Artificial intelligence based beam management

Abstract

Systems, methods, and devices for wireless communication are described. One example method includes configuring, by a network device, at least one of whether a prediction result associated with a beam or a reference signal (RS) can be used for a function, which beam's prediction result is used for the function, or a margin applied when the prediction result is used for the function. Another example method includes determining, by a wireless device, that a failure occurs or a first condition is met. The method also includes performing, by the wireless device and based on the determination, a corresponding operation. Among other benefits, the disclosed methods improve the efficiency of beam management.

Description

Technical Field

[0001] This patent document generally relates to wireless communication. Background Technology

[0002] Mobile communication technologies are driving the world toward an increasingly interconnected and networked society. Compared to existing wireless networks, next-generation systems and wireless communication technologies will need to support a wider range of use case characteristics and provide more complex and granular access requirements and flexibility.

[0003] LTE (Long Term Evolution) is a wireless communication standard developed by the 3rd Generation Partnership Project (3GPP) for mobile devices and data terminals. LTE-Advanced (LTE-A) is a wireless communication standard that enhances the LTE standard. The fifth-generation wireless system, known as 5G, advances the LTE and LTE-A wireless standards and aims to support higher data rates, massive connectivity, ultra-low latency, high reliability, and other emerging service requirements. Summary of the Invention

[0004] This patent document discloses a method for using beamforming results predicted by artificial intelligence (AI) and / or machine learning (ML). The base station (BS) indicates whether and how the predicted beamforming results are used for a function. However, if a failure occurs, the user equipment (UE) stops applying the predicted beamforming results, disables active AI / ML functions or models, and / or begins collecting data for AI / ML functions, or model training, testing, validation, or performance monitoring. Among other benefits, this method can also improve the efficiency of beamforming management.

[0005] The first example wireless communication method includes a network device configuring at least one of the following: whether the prediction result associated with a beam or reference signal (RS) can be used for a function; which beam's prediction result is used for a function; or a margin applied when the prediction result is used for a function.

[0006] The second example wireless communication method includes determining, by a wireless device, that a failure has occurred or a first condition has been met. The method also includes, by the wireless device and based on that determination, performing at least one of the following operations: ceasing the use of prediction results associated with the beam or reference signal (RS) for a function; deactivating an active artificial intelligence (AI) or machine learning (ML) function or model that outputs the prediction results; switching to an AI or ML function or model; or initiating data collection.

[0007] The third example wireless communication method includes receiving conditions for data collection by a wireless device. The method also includes the wireless device determining, based on the conditions, whether to perform data collection.

[0008] The fourth example wireless communication method includes a wireless device determining the prediction accuracy of a prediction result associated with a beam or reference signal (RS). The method also includes the wireless device determining, based on the prediction accuracy, whether to record or report the collected data.

[0009] Note that if a patent document discloses a method for transmitting information from a first device to a second device, it will be understood that a method for receiving information from the first device by the second device is also disclosed. Similarly, if a method for receiving a message from the second device is disclosed, it will be understood that the message is transmitted from the second device to the first device.

[0010] In yet another example embodiment, an apparatus configured or operable to perform the methods described above is disclosed. The apparatus includes at least one processor configured to implement the methods described above.

[0011] In yet another example embodiment, the above method is embodied in processor-executable code and stored in a non-transitory computer-readable storage medium. When executed by a processor, the code included in the computer-readable storage medium causes the processor to implement the method described in this patent document.

[0012] The above and other aspects, and their embodiments, are described in more detail in the accompanying drawings, description, and claims. Attached Figure Description

[0013] Figure 1 This is an example flowchart for configuring whether or how the predicted beam results are used for a function.

[0014] Figure 2 This is an example flowchart for performing actions when a failure occurs.

[0015] Figure 3 This is a sample flowchart for performing data collection.

[0016] Figure 4 This is an example flowchart used to record or report the collected data.

[0017] Figure 5 An example block diagram of a hardware platform that may be part of a network device or a wireless device is shown.

[0018] Figure 6 Example wireless communication including a base station (BS) and a user equipment (UE) based on some implementations of the disclosed technology is shown. Detailed Implementation

[0019] The example headings in the following sections are used to facilitate understanding of the disclosed subject matter and do not in any way limit the scope of the claimed subject matter. Therefore, one or more features of one example section may be combined with one or more features of another example section. Furthermore, the term "5G" is used for clarity, but the technologies disclosed in this document are not limited to 5G technology and can also be used in wireless systems implementing other protocols.

[0020] I. Introduction

[0021] This patent document discloses whether and how beamforming results predicted by artificial intelligence (AI) and / or machine learning (ML) are used to indicate, configure, or specify functions. This patent document also discloses the actions to be performed when a failure occurs. Among other benefits, this method can improve the efficiency of beam management.

[0022] The convergence of communication networks and artificial intelligence (AI) technologies is progressing, building upon previous work on AI-enabled network architectures and leveraging AI / machine learning (ML) to achieve 5GC and air interface intelligence through closed-loop processes such as data collection, ML model training, analytical inference, and data analysis. AI / ML for the NR air interface is being discussed in 3GPP. Discussions include collecting data from UEs for ML model training and supporting a common 5GS AI / ML framework for AI / ML radio implementation.

[0023] Currently, 3GPP research focuses on realizing potential enhancements to AI / ML for beam management. During the research, BM-Case 1 (spatial domain downlink beam prediction) and BM-Case 2 (temporal downlink beam prediction) were selected.

[0024] BM-Case 1: Spatial domain downlink beam prediction for beam set A based on measurement results from beam set B. AI / ML model training and inference can be performed on the NW side or the UE side. Set A and set B are different, or set B is a subset of set A. AI / ML model input can be L1-RSRP measurements based on set B, L1-RSRP measurements based on set B and auxiliary information, or CIR based on set B, or L1-RSRP measurements based on set B and the corresponding DL Tx and / or Rx beam IDs.

[0025] BM-Case 2: Temporal downlink beam prediction for beam set A based on historical measurements of beam set B. AI / ML model training and inference are performed on the NW or UE side. Set A and set B are different (set B is not a subset of set A), or set B is a subset of set A (set A and set B are not the same), or set A and set B are the same. The AI / ML model input can be the measurements of K (K≥1) most recent measurement instances, with the following alternatives: Alt 1): L1-RSRP measurements based solely on set B; Alt 2): L1-RSRP measurements based on set B and auxiliary information; Alt 3): L1-RSRP measurements based on set B and the corresponding DLTx and / or Rx beam IDs. F predictions for F future time instances can be obtained based on the output of the AI / ML model, where each prediction is for each time instance. At least F=1.

[0026] Many issues regarding AI / ML-based functions remain to be addressed, such as the application of predictive results (the output of AI / ML functions / models based on input data) and the conditions for data collection for AI / ML model training, testing, validation, or performance monitoring. In this patent document, we provide some solutions to these problems.

[0027] In all embodiments of this patent document, the prediction result refers to the output of an artificial intelligence (AI) or machine learning (ML) function or AI / ML model. The prediction result of beam set A is the output of the AI / ML function of the AI / ML model based on the measurement results of beam set B.

[0028] In this patent document, the UE performs measurements on the RS transmitted on the beam to evaluate the beam quality. Therefore, the prediction result of the beam also implies the prediction result of the RS, and vice versa.

[0029] In this patent document, the predicted result or the predicted result refers to the reasoning result of the AI / ML function or the AI / ML model.

[0030] II. Example 1

[0031] Example 1 describes the application of the prediction results.

[0032] This example describes how to apply the prediction results of DL beam / reference signal in various functions.

[0033] - BS can be configured to determine whether the prediction results can be used for the following functions:

[0034] - Beam Failure Detection (BFD) Function

[0035] - Beam Failure Recovery (BFR) function

[0036] - Radio Link Failure (RLF) Detection

[0037] - RACH resource selection

[0038] - Beam selection for paging / SI messages / reception

[0039] - SSB selection for CG-based Small Data Transfer (SDT)

[0040] - L3 Community-Level Quality Derivation

[0041] - L3 beam quality derivation

[0042] The BS can be configured to use the prediction results of which beams can be used for one or more of various functions.

[0043] The Baseline Controller (BS) can be configured to apply margins when prediction results are used in one or more of various functions. The User Experience Controller (UE) can indicate the accuracy of the AI / ML function / model. The BS can determine the margins based on the accuracy indicated by the UE.

[0044] Previously, this function was performed based on measurements of the downlink beam / reference signal.

[0045] This embodiment provides a method for performing these functions based on beam prediction results.

[0046] In this embodiment:

[0047] The BS can configure whether the prediction results can be used for one or more functions by including indicators in the cell's system information or dedicated RRC message, or MAC CE or DCI.

[0048] Indicators can indicate whether the prediction results of a beam or reference signal can be used for the indicated function.

