165] Regarding the determination of the order of the A-beam set during model training (e.g., during the IoDT training process at 1110), for offline data sharing, A L1-RSRPs can be provided to UE 115-f, where each L1-RSRP should also be accompanied by a resource ID, such that the A L1-RSRPs are ordered as {1st, 2nd, ..., Ath} L1-RSRPs. UE 115-f can use the channel characteristics based on the {1st, 2nd, ..., Ath} L1-RSRP identifiers to determine the baseline truth for the {1st, 2nd, ..., Ath} AI model output features (or feature groups) associated with the corresponding AI model ID. Reserved values ​​can be used to represent unmeasured or unreported L1-RSRPs.

[0166] As described herein, the order of beam set A should be consistent between training (e.g., at 1110) and inference (e.g., at 1120-1140) and between inferences involving different network entities. During model inference, UE115-f can be configured with a set of SSBs, CSI-RSs, or virtual resources comprising A SSBs, CSI-RSs, or virtual resources (where each SSB, CSI-RS, or virtual resource has its own serving cell-specific configuration resource ID). A Each SSB, CSI-RS, or virtual resource is associated with an entry {1st, 2nd, ..., 5th} within a resource set, therefore A Each SSB, CSI-RS, or virtual resource is sorted as follows: A Within an SSB, CSI-RS, or virtual resource set {the 1st, 2nd, ..., 1st} A} SSB, CSI-RS, or virtual resources. UE 115-f can use {the 1st, 2nd, ..., 3rd} SSB, CSI-RS, or virtual resources. A The AI ​​model outputs features (or feature groups) to derive values ​​from the output values ​​provided by the features (or feature groups) of the {1st, 2nd, ..., 3rd} features. A Beam prediction results associated with each SSB, CSI-RS, or virtual resource. Virtual resources can be used to identify spatial beam prediction targets for the actual transmitted set A beams. Such sets of SSBs, CSI-RS, or virtual resources can be configured as a set of "predicted resources" set in the CSI report (e.g., in control signaling at 1130), where UE 115-f is configured to provide feedback on the AI ​​model ID and beam prediction results for each SSB, CSI-RS, or virtual resource via a corresponding CSI report (e.g., transmitted at 1140).

[0167] Figure 12An example of AI model training and inference diagram 1200 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. AI model training and inference diagram 1200 may be implemented by or can be implemented by aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400 or wireless communication system 500 or AI model ID registration scenario 900.

[0168] The AI ​​model (e.g., AI / ML model ID #X) can be trained in training process 1205 and used in inference process 1210. For example... Figure 12 As shown, the data collection process in the training process 1205 across different network entities 105 should follow the same resource quantity and order consistency. Similarly, the inference process 1210 across different network entities 105 should follow the same resource quantity and order consistency.

[0169] like Figure 12 As shown, the inputs to the AI ​​model used for training process 1205 (e.g., measurements of set B beams) should be consistent with the inputs to the AI ​​model used for inference process 1210 (e.g., measurements of set B beams), and the outputs of the AI ​​model used for training process 1205 (e.g., predicted values ​​of set A beams) should be consistent with the outputs of the AI ​​model used for inference process 1210 (e.g., predicted values ​​of set A beams).

[0170] Figure 13 An example of an AI model training diagram 1300 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The AI ​​model training diagram 1300 may be implemented by or can be implemented by aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400 or wireless communication system 500 or AI model ID registration scenario 900.

[0171] like Figure 13As shown, the SSB can be the first set of resources, and the input to the AI ​​model used for training can be the L1-RSRP measurement of the indicated SSB. The output of the AI ​​model can be compared with a benchmark ground truth dataset used for AI model training. For example, the benchmark ground truth dataset used for AI model training can be the L1-RSRP measurement of the indicated CSI-RS resource, where the CSI-RS resource can be indicated in the RRC configuration 1310 of the non-zero power (NZP) CSI-RS resource. For example, the RRC configuration 1310 can include IE NZP-CSI-RS resources, which include a resourceMapping field indicating the CSI-RS resource mapping, a ScramblingID field indicating the scrambling ID, and a qcl-infoPeriodicCSI-RS field indicating the Transmit Configuration Indicator (TCI) status ID. For example, the qcl-infoPeriodicCSI-RS field can indicate that a given set of A-beam CSI-RS has a type D QCL relationship with a set of B-beam SSBs. The QCL relationship between set A beams and set B beams can be configured using existing QCL indication frameworks (e.g., via TCI states). UE 115 can identify the receive beam of each CSI-RS based on the configured TCI state ID indicated in the qcl-infoPeriodicCSI-RS field. UE 115 can identify the frequency and time resource occupancy of CSI-RS based on the configured CSI-RS resource mapping. After channel estimation and equalization, UE 115 can measure the identified resource elements and descramble the CSI-RS using the indicated scrambling ID. UE 115 can generate measurements for the indicated CSI-RS (e.g., calculate L1-RSRP).

[0172] Figure 14 An example of an AI model inference diagram 1400 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The AI ​​model inference diagram 1400 may be implemented by or can be implemented by aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400 or wireless communication system 500 or AI model ID registration scenario 900.

[0173] Figure 14 An example is shown where the input to the AI ​​model is the L1-RSRp measurement of SSB and the output of the AI ​​model is the L1-RSRp of CSI-RS. Figure 14 In this context, existing QCL indication frameworks (e.g., via TCI states) can be used to configure the QCL relationship between set A beams (e.g., CSI-RS) and set B beams (e.g., SSB). See reference... Figure 12As described, the inputs of the AI ​​model in the AI ​​model inference diagram 1400 (e.g., the L1-RSRP measurement of the SSB) should be consistent with the inputs of the AI ​​model used in the training process of the AI ​​model, and the outputs of the AI ​​model in the AI ​​model inference diagram 1400 (e.g., the predicted L1-RSRP value of CSI-RS) should be consistent with the outputs of the AI ​​model used in the training process of the AI ​​model. For example, network entity 105 (e.g., a gNB provider) can guarantee consistency at the resource level (resource quantity, order, QCL relationship, or beamform).

[0174] Figure 15 An example of an AI model training and inference scenario 1500 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. The AI ​​model training and inference scenario 1500 may be implemented by or can be implemented by aspects of wireless communication system 100, wireless communication system 200, wireless communication system 400, or wireless communication system 500, or AI model ID registration scenario 900.

[0175] As described herein, the inputs of the AI ​​model used in the training and inference scenarios 1500 (e.g., L1-RSRP measurements of SSB) should be consistent with the inputs of the AI ​​model used in the inference process, and the outputs of the AI ​​model used in the training and inference scenarios 1500 (e.g., predicted L1-RSRP values ​​of CSI-RS) should be consistent with the outputs of the AI ​​model used in the inference process. Given the same AI model ID (e.g., model ID #X), the quantity, order, D-type QCL relation, and beamform at the resource level should be consistent during training and inference.

[0176] For example, for RRC, quantity consistency, beam shape consistency, and / or QCL consistency can be configured for training and inference. For quantity consistency, the same number of beams (e.g., SSBs, CSI-RSs, or virtual resources) in set A and set B can be consistent during training and inference in an RRC configuration. Beam shape consistency can refer to the relative pointing direction and beamwidth difference between the physical beams of two resources remaining the same between two sets of SSB resources used for training and inference, between two sets of CSI-RS resources used for training and inference, or between two sets of virtual resources used for training and inference. QCL consistency can mean that if a type D QCL source with CRI=r corresponds to SSBRI=r, then a type D QCL source with PRI=r should also correspond to SSBRI=r.

[0177] For example, the inputs and outputs used for offline model training of an AI model may require mandatory sequential consistency with the online model inference process of the same AI model. For instance, to maintain sequential consistency, UE 115 should use the L1-RSRP with respect to SSBRI=k to determine the input value for the k-th input feature during training and inference; UE 115 should use the L1-RSRP with respect to CRI=k as the label for the k-th output feature during training, and UE 115 should use the value from the k-th output feature to derive the prediction result for PRI=k during inference.

[0178] Figure 16 A block diagram 1600 of a device 1605 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. Device 1605 may be an example of aspects of UE 115 as described herein. Device 1605 may include a receiver 1610, a transmitter 1615, and a communication manager 1620. Device 1605, or one or more components of device 1605 (e.g., receiver 1610, transmitter 1615, and communication manager 1620), may include at least one processor that may be coupled to at least one memory to individually or jointly support or implement the described techniques. Each of these components may communicate with each other (e.g., via one or more buses).

[0179] Receiver 1610 may provide components for receiving information (such as packets, user data, control information, or any combination thereof) associated with various information channels (e.g., control channels, data channels, information channels related to beam correspondence between AI model training and inference). The information may be transmitted to other components of device 1605. Receiver 1610 may utilize a single antenna or a collection of multiple antennas.

[0180] Transmitter 1615 may provide components for transmitting signals generated by other components of device 1605. For example, transmitter 1615 may transmit information associated with various information channels, such as control channels, data channels, and information channels related to beam correspondence between AI model training and inference, including packets, user data, control information, or any combination thereof. In some examples, transmitter 1615 may be co-located with receiver 1610 in a transceiver module. Transmitter 1615 may utilize a single antenna or a collection of multiple antennas.

[0181] The communication manager 1620, receiver 1610, transmitter 1615, or various combinations thereof, or various components thereof, may be examples of components for performing various aspects of the beam correspondence between AI model training and inference as described herein. For example, the communication manager 1620, receiver 1610, transmitter 1615, or various combinations thereof, or components thereof, may be able to perform one or more of the functions described herein.

[0182] In some examples, the communication manager 1620, receiver 1610, transmitter 1615, or various combinations or components thereof may be implemented in hardware (e.g., in communication management circuitry). The hardware may include at least one of the following: a processor, digital signal processor (DSP), central processing unit (CPU), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA) or other programmable logic device, microcontroller, discrete gate or transistor logic component, discrete hardware component, or any combination thereof, configured as or otherwise individually or collectively to support components for performing the functions described herein. In some examples, at least one processor and at least one memory coupled to said at least one processor may be configured to perform one or more of the functions described herein (e.g., instructions stored in at least one memory are executed individually or collectively by one or more processors).

[0183] Additionally or alternatively, the communication manager 1620, receiver 1610, transmitter 1615, or various combinations or components thereof may be implemented in code (e.g., as communication management software or firmware) executed by at least one processor. If implemented in code executed by at least one processor, the functionality of the communication manager 1620, receiver 1610, transmitter 1615, or various combinations or components thereof may be performed by (e.g., a general-purpose processor, DSP, CPU, ASIC, FPGA, microcontroller, or any combination of these or other programmable logic devices configured, either individually or collectively, as a component for performing the functions described in this disclosure).