[0049] If a margin is configured, it is considered along with the beam prediction results when the beam prediction is used for functionality. For example, beam quality is considered as (prediction result + margin) or (prediction result - margin). The prediction result can be the beam's RSRP, RSRQ, or SINR. The margin can be RSRQP, RSRQ, or SINR margin. This margin can be indicated in system information, RRC messages, MAC CE, or DCI.

[0050] The UE can indicate the accuracy information of the AI / ML function or AI / ML model used for beam prediction. Accuracy information may include one of the following:

[0051] - The probability that the output result is within N1 dB of the measurement result.

[0052] - Based on the measurement results, the probability that the predicted optimal beam is within the first N2 beams.

[0053] - Based on the measurement results, the predicted probability that the top N3 beams include the optimal beam.

[0054] - Based on the measurement results, the predicted optimal beam is the probability that it is the optimal beam.

[0055] The BS can configure RSRP / RSRQ / SINR margins based on the accuracy reported by the UE.

[0056] To configure which beam's prediction results can be used for the indicated function:

[0057] The BS can be configured so that only the prediction results of the first N predicted beams can be used for the indicated function. The BS is configured to do this by including an indicator or the value of N in the system information, RRC message, MAC CE, or DCI. If configured this way, the UE will only consider using the first N predicted beam results for the indicated function.

[0058] The BS can be configured so that only predictions greater than a threshold are used for the indicated function. The BS is configured to do this by specifying a threshold. The threshold can be an RSRP, RSRQ, or SINR threshold.

[0059] BS can be configured with a list of beams, and its prediction results can be used for the indicated functions.

[0060] As a concrete example:

[0061] The BS can indicate the following configurations in the system information:

[0062] The prediction results of the indicator beam can be used for RACH resource selection and beam failure recovery processes.

[0063] The prediction results of the indicator beam cannot be used for RLF detection.

[0064] The prediction results of the first N predicted beams can be used for RACH resource selection and beam failure recovery.

[0065] RSRP margin used for beam prediction results.

[0066] The UE then determines the beam corresponding to the first N prediction results and performs beam selection based on the prediction results of the determined beam. When evaluating the quality of the beam, the configured RSRP margin is considered by subtracting the margin from the prediction results of the beam (i.e., predicted RSRP - RSRP margin).

[0067] The UE can perform beam failure assessment based on the prediction result of one of the top N predicted beams. When assessing beam quality, the configured RSRP margin is also taken into account; for example, the UE considers (prediction result - configured RSRP margin) as beam quality.

[0068] The following section provides more methods for specific functionalities. The methods provided above also apply.

[0069] For BFD functionality:

[0070] The BS can be configured with a list of reference signals, and the prediction results can be used for beam failure detection. The UE performs beam failure detection based on the prediction results of the configured beams.

[0071] In NR, the BS configures a list of reference signals (RS) for beam failure detection (BFD). If not provided by the BS, the UE can use the RS corresponding to the active TCI state for PDCCH (Physical Downlink Control Channel) reception. In NR, RS for BFD can be configured for each DL BWP. The TCI state for PDCCH reception on the CORESET of the active DL BWP is also specified. The UE can declare a beam failure state when the number of beam failure instances indicated from the physical layer reaches a configured threshold before the configured timer expires.

[0072] In this embodiment:

[0073] A BS can indicate one or more RSs configured in a list of RSs used for BFD, and their predictions can be used for the BFD function. The BS can indicate an index of the RS, such as an SSB index or a CSI-RS index, or an ID used to identify the RS used for BFD (e.g., in NR, the ID is RadioLinkMonitoringRS), or include an indicator in each entry of the list of RSs used for BFD.

[0074] The BS can configure a list of RSs for BFD. The prediction results of each RS in the list can be used for the BFD function. When the list of RSs for BFD is configured, the UE can perform BFD based solely on the RSs in that list, or the UE can perform BFD based on both the RSs in that list and the RSs configured for BFD in the traditional RSs.

[0075] Or it can be used to determine whether a beam failure has been declared.

[0076] The BS can be configured to use a threshold or threshold offset when the prediction result is applied to the BFD function. The UE applies the threshold or threshold offset to the BFD function.

[0077] When using the prediction results of RS to evaluate the beam quality of RS used for BFD, a configured threshold or threshold offset is applied.

[0078] If a threshold is configured, it is separate from the traditional threshold applied when RS measurements are used for BFD functionality. The UE evaluates RS based on the configured threshold.

[0079] If a threshold offset is configured, the UE evaluates RS quality based on the threshold offset and the conventional threshold when the prediction results are used in the BFD process. As a specific example, when evaluating RS quality, the UE compares the predicted RS result with (threshold offset + conventional threshold).

[0080] For the BFR function, the BS can be configured with one or more beams, and its prediction results can be used for candidate beam selection for the BFR function.

[0081] BS can be configured to apply a threshold or threshold offset when the prediction results are used for the BFR function.

[0082] previously:

[0083] After a beam failure is detected on the PCell, the UE can trigger beam failure recovery by initiating a random access (RACH) procedure on the serving cell. If the RACH procedure involves a contention-based random access procedure, the UE can include an indication of the beam failure on the PCell in the BFR MACCE (Media Access Control Control Element).

[0084] Upon detecting a beam failure on an SCell, the UE can trigger a beam failure recovery procedure, in which the UE transmits a BFR MAC CE for that SCell. The UE can select a suitable / candidate beam (if available) for that SCell and indicate the beam's identifier and information about the beam failure in the BFR MAC CE. The beam indicated in the MAC CE is selected from a list of candidate beams configured by the BS. The indicated beam is selected based on a comparison between the corresponding beam's measurement results and a configured threshold.

[0085] In this embodiment, the BFR function includes at least one of the following:

[0086] 1. When a UE initiates a RACH procedure for BFR, the UE selects one or more beams from a candidate beam list and selects RACH resources based on the selected candidate beams. In NR, this function is applied when a beam failure is declared in the PCell or PSCell.

[0087] 2. Select one or more candidate beams and indicate the selected candidate beam information to the BS. In NR, this function is performed when beam declaration for SCell fails. The selected candidate beam information is indicated to the BS by transmitting a MAC CE to the BS.

[0088] If the RS is configured to use its prediction results for candidate beam selection, the UE can apply the beam prediction results when selecting candidate beams.

[0089] In addition to the methods listed above, the BS can also configure the beam list using one of the following methods:

[0090] 1. A list of beams configured in the BS configuration. Each entry in the list includes the beam's identifier. This beam is also configured in the list of candidate beams for the BFR.

[0091] 2. A list of beam indices configured for BS, where each entry refers to the index of an entry in the candidate beam list used for BFR.

[0092] 3. BS configures another list of candidate beams for BFR, which includes beams whose predictions can be used for BFR functions.

[0093] If the RS configuration applies a threshold or threshold offset when the prediction results are used for the BFR function, the UE performs candidate beam selection based on the threshold or threshold offset.

[0094] If the BS is configured with a threshold, the UE will compare the prediction results of the candidate beams with the threshold, rather than with the traditional threshold applied when the measurement results are used for the BFR function.

[0095] If the BS is configured with a threshold offset, the UE will compare the prediction results of the candidate beam with a combination of the threshold offset and the conventional threshold (e.g., conventional threshold + threshold offset).

[0096] The UE can instruct the BS whether the selected candidate beam is based on the prediction results.

[0097] The UE can indicate:

[0098] 1. A list of selected candidate beams based on the prediction results.

[0099] 2. A flag used to indicate the selection of some or all of the selected candidate beams based on the prediction results.

[0100] 3. A flag used for each selected candidate beam to indicate whether the corresponding beam was selected based on the prediction results.

[0101] The UE can indicate this in a BFR MAC CE, or RRC message, or PUCCH signaling.

[0102] For RACH resource selection, the BS can be configured to use its prediction results for RS / beams in RACH resource selection. The BS can also be configured to apply a threshold when the beam prediction results are used for RACH resource selection.

[0103] Previously, RACH resource selection could be based on the quality of the downlink beam / RS. The UE selects the beam based on the measurement results of the DL beam and a threshold.

[0104] In this embodiment, if the BS is configured with a beam whose prediction results can be used for RACH resource selection, the UE can use the configured BS prediction results for RACH resource selection.

[0105] The beam can be configured using one of the following methods:

[0106] - A list indicating beam indices

[0107] - Indicator bitmap, where each bit represents a beam / RS. For example, each bit represents an SSB.

[0108] - Indicates that the top N predicted beams can be candidate beams for RACH resource selection.

[0109] - Indicates the top N predicted beams whose predictions can be used for RACH resource selection.

[0110] When using prediction results, the BS can configure a separate threshold or threshold offset for RO selection. When a threshold or threshold offset is configured, the UE applies the threshold or threshold offset in RACH resource selection.

[0111] The BS can indicate whether the method of this embodiment is applied to the purpose of initiating a RACH procedure. For example, the BS can indicate that it is applied to a RACH procedure initiated for SDT or beam failure recovery, but not to an SI request.