[0184] In some examples, the communication manager 1620 may be configured to use a receiver 1610, a transmitter 1615, or both, or otherwise cooperate with them to perform various operations (e.g., receiving, acquiring, monitoring, outputting, transmitting). For example, the communication manager 1620 may receive information from the receiver 1610, transmit information to the transmitter 1615, or integrate with the receiver 1610, the transmitter 1615, or both to acquire information, output information, or perform various other operations as described herein.

[0185] The communication manager 1620 may support wireless communication according to examples disclosed herein. For example, the communication manager 1620 may be capable of, configured to, or operable to support components for receiving a first message from a network entity, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction. The communication manager 1620 may be capable of, configured to, or operable to support components for sending a second message to a network entity, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. The communication manager 1620 may be capable of, configured to, or operable to support components for receiving control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set. The communication manager 1620 is capable of, configured to, or operable to support components for receiving a set of reference signals via a set of beam measurement resources. The communication manager 1620 is capable of, configured to, or operable to support components for transmitting a report indicating a set of predicted values ​​for the beam set generated using an AI model and based on the set of reference signals.

[0186] By including or configuring a communication manager 1620 according to an example as described herein, device 1605 (e.g., controlling receiver 1610, transmitter 1615, communication manager 1620, or a combination thereof, or at least one processor otherwise coupled to them) can support techniques for more efficient use of communication resources.

[0187] Figure 17 A block diagram 1700 of a device 1705 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. Device 1705 may be an example of aspects of device 1605 or UE 115 as described herein. Device 1705 may include receiver 1710, transmitter 1715, and communication manager 1720. Device 1705, or one or more components of device 1705 (e.g., receiver 1710, transmitter 1715, and communication manager 1720), may include at least one processor that may be coupled to at least one memory to support the described techniques. Each of these components may communicate with each other (e.g., via one or more buses).

[0188] Receiver 1710 may provide components for receiving information (such as packets, user data, control information, or any combination thereof) associated with various information channels (e.g., control channels, data channels, information channels related to beam correspondence between AI model training and inference). The information may be transmitted to other components of device 1705. Receiver 1710 may utilize a single antenna or a collection of multiple antennas.

[0189] Transmitter 1715 may provide components for transmitting signals generated by other components of device 1705. For example, transmitter 1715 may transmit information associated with various information channels, such as control channels, data channels, and information channels related to beam correspondence between AI model training and inference, including packets, user data, control information, or any combination thereof. In some examples, transmitter 1715 may be co-located with receiver 1710 in a transceiver module. Transmitter 1715 may utilize a single antenna or a collection of multiple antennas.

[0190] Device 1705 or its various components may be examples of parts for performing various aspects of beam correspondence between AI model training and inference as described herein. For example, communication manager 1720 may include a supported AI model manager 1725, a beam prediction scheduling manager 1730, a reference signal receiving manager 1735, a beam prediction reporting manager 1740, or any combination thereof. Communication manager 1720 may be an example of aspects of communication manager 1620 as described herein. In some examples, communication manager 1720 or its various components may be configured to use receiver 1710, transmitter 1715, or both, or otherwise cooperate with them to perform various operations (e.g., receiving, acquiring, monitoring, outputting, transmitting). For example, communication manager 1720 may receive information from receiver 1710, transmit information to transmitter 1715, or integrate in combination with receiver 1710, transmitter 1715, or both to acquire information, output information, or perform various other operations as described herein.

[0191] Communication manager 1720 may support wireless communication according to examples disclosed herein. Supported AI model manager 1725 is capable of, configured to, or operable to support components for receiving a first message from a network entity, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction. Supported AI model manager 1725 is capable of, configured to, or operable to support components for sending a second message to a network entity, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. Beam prediction scheduling manager 1730 is capable of, configured to, or operable to support components for receiving control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifier and the outputs of the AI ​​model to the beam set. The reference signal receiver manager 1735 is capable of, configured to, or operable to support components for receiving a set of reference signals via a set of beam measurement resources. The beam prediction report manager 1740 is capable of, configured to, or operable to support components for sending reports indicating a set of predicted values ​​for the beam set generated using an AI model and based on the set of reference signals.

[0192] Figure 18 A block diagram 1800 illustrates a communication manager 1820 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure. The communication manager 1820 may be an example of aspects of the communication manager 1620, communication manager 1720, or both as described herein. The communication manager 1820 or its various components may be examples of components for performing various aspects of the beam correspondence between AI model training and inference as described herein. For example, the communication manager 1820 may include a supported AI model manager 1825, a beam prediction scheduling manager 1830, a reference signal receiving manager 1835, a beam prediction reporting manager 1840, an AI model training manager 1845, an AI model validation dataset manager 1850, an AI model validation manager 1855, or any combination thereof. Each of these components, or its components or sub-components (e.g., one or more processors, one or more memories), may communicate directly or indirectly with each other (e.g., via one or more buses).

[0193] Communication manager 1820 may support wireless communication according to examples disclosed herein. Supported AI model manager 1825 is capable of, configured to, or operable to support components for receiving a first message from a network entity, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction. In some examples, supported AI model manager 1825 is capable of, configured to, or operable to support components for sending a second message to a network entity, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. Beam prediction scheduling manager 1830 is capable of, configured to, or operable to support components for receiving control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifier and the outputs of the AI ​​model to the beam set. The reference signal receiver manager 1835 is capable of, configured to, or operable to support components for receiving a set of reference signals via a set of beam measurement resources. The beam prediction report manager 1840 is capable of, configured to, or operable to support components for sending reports indicating a set of predicted values ​​for the beam set generated using an AI model and based on the set of reference signals.

[0194] In some examples, the first message is one of system information, RRC message, MAC-CE, or DCI.

[0195] In some examples, the second message is one of an RRC message, a MAC-CE message, or a UCI message.

[0196] In some examples, to indicate the beam measurement resource set, control signaling indicates the CSI-RS index set or the SSB index set. In some examples, the AI ​​model identifier indicates the mapping from the CSI-RS index set or the SSB index set to the input of the AI ​​model.

[0197] In some examples, the AI ​​model identifier indicates the corresponding beamwidth and corresponding beam direction for each beam in the beam measurement resource set and for each beam in the beam set.

[0198] In some examples, the AI ​​model identifier indicates the QCL relationship between beams in the beam measurement resource set and beams in the beam set.

[0199] In some examples, the AI ​​model identifier indicates the relative timing between the set of beam measurement resources and the set of predicted values.

[0200] In some examples, based on the set of predicted values, the report also indicates the pre-decoding matrix indicator, channel quality indicator, channel impulse response, rank indicator, angle of arrival, downlink departure angle, or departure direction.

[0201] In some examples, control signaling also schedules CSI reports. In some examples, the reports are included within the CSI reports.

[0202] In some examples, the AI ​​model training manager 1845 is capable of, configured to, or operable to support components for performing a training process of an AI model with network entities using a set of training beam measurement resources to predict a set of training measurement results for the training beam set, wherein the set of beam measurement resources corresponds to the set of training beam measurement resources, and wherein the beam set corresponds to the set of training beams.

[0203] In some examples, the training beam measurement resource set includes a first CSI-RS set or a first SSB set with a first index value. In some examples, the beam measurement resource set includes a second CSI-RS set or a second SSB set with a first index value. In some examples, the same mapping from the first index value to the input value of the AI ​​model is used to predict the training measurement result set and generate a predicted value set.

[0204] In some examples, the AI ​​model validation dataset manager 1850 is capable of, configured to, or operable to support components for receiving a validation beam measurement resource set and a corresponding benchmark true prediction target for a second beam set used for an AI model from a network entity. In some examples, the reference signal receiver manager 1835 is capable of, configured to, or operable to support components for receiving a second reference signal set via the validation beam measurement resource set. In some examples, the AI ​​model validation manager 1855 is capable of, configured to, or operable to support components for sending a second report indicating that a second set of predicted values ​​for the second beam set generated using the AI ​​model based on the reference signal set satisfies an accuracy metric regarding the corresponding benchmark true prediction target, wherein the reception of control signaling is based on the transmission of the second report.

[0205] Figure 19A diagram of a system 1900 including a device 1905 supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown. Device 1905 may be an example of device 1605, device 1705, or UE 115 as described herein, or may include components thereof. Device 1905 may communicate with one or more network entities 105, one or more UEs 115, or any combination thereof (e.g., wirelessly). Device 1905 may include components for bidirectional voice and data communication, including components for transmitting and receiving communications, such as a communication manager 1920, an input / output (I / O) controller 1910, a transceiver 1915, an antenna 1925, at least one memory 1930, code 1935, and at least one processor 1940. These components may communicate electronically via one or more buses (e.g., bus 1945) or be coupled in other ways (e.g., operational ground, communication ground, functional ground, electronic ground, electrical ground).

[0206] I / O controller 1910 can manage the input and output signals of device 1905. I / O controller 1910 can also manage peripheral devices not integrated into device 1905. In some cases, I / O controller 1910 can represent a physical connection or port to an external peripheral device. In some cases, I / O controller 1910 can utilize an operating system, such as iOS. ® ANDROID ® MS-DOS ® MS-WINDOWS ® OS / 2 ® UNIX ® LINUX ® Or another known operating system. Additionally or alternatively, the I / O controller 1910 may represent or interact with a modem, keyboard, mouse, touchscreen, or similar device. In some cases, the I / O controller 1910 may be implemented as part of one or more processors, such as at least one processor 1940. In some cases, a user may interact with the device 1905 via the I / O controller 1910 or via hardware components controlled by the I / O controller 1910.

[0207] In some cases, device 1905 may include a single antenna 1925. However, in other cases, device 1905 may have more than one antenna 1925, which may be capable of concurrently transmitting or receiving multiple wireless transmissions. Transceiver 1915 may communicate bidirectionally via one or more antennas 1925 as described herein, or via a wired or wireless link. For example, transceiver 1915 may represent a wireless transceiver and may communicate bidirectionally with another wireless transceiver. Transceiver 1915 may also include a modem for: modulating packets; providing the modulated packets to one or more antennas 1925 for transmission; and demodulating packets received from one or more antennas 1925. Transceiver 1915, or transceiver 1915 and one or more antennas 1925, may be an example of transmitter 1615, transmitter 1715, receiver 1610, receiver 1710, or any combination thereof or components thereof as described herein.

[0208] At least one memory 1930 may include random access memory (RAM) and read-only memory (ROM). At least one memory 1930 may store computer-readable, computer-executable code 1935, including instructions that, when executed by at least one processor 1940, cause device 1905 to perform the various functions described herein. Code 1935 may be stored in a non-transitory computer-readable medium such as system memory or another type of memory. In some cases, code 1935 may not be directly executable by at least one processor 1940, but may enable a computer (e.g., when compiled and executed) to perform the functions described herein. In some cases, at least one memory 1930 may contain a basic I / O system (BIOS), etc., which controls basic hardware or software operations, such as interaction with peripheral components or devices.