[0112] For RLF detection, SI message reception, and paging message reception:

[0113] The BS can configure a list of SSBs, and its prediction results can be used for SSB selection for SI message / paging message reception or for evaluation of the corresponding beam for RLF detection.

[0114] The BS can be configured to select only the top N predicted beams for SI message / paging message reception based on its prediction results.

[0115] The BS can configure an RSRP threshold, which is used when the UE selects the SSB for SI message / paging message reception based on the prediction result.

[0116] The BS can configure an RSRP threshold, which is used when the UE evaluates the beam quality used for RLF detection.

[0117] Regarding SDT resource selection:

[0118] Previously, the BS could configure Small Data Transmission (SDT) based on Configuration Grant (CG). Each CG resource was configured with a configured uplink grant and associated with a list of SSBs. When the UE used a CG resource for small data transmission, the UE selected an SSB based on the configured SSB list and a configured RSRP threshold. More specifically, the UE selected an SSB whose SS-RSRP was higher than the (configured) cg-SDT-RSRP-ThresholdSSB among the SSBs associated with the configured uplink grant.

[0119] In this embodiment:

[0120] The BS can configure a subset of the SSB list associated with the configured authorization configuration. The prediction results of the subset of SSBs can be used for SSB selection in CG-based SDT.

[0121] The BS can configure an RSRP threshold, which is used when the UE selects an SSB for SDT based on the prediction results.

[0122] Using the method provided in this embodiment, the BS can configure which functions the beam prediction results can be applied to, and how the prediction results are applied to these functions. Furthermore, the BS can configure the beam / reference signal, so that its prediction results can be used for specific processes when evaluating the quality of the beam / reference signal. This approach provides fine-grained control over UE behavior when applying prediction results.

[0123] III. Example 2

[0124] Example 2 describes the conditions for disabling the AI ​​model and prediction results, as well as for starting / stopping data collection.

[0125] This embodiment relates to stopping the application of prediction results, disabling / switching AI / ML functions or models, and / or starting / stopping data collection when certain conditions are met.

[0126] In the event of a failure on the UE side, the UE can perform one of the following first actions:

[0127] - Stop using the beam prediction results for the functions shown in Example 1.

[0128] - Disable active AI / ML models or AI / ML functions that output prediction results.

[0129] - Switch to another AI / ML model.

[0130] - Start collecting data.

[0131] Failure includes one of the following:

[0132] - Failed during radio link detection

[0133] - Failure during beam failure detection

[0134] - Failed during RACH

[0135] - Failed during the switchover process

[0136] - Failure during SDT process

[0137] - Failure in performance monitoring of AI / ML functions or AI / ML models

[0138] If the previous failure is recovered or handled, or if the configured conditions are not met, the UE may perform one of the following second actions.

[0139] - Begin using the beam prediction results for the functions listed in Example 1.

[0140] - Activate AI / ML models or AI / ML functions that were disabled when activation failure occurs.

[0141] - Stop collecting data.

[0142] When a failure occurs, or after a failure is recovered or handled, the UE can indicate to the BS the application of beam prediction results, or the state of the AI / ML model. The UE can indicate to the BS whether to use beam prediction results when a failure occurs, or whether to use beam prediction results during failure detection.

[0143] The BS can instruct the UE to activate the AI / ML function or AI / ML model, or start using the beam prediction results, or start data collection, or stop data collection upon receiving an instruction.

[0144] The data collected in this embodiment can be used for model training, model validation, testing, or performance monitoring. The BS can be configured for the duration, interval, or amount of data collection.

[0145] If the beam prediction results are used when a failure occurs or during failure detection, the UE can apply the method in this embodiment.

[0146] The collected data may include the following: predictions that led to failure, predictions used during failure, or predictions used in the failure detection process.

[0147] The UE can perform data collection according to the configured data collection settings.

[0148] Data collection configurations can include one of the following:

[0149] - Duration during which the UE performs data collection.

[0150] - The interval between data collection.

[0151] - The maximum amount of data to be collected.

[0152] The collected data includes at least one of the following:

[0153] - Information about beam set A, including beam identification, or measurement results or sorting information (the order of beams when all beams in set A are sorted), whether the beam is the best beam, and whether the beam is within N dB of the best beam.

[0154] - The prediction results of the beams in set A, if the prediction results of set A are used.

[0155] - Information and measurement results of beam set B.

[0156] - Information about whether the prediction or measurement results were used for a specific beam.

[0157] - Information about the model that generates the predicted beam.

[0158] The UE can indicate the availability of the collected data to the BS.

[0159] The UE can transmit the collected data to the BS.

[0160] The BS can instruct the UE whether to apply the method in this embodiment via system information, RRC messages, MAC CE, or DCI.

[0161] For RLF:

[0162] Previously: In NR, after receiving a continuous desynchronization indication (N310) from the physical layer, the UE starts a radio problem timer (T310). While timer T310 is running, after receiving a continuous synchronization indication (N311) from the physical layer, the UE stops timer T310. After T310 expires, the UE declares an RLF. The values ​​of N310, N311, and T310 can be configured by the BS to the UE via an RRC message with IE (Information Element) RLF-TimersAndConstants. N310 can define the maximum number of continuous "desynchronization" indications received from lower layers for the PCell. N311 can define the maximum number of continuous "synchronization" indications received from lower layers for the PCell.

[0163] When an RLF is declared, the UE can select a suitable cell and then initiate an RRC reconstruction procedure. If no suitable cell is found within a certain period after the RLF is declared, the UE can enter the RRC_IDLE state.

[0164] In this embodiment:

[0165] When the UE receives a first number of consecutive asynchronous indications configured from the physical layer, or when timer T310 starts, or when an RLF is declared, the UE performs one of the first actions.

[0166] When the UE receives a second number of continuous synchronization indications configured from the physical layer, or when timer T310 stops, the UE performs at least one of the second actions.

[0167] The BS can configure a first number of values ​​for the continuous asynchronous indication and a second number of values ​​for the continuous synchronous indication.

[0168] When an RLF is declared, in the RRC Reconstruction Complete message, or the RRC Reconstruction Post-Reconstruction RRC message or MAC CE, the UE indicates to the BS the availability of collected data, or whether to use or which beam prediction results to detect the RLF.

[0169] If an RLF is not declared, the UE indicates to the BS, in an RRC message or MAC CE, the availability of collected data, or whether to use, or which beam's prediction results to use, to detect the RLF.

[0170] If the UE performs data collection during this process, the collected data includes the following data, which contains prediction results for detecting RLF.

[0171] In the event of a MAC reset, all models are deactivated.

[0172] For beam failure:

[0173] Previously: In NR, upon receiving a beam failure instance indication from a lower layer of the serving cell, the UE starts or restarts the beam failure detection timer for the serving cell and increments the beam failure counter by 1. If the beam failure detection timer expires, the UE resets the beam failure counter to zero. If the beam failure counter is greater than or equal to the configured maximum beam failure instance count, the UE considers a beam failure to have occurred.

[0174] When a beam failure occurs, if the serving cell is a PCell or PSCell, the UE can initiate a BFR procedure by initiating a RACH for BFR; or if the serving cell is an SCell, the UE can send a MAC CE to the BS. If the serving cell is a PSCell and the SCG is disabled, the UE indicates the beam failure to the RRC layer. The UE can also indicate to the BS that a beam failure occurred in the PSCell.

[0175] The UE considers the beam failure recovery process to have been successfully completed and resets the beam failure counter to zero under the following conditions:

[0176] If the serving cell is a PCell or PSCell, and the random access procedure initiated for the BFR is successfully completed; if the serving cell is an SCell, and a PDCCH addressed to the C-RNTI is received, the PDCCH indicating uplink grant for the new transmission, and the PDCCH is for a HARQ process that transmits a MAC CE for the BFR, the MAC CE for the BFR containing beam failure recovery information for the serving cell.

[0177] In this embodiment:

[0178] When a beam failure instance indication is received from a lower layer, or the beam failure detection timer is started or restarted, or the beam failure detection counter is greater than or equal to the configured maximum beam failure instance count or the configured threshold, or the UE believes that a beam failure has occurred in the serving cell, the UE performs at least one of the first actions for the serving cell.

[0179] When the UE believes that the beam failure recovery process has been successfully completed, or when the beam failure detection timer expires, the UE performs at least one of the second actions for the serving cell.

[0180] In the case of performing a BFR procedure, in the BFR MAC CE, or in the RRC message after the BFR procedure is successfully completed, the UE indicates to the BS the availability of the collected data, and / or whether to use or which beam prediction results to detect beam failure.

[0181] If the UE performs data collection during this process, the collected data includes the following data, which contains prediction results for detecting beam failure.

[0182] For RACH failure:

[0183] During random access procedures, there are situations where the RACH process fails to complete successfully or random access problems occur. For example, in contention-based RACH, if the random access response is not received successfully, or if contention resolution is considered unsuccessful, the UE can increment a counter by 1. If the counter exceeds the configured maximum count, the UE considers the RACH process to have failed. Otherwise, the UE initiates another preamble transmission to continue the RACH process.

[0184] In this embodiment:

[0185] When the RACH procedure is not successfully completed, or a random access problem occurs, or the random access response is not received successfully, or the contention resolution is deemed unsuccessful, or the number of preamble transmissions reaches the configured threshold, the UE performs at least one of the first actions.

[0186] When the UE has performed one of the first actions and the RACH procedure is considered to have been successfully completed, or the random access response is successfully received, or the contention resolution is considered to have been successful, the UE performs at least one of the second actions.

[0187] In this situation, the UE can indicate the availability of the collected data to the BS in a MAC CE or RRC message.

[0188] The UE can indicate to the BS information about the beams used for RACH resource selection. This information may include the beam index and whether the beam prediction result was used for RACH resource selection.

[0189] If the UE performs data collection during this process, the collected data includes data containing prediction results for RACH resource selection.

[0190] For switching failures:

[0191] previously:

[0192] The UE starts at time T304 in one of the following situations:

[0193] - After receiving a handover command message for the MCG (primary cell group) (e.g., RRCReconfiguration message, including reconfigurationWithSync in NR).

[0194] - After receiving the handover command message for SCG (Secondary Cell Group).

[0195] - After a conditional reconfiguration is executed, i.e. when the application stores the switch command message.

[0196] - For MCG and SCG, when the LTM cell switching procedure is triggered by an indication from a lower layer, and for MCG, when the LTM cell switching procedure is performed after cell selection is executed while timer T311 is running.

[0197] When timer T304 expires, the UE considers the handover to have failed, and:

[0198] For MCG T304, in the case of handover from NR or within NR, or handover without performing a random access procedure, or LTM cell switching procedure, the UE initiates an RRC reconstruction procedure.

[0199] For SCG T304, the network is notified of the synchronization failure and reconfiguration by initiating the SCG failure information procedure.

[0200] In this embodiment, in addition to the above list of actions, when timer T304 expires, the UE also performs at least one of the following actions: either the first action for the serving cell of the MCG is performed when T304 of the MSG expires, or the first action for the serving cell of the SCG is performed when T304 of the SCG expires.

[0201] If T304 for the MCG expires, the UE can initiate data collection for the serving cell of the MCG; or if T304 for the SCG expires, the UE can initiate data collection for the serving cell of the SCG.

[0202] If the UE performs data collection during this process, the collected data includes the prediction results used during the handover process.

[0203] After the RRC reconstruction process is successfully completed, the UE can indicate to the BS beam whether or which beam to use during the handover.

[0204] For conditions related to performance monitoring of AI / ML functions or models:

[0205] Performance monitoring is used to monitor the performance of activated or deactivated AI / ML features or models based on similar-level performance metrics or metrics based on the output of AI / ML features or models.

[0206] Link-level performance metrics may include BLER (Block Error Rate) or throughput on the UE side. The BS can configure BLER thresholds or throughput thresholds. If the measured BLER or throughput on the UE side does not meet the performance conditions / requirements, i.e., the measured BLER is greater than the BLER threshold, or the measured throughput is less than the throughput threshold, the UE performs at least one of the first actions. If the measured BLER again becomes less than the BLER threshold, and / or the measured throughput again becomes greater than the throughput threshold, the UE performs at least one of the second actions.

[0207] Metrics based on the output of AI / ML functions or AI / ML models include one or more of the following:

[0208] - The number, probability, or ratio of predicted results within N1 dB of the measurement results.

[0209] - The number, probability, or ratio of the predicted best beam within the first N2 beams of the measurement results.

[0210] - The predicted top N3 beams include the number, probability, or ratio of the best beams in the measurement results.

[0211] - The predicted best beam is the number, probability, or ratio of the best beams in the measurement results.

[0212] The quantity, ratio, and probability of each metric can be calculated based on predictions generated during a time period. This time period can be configured to be periodic, and it occurs periodically. The time period can be a period preceding a specific point in time, for example, when calculating quantity, probability, or ratio.

[0213] BS allows you to configure the length of the time period.

[0214] BS can be configured with values ​​for N1, N2, or N3.

[0215] BS can configure quantity thresholds, probability thresholds, or ratio thresholds for accuracy metrics.

[0216] When one or more of the probabilities listed above are greater than the configured probability threshold, or one or more of the ratios listed above are greater than the configured ratio threshold, or one or more of the quantities listed above are greater than the configured quantity threshold, the UE performs at least one of the first actions, in particular, initiating data collection.

[0217] When one or more of the probabilities listed above become less than the configured probability threshold, or one or more of the ratios listed above become less than the configured ratio threshold, or one or more of the quantities listed above again become less than the configured numerical threshold, the UE performs at least one of the second actions.

[0218] If the UE performs data collection during this process, the collected data includes data for performance monitoring, or prediction results for performance monitoring metric calculations, or data collected during the performance monitoring period, or data collected within the period.

[0219] Using the method provided in this embodiment, when a failure occurs in the serving cell, the UE stops using the prediction results or switches to another AI / ML model. When the failure is recovered or processed, or when indicated by the BS, the UE resumes using the beam prediction results. Therefore, in the event of a failure, based on the assumption that the failure may be caused by inaccurate outputs of the AI / ML model, the prediction results inferred by the AI / ML model are not applied. When the failure is processed or recovered, the UE continues to use the prediction results or activates the disabled AI / ML model, if indicated by the network.

[0220] Furthermore, when AI / ML model training, validation, testing, or performance monitoring fails, the user experience (UE) initiates data collection. This approach is based on the assumption that the AI / ML model's accuracy is insufficient in the event of failure. Therefore, the data collected during failures is valuable for readjusting the AI / ML model.

[0221] IV. Example 3

[0222] Example 3 describes the conditions for performing data collection.

[0223] This example pertains to the conditions used for data collection.

[0224] The network node configures a set of conditions for data collection for the UE.

[0225] When one or more conditions are met, network nodes can be configured to perform data collection.

[0226] When one or more conditions are not met, network nodes can be configured to perform data collection.

[0227] The UE may or may not perform data collection depending on its configuration.

[0228] If data collection is performed, the UE can record information related to association conditions within the collected data. Association conditions are one or more configuration conditions that are met when data is collected.

[0229] The UE can indicate the availability and association conditions of the collected data to the network node.

[0230] Network nodes can request collected data associated with one or more conditions.

[0231] The application requirements include at least one or a combination of the following conditions:

[0232] - UE altitude-related conditions, whether it is the UE's altitude or its altitude relative to the ground.

[0233] - Codebook-related conditions.

[0234] - Conditions related to UE orientation change.

[0235] - UE mobility information, including UE speed, or static, low-speed, and high-speed conditions as specified in the specification. UE mobility can be in the vertical or horizontal direction.

[0236] The collected data is used at least for AI / ML model training, testing, validation, or performance monitoring.

[0237] In this embodiment, the network node can be a BS, an OAM service, a core network node, or an OTT (over-the-top) server. The network node differs in different steps of this embodiment.

[0238] The altitude-related condition specifies the altitude range at which the UE performs data collection. Altitude information may include at least the following:

[0239] - Height range

[0240] - Minimum height threshold

[0241] - Maximum height threshold

[0242] - High lag

[0243] The UE performs data collection within the configured altitude range indicated by the altitude information, or the UE stops / does not perform data collection within the configured altitude range.

[0244] The data collected when the UE is within the altitude range includes information about the associated altitude conditions.

[0245] Codebook-related conditions can include at least the following:

[0246] - Identifier for the codebook configuration.

[0247] - Codebook type, such as type 1 and type 2 as defined in the specification.

[0248] - Codebook subtypes, such as typeI-SinglePanel, typeI-MultiPanel, and type II as specified in the specification.

[0249] - Codebook mode.

[0250] - Codebook parameters related to the number of beams, number of panels, limits, and number of antenna ports as defined in the specification.

[0251] When the configured codebook-related conditions are met, the network node can configure whether to perform data collection. If the codebook currently used by the UE meets the configured codebook-related conditions, the UE will perform data collection or not perform data collection according to the network node's configuration.

[0252] If data collection is performed when the configured codebook-related conditions are met, the UE can record codebook-related information in the collected data.

[0253] Orientation-related conditions may include at least one of the following:

[0254] - The maximum value of the azimuth change. For example, if the angle between the UE's initial azimuth and the UE's current azimuth is less than the configured threshold.

[0255] - Minimum change in UE orientation. For example, if the angle between the UE's initial orientation and its current orientation is greater than a configured threshold.

[0256] - Maximum rate of azimuth change. For example, if the current angular velocity of the UE's azimuth or the maximum angular velocity of the UE's azimuth during a time period is less than the configured azimuth change rate threshold.