[0209] At least one processor 1940 may include intelligent hardware devices (e.g., general-purpose processors, DSPs, CPUs, microcontrollers, ASICs, FPGAs, programmable logic devices, discrete gate or transistor logic components, discrete hardware components, or any combination thereof). In some cases, at least one processor 1940 may be configured to operate a memory array using a memory controller. In some other cases, the memory controller may be integrated into at least one processor 1940. At least one processor 1940 may be configured to execute computer-readable instructions stored in memory (e.g., at least one memory 1930) to cause device 1905 to perform various functions (e.g., functions or tasks supporting beam correspondence between AI model training and inference). For example, device 1905 or components of device 1905 may include at least one processor 1940 and at least one memory 1930 coupled to or coupled to at least one processor 1940, wherein at least one processor 1940 and at least one memory 1930 are configured to perform the various functions described herein. In some examples, at least one processor 1940 may include multiple processors, and at least one memory 1930 may include multiple memories. One or more of a plurality of processors may be coupled to one or more of a plurality of memories, which may be configured individually or collectively to perform the various functions described herein. In some examples, at least one processor 1940 may be a component of a processing system, which may refer to a machine (such as a series of machines), circuitry (including, for example, one or both of processor circuitry (which may include at least one processor 1940) and memory circuitry (which may include at least one memory 1930)) or system of components that receive or receive input and process the input to produce, generate or obtain a set of outputs. The processing system may be configured to perform one or more of the functions described herein. Thus, at least one processor 1940 or a processing system including at least one processor 1940 may be configured, capable of being configured, or operable to cause device 1905 to perform one or more of the functions described herein. Furthermore, as described herein, “configured to,” “capable of being configured,” and “operable to” are used interchangeably and may be associated with the ability to perform one or more of the functions described herein when executing code stored in at least one memory 1930 or otherwise.

[0210] The communication manager 1920 may support wireless communication according to examples disclosed herein. For example, the communication manager 1920 may be capable of, configured to, or operable to support components for receiving a first message from a network entity, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction. The communication manager 1920 may be capable of, configured to, or operable to support components for sending a second message to a network entity, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. The communication manager 1920 may be capable of, configured to, or operable to support components for receiving control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifiers and the outputs of the AI ​​model to the beam set. The communication manager 1920 is capable of, configured to, or operable to support components for receiving a set of reference signals via a set of beam measurement resources. The communication manager 1920 is capable of, configured to, or operable to support components for transmitting a report indicating a set of predicted values ​​for the beam set generated using an AI model and based on the set of reference signals.

[0211] By including or configuring a communication manager 1920 according to an example as described herein, device 1905 can support techniques for improving communication reliability, more efficient use of communication resources, improved coordination between devices, and improved utilization of processing power.

[0212] In some examples, the communication manager 1920 may be configured to perform various operations (e.g., receiving, monitoring, transmitting) using a transceiver 1915, one or more antennas 1925, or any combination thereof, or otherwise cooperating with them. Although the communication manager 1920 is illustrated as a separate component, in some examples, one or more functions described with reference to the communication manager 1920 may be supported or executed by at least one processor 1940, at least one memory 1930, code 1935, or any combination thereof. For example, code 1935 may include instructions that can be executed by at least one processor 1940 to cause the device 1905 to perform various aspects of the beam correspondence between AI model training and inference as described herein, or at least one processor 1940 and at least one memory 1930 may be otherwise configured to perform or support such operations individually or jointly.

[0213] Figure 20A block diagram 2000 of a device 2005 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. Device 2005 may be an example of aspects of network entity 105 as described herein. Device 2005 may include receiver 2010, transmitter 2015, and communication manager 2020. Device 2005, or one or more components of device 2005 (e.g., receiver 2010, transmitter 2015, and communication manager 2020), may include at least one processor that may be coupled to at least one memory to individually or jointly support or implement the described techniques. Each of these components may communicate with each other (e.g., via one or more buses).

[0214] Receiver 2010 may provide components for acquiring (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 passed to other components of device 2005. In some examples, receiver 2010 may support acquiring information by receiving signals via one or more antennas. Additionally or alternatively, receiver 2010 may support acquiring information by receiving signals via one or more wired (e.g., electrical, fiber optic) interfaces, wireless interfaces, or any combination thereof.

[0215] Transmitter 2015 may provide components for outputting (e.g., transmitting, providing, conveying, transmitting) information generated by other components of device 2005. For example, transmitter 2015 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, transmitter 2015 may support outputting information by transmitting signals via one or more antennas. Additionally or alternatively, transmitter 2015 may support outputting information by transmitting signals via one or more wired (e.g., electrical, fiber optic) interfaces, wireless interfaces, or any combination thereof. In some examples, transmitter 2015 and receiver 2010 may be co-located in a transceiver, which may include or be coupled to a modem.

[0216] The communication manager 2020, receiver 2010, transmitter 2015, or various combinations thereof, or various components thereof, may be examples of components used to perform various aspects of the beam correspondence between AI model training and inference as described herein. For example, the communication manager 2020, receiver 2010, transmitter 2015, or various combinations thereof, or components thereof, may be able to perform one or more of the functions described herein.

[0217] In some examples, the communication manager 2020, receiver 2010, transmitter 2015, or various combinations or components thereof may be implemented in hardware (e.g., in communication management circuitry). The hardware may include at least one of a processor, DSP, CPU, ASIC, FPGA, or other programmable logic device, microcontroller, discrete gate or transistor logic unit, discrete hardware component, or any combination thereof, configured as or otherwise individually or collectively to support components for performing the functions described herein. In some examples, at least one processor and at least one memory coupled to said at least one processor may be configured to perform one or more of the functions described herein (e.g., instructions stored in at least one memory are executed individually or collectively by one or more processors).

[0218] Additionally or alternatively, the communication manager 2020, receiver 2010, transmitter 2015, or various combinations or components thereof may be implemented in code (e.g., as communication management software or firmware) executed by at least one processor. If implemented in code executed by at least one processor, the functionality of the communication manager 2020, receiver 2010, transmitter 2015, or various combinations or components thereof may be performed by (e.g., a general-purpose processor, DSP, CPU, ASIC, FPGA, microcontroller, or any combination of these or other programmable logic devices configured, either individually or collectively, as a component for performing the functions described in this disclosure).

[0219] In some examples, the communication manager 2020 may be configured to use the receiver 2010, the transmitter 2015, or both, or otherwise cooperate with them to perform various operations (e.g., receiving, acquiring, monitoring, outputting, transmitting). For example, the communication manager 2020 may receive information from the receiver 2010, transmit information to the transmitter 2015, or integrate with the receiver 2010, the transmitter 2015, or both to acquire information, output information, or perform various other operations as described herein.

[0220] The Communication Manager 2020 may support wireless communication according to examples disclosed herein. For example, the Communication Manager 2020 may be capable of, configured to, or operable to support components for sending a first message to a UE, the first message indicating a first subset of AI model identifiers supported by a network entity from a predefined set of multiple AI model identifiers associated with beam prediction. The Communication Manager 2020 may be capable of, configured to, or operable to support components for receiving a second message from a UE, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. The Communication Manager 2020 may be capable of, configured to, or operable to support components for sending control signaling to a UE, the control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifier and the outputs of the AI ​​model to the beam set. The Communication Manager 2020 is capable of, configured to, or operable to support components for transmitting a set of reference signals via a set of beam measurement resources. The Communication Manager 2020 is also capable of, configured to, or operable to support components for receiving reports from the UE indicating a set of predicted values ​​for the beam set based on an AI model identifier and the set of reference signals.

[0221] By including or configuring a communication manager 2020 according to an example as described herein, device 2005 (e.g., controlling receiver 2010, transmitter 2015, communication manager 2020 or a combination thereof or at least one processor otherwise coupled to them) can support techniques for more efficient use of communication resources.

[0222] Figure 21 A block diagram 2100 of a device 2105 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure is shown. Device 2105 may be an example of aspects of device 2005 or network entity 105 as described herein. Device 2105 may include receiver 2110, transmitter 2115, and communication manager 2120. Device 2105, or one or more components of device 2105 (e.g., receiver 2110, transmitter 2115, and communication manager 2120), may include at least one processor that may be coupled to at least one memory to support the described techniques. Each of these components may communicate with each other (e.g., via one or more buses).

[0223] Receiver 2110 may provide components for acquiring (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 passed to other components of device 2105. In some examples, receiver 2110 may support acquiring information by receiving signals via one or more antennas. Additionally or alternatively, receiver 2110 may support acquiring information by receiving signals via one or more wired (e.g., electrical, fiber optic) interfaces, wireless interfaces, or any combination thereof.

[0224] Transmitter 2115 may provide components for outputting (e.g., transmitting, providing, conveying, transmitting) information generated by other components of device 2105. For example, transmitter 2115 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, transmitter 2115 may support outputting information by transmitting signals via one or more antennas. Additionally or alternatively, transmitter 2115 may support outputting information by transmitting signals via one or more wired (e.g., electrical, fiber optic) interfaces, wireless interfaces, or any combination thereof. In some examples, transmitter 2115 and receiver 2110 may be co-located in a transceiver, which may include or be coupled to a modem.

[0225] Device 2105 or its various components may be examples of parts for performing various aspects of beam correspondence between AI model training and inference as described herein. For example, communication manager 2120 may include a supported AI model manager 2125, a beam prediction scheduling manager 2130, a reference signal transmission manager 2135, a beam prediction report manager 2140, or any combination thereof. Communication manager 2120 may be examples of aspects of communication manager 2020 as described herein. In some examples, communication manager 2120 or its various components may be configured to use or otherwise cooperate with receiver 2110, transmitter 2115, or both to perform various operations (e.g., receiving, acquiring, monitoring, outputting, transmitting). For example, communication manager 2120 may receive information from receiver 2110, transmit information to transmitter 2115, or be integrated in combination with receiver 2110, transmitter 2115, or both to acquire information, output information, or perform various other operations as described herein.

[0226] Communication manager 2120 may support wireless communication according to examples disclosed herein. Supported AI model manager 2125 is capable of, configured to, or operable to support components for sending a first message to the UE, the first message indicating a first subset of AI model identifiers supported by a network entity from a predefined set of multiple AI model identifiers associated with beam prediction. Supported AI model manager 2125 is capable of, configured to, or operable to support components for receiving a second message from the UE, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. Beam prediction scheduling manager 2130 is capable of, configured to, or operable to support components for sending control signaling to the UE, the control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifier and the outputs of the AI ​​model to the beam set. Reference signal transmission manager 2135 is capable of, configured to, or operable to support components for transmitting a set of reference signals via a set of beam measurement resources. Beam prediction report manager 2140 is capable of, configured to, or operable to support components for receiving a report from the UE indicating a set of predicted values ​​for the beam set based on an AI model identifier and the set of reference signals.