[0257] - Minimum rate of azimuth change. For example, if the current angular velocity of the UE's azimuth or the maximum angular velocity of the UE's azimuth during a time period is greater than the configured azimuth change rate threshold.

[0258] - Maximum average rate of azimuth change. For example, if the average rate of change of the UE's azimuth angle or angular velocity during a time period is less than the configured average rate of change of azimuth change threshold.

[0259] - Minimum average rate of azimuth change. For example, if the average rate of change of the UE's azimuth angle or angular velocity during a time period is greater than the configured average rate of azimuth change.

[0260] Network nodes can configure whether the UE performs data collection when orientation-related conditions are met. When orientation-related conditions are met, the UE performs or does not perform data collection according to the network node's configuration.

[0261] If data collection is performed when the orientation conditions are met, the UE can record orientation-related information in the collected data.

[0262] If the UE has already collected data, when the UE indicates the availability of the collected data, the UE can also indicate information about the associated conditions for collecting the data.

[0263] When a network node requests data to be collected by a UE, the network node can request data associated with one or more specific conditions by indicating specific conditions in the request message. For example, in a request for collected data, the network node indicates altitude information, which is used to indicate that the collected data will be requested when the indicated altitude information is met.

[0264] In messages used for availability indications or data requests, the condition information can be one of the following:

[0265] - The condition ID configured by the network node along with the condition.

[0266] - Altitude information of the UE within the altitude range when collecting data.

[0267] - Codebook information used when collecting data.

[0268] - Information related to the orientation conditions that were met when collecting data.

[0269] - Altitude information of the UE when the data is collected, even if the UE is not within the altitude range.

[0270] - Codebook information not used during data collection.

[0271] - Information related to location conditions that were not met during data collection.

[0272] The collected data can be associated with one or more conditions.

[0273] Using the method provided in this embodiment, a network node can configure a set of conditions that are evaluated when the UE decides whether to perform data collection for a specific purpose (e.g., AI / ML model training, testing, validation, performance monitoring). This method enables the network node to control data collection behavior. In some cases, the network node only needs to collect data in certain scenarios, such as when the UE is above a height threshold, or when a specific codebook is used, or when the average angular velocity is less than a configured value.

[0274] Furthermore, this method can associate the collected data with application conditions. Therefore, the UE can specify the data and the associated application conditions, and network nodes can request data associated with specific application conditions. This provides flexibility in the data collection request process.

[0275] This embodiment also provides some creative application conditions: UE height, codebook, and UE orientation change related conditions.

[0276] V. Example 4

[0277] Example 4 describes the conditions for omitting data from the UE-side model.

[0278] This embodiment concerns how to perform the selection / omission of collected data, or whether to record data based on the accuracy of AI / ML predictions based on the collected data.

[0279] The UE records the collected data based on the following methods, or decides to select the data for reporting, or decides to omit the collected data.

[0280] The UE validates the collected data using relevant AI / ML models by comparing the beam measurement results with the prediction results. The UE determines the prediction accuracy based on the comparison results.

[0281] If the prediction accuracy is higher than the configured threshold, the UE can omit the collected data for reporting or not record the collected data.

[0282] If the prediction accuracy is lower than the configured threshold, the UE can choose to use the collected data for reporting or record the collected data in local storage.

[0283] Prediction accuracy includes one of the following:

[0284] - The predicted result is within N1 dB of the measured result.

[0285] - The predicted best beam is within the first N2 beams of the measurement results.

[0286] - The predicted top N3 beams include the best beams from the measurement results.

[0287] - The predicted optimal beam is the optimal beam in the measurement results.

[0288] Network nodes can be configured with values ​​for N1, N2, and N3.

[0289] Network nodes can be configured to apply one or more precision parameters.

[0290] Network nodes can configure these settings via system information, RRC messages, MAC CE, or DCI.

[0291] The collected data can be used for AI / ML model training, readjustment, testing, validation, or performance monitoring.

[0292] This method allows network nodes to be configured with criteria for selecting whether to report or record collected data. The approach is based on the idea that if the AI / ML model's output is insufficient (below a configured threshold or condition) based on the collected data (measurements using beam set B as input), then the collected data is valuable for model training, testing, validation, or performance monitoring. Otherwise, if the collected data is within the model's expected accuracy, then there is no need to report or record it.

[0293] This method saves signaling and radio resources used for data reporting, as well as UE local storage used for recording the collected data.

[0294] The patent document discloses the following, among other things:

[0295] BS indicates whether and how the prediction results are used for a specific function.

[0296] In the event of a failure on the UE side, the UE stops applying the beam prediction results / disables the active AI / ML model, and / or initiates execution data collection.

[0297] Conditions for triggering / stopping data collection include changes in altitude, UE orientation, and codebook information.

[0298] Based on the verification results, the data collected for the UE-side model is omitted / selected.

[0299] Figure 1 This is an example flowchart for configuring whether or how predicted beam results are used for a function. Operation 102 includes configuring at least one of the following by the network device: whether the prediction results associated with a beam or reference signal (RS) are available for use in the function; which beam's prediction results are used in the function; or the margin applied when the prediction results are used in the function. In some embodiments, the method may be implemented according to embodiments 1-4. In some embodiments, further steps of performing the method may be based on better system performance than conventional protocols.

[0300] In some embodiments, configuring whether the prediction results can be used for the function includes at least one of the following: the network device is configured to allow only the prediction results of the first N predicted beams to be used for the function; the network device is configured to allow only the prediction results greater than a threshold to be used for the function, wherein the threshold includes at least one of a reference signal received power (RSRP) threshold, a reference signal received quality (RSRQ) threshold, or a signal interference to noise ratio (SINR) threshold; or the network device is configured with a list of beams whose prediction results can be used for the function.

[0301] In some embodiments, the function includes at least one of the following: beam failure detection (BFD); beam failure recovery (BFR); radio link failure (RLF) detection; random access channel (RACH) resource selection; beam selection for receiving paging messages or system information (SI) messages; synchronization signal block (SSB) selection for small data transmission (SDT) based on configuration grant (CG); layer 3 (L3) cell-level quality derivation; or L3 beam quality derivation.

[0302] In some embodiments, the network device is configured or indicates at least one of the following, whose prediction results may be used for the beam failure detection (BFD) function: a list of RSs; one or more RSs configured in the list of RSs; or a list of RSs of another.

[0303] In some embodiments, the network device is configured with a threshold or threshold offset that is used when the prediction results are applied to the beam failure detection (BFD) function.

[0304] In some embodiments, the network device is configured or indicates at least one of the following, the prediction results of which can be used for candidate beam selection for the beam failure recovery (BFR) function: one or more beams; a list of beams, wherein each entry in the list of beams includes a beam identifier and corresponds to an entry in the candidate beam list for the BFR function; a list of beam indices, wherein each entry in the list of beam indices corresponds to an entry in the candidate beam list for the BFR function; or another candidate beam list.

[0305] In some embodiments, the network device receives from the wireless device an indication regarding whether the selection of candidate beams for the BFR function is based on prediction results, wherein the indication includes at least one of the following: a list of selected candidate beams selected based on prediction results; a flag indicating some or all of the selected candidate beams selected based on prediction results; or a flag for each selected candidate beam indicating whether the corresponding beam is selected based on prediction results.

[0306] In some embodiments, the network device is configured with a threshold or threshold offset that is used when the prediction result is applied to the beam failure recovery (BFR) function, wherein the wireless device performs candidate beam selection based on the threshold or threshold offset.

[0307] In some embodiments, the network device configures or indicates at least one of the following, whose prediction results can be used for Random Access Channel (RACH) resource selection: a beam index list; a bitmap, wherein each bit represents a beam or RS; the top N predicted beams that can be used as candidate beams for RACH resource selection; or the top N predicted beams.

[0308] In some embodiments, the network device is configured with a threshold or threshold offset that is used when the prediction results are applied to the selection of random access channel (RACH) resources.

[0309] In some embodiments, the network device indicates whether the prediction result is applied to the purpose of initiating the Random Access Channel (RACH) procedure.

[0310] In some embodiments, the network device is configured or indicates at least one of the following, the prediction results of which can be used for radio link failure (RLF) detection, paging message reception, or system information (SI) message reception: a list of synchronization signal blocks (SSBs); or the predicted top N beams.

[0311] In some embodiments, the network device is configured with a reference signal received power (RSRP) threshold, which is used when the wireless device selects a synchronization signal block (SSB) for system information (SI) message reception or paging message reception based on prediction results, or when the wireless device evaluates the beam quality for radio link failure (RLF) detection.

[0312] In some embodiments, the network device is configured or indicates the following, the prediction of which can be used for Small Data Transmission (SDT) resource selection: a subset of the list of Synchronization Signal Blocks (SSBs) associated with the configured Grant (CG) configuration.

[0313] In some embodiments, the network device is configured with a reference signal received power (RSRP) threshold, which is used when the wireless device selects a synchronization signal block (SSB) for small data transmission (SDT) based on prediction results.

[0314] In some embodiments, the margin is associated with accuracy information of an artificial intelligence (AI) or machine learning (ML) function or model.

[0315] In some embodiments, the accuracy information includes at least one of the following: the probability that the output result is within N1 dB of the measurement result; the probability that the predicted best beam is within the first N2 beams based on the measurement result; the probability that the predicted first N3 beams include the best beam based on the measurement result; or the probability that the predicted best beam is the best beam based on the measurement result.

[0316] Figure 2This is an example flowchart for performing operations upon failure. Operation 202 includes the wireless device determining that a failure has occurred or a first condition has been met. Operation 204 includes the wireless device, based on this determination, performing at least one of the following operations: stopping the use of prediction results associated with the beam or reference signal (RS) for the function; deactivating active artificial intelligence (AI) or machine learning (ML) functions or models that output prediction results; switching to AI or ML functions or models; or initiating data collection. In some embodiments, the method may be implemented according to embodiments 1-4. In some embodiments, further steps in performing the method may be based on better system performance than conventional protocols.

[0317] In some embodiments, failure includes at least one of the following: failure during radio link detection (RLD); failure during beam failure detection (BFD); failure during random access channel (RACH) procedures; failure during handover procedures; failure during small data transmission (SDT); or failure during performance monitoring for AI or ML functions or models.

[0318] In some embodiments, the failure is recovered or handled, or a second condition is met, and the method further includes the wireless device performing at least one of the following operations: starting to use the prediction results associated with the beam or RS for the function; activating the AI ​​or ML function or model that was deactivated when the failure occurred; or stopping data collection.

[0319] In some embodiments, the method further includes transmitting from the wireless device to the network device at least one of the following indications: application of the prediction result; state of the AI ​​or ML function or model when a failure occurs; state of the AI ​​or ML function or model after the failure is recovered or processed; whether the prediction result is used when a failure occurs; or whether the prediction result is used during failure detection.

[0320] In some embodiments, the method further includes receiving at least one of the following instructions from a wireless device and from a network device: an instruction to activate an AI or ML function or model; an instruction to initiate the use of prediction results; an instruction to initiate data collection; or an instruction to stop data collection.

[0321] In some embodiments, data is collected if the prediction result is used when a failure occurs, or if the prediction result is used during failure detection, wherein the data includes at least one of the following: the prediction result that caused the failure; the prediction result used during the failure; the prediction result used during failure detection; the prediction result of a set of beams A, if the prediction result of a set of beams A is used; or information about whether the prediction result or the measurement result is used for a particular beam.

[0322] In some embodiments, for a radio link failure (RLF), the first condition includes at least one of the following: when a configured first number of consecutive asynchronous indications are received from the physical layer; when timer T310 is started; or when an RLF is declared. The wireless device indicates to the network device the availability of the collected data, or the wireless device indicates to the network device whether or which predictions are used for RLF detection, and the collected data includes predictions for RLF detection.

[0323] In some embodiments, for a radio link failure (RLF), the second condition includes at least one of the following: when a configured second number of continuous synchronization indications are received from the physical layer; or when timer T310 is stopped. The wireless device indicates to the network device the availability of the collected data, or the wireless device indicates to the network device whether or which predictions are used for RLF detection, and the collected data includes predictions for RLF detection.

[0324] In some embodiments, for beam failure, the first condition includes at least one of the following: when a beam failure instance indication is received from a lower layer; a beam failure detection timer is started or restarted; a beam failure detection counter is greater than or equal to a configured maximum count of beam failure instances or a configured threshold; or the wireless device determines that a beam failure has occurred in the serving cell. The wireless device indicates to the network device the availability of collected data, or the wireless device indicates to the network device whether or not to use, and which prediction results are used, for beam failure detection, and the collected data includes prediction results for beam failure detection.

[0325] In some embodiments, for beam failure, the second condition includes at least one of the following: when the wireless device determines that the beam failure recovery process has been successfully completed; or when the beam failure detection timer expires. The wireless device indicates to the network device the availability of the collected data, or the wireless device indicates to the network device whether or which predictions are used for beam failure detection, and the collected data includes predictions for beam failure detection.

[0326] In some embodiments, for a Random Access Channel (RACH) procedure failure, a first condition includes at least one of the following: the RACH procedure did not complete successfully; a random access problem occurred; a random access response was not received successfully; contention resolution was unsuccessful; or the number of preamble transmissions reached a configured threshold. The wireless device indicates to the network device the availability of collected data, or the wireless device indicates to the network device whether or which predictions are used for RACH resource selection, and the collected data includes predictions for RACH resource selection.

[0327] In some embodiments, for a failed Random Access Channel (RACH) procedure, the second condition includes at least one of the following: the RACH procedure was successfully completed; a random access response was successfully received; or contention resolution was successful. The wireless device indicates to the network device the availability of the collected data, or the wireless device indicates to the network device whether or which predictions are used for RACH resource selection, and the collected data includes predictions for RACH resource selection.

[0328] In some embodiments, for a handover failure, the first condition includes at least one of the following: timer T304 for the primary cell group (MCG) expires; or timer T304 for the secondary cell group (SCG) expires. The wireless device indicates to the network device the availability of the collected data, or the wireless device indicates to the network device whether or which prediction results are used during the handover, and the collected data includes prediction results used during the handover.

[0329] In some embodiments, for performance monitoring of AI or ML functions or models, a first condition includes at least one of the following: one or more probabilities are greater than a configured probability threshold; one or more ratios are greater than a configured ratio threshold; or one or more quantities are greater than a configured quantity threshold. The collected data includes data used for performance monitoring, prediction results used for performance monitoring metric calculations, data collected during the performance monitoring period, or data collected within a predetermined time period.

[0330] In some embodiments, for performance monitoring of AI or ML functions or models, the second condition includes at least one of the following: one or more probabilities are less than a configured probability threshold; one or more ratios are less than a configured ratio threshold; or one or more quantities are less than a configured quantity threshold. The collected data includes data used for performance monitoring, prediction results used for performance monitoring metric calculations, data collected during the performance monitoring period, or data collected within a predetermined time period.

[0331] In some embodiments, one or more probabilities, one or more ratios, or one or more quantities are calculated based on predictions generated during a time period, wherein the time period is configured to be periodic and occurs periodically.

[0332] Figure 3 This is an example flowchart for performing data collection. Operation 302 includes receiving conditions for data collection by the wireless device. Operation 304 includes determining, by the wireless device and based on the conditions, whether to perform data collection. In some embodiments, the method may be implemented according to embodiments 1-4. In some embodiments, further steps in performing the method may be based on better system performance than conventional protocols.

[0333] In some embodiments, the wireless device performs data collection, wherein the wireless device records information related to association conditions, and wherein the wireless device indicates to the network device the availability of the collected data and the information related to association conditions.

[0334] In some embodiments, the condition includes at least one of the following: height-related conditions of the wireless device; codebook-related conditions of the wireless device; orientation change-related conditions of the wireless device; or mobility information of the wireless device.

[0335] In some embodiments, codebook-related conditions include at least one of the following: an identifier of the codebook configuration; a codebook type; a codebook subtype; a codebook mode; or codebook parameters related to the number of beams, the number of panels, limits, or the number of antenna ports.

[0336] In some embodiments, the azimuth change-related conditions include at least one of the following: maximum azimuth change; minimum azimuth change; maximum rate of azimuth change; minimum rate of azimuth change; maximum average rate of azimuth change; or minimum average rate of azimuth change.

[0337] In some embodiments, the collected data is used at least for training, testing, validating, or monitoring the performance of artificial intelligence (AI) or machine learning (ML) functions or models.

[0338] In some embodiments, the method further includes receiving a request from a network device by a wireless device for collecting data associated with one or more specific conditions; and transmitting the collected data by the wireless device in response to the request.

[0339] Figure 4 This is an example flowchart for recording or reporting collected data. Operation 402 includes the wireless device determining the prediction accuracy of the prediction result associated with the beam or reference signal (RS). Operation 404 includes the wireless device determining, based on the prediction accuracy, whether to record or report the collected data. In some embodiments, the method may be implemented according to embodiments 1-4. In some embodiments, further steps in performing the method may be based on better system performance than conventional protocols.

[0340] In some embodiments, the prediction accuracy is greater than a configured threshold, and the wireless device does not log the collected data in local storage or report the collected data.

[0341] In some embodiments, the prediction accuracy is less than a configured threshold, and the wireless device logs the collected data in local storage or reports the collected information.