[0227] Figure 22 A block diagram 2200 is shown of a communication manager 2220 supporting beam correspondence between AI model training and inference according to one or more aspects of this disclosure. The communication manager 2220 may be an example of aspects of the communication manager 2020, communication manager 2120, or both as described herein. The communication manager 2220 or its various components may be examples of components for performing various aspects of the beam correspondence between AI model training and inference as described herein. For example, the communication manager 2220 may include a supported AI model manager 2225, a beam prediction scheduling manager 2230, a reference signal transmission manager 2235, a beam prediction report manager 2240, an AI model training manager 2245, an AI model validation dataset manager 2250, an AI model validation manager 2255, or any combination thereof. These components, or each of their components or sub-components (e.g., one or more processors, one or more memories), may communicate directly or indirectly with each other (e.g., via one or more buses), and such communication may include communication within protocol layers of the protocol stack, communication associated with logical channels of the protocol stack (e.g., between protocol layers of the protocol stack, within devices, components, or virtualization components associated with network entity 105, between devices, components, or virtualization components associated with network entity 105), or any combination thereof.

[0228] Communication manager 2220 may support wireless communication according to examples disclosed herein. Supported AI model manager 2225 is capable of, configured to, or operable to support components for sending a first message to the UE, the first message indicating a first subset of AI model identifiers supported by a network entity from a predefined set of multiple AI model identifiers associated with beam prediction. In some examples, supported AI model manager 2225 is capable of, configured to, or operable to support components for receiving a second message from the UE, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. Beam prediction scheduling manager 2230 is capable of, configured to, or operable to support components for sending control signaling to the UE, the control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifier and the outputs of the AI ​​model to the beam set. Reference signal transmission manager 2235 is capable of, configured to, or operable to support components for transmitting a set of reference signals via a set of beam measurement resources. Beam prediction report manager 2240 is capable of, configured to, or operable to support components for receiving a report from the UE indicating a set of predicted values ​​for the beam set based on an AI model identifier and the set of reference signals.

[0229] In some examples, the first message is one of system information, RRC message, MAC-CE, or DCI.

[0230] In some examples, the second message is one of an RRC message, a MAC-CE message, or a UCI message.

[0231] In some examples, to indicate the beam measurement resource set, control signaling indicates the CSI-RS index set or the SSB index set. In some examples, the AI ​​model identifier indicates the mapping of the CSI-RS index set or the SSB index set to the input of the AI ​​model associated with the AI ​​model identifier.

[0232] In some examples, the AI ​​model identifier indicates the corresponding beamwidth and corresponding beam direction for each beam in the beam measurement resource set and for each beam in the beam set.

[0233] In some examples, the AI ​​model identifier indicates the QCL relationship between beams in the beam measurement resource set and beams in the beam set.

[0234] In some examples, the AI ​​model identifier indicates the relative timing between the set of beam measurement resources and the set of predicted values.

[0235] In some examples, based on the set of predicted values, the report also indicates the pre-decoding matrix indicator, channel quality indicator, channel impulse response, rank indicator, angle of arrival, downlink departure angle, or departure direction.

[0236] In some examples, control signaling also schedules CSI reports. In some examples, the reports are included within the CSI reports.

[0237] In some examples, the AI ​​model training manager 2245 is capable of, configured to, or operable to support components for performing a training process of an AI model with the UE using a set of training beam measurement resources to predict a set of training measurement results for the training beam set, wherein the set of beam measurement resources corresponds to a set of training beam measurement resources, and wherein the beam set corresponds to a set of training beams.

[0238] In some examples, the training beam measurement resource set includes a first CSI-RS set or a first SSB set with a first index value. In some examples, the beam measurement resource set includes a second CSI-RS set or a second SSB set with a first index value. In some examples, the same mapping from the first index value to the input value of the AI ​​model is used to predict the training measurement result set and generate a predicted value set.

[0239] In some examples, the AI ​​model validation dataset manager 2250 is capable of, configured to, or operable to support components for transmitting to the UE a validation beam measurement resource set and a corresponding reference truth prediction target for a second beam set used for the AI ​​model. In some examples, the reference signal transmission manager 2235 is capable of, configured to, or operable to support components for transmitting a second reference signal set via the validation beam measurement resource set. In some examples, the AI ​​model validation manager 2255 is capable of, configured to, or operable to support components for receiving a second report from the UE indicating that a second set of predicted values ​​for a second beam set generated at the UE using the AI ​​model based on the reference signal set satisfies an accuracy metric regarding the corresponding reference truth prediction target, wherein the transmission of control signaling is based on the reception of the second report.

[0240] Figure 23A diagram of a system 2300 including a device 2305 supporting beam correspondence between AI model training and inference, according to one or more aspects of this disclosure, is shown. Device 2305 may be an example of device 2005, device 2105, or network entity 105 as described herein, or a component including such devices. Device 2305 may communicate with one or more network entities 105, one or more UEs 115, or any combination thereof, and this communication may include communication via one or more wired interfaces, one or more wireless interfaces, or any combination thereof. Device 2305 may include components supporting output and acquisition of communication, such as a communication manager 2320, a transceiver 2310, an antenna 2315, at least one memory 2325, code 2330, and at least one processor 2335. These components may communicate electronically or otherwise (e.g., operative ground, communication ground, functional ground, electronic ground, electrical ground) via one or more buses (e.g., bus 2340).

[0241] Transceiver 2310 may support bidirectional communication via a wired link, a wireless link, or both, as described herein. In some examples, transceiver 2310 may include a wired transceiver and may communicate bidirectionally with another wired transceiver. Additionally or alternatively, in some examples, transceiver 2310 may include a wireless transceiver and may communicate bidirectionally with another wireless transceiver. In some examples, device 2305 may include one or more antennas 2315 that may be capable of transmitting or receiving wireless transmissions (e.g., concurrently). Transceiver 2310 may also include a modem for: modulating a signal; providing the modulated signal for transmission (e.g., via one or more antennas 2315, via a wired transmitter); receiving the modulated signal (e.g., from one or more antennas 2315, from a wired receiver); and demodulating the signal. In some embodiments, transceiver 2310 may include one or more interfaces, such as one or more interfaces coupled to one or more antennas 2315 configured to support various receive or acquire operations, or one or more interfaces coupled to one or more antennas 2315 configured to support various transmit or output operations, or combinations thereof. In some embodiments, transceiver 2310 may include one or more processors or one or more memory components, or be configured to couple to such processors or memory components, which are operable to perform or support operations based on received or acquired information or signals, or generate information or other signals for transmission or other output, or any combination thereof. In some embodiments, transceiver 2310, or transceiver 2310 and one or more antennas 2315, or transceiver 2310 and one or more antennas 2315 and one or more processors or one or more memory components (e.g., at least one processor 2335, at least one memory 2325, or both), may be included in a chip or chip assembly mounted in device 2305. In some examples, transceiver 2310 may be able to operate to support communication via one or more communication links (e.g., communication link 125, backhaul communication link 120, midhaul communication link 162, and fronthaul communication link 168).

[0242] At least one memory 2325 may include RAM, ROM, or any combination thereof. At least one memory 2325 may store computer-readable, computer-executable code 2330 including instructions that, when executed by one or more processors of at least one processor 2335, cause device 2305 to perform the various functions described herein. Code 2330 may be stored in a non-transitory computer-readable medium such as system memory or another type of memory. In some cases, code 2330 may not be directly executable by one of the at least one processor 2335, but may enable a computer (e.g., when compiled and executed) to perform the functions described herein. In some cases, at least one memory 2325 may contain a BIOS, etc., which controls basic hardware or software operations, such as interaction with peripheral components or devices. In some examples, at least one processor 2335 may include multiple processors, and at least one memory 2325 may include multiple memories. One or more of the multiple processors may be coupled to one or more of the multiple memories, which may be configured individually or collectively to perform the various functions described herein (e.g., as part of a processing system).

[0243] At least one processor 2335 may include intelligent hardware devices (e.g., general-purpose processors, DSPs, ASICs, CPUs, FPGAs, microcontrollers, programmable logic devices, discrete gate or transistor logic components, discrete hardware components, or any combination thereof). In some cases, at least one processor 2335 may be configured to operate a memory array using a memory controller. In some other cases, the memory controller may be integrated into one or more processors in at least one processor 2335. At least one processor 2335 may be configured to execute computer-readable instructions stored in memory (e.g., one or more memories in at least one memory 2325) to cause device 2305 to perform various functions (e.g., functions or tasks supporting beam correspondence between AI model training and inference). For example, device 2305 or components of device 2305 may include at least one processor 2335 and at least one memory 2325 coupled to one or more processors in at least one processor 2335, wherein at least one processor 2335 and at least one memory 2325 are configured to perform the various functions described herein. At least one processor 2335 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 can host functions (e.g., by executing code 2330) to perform the functions of device 2305. At least one processor 2335 may be any one or more suitable processors capable of executing scripts or instructions of one or more software programs stored in device 2305 (such as within one or more memories of at least one memory 2325). In some examples, at least one processor 2335 may include multiple processors, and at least one memory 2325 may include multiple memories. One or more of the multiple processors may be coupled to one or more of the multiple memories, which may be configured individually or collectively to perform the various functions described herein. In some examples, at least one processor 2335 may be a component of a processing system, which may refer to a system of machines (such as a series of machines), circuits (including, for example, one or both of processor circuitry (which may include at least one processor 2335) and memory circuitry (which may include at least one memory 2325)) or components that receive or acquire input and process the input to produce, generate, or acquire a set of outputs. The processing system can be configured to perform one or more of the functions described herein. Therefore, at least one processor 2335 or a processing system including at least one processor 2335 can be configured, configured to, or operated to cause the device 2305 to perform one or more of the functions described herein.Furthermore, as described herein, “configured to,” “capable of being configured to,” and “capable of operating to” are used interchangeably and may be associated with the ability to perform one or more of the functions described herein when executing code stored in at least one memory 2325 or otherwise.

[0244] In some examples, bus 2340 may support communication at protocol layers (e.g., within a protocol layer) in a protocol stack. In some examples, bus 2340 may support communication associated with logical channels of a protocol stack (e.g., between protocol layers of the protocol stack), which may include communication performed within components of device 2305, or communication performed between different components of device 2305 that are co-addressable or may be located in different locations (e.g., where device 2305 may refer to a system in which one or more of communication manager 2320, transceiver 2310, at least one memory 2325, code 2330 and at least one processor 2335 may be located in one component of different components or partitioned between different components).

[0245] In some examples, the communication manager 2320 can manage (e.g., via one or more wired or wireless backhaul links) various aspects of communication with the core network 130. For example, the communication manager 2320 can manage the transfer of data communications by client devices such as one or more UEs 115. In some examples, the communication manager 2320 can manage communication with other network entities 105 and may include a controller or scheduler for cooperatively controlling communication with UE 115 with other network entities 105. In some examples, the communication manager 2320 may support the X2 interface within LTE / LTE-A wireless communication network technology to provide communication between network entities 105.