[0342] Figure 5An example block diagram of a hardware platform 500 is shown, which may be part of a network device (e.g., a base station (BS), transmit and receive point (TRP), or radio access network (RAN)) or a wireless device (e.g., a user equipment (UE)). The hardware platform 500 includes at least one processor 510 and a memory 505 storing instructions thereon. When the instructions are executed by the processor 510, the hardware platform 500 is configured to perform its functions. Figure 1-4 The operations described in the various embodiments described in this patent document are as follows: Transmitter 515 transmits or sends information or data to another device. For example, a network device transmitter may send a message to a user equipment. Receiver 520 receives information or data transmitted or sent by another device. For example, a user equipment may receive a message from a network device. For example, a UE, wireless device, or network device as described in this document may be implemented using hardware platform 500.

[0343] The above implementation method will be applied to wireless communication. Figure 6 An example of a wireless communication system (e.g., a 5G or NR cellular network) including a base station 620 and one or more user equipments (UEs) 611, 612, and 613 is illustrated. In some embodiments, the UE accesses the BS (e.g., the network) using a communication link to the network (sometimes referred to as the uplink direction, as depicted by dashed arrows 631, 632, and 633), which then enables subsequent communication from the BS to the UE (e.g., as shown in the direction from the network to the UE, sometimes referred to as the downlink direction, as shown by arrows 641, 642, and 643). In some embodiments, the BS sends information to the UE (sometimes referred to as the downlink direction, as depicted by arrows 641, 642, and 643), which then enables subsequent communication from the UE to the BS (e.g., as shown in the direction from the UE to the BS, sometimes referred to as the uplink direction, as shown by dashed arrows 631, 632, and 633). The UE can be, for example, a smartphone, tablet, mobile computer, machine-to-machine (M2M) device, and Internet of Things (IoT) device. The UE described in this document can be communicatively coupled to... Figure 6 The base station 620 is depicted in the image.

[0344] Those skilled in the art will understand that this patent document discloses methods for improving the efficiency of beam management, among other benefits. The disclosed methods include whether and how beaming results predicted by artificial intelligence (AI) and / or machine learning (ML) are used for indicated, configured, or designated functions. The disclosed methods also include what actions to take when a failure occurs.

[0345] Some embodiments described herein are described in the general context of methods or processes that may be implemented in one embodiment as a computer program product embodied in a computer-readable medium, including computer-executable instructions, such as program code, that are executed by a computer in a networked environment. The computer-readable medium may include removable and non-removable storage devices, including but not limited to read-only memory (ROM), random access memory (RAM), optical disc (CD), digital versatile optical disc (DVD), etc. Therefore, the computer-readable medium may include non-transitory storage media. Typically, program modules may include routines, programs, objects, components, data structures, etc., that perform a particular task or implement a particular abstract data type. Computer or processor-executable instructions, associated data structures, and program modules represent examples of program code for performing steps of the methods disclosed herein. A particular sequence of such executable instructions or associated data structures represents examples of corresponding actions for implementing the functionality described in such steps or processes.

[0346] Some of the disclosed embodiments can be implemented as devices or modules using hardware circuitry, software, or a combination thereof. For example, hardware circuitry implementations may include discrete analog and / or digital components, which may be integrated, for example, as part of a printed circuit board. Alternatively or additionally, the disclosed components or modules may be implemented as application-specific integrated circuits (ASICs) and / or field-programmable gate arrays (FPGAs) devices. Some implementations may additionally or alternatively include digital signal processors (DSPs), which are dedicated microprocessors with an architecture optimized for the operational requirements of digital signal processing associated with the disclosed functions of this application. Similarly, various components or sub-components within each module can be implemented using software, hardware, or firmware. Interconnectivity between modules and / or components within modules can be provided using any of the connection methods and media known in the art, including but not limited to communication over the Internet, wired, or wireless networks using appropriate protocols.

[0347] While this document contains numerous details, these details should not be construed as limiting the scope of the claimed invention or any potentially claimed content, but rather as descriptions of features specific to particular embodiments. Certain features described herein may also be implemented in combination in a single embodiment within the context of an individual embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually in multiple embodiments or in any suitable sub-combination. Furthermore, although the foregoing features may be described as functioning in certain combinations, and even initially claimed in this way, in some cases one or more features from said combination may be removed from said combination, and said combination may involve sub-combinations or variations thereof. Similarly, although operations are depicted in a specific order in the drawings, this should not be construed as requiring these operations to be performed in the specific order or sequence shown, or requiring all illustrated operations to be performed to obtain the desired result.

[0348] Only some implementation methods and examples are described, and other implementations, enhancements and variations may be made based on the content described and illustrated in this patent document.

Claims

1. A method for wireless communication, comprising: The network device is configured with at least one of the following: Whether the prediction results associated with the beam or reference signal (RS) can be used for functionality; Which beam prediction result was used for the function; or The margin applied when the forecast results are used in a function.

2. The method according to claim 1, wherein, Whether the prediction results can be used for a function includes at least one of the following: The network device is configured such that only the prediction results of the first N predicted beams can be used for the function. The network device is configured such that only predictions greater than a threshold can be used for the function, wherein the threshold includes at least one of a Reference Signal Received Power (RSRP) threshold, a Reference Signal Received Quality (RSRQ) threshold, or a Signal-to-Interference-to-Noise Ratio (SINR) threshold; or The network device is configured with a list of beams whose prediction results can be used for the function.

3. The method according to claim 1 or 2, wherein, The function includes at least one of the following: Beam Failure Detection (BFD) function; Beam Failure Recovery (BFR) function; Radio link failure (RLF) detection; Random Access Channel (RACH) resource selection; Beam selection for receiving paging messages or system information (SI) messages; Synchronization Signal Block (SSB) selection for Small Data Transmission (SDT) based on Configuration Grant (CG); L3 (Layer 3) cell-level quality derivation; or L3 beam quality derivation.

4. The method according to any one of claims 1-3, wherein, The network device is configured or indicates at least one of the following, and its prediction results can be used for beam failure detection (BFD) functionality: List of RS; One or more RSs configured in the list of RSs; or Another list of RS.

5. The method according to any one of claims 1-4, wherein, The network device is configured with a threshold or threshold offset to use when the prediction result is applied to the beam failure detection (BFD) function.

6. The method according to any one of claims 1-5, wherein, The network device is configured or indicates at least one of the following, and its prediction results can be used for candidate beam selection for the beam failure recovery (BFR) function: One or more beams; A list of beams, wherein each entry in the list of beams includes a beam identifier and corresponds to an entry in the list of candidate beams for the BFR function; A list of beam indices, wherein each entry in the list of beam indices corresponds to an entry in the candidate beam list for the BFR function; or Another list of candidate beams.

7. The method according to claim 6, wherein, The network device receives from the wireless device an indication regarding whether the candidate beam selection for the BFR function is based on the prediction result, wherein the indication includes at least one of the following: A list of selected candidate beams selected based on the prediction results; A flag used to indicate part or all of the selected candidate beams chosen based on the prediction results; or A flag used for each selected candidate beam to indicate whether the corresponding beam is selected based on the prediction result.

8. The method according to any one of claims 1-7, wherein, The network device is configured with a threshold or threshold offset to use when the prediction result is applied to the beam failure recovery (BFR) function, and wherein the wireless device performs candidate beam selection based on the threshold or threshold offset.

9. The method according to any one of claims 1-8, wherein, The network device is configured or indicates at least one of the following, and its prediction result can be used for random access channel (RACH) resource selection: Beam index list; Bitmap, where each bit represents a beam or RS; The predicted top N beams can be used as candidate beams for the RACH resource selection; or The predicted first N beams.

10. The method according to any one of claims 1-9, wherein, The network device is configured with a threshold or threshold offset to use when the prediction result is applied to the selection of random access channel (RACH) resources.

11. The method according to any one of claims 1-10, wherein, The network device indicates whether the prediction result is applied to the purpose of initiating the Random Access Channel (RACH) procedure.

12. The method according to any one of claims 1-11, wherein, The network device is configured or indicates at least one of the following, the prediction results of which can be used for radio link failure (RLF) detection, paging message reception, or system information (SI) message reception: A list of Synchronization Signal Blocks (SSBs); or The predicted first N beams.

13. The method according to any one of claims 1-12, wherein, The network device is configured with a reference signal received power (RSRP) threshold, which is used when the wireless device selects a synchronization signal block (SSB) for system information (SI) message reception or paging message reception based on the prediction result, or when the wireless device evaluates the quality of the beam used for radio link failure (RLF) detection.

14. The method according to any one of claims 1-13, wherein, The network device is configured or indicates at least one of the following, and the prediction result can be used for Small Data Transmission (SDT) resource selection: A subset of the list of synchronization signal blocks (SSBs) associated with the configured authorization (CG) configuration.

15. The method according to any one of claims 1-14, wherein, The network device is configured with a reference signal received power (RSRP) threshold to use when the wireless device selects a synchronization signal block (SSB) for small data transmission (SDT) based on the prediction result.

16. The method according to any one of claims 1-15, wherein, The margin is associated with the accuracy information of artificial intelligence (AI) or machine learning (ML) functions or models.