[0246] The communication manager 2320 may support wireless communication according to examples disclosed herein. For example, the communication manager 2320 may be capable of, configured to, or operable to support components for sending a first message to a UE, the first message indicating a first subset of AI model identifiers supported by a network entity from a predefined set of multiple AI model identifiers associated with beam prediction. The communication manager 2320 may be capable of, configured to, or operable to support components for receiving a second message from a UE, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. The communication manager 2320 may be capable of, configured to, or operable to support components for sending control signaling to a UE, the control signaling indicating AI model identifiers included in both the first and second subsets, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the inputs of the AI ​​model associated with the AI ​​model identifier and the outputs of the AI ​​model to the beam set. The communication manager 2320 is capable of, configured to, or operable to support components for transmitting a set of reference signals via a set of beam measurement resources. The communication manager 2320 is capable of, configured to, or operable to support components for receiving a report from the UE indicating a set of predicted values ​​for the beam set based on an AI model identifier and the set of reference signals.

[0247] By including or configuring a communication manager 2320 according to an example as described herein, device 2305 can support techniques for improving communication reliability, more effectively utilizing communication resources, improving coordination between devices, and improving the utilization of processing power.

[0248] In some examples, the communication manager 2320 may be configured to use or otherwise cooperate with transceiver 2310, one or more antennas 2315 (e.g., where applicable) or any combination thereof to perform various operations (e.g., receiving, acquiring, monitoring, outputting, transmitting). Although the communication manager 2320 is illustrated as a separate component, in some examples, one or more functions described with reference to the communication manager 2320 may be supported or performed by transceiver 2310, one or more processors in at least one processor 2335, one or more memories in at least one memory 2325, code 2330, or any combination thereof (e.g., by a processing system including at least a portion of at least one processor 2335, at least one memory 2325, code 2330, or any combination thereof). For example, code 2330 may include instructions that can be executed by one or more processors of at least one processor 2335 to cause device 2305 to perform various aspects of the beam correspondence between AI model training and inference as described herein, or at least one processor 2335 and at least one memory 2325 may be otherwise configured to perform or support such operations individually or jointly.

[0249] Figure 24 A flowchart illustrating a method 2400 for supporting beam correspondence between AI model training and inference according to various aspects of this disclosure is shown. The operation of method 2400 can be implemented by a UE or its components as described herein. For example, the operation of method 2400 can be implemented by, as referenced... Figures 1 to 19 The UE 115 described herein is used to perform this function. In some examples, the UE can execute a set of instructions to control the functional elements of the UE to perform the described function. Additionally or alternatively, the UE may use dedicated hardware to perform aspects of the described function.

[0250] At 2405, the method may include receiving a first message from a network entity indicating a first subset of AI model identifiers supported by the network entity from a predefined set of multiple AI model identifiers associated with beam prediction. Operation of box 2405 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2405 may be derived from references... Figure 18 The described supported AI model manager 1825 is used to execute this.

[0251] At 2410, the method may include sending a second message to a network entity, the second message indicating a second subset of AI model identifiers supported by the UE from a predefined set of multiple AI model identifiers. The operation of block 2410 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2410 may be provided by reference to... Figure 18The described supported AI model manager 1825 is used to execute this.

[0252] At 2415, the method may include receiving control signaling indicating an AI model identifier included in both a first subset and a second subset, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the input of the AI ​​model associated with the AI ​​model identifier and the output of the AI ​​model to the beam set. The operation of block 2415 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2415 may be provided by reference to [reference needed]. Figure 18 The beam prediction scheduling manager 1830 described is used to perform this.

[0253] At 2420, the method may include receiving a set of reference signals via a beam measurement resource set. The operation of block 2420 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2420 may be provided by reference to... Figure 18 The reference signal receiver manager 1835 described herein shall be used to perform this action.

[0254] At 2425, the method may include sending a report indicating a set of predicted values ​​for the beam set generated using an AI model and based on a set of reference signals. The operation of box 2425 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2425 may be provided by reference... Figure 18 The described beam prediction report manager 1840 is used to perform this.

[0255] Figure 25 A flowchart illustrating a method 2500 for supporting beam correspondence between AI model training and inference according to various aspects of this disclosure is shown. The operation of method 2500 can be implemented by a network entity or its components as described herein. For example, the operation of method 2500 can be implemented by, as referenced... Figures 1 to 15 as well as Figures 20 to 23 The network entity described herein performs the function. In some examples, the network entity may execute a set of instructions to control the functional elements of the network entity to perform the described function. Additionally or alternatively, the network entity may use dedicated hardware to perform aspects of the described function.

[0256] At 2505, the method may include sending a first message to the UE indicating a first subset of AI model identifiers supported by a network entity from a predefined set of multiple AI model identifiers associated with beam prediction. Operation of box 2505 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2505 may be provided by reference to... Figure 22 The described supported AI model manager 2225 is used to execute this.

[0257] At 2510, the method may include receiving a second message from the UE indicating a second subset of AI model identifiers supported by the UE from a predefined set of a plurality of AI model identifiers. Operation of box 2510 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2510 may be provided by reference to [reference needed]. Figure 22 The described supported AI model manager 2225 is used to execute this.

[0258] At 2515, the method may include sending control signaling to the UE, the control signaling indicating an AI model identifier included in both a first subset and a second subset, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the input of the AI ​​model associated with the AI ​​model identifier and the output of the AI ​​model to the beam set. The operation of block 2515 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2515 may be provided by reference to [reference needed]. Figure 22 The beam prediction scheduling manager 2230 described herein is used to perform this.

[0259] At 2520, the method may include transmitting a set of reference signals via a beam measurement resource set. The operation of block 2520 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2520 may be provided by reference to... Figure 22 The reference signal sending manager 2235 described herein shall be executed.

[0260] At 2525, the method may include receiving a report from the UE indicating a set of predicted values ​​for a beam set based on an AI model identifier and a set of reference signals. Operation of block 2525 may be performed according to examples as disclosed herein. In some examples, aspects of the operation of 2525 may be determined by reference to... Figure 22 The beam prediction report manager 2240 described herein is used to perform this.

[0261] The following provides an overview of the various aspects of this disclosure.

[0262] Aspect 1: A method for wireless communication at a UE, the method comprising: receiving a first message from a network entity, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of AI model identifiers associated with beam prediction; sending a second message to the network entity, the second message indicating a second subset of AI model identifiers supported by the UE from the predefined set of AI model identifiers; receiving control signaling indicating AI model identifiers included in both the first subset and the second subset, wherein the control signaling further indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to inputs of an AI model associated with the AI ​​model identifiers and outputs of the AI ​​model to the beam set; receiving a set of reference signals via the beam measurement resource set; and sending a report indicating a set of predicted values ​​for the beam set generated using the AI ​​model and at least in part based on the set of reference signals.

[0263] Aspect 2: According to the method of aspect 1, the first message is one of system information, RRC message, MAC-CE or DCI.

[0264] Aspect 3: The method according to any one of Aspects 1 to 2, wherein the second message is one of RRC message, MAC-CE or UCI.

[0265] Aspect 4: The method according to any one of Aspects 1 to 3, wherein, in order to indicate the beam measurement resource set, the control signaling indicates a CSI-RS index set or an SSB index set, and the AI ​​model identifier indicates a mapping of the CSI-RS index set or the SSB index set to the input of the AI ​​model.

[0266] Aspect 5: The method according to any one of Aspects 1 to 4, wherein the AI ​​model identifier indicates each beam in the beam measurement resource set and the corresponding beamwidth and corresponding beam direction of each beam in the beam set.

[0267] Aspect 6: The method according to any one of Aspects 1 to 5, wherein the AI ​​model identifier indicates the QCL relationship between beams in the beam measurement resource set and beams in the beam set.

[0268] Aspect 7: The method according to any one of Aspects 1 to 6, wherein the AI ​​model identifier indicates the relative timing between the beam measurement resource set and the predicted value set.

[0269] Aspect 8: The method according to any one of Aspects 1 to 7, wherein the report is based at least in part on the set of predicted values, and the report also indicates a pre-decoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink departure angle, or a departure direction.

[0270] Aspect 9: The method according to any one of Aspects 1 to 8, wherein the control signaling further schedules a CSI report, and the report is included within the CSI report.

[0271] Aspect 10: The method according to any one of Aspects 1 to 9, the method further comprising: using a training beam measurement resource set together with the network entity to perform a training process of the AI ​​model to predict a training measurement result set of the training beam set, wherein the beam measurement resource set corresponds to the training beam measurement resource set, and wherein the beam set corresponds to the training beam set.

[0272] Aspect 11: According to the method of aspect 10, wherein the training beam measurement resource set includes a first CSI-RS set or a first SSB set having a first index value, the beam measurement resource set includes a second CSI-RS set or a second SSB set having the first index value, and the same mapping from the first index value to the input value of the AI ​​model is used to predict the training measurement result set and generate the predicted value set.

[0273] Aspect 12: The method according to any one of Aspects 1 to 11, the method further comprising: receiving from the network entity a set of verification beam measurement resources for a second beam set for the AI ​​model and a corresponding reference truth prediction target; receiving a second set of reference signals via the set of verification beam measurement resources; and sending a second report indicating that a second set of predictions for the second beam set generated using the AI ​​model, based at least in part on the set of reference signals, satisfies an accuracy metric with respect to the corresponding reference truth prediction target, wherein the reception of control signaling is based at least in part on the transmission of the second report.

[0274] Aspect 13: A method for wireless communication at a network entity, the method comprising: sending a first message to a UE, the first message indicating a first subset of AI model identifiers supported by the network entity from a predefined set of a plurality of AI model identifiers associated with beam prediction; receiving a second message from the UE, the second message indicating a second subset of AI model identifiers supported by the UE from the predefined set of a plurality of AI model identifiers; sending control signaling to the UE, the control signaling indicating AI model identifiers included in both the first subset and the second subset, wherein the control signaling further indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to inputs of an AI model associated with the AI ​​model identifiers and outputs of the AI ​​model to the beam set; sending a set of reference signals via the beam measurement resource set; and receiving a report from the UE, the report indicating a set of predicted values ​​for the beam set based at least in part on the AI ​​model identifiers and at least in part on the set of reference signals.

[0275] Aspect 14: The method according to aspect 13, wherein the first message is one of system information, RRC message, MAC-CE or DCI.

[0276] Aspect 15: The method according to any one of Aspects 13 to 14, wherein the second message is one of an RRC message, a MAC-CE message, or a UCI message.

[0277] Aspect 16: The method according to any one of Aspects 13 to 15, wherein, in order to indicate the beam measurement resource set, the control signaling indicates a CSI-RS index set or an SSB index set, and the AI ​​model identifier indicates a mapping of the CSI-RS index set or the SSB index set to the input of the AI ​​model associated with the AI ​​model identifier.