17. The method according to claim 16, wherein, The accuracy information includes at least one of the following: The probability that the output result is within N1 dB of the measurement result; Based on the measurement results, the predicted probability that the best beam is within the first N2 beams; Based on the measurement results, the predicted probability that the top N3 beams include the optimal beam; or Based on the measurement results, the predicted optimal beam is the probability that it is the optimal beam.

18. A method for wireless communication, comprising: The wireless device determines whether a failure has occurred or the first condition has been met. as well as The wireless device performs at least one of the following operations based on the determination: Stop using prediction results associated with the beam or reference signal (RS) for functionality; Disable active artificial intelligence (AI) or machine learning (ML) functions or models that output predictive results; Switch to AI or ML features or models; or Start collecting data.

19. The method according to claim 18, wherein, The failure includes at least one of the following: Failure during Radio Link Detection (RLD); Failure during Beam Failure Detection (BFD); Failure during the Random Access Channel (RACH) procedure; Failure during the handover process; Failure during Small Data Transfer (SDT); or Failure during performance monitoring of AI or ML functions or models.

20. The method according to claim 18 or 19, wherein, The failure is recovered or handled, or the second condition is met, and the wireless device performs at least one of the following: Begin using the prediction results associated with the beam or RS for functionality; Activate the AI ​​or ML features or models that were deactivated when the failure occurred; or Stop collecting data.

21. The method according to any one of claims 18-20, further comprising transmitting, from the wireless device to the network device, at least one of the following instructions: Application of prediction results; The state of the AI ​​or ML function or model when the aforementioned failure occurs; The state of the AI ​​or ML function or model after the failure is recovered or handled; Whether to use the prediction result when the aforementioned failure occurs; or Whether to use the prediction results during the failure detection process.

22. The method according to any one of claims 18-21, further comprising receiving, from the wireless device and from the network device, at least one of the following instructions: Instructions to activate AI or ML features or models; Instructions to initiate the use of prediction results; Instructions to initiate data collection; or Instructions to stop data collection.

23. The method according to any one of claims 18-22, wherein, The data is collected if the prediction result is used when the failure occurs, or if the prediction result is used during the failure detection process, and wherein the data includes at least one of the following: The prediction result that led to the failure; The prediction results used during the failure; The prediction results used in the failure detection process; The prediction result of the beam set A, if the prediction result of the beam set A is used; or Whether the prediction or measurement results are used for information about a specific beam.

24. The method according to any one of claims 18-23, wherein, For radio link failure (RLF), the first condition includes at least one of the following: When the first number of consecutive asynchronous indications configured are received from the physical layer; When timer T310 is started; or When the RLF is declared; Wherein, the wireless device indicates to the network device the availability of the collected data, or wherein the wireless device indicates to the network device whether to use or which prediction results to use for the detection of the RLF, and wherein the collected data includes prediction results for the detection of the RLF.

25. The method according to any one of claims 20-24, wherein, For radio link failure (RLF), the second condition includes at least one of the following: When a second number of consecutive synchronization indications configured are received from the physical layer; or When timer T310 stops; Wherein, the wireless device indicates to the network device the availability of the collected data, or wherein the wireless device indicates to the network device whether to use or which prediction results to use for the detection of the RLF, and wherein the collected data includes prediction results for the detection of the RLF.

26. The method according to any one of claims 18-25, wherein, For beam failure, the first condition includes at least one of the following: When a beam failure instance indication is received from a lower layer; The beam failure detection timer has been started or restarted; The beam failure detection counter is greater than or equal to the configured maximum count of beam failure instances or the configured threshold; or The wireless device determines that a beam failure has occurred in the serving cell; Wherein, the wireless device indicates to the network device the availability of the collected data, or wherein the wireless device indicates to the network device whether to use or which prediction results to use for the detection of the beam failure, and wherein the collected data includes prediction results for the detection of the beam failure.

27. The method according to any one of claims 20-26, wherein, For beam failure, the second condition includes at least one of the following: When the wireless device determines that the beam failure recovery process has been successfully completed; or Beam failure detection timer expired; Wherein, the wireless device indicates to the network device the availability of the collected data, or wherein the wireless device indicates to the network device whether to use or which prediction results to use for the detection of the beam failure, and wherein the collected data includes prediction results for the detection of the beam failure.

28. The method according to any one of claims 18-27, wherein, For a random access channel (RACH) procedure to fail, the first condition includes at least one of the following: The RACH process failed to complete. A random access problem occurred; Random access response received unsuccessfully; The competition was not resolved successfully; or The number of preamble transmissions has reached the configured threshold; Wherein, the wireless device indicates to the network device the availability of the collected data, or wherein the wireless device indicates to the network device whether to use or which prediction results to use for RACH resource selection, and wherein the collected data includes prediction results for the RACH resource selection.

29. The method according to any one of claims 20-28, wherein, For a Random Access Channel (RACH) procedure to fail, the second condition includes at least one of the following: The RACH process was successfully completed; Random access response received successfully; or The competition was successfully resolved; Wherein, the wireless device indicates to the network device the availability of the collected data, or wherein the wireless device indicates to the network device whether to use or which prediction results to use for RACH resource selection, and wherein the collected data includes prediction results for the RACH resource selection.

30. The method according to any one of claims 18-29, wherein, For a handover failure, the first condition includes at least one of the following: Timer T304 for the primary cell group (MCG) has expired; or Timer T304 for the secondary cell group (SCG) has expired; Wherein, the wireless device indicates to the network device the availability of the collected data, or wherein the wireless device indicates to the network device whether or which prediction results are used during the handover, and wherein the collected data includes prediction results used during the handover.

31. The method according to any one of claims 18-30, wherein, For performance monitoring of AI or ML functions or models, the first condition includes at least one of the following: One or more probabilities are greater than the configured probability threshold; One or more ratios are greater than the configured ratio threshold; or One or more quantities exceed the configured quantity threshold; Furthermore, the collected data includes data used for performance monitoring, prediction results used for performance monitoring metric calculations, data collected during the performance monitoring period, or data collected within a predetermined period.

32. The method according to any one of claims 20-31, wherein, For performance monitoring of AI or ML functions or models, the second condition includes at least one of the following: One or more probabilities are less than the configured probability threshold; One or more ratios are less than the configured ratio threshold; or One or more quantities are less than the configured quantity threshold; Furthermore, the collected data includes data used for performance monitoring, prediction results used for performance monitoring metric calculations, data collected during the performance monitoring period, or data collected within a predetermined period.

33. The method according to claim 31 or 32, wherein, The one or more probabilities, the one or more ratios, or the one or more quantities are calculated based on predictions generated during a time period, wherein the time period is configured to be periodic and occurs periodically.

34. A method for wireless communication, comprising: Conditions for data collection are received by the wireless device; as well as The wireless device determines whether to perform the data collection based on the conditions.

35. The method according to claim 34, wherein, The wireless device performs the data collection, wherein the wireless device records information related to the association conditions, and wherein the wireless device indicates to the network device the availability of the collected data and the information related to the association conditions.

36. The method according to claim 34 or 35, wherein, The conditions include at least one of the following: The highly relevant conditions of the wireless device; The codebook-related conditions of the wireless device; The location change related conditions of the wireless device; or The mobility information of the wireless device.

37. The method of claim 36, wherein, The codebook-related conditions include at least one of the following: Identifiers for codebook configuration; Codebook type; Codebook subtype; Codebook mode; or Codebook parameters related to the number of beams, number of panels, limits, or number of antenna ports.

38. The method according to claim 36 or 37, wherein, The conditions related to the change of orientation include at least one of the following: The maximum value of the change in orientation; Minimum value of azimuth change; The maximum rate of change of orientation; The minimum rate of change of orientation; The maximum average rate of change of orientation; or The minimum average rate of change of orientation.

39. The method according to any one of claims 34-38, wherein, The collected data has been used at least for the training, testing, validation, or performance monitoring of artificial intelligence (AI) or machine learning (ML) functions or models.

40. The method according to any one of claims 34-39, further comprising: The wireless device receives a request from the network device for collecting data associated with one or more specific conditions; as well as The collected data is transmitted by the wireless device in response to the request.

41. A method for wireless communication, comprising: The prediction accuracy of the prediction results associated with the beam or reference signal (RS) is determined by the wireless device; as well as The wireless device determines whether to record or report the collected data based on the prediction accuracy.

42. The method according to claim 41, wherein, The prediction accuracy is greater than a configured threshold, and the wireless device does not record the collected data in local storage or report the collected data.

43. The method according to claim 41, wherein, The prediction accuracy is less than a configured threshold, and the wireless device records the collected data in local storage or reports the collected data.

44. An apparatus for wireless communication, comprising a processor, wherein, The processor is configured to implement the method according to any one or more of claims 1 to 43.

45. A computer-readable program storage medium having code stored thereon, which, when executed by a processor, causes the processor to perform the method of any one or more of claims 1 to 43.

Images