[0278] Aspect 17: The method according to any one of Aspects 13 to 16, wherein the AI ​​model identifier indicates each beam in the beam measurement resource set and the corresponding beamwidth and corresponding beam direction of each beam in the beam set.

[0279] Aspect 18: The method according to any one of Aspects 13 to 17, wherein the AI ​​model identifier indicates the QCL relationship between beams in the beam measurement resource set and beams in the beam set.

[0280] Aspect 19: The method according to any one of Aspects 13 to 18, wherein the AI ​​model identifier indicates the relative timing between the beam measurement resource set and the predicted value set.

[0281] Aspect 20: The method according to any one of Aspects 13 to 19, wherein the report is based at least in part on the set of predicted values, and the report also indicates a pre-decoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink departure angle, or a departure direction.

[0282] Aspect 21: The method according to any one of Aspects 13 to 20, wherein the control signaling further schedules a CSI report, and the report is included within the CSI report.

[0283] Aspect 22: The method according to any one of Aspects 13 to 21, the method further comprising: using a training beam measurement resource set together with the UE to perform a training process of the AI ​​model to predict a training measurement result set of the training beam set, wherein the beam measurement resource set corresponds to the training beam measurement resource set, and wherein the beam set corresponds to the training beam set.

[0284] Aspect 23: According to the method of aspect 22, wherein the training beam measurement resource set includes a first CSI-RS set or a first SSB set having a first index value, the beam measurement resource set includes a second CSI-RS set or a second SSB set having the first index value, and the same mapping from the first index value to the input value of the AI ​​model is used to predict the training measurement result set and generate the predicted value set.

[0285] Aspect 24: The method according to any one of Aspects 13 to 23, the method further comprising: sending to the UE a set of verification beam measurement resources for a second beam set of the AI ​​model and a corresponding reference truth prediction target; sending a second set of reference signals via the set of verification beam measurement resources; and receiving from the UE a second report indicating that a second set of prediction values ​​for the second beam set generated at the UE using the AI ​​model based at least in part on the set of reference signals satisfies an accuracy metric with respect to the corresponding reference truth prediction target, wherein the transmission of control signaling is based at least in part on the reception of the second report.

[0286] Aspect 25: A UE for wireless communication, the UE comprising: one or more memories storing processor-executable code; and one or more processors coupled to the one or more memories and capable of operating individually or jointly to execute the code to cause the UE to perform a method according to any one of Aspects 1 to 12.

[0287] Aspect 26: A UE for wireless communication, the UE comprising at least one component for performing a method according to any one of aspects 1 to 12.

[0288] Aspect 27: A non-transitory computer-readable medium storing code for wireless communication, said code including instructions executable by a processor to perform the method according to any one of aspects 1 to 12.

[0289] Aspect 28: A network entity for wireless communication, the network entity comprising: one or more memories storing processor-executable code; and one or more processors coupled to the one or more memories and capable of operating individually or jointly to execute the code to cause the network entity to perform a method according to any one of Aspects 13 to 24.

[0290] Aspect 29: A network entity for wireless communication, the network entity comprising at least one component for performing the method according to any one of aspects 13 to 24.

[0291] Aspect 30: A non-transitory computer-readable medium storing code for wireless communication, said code including instructions executable by a processor to perform a method according to any one of aspects 13 to 24.

[0292] It should be noted that the methods described herein describe possible specific implementations, and the operations and steps can be rearranged or otherwise modified, and other specific implementations are also possible. Furthermore, aspects from two or more of these methods can be combined.

[0293] While aspects of LTE, LTE-A, LTE-A Pro, or NR systems may be described for illustrative purposes, and the terms LTE, LTE-A, LTE-A Pro, or NR may be used in most of the description, the techniques described herein are also applicable to networks outside of LTE, LTE-A, LTE-A Pro, or NR networks. For example, the techniques described are applicable to a variety of other wireless communication systems, such as Ultra Mobile Broadband (UMB), IEEE 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802.20, Flash-OFDM, and other systems and radio technologies not explicitly mentioned herein.

[0294] The information and signals described herein can be represented using any of a variety of different techniques and skills. For example, data, instructions, commands, information, signals, bits, symbols, and chips mentioned throughout the description can be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or optical particles, or any combination thereof.

[0295] The various exemplary blocks and components described 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 unit, discrete hardware component, or any combination thereof, designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but in alternative embodiments, a processor may be any processor, controller, microcontroller, or state machine. A 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 cooperating with a DSP core, or any other such configuration). Any function or operation described herein that can be performed by a processor may be performed by multiple processors capable of performing the described functions or operations individually or jointly.

[0296] The functionality 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 functionality 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 this disclosure and the appended claims. For example, due to the nature of software, the functionality described herein can be implemented using software executed by a processor, hardware, firmware, hardwiring, or any combination of these. Features implementing the functionality can also be physically located in various locations, including portions distributed such that the functionality is implemented at different physical locations.

[0297] Computer-readable media includes both non-transitory computer storage media and communication media, encompassing any medium that facilitates the transfer of a computer program from one location to another. Non-transitory storage media can be any available medium accessible by a general-purpose or special-purpose computer. By way of example, and not limitation, non-transitory computer-readable media may include RAM, ROM, electrically erasable programmable ROM (EEPROM), flash memory, compressed optical disc (CD) ROM or other optical disc storage devices, magnetic disk storage devices 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 accessible by a general-purpose or special-purpose computer or a general-purpose or special-purpose processor. Furthermore, any connection is appropriately referred to as computer-readable media. 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 coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included within the definition of computer-readable media. As used herein, disks and optical discs include CDs, laser discs, optical discs, digital multifunction discs (DVDs), floppy disks, and Blu-ray discs. Disks can magnetically reproduce data, and optical discs can optically reproduce data using lasers. Combinations of the above are also included within the scope of computer-readable media. Any function or operation described herein that can be performed by memory can be performed by multiple memories capable of performing the described function or operation individually or jointly.

[0298] As used herein, the word "or" 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") in the claims indicates an inclusive list, such that a list of at least one of, for example, 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). Furthermore, 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" could be based on both condition A and condition B without departing from the scope of this 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".

[0299] As used herein, including in claims, the article “a” preceding a noun is open-ended and is understood to refer to “at least one” or “one or more” of those nouns. Therefore, the terms “a,” “at least one,” “one or more,” and “at least one of one or more” are interchangeable. For example, where a claim enumerates “components” performing one or more functions, each of the individual functions may be performed by a single component or by any combination of multiple components. Thus, the term “component” having a characteristic or performing a function may refer to “at least one of one or more components” having a particular characteristic or performing a particular function. Subsequent references to a component introduced with the article “a” using the terms “the” or “the” can refer to any or all of the one or more components. For example, a component introduced with the article “a” can be understood to mean “one or more components,” and subsequent reference to “the component” in a claim can be understood as equivalent to referring to “at least one of the one or more components.” Similarly, subsequent references to a component introduced with the terms “the” or “the” as “one or more components” can refer to any or all of the one or more components. For example, reference to "the one or more components" in the subsequent claims can be understood as equivalent to reference to "at least one of the one or more components".

[0300] The term "determine" encompasses a variety of actions, and therefore, "determine" can include calculation, computation, processing, derivation, investigation, lookup (such as by searching in a table, database, or other data structure), identification, and similar actions. Furthermore, "determine" can include receiving (e.g., receiving information), accessing (e.g., accessing data stored in memory), etc. Moreover, "determine" can include parsing, acquiring, selecting, choosing, creating, and other similar actions.

[0301] In the accompanying drawings, similar components or features may have the same reference numerals. Furthermore, various components of the same type can be distinguished by adding a dash after the reference numeral and a second reference numeral to differentiate between similar components. If only the first reference numeral is used in the description, the description can be applied to any of the similar components having the same first reference numeral, regardless of the second reference numeral or other subsequent reference numerals.

[0302] The description herein, illustrated with reference to the accompanying drawings, describes an example configuration and does not represent all achievable examples or those within the scope of the claims. The term "example" as used herein means "serving as an example, instance, or illustration," not "preferred" or "advantageous over other examples." The detailed description includes specific details used to provide an understanding of the described techniques. However, these techniques can be practiced without these specific details. In some cases, known structures and devices are shown in block diagram form to avoid obscuring the concept of the described examples.

[0303] The description herein is provided to enable those skilled in the art to implement or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the scope of this disclosure. Therefore, this disclosure is not limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein. Claims (as amended under Article 19 of the Treaty) 1. A user equipment (UE), the user equipment (UE) comprising: One or more memories, wherein the one or more memories store processor-executable code; and One or more processors, coupled to one or more memories and capable of operating individually or jointly to execute the code to enable the UE: Receive a first message from a network entity, the first message indicating a first subset of artificial intelligence model identifiers supported by the network entity from a predefined set of multiple artificial intelligence model identifiers associated with beam prediction; Send a second message to the network entity, the second message indicating a second subset of artificial intelligence model identifiers supported by the UE from the predefined set of multiple artificial intelligence model identifiers; Receive control signaling, the control signaling indicating an artificial intelligence model identifier included in both the first subset and the second subset. The control signaling further indicates a beam measurement resource set, and the control signaling indicates a mapping of the beam measurement resource set to the input of an artificial intelligence model associated with the artificial intelligence model identifier and the output of the artificial intelligence model to the beam set; Receive a set of reference signals via the beam measurement resource set; and A report is sent indicating a set of predicted values ​​for the beam set generated using the artificial intelligence model and at least in part based on the set of reference signals. 2. The UE according to claim 1, wherein the first message is one of system information, radio resource control message, media access control (MAC) control element, or downlink control information. 3. The UE according to claim 1, wherein the second message is one of a radio resource control message, a media access control (MAC) control element, or uplink control information. 4. The UE according to claim 1, wherein: To indicate the beam measurement resource set, the control signaling indicates the channel state information reference signal index set or the synchronization signal block index set, and The artificial intelligence model identifier indicates the mapping of the channel state information reference signal index set or the synchronization signal block index set to the input of the artificial intelligence model. 5. The UE of claim 1, wherein the artificial intelligence model identifier indicates the corresponding beamwidth and corresponding beam direction of each beam in the beam measurement resource set and each beam in the beam set. 6. The UE of claim 1, wherein the artificial intelligence model identifier indicates a quasi-co-location relationship between beams in the beam measurement resource set and beams in the beam set. 7. The UE of claim 1, wherein the artificial intelligence model identifier indicates the relative timing between the beam measurement resource set and the predicted value set. 8. The UE of claim 1, wherein the report further indicates a pre-decoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink departure angle, or a departure direction, based at least in part on the set of predicted values. 9. The UE according to claim 1, wherein: The control signaling also schedules channel state information reports, and The report is included in the channel status information report. 10. The UE of claim 1, wherein the one or more processors are individually or jointly capable of further operating to execute the code to cause the UE to: The training process of the artificial intelligence model is performed using a set of training beam measurement resources together with the network entity to predict a set of training measurement results for the training beam set, wherein the set of beam measurement resources corresponds to the training beam set, and wherein the beam set corresponds to the training beam set. 11. The UE according to claim 10, wherein: The training beam measurement resource set includes a first channel state information reference signal set or a first synchronization signal block set with a first index value. The beam measurement resource set includes a second channel state information reference signal set or a second synchronization signal block set having the first index value, and The same mapping from the first index value to the input value of the artificial intelligence model is used to predict the set of training measurement results and generate the set of predicted values. 12. The UE of claim 1, wherein the one or more processors are individually or jointly capable of further operating to execute the code to cause the UE to: Receive from the network entity a set of verification beam measurement resources for the second beam set used in the artificial intelligence model and a corresponding benchmark truth prediction target; Receive a second reference signal set via the verification beam measurement resource set; and A second report is sent, indicating that a second set of predicted values ​​for the second beam set, generated using the artificial intelligence model based at least in part on the reference signal set, satisfies an accuracy metric regarding the corresponding benchmark true value prediction target, wherein the reception of control signaling is at least in part based on the transmission of the second report. 13. A network entity, the network entity comprising: One or more memories, wherein the one or more memories store processor-executable code; and One or more processors, coupled to one or more memories and capable of operating individually or jointly to execute the code to enable the network entity: Send a first message to the user equipment (UE), the first message indicating a first subset of artificial intelligence model identifiers supported by the network entity from a predefined set of multiple artificial intelligence model identifiers associated with beam prediction; The UE receives a second message indicating a second subset of artificial intelligence model identifiers supported by the UE from the predefined set of multiple artificial intelligence model identifiers; Send control signaling to the UE, the control signaling indicating an artificial intelligence model identifier included in both the first subset and the second subset, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the input of the artificial intelligence model associated with the artificial intelligence model identifier and the output of the artificial intelligence model to the beam set; Transmit a set of reference signals via the set of beam measurement resources; and The UE receives a report indicating a set of predicted values ​​for the beam set that are at least partially based on the artificial intelligence model identifier and at least partially based on the reference signal set. 14. The network entity of claim 13, wherein the first message is one of system information, radio resource control message, media access control (MAC) control element, or downlink control information. 15. The network entity of claim 13, wherein the second message is one of a radio resource control message, a media access control (MAC) control element, or uplink control information. 16. The network entity according to claim 13, wherein: To indicate the beam measurement resource set, the control signaling indicates the channel state information reference signal index set or the synchronization signal block index set, and The artificial intelligence model identifier indicates the mapping of the channel state information reference signal index set or the synchronization signal block index set to the input of the artificial intelligence model associated with the artificial intelligence model identifier. 17. The network entity of claim 13, wherein the artificial intelligence model identifier indicates the respective beamwidth and respective beam direction of each beam in the beam measurement resource set and each beam in the beam set. 18. The network entity of claim 13, wherein the artificial intelligence model identifier indicates a quasi-co-location relationship between beams in the beam measurement resource set and beams in the beam set. 19. The network entity of claim 13, wherein the artificial intelligence model identifier indicates the relative timing between the beam measurement resource set and the predicted value set. 20. The network entity of claim 13, wherein the report further indicates a pre-decoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink departure angle, or a departure direction, based at least in part on the set of predicted values. 21. The network entity according to claim 13, wherein: The control signaling also schedules channel state information reports, and The report is included in the channel status information report. 22. The network entity of claim 13, wherein the one or more processors are individually or jointly further operable to execute the code to cause the network entity to: The training process of the artificial intelligence model is performed together with the UE using a training beam measurement resource set to predict a training measurement result set of the training beam set, wherein the beam measurement resource set corresponds to the training beam measurement resource set, and wherein the beam set corresponds to the training beam set. 23. The network entity according to claim 22, wherein: The training beam measurement resource set includes a first channel state information reference signal set or a first synchronization signal block set with a first index value. The beam measurement resource set includes a second channel state information reference signal set or a second synchronization signal block set having the first index value, and The same mapping from the first index value to the input value of the artificial intelligence model is used to predict the set of training measurement results and generate the set of predicted values. 24. The network entity of claim 13, wherein the one or more processors are individually or jointly further operable to execute the code to cause the network entity to: Send the verification beam measurement resource set and the corresponding reference true value prediction target of the second beam set for the artificial intelligence model to the UE; Transmit a second set of reference signals via the set of verified beam measurement resources; and A second report is received from the UE, the second report indicating that a second set of predicted values ​​for the second beam set generated at the UE using the artificial intelligence model based at least in part on the reference signal set satisfies an accuracy metric with respect to the corresponding benchmark truth prediction target, wherein the transmission of control signaling is based at least in part on the reception of the second report. 25. A method for conducting wireless communication at a user equipment (UE), the method comprising: Receive a first message from a network entity, the first message indicating a first subset of artificial intelligence model identifiers supported by the network entity from a predefined set of multiple artificial intelligence model identifiers associated with beam prediction; Send a second message to the network entity, the second message indicating a second subset of artificial intelligence model identifiers supported by the UE from the predefined set of multiple artificial intelligence model identifiers; Receive control signaling, the control signaling indicating an artificial intelligence model identifier included in both the first subset and the second subset, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the input of the artificial intelligence model associated with the artificial intelligence model identifier and the output of the artificial intelligence model to the beam set; Receive a set of reference signals via the beam measurement resource set; and A report is sent indicating a set of predicted values ​​for the beam set generated using the artificial intelligence model and at least in part based on the set of reference signals. 26. The method of claim 25, wherein: To indicate the beam measurement resource set, the control signaling indicates the channel state information reference signal index set or the synchronization signal block index set, and The artificial intelligence model identifier indicates the mapping of the channel state information reference signal index set or the synchronization signal block index set to the input of the artificial intelligence model. 27. The method of claim 25, wherein the artificial intelligence model identifier indicates the corresponding beamwidth and corresponding beam direction of each beam in the beam measurement resource set and each beam in the beam set. 28. The method of claim 25, wherein the artificial intelligence model identifier indicates a quasi-co-location relationship between beams in the beam measurement resource set and beams in the beam set. 29. A method for conducting wireless communication at a network entity, the method comprising: Send a first message to the user equipment (UE), the first message indicating a first subset of artificial intelligence model identifiers supported by the network entity from a predefined set of multiple artificial intelligence model identifiers associated with beam prediction; The UE receives a second message indicating a second subset of artificial intelligence model identifiers supported by the UE from the predefined set of multiple artificial intelligence model identifiers; Send control signaling to the UE, the control signaling indicating an artificial intelligence model identifier included in both the first subset and the second subset, wherein the control signaling also indicates a beam measurement resource set, and wherein the control signaling indicates a mapping of the beam measurement resource set to the input of the artificial intelligence model associated with the artificial intelligence model identifier and the output of the artificial intelligence model to the beam set; Transmit a set of reference signals via the set of beam measurement resources; and The UE receives a report indicating a set of predicted values ​​for the beam set that are at least partially based on the artificial intelligence model identifier and at least partially based on the reference signal set. 30. The method of claim 29, wherein: To indicate the beam measurement resource set, the control signaling indicates the channel state information reference signal index set or the synchronization signal block index set, and The artificial intelligence model identifier indicates the mapping of the channel state information reference signal index set or the synchronization signal block index set to the input of the artificial intelligence model associated with the artificial intelligence model identifier.

Claims

1. A user equipment (UE), the UE comprising: one or more memories storing processor-executable code; and one or more processors coupled with the one or more memories and capable, individually or collectively, of executing the code to cause the UE to: receive, from a network entity, a first message indicating a first subset of artificial intelligence model identifiers, of a predefined set of artificial intelligence model identifiers associated with beam prediction, that are supported by the network entity; send, to the network entity, a second message indicating a second subset of artificial intelligence model identifiers, of the predefined set of artificial intelligence model identifiers, that are supported by the UE; receive control signaling indicating an artificial intelligence model identifier that is included in both the first subset and the second subset, wherein the control signaling further indicates a set of beam measurement resources, and wherein the control signaling indicates a mapping of the set of beam measurement resources to an input of an artificial intelligence model associated with the artificial intelligence model identifier and an output of the artificial intelligence model to a set of beams; receive, via the set of beam measurement resources, a set of reference signals; and send a report indicating a set of predicted values for the set of beams generated using the artificial intelligence model and based at least in part on the set of reference signals.

2. The UE of claim 1, wherein the first message is one of system information, a radio resource control message, a medium access control (MAC) control element, or downlink control information.

3. The UE of claim 1, wherein the second message is one of a radio resource control message, a medium access control (MAC) control element, or uplink control information.

4. The UE of claim 1, wherein: to indicate the set of beam measurement resources, the control signaling indicates a set of channel state information reference signal indices or a set of synchronization signal block indices, and the artificial intelligence model identifier indicates a mapping of the set of channel state information reference signal indices or the set of synchronization signal block indices to the input of the artificial intelligence model.

5. The UE of claim 1, wherein the artificial intelligence model identifier indicates a respective beam width and a respective beam direction for each beam of the set of beam measurement resources and each beam of the set of beams.

6. The UE of claim 1, wherein the artificial intelligence model identifier indicates a quasi-co-location relationship between a beam of the set of beam measurement resources and a beam of the set of beams.

7. The UE of claim 1, wherein the artificial intelligence model identifier indicates a relative timing between the set of beam measurement resources and the set of predicted values.

8. The UE of claim 1, wherein the report further indicates a precoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink angle of departure, or a departure direction based at least in part on the set of predicted values.

9. The UE of claim 1, wherein: the control signaling further schedules a channel state information report, and the report is included within the channel state information report.

10. The UE of claim 1, wherein the one or more processors are further capable of individually or collectively operating to execute the code to cause the UE to: perform, with the network entity, a training procedure of the artificial intelligence model to predict a set of training measurement results for a set of training beams using a set of training beam measurement resources, wherein the set of beam measurement resources corresponds to the set of training beam measurement resources, and wherein the set of beams corresponds to the set of training beams.

11. The UE of claim 10, wherein: the set of training beam measurement resources includes a first set of channel state information reference signals or a first set of synchronization signal blocks having a first index value, the set of beam measurement resources includes a second set of channel state information reference signals or a second set of synchronization signal blocks having the first index value, and a same mapping of the first index value to an input value of the artificial intelligence model is used to predict the set of training measurement results and generate the set of predicted values.

12. The UE of claim 1, wherein the one or more processors are further capable of individually or collectively operating to execute the code to cause the UE to: receive, from the network entity, a set of validation beam measurement resources and corresponding ground truth prediction targets for a second set of beams of the artificial intelligence model; receive a second set of reference signals via the set of validation beam measurement resources; and transmit a second report indicating that a second set of predicted values for the second set of beams generated using the artificial intelligence model based at least in part on the set of reference signals satisfy an accuracy metric with respect to the corresponding ground truth prediction targets, wherein the reception of the control signaling is based at least in part on the transmission of the second report.

13. A network entity, the network entity comprising: one or more memories storing processor-executable code; and one or more processors coupled with the one or more memories and capable of individually or collectively operating to execute the code to cause the network entity to: transmit, to a user equipment (UE), a first message indicating a first subset of artificial intelligence model identifiers of a predefined set of multiple artificial intelligence model identifiers associated with beam prediction that are supported by the network entity; receive, from the UE, a second message indicating a second subset of artificial intelligence model identifiers of the predefined set of multiple artificial intelligence model identifiers that are supported by the UE; transmitting control signaling to the UE, the control signaling indicating an artificial intelligence model identifier included in both the first subset and the second subset, wherein the control signaling further indicates a set of beam measurement resources, and wherein the control signaling indicates a mapping of the set of beam measurement resources to an input of an artificial intelligence model associated with the artificial intelligence model identifier and an output of the artificial intelligence model to a set of beams; transmitting a set of reference signals via the set of beam measurement resources; and receiving a report from the UE, the report indicating a set of predicted values for the set of beams based at least in part on the artificial intelligence model identifier and based at least in part on the set of reference signals.

14. The network entity of claim 13, wherein the first message is one of system information, a radio resource control message, a medium access control (MAC) control element, or downlink control information.

15. The network entity of claim 13, wherein the second message is one of a radio resource control message, a medium access control (MAC) control element, or uplink control information.

16. The network entity of claim 13, wherein: to indicate the set of beam measurement resources, the control signaling indicates a set of channel state information reference signal indices or a set of synchronization signal block indices, and the artificial intelligence model identifier indicates a mapping of the set of channel state information reference signal indices or the set of synchronization signal block indices to the input of an artificial intelligence model associated with the artificial intelligence model identifier.

17. The network entity of claim 13, wherein the artificial intelligence model identifier indicates a respective beam width and a respective beam direction for each beam in the set of beam measurement resources and each beam in the set of beams.

18. The network entity of claim 13, wherein the artificial intelligence model identifier indicates a quasi-co-location relationship between a beam in the set of beam measurement resources and a beam in the set of beams.

19. The network entity of claim 13, wherein the artificial intelligence model identifier indicates a relative timing between the set of beam measurement resources and the set of predicted values.

20. The network entity of claim 13, wherein based at least in part on the set of predicted values, the report further indicates a precoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink angle of departure, or a departure direction.

21. The network entity of claim 13, wherein: the control signaling further schedules a channel state information report, and the report is included within the channel state information report.

22. The network entity of claim 13, wherein the one or more processors are further capable of individually or collectively executing the code to cause the network entity to: performing, with the UE, a training procedure of the artificial intelligence model using a set of training beam measurement resources to predict a set of training measurement results for a set of training beams, wherein the set of beam measurement resources corresponds to the set of training beam measurement resources, and wherein the set of beams corresponds to the set of training beams.

23. The network entity of claim 22, wherein: the set of training beam measurement resources comprises a first set of channel state information reference signals or a first set of synchronization signal blocks having a first index value, the set of beam measurement resources comprises a second set of channel state information reference signals or a second set of synchronization signal blocks having the first index value, and a same mapping of the first index value to an input value of the artificial intelligence model is used to predict the set of training measurement results and generate the set of predicted values.

24. The network entity of claim 13, wherein the one or more processors are further individually or collectively capable of operating to execute the code to cause the network entity to: transmit, to the UE, a set of validation beam measurement resources and corresponding ground truth prediction targets for a second set of beams of the artificial intelligence model; transmit, via the set of validation beam measurement resources, a second set of reference signals; and receive, from the UE, a second report indicating that a second set of predicted values for the second set of beams generated at the UE using the artificial intelligence model based at least in part on the set of reference signals satisfy an accuracy metric with respect to the corresponding ground truth prediction targets, wherein the transmission of the control signaling is based at least in part on the reception of the second report.

25. A method for wireless communication at a user equipment (UE), the method comprising: receiving, from a network entity, a first message indicating a first subset of artificial intelligence model identifiers, of a predefined set of multiple artificial intelligence model identifiers associated with beam prediction, that are supported by the network entity; transmitting, to the network entity, a second message indicating a second subset of artificial intelligence model identifiers, of the predefined set of multiple artificial intelligence model identifiers, that are supported by the UE; receiving control signaling indicating an artificial intelligence model identifier that is included in both the first subset and the second subset, wherein the control signaling further indicates a set of beam measurement resources, and wherein the control signaling indicates a mapping of the set of beam measurement resources to an input of an artificial intelligence model associated with the artificial intelligence model identifier and an output of the artificial intelligence model to a set of beams; receiving, via the set of beam measurement resources, a set of reference signals; and transmitting a report indicating a set of predicted values for the set of beams generated using the artificial intelligence model and based at least in part on the set of reference signals.

26. The method of claim 25, wherein the first message is one of system information, a radio resource control message, a medium access control (MAC) control element, or downlink control information.

27. The method of claim 25, wherein the second message is one of a radio resource control message, a medium access control (MAC) control element, or uplink control information.

28. The method of claim 25, wherein: to indicate the set of beam measurement resources, the control signaling indicates a set of channel state information reference signal indices or a set of synchronization signal block indices, and the artificial intelligence model identifier indicates a mapping of the set of channel state information reference signal indices or the set of synchronization signal block indices to the input of the artificial intelligence model.

29. The method of claim 25, wherein the artificial intelligence model identifier indicates a respective beam width and a respective beam direction of each beam in the set of beam measurement resources and each beam in the set of beams.

30. The method of claim 25, wherein the artificial intelligence model identifier indicates a quasi-co-location relationship between a beam in the set of beam measurement resources and a beam in the set of beams.

31. The method of claim 25, wherein the artificial intelligence model identifier indicates a relative timing between the set of beam measurement resources and the set of predicted values.

32. The method of claim 25, wherein based at least in part on the set of predicted values, the report further indicates a precoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink angle of departure, or a departure direction.

33. The method of claim 25, wherein: the control signaling further schedules a channel state information report, and the report is included within the channel state information report.

34. The method of claim 25, further comprising: performing, with the network entity, a training procedure of the artificial intelligence model using a set of training beam measurement resources to predict a set of training measurement results for a set of training beams, wherein the set of beam measurement resources corresponds to the set of training beam measurement resources, and wherein the set of beams corresponds to the set of training beams.

35. The method of claim 34, wherein: the set of training beam measurement resources includes a first set of channel state information reference signals or a first set of synchronization signal blocks having a first index value, the set of beam measurement resources includes a second set of channel state information reference signals or a second set of synchronization signal blocks having the first index value, and a same mapping of the first index value to an input value of the artificial intelligence model is used to predict the set of training measurement results and generate the set of predicted values.

36. The method of claim 25, further comprising: receiving, from the network entity, a set of validation beam measurement resources for a second set of beams of the artificial intelligence model and a corresponding benchmark ground truth prediction target; receiving a second set of reference signals via the set of validation beam measurement resources; and transmitting a second report indicating that a second set of predicted values for the second set of beams generated using the artificial intelligence model based at least in part on the set of reference signals satisfies an accuracy metric with respect to the corresponding benchmark true value prediction target, wherein the reception of the control signaling is based at least in part on the transmission of the second report.

37. A method for wireless communication at a network entity, the method comprising: transmitting, to a user equipment (UE), a first message indicating a first subset of artificial intelligence model identifiers, of a predefined set of artificial intelligence model identifiers associated with beam prediction, that are supported by the network entity; receiving, from the UE, a second message indicating a second subset of artificial intelligence model identifiers, of the predefined set of artificial intelligence model identifiers, that are supported by the UE; transmitting, to the UE, control signaling indicating an artificial intelligence model identifier that is included in both the first subset and the second subset, wherein the control signaling further indicates a set of beam measurement resources, and wherein the control signaling indicates a mapping of the set of beam measurement resources to an input of an artificial intelligence model associated with the artificial intelligence model identifier and an output of the artificial intelligence model to a set of beams; transmitting, via the set of beam measurement resources, a set of reference signals; and receiving, from the UE, a report indicating a set of predicted values for the set of beams based at least in part on the artificial intelligence model identifier and based at least in part on the set of reference signals.

38. The method of claim 37, wherein the first message is one of system information, a radio resource control message, a medium access control (MAC) control element, or downlink control information.

39. The method of claim 37, wherein the second message is one of a radio resource control message, a medium access control (MAC) control element, or uplink control information.

40. The method of claim 37, wherein: to indicate the set of beam measurement resources, the control signaling indicates a set of channel state information reference signal indices or a set of synchronization signal block indices, and the artificial intelligence model identifier indicates a mapping of the set of channel state information reference signal indices or the set of synchronization signal block indices to the input of an artificial intelligence model associated with the artificial intelligence model identifier.

41. The method of claim 37, wherein the artificial intelligence model identifier indicates a respective beam width and a respective beam direction of each beam in the set of beam measurement resources and each beam in the set of beams.

42. The method of claim 37, wherein the artificial intelligence model identifier indicates a quasi-co-location relationship between a beam in the set of beam measurement resources and a beam in the set of beams.

43. The method of claim 37, wherein the artificial intelligence model identifier indicates a relative timing between the set of beam measurement resources and the set of predicted values.

44. The method of claim 37, wherein the report further indicates a precoding matrix indicator, a channel quality indicator, a channel impulse response, a rank indicator, an angle of arrival, a downlink angle of departure, or a departure direction based at least in part on the set of predicted values.

45. The method of claim 37, wherein: the control signaling further schedules a channel state information report, and the report is included within the channel state information report.

46. The method of claim 37, further comprising: performing, with the UE, a training procedure of the artificial intelligence model to predict a set of training measurement results for a set of training beams using a set of training beam measurement resources, wherein the set of beam measurement resources corresponds to the set of training beam measurement resources, and wherein the set of beams corresponds to the set of training beams.

47. The method of claim 46, wherein: the set of training beam measurement resources includes a first set of channel state information reference signals or a first set of synchronization signal blocks having a first index value, the set of beam measurement resources includes a second set of channel state information reference signals or a second set of synchronization signal blocks having the first index value, and a same mapping of the first index value to an input value of the artificial intelligence model is used to predict the set of training measurement results and generate the set of predicted values.

48. The method of claim 37, further comprising: transmitting, to the UE, a set of validation beam measurement resources for a second set of beams of the artificial intelligence model and a corresponding ground truth prediction target; transmitting a second set of reference signals via the set of validation beam measurement resources; and receiving, from the UE, a second report indicating that a second set of predicted values for the second set of beams generated at the UE using the artificial intelligence model based at least in part on the second set of reference signals satisfies an accuracy metric with respect to the corresponding ground truth prediction target, wherein the transmission of the control signaling is based at least in part on the reception of the second report.

Images