System and method for predictive timing advance for mobility enhancement based on artificial intelligence / machine learning

Abstract

Systems and methods for using various artificial intelligence (AI) / machine learning (ML) models with respect to various mobility aspects are described herein. Generation and use of L3 beam level measurement predictions, L3 cell level measurement predictions, L1 measurement predictions, network-based timing advance (TA) value predictions, and UE-based TA value predictions are discussed using corresponding ML models. Various examples of inputs that can be used with respect to these ML models are discussed. Various of these predictions are discussed to be used within mobility environments including Layer 3-based handover, Layer 1 / Layer 2 triggered mobility (LTM), and Conditional Handover (CHO).

Description

Technical Field

[0001] This application relates in general to wireless communication systems, including wireless communication systems capable of performing measurement and / or timing advance (TA) prediction. Background Technology

[0002] Wireless mobile communication technologies use various standards and protocols to transmit data between base stations and wireless communication devices. For example, wireless communication system standards and protocols may include, for instance, 3GPP Long Term Evolution (LTE) (e.g., 4G), 3GPP New Radio (NR) (e.g., 5G), and the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard for Wireless Local Area Networks (WLANs) (often referred to as Wi-Fi within the industry organization). ® ).

[0003] As envisioned by 3GPP, different wireless communication system standards and protocols can use various radio access networks (RANs) for communication between RAN base stations (sometimes referred to as RAN nodes, network nodes, or simply nodes) and wireless communication equipment called user equipment (UEs). 3GPP RANs can include, for example, Global System for Mobile Communications (GSM), Enhanced Data Rate GSM Evolution (EDGE) RAN (GERAN), Universal Terrestrial Radio Access Network (UTRAN), Evolved Universal Terrestrial Radio Access Network (E-UTRAN), and / or Next Generation Radio Access Network (NG-RAN).

[0004] Each RAN can use one or more Radio Access Technologies (RATs) to perform communication between the base station and the UE. For example, GERAN implements the GSM and / or EDGE RAT, UTRAN implements the Universal Mobile Telecommunications System (UMTS) RAT or other 3GPP RATs, E-UTRAN implements the LTE RAT (sometimes simply referred to as LTE), and NG-RAN implements the NR RAT (this NR RAT is sometimes referred to herein as the 5G RAT, 5G NR RAT, or simply NR). In some deployments, E-UTRAN may also implement the NR RAT. In some deployments, NG-RAN may also implement the LTE RAT.

[0005] The base stations used by a RAN can correspond to that RAN. An example of an E-UTRAN base station is an Evolved Universal Terrestrial Radio Access Network (E-UTRAN) Node B (also commonly referred to as Evolved Node B, Enhanced Node B, eNodeB, or eNB). An example of an NG-RAN base station is a Next Generation Node B (sometimes also called gNode B or gNB).

[0006] The RAN provides communication services to external entities through its connection with the core network (CN). For example, E-UTRAN can utilize the evolved packet core (EPC), while NG-RAN can utilize the 5G core network (5GC). Attached Figure Description

[0007] To facilitate the identification of any particular element or action in the discussion, one or more of the most significant digits in the figure reference numerals refer to the figure number in which the element was first introduced.

[0008] Figure 1 Example frameworks for using AI and / or ML in wireless communication system environments are illustrated.

[0009] Figure 2 An L3 measurement framework that can be used in a wireless communication system according to the implementation scheme discussed herein is illustrated.

[0010] Figure 3 A flowchart illustrating a UE-side process for L3 measurement prediction, as shown in the embodiment of this document, is provided.

[0011] Figure 4 A flowchart illustrating a two-sided process for L3 measurement prediction, such as between a UE and a network, is shown according to an embodiment of this document.

[0012] Figure 5A A first mechanism for time prediction of L3 cell-level measurements is illustrated according to the implementation scheme discussed herein.

[0013] Figure 5B A second mechanism for time prediction of L3 cell-level measurements is illustrated according to the implementation scheme discussed herein.

[0014] Figure 5C A third mechanism for time prediction of L3 cell-level measurements is illustrated according to the implementation scheme discussed herein.

[0015] Figure 5D A fourth mechanism for time prediction of L3 cell-level measurements is illustrated according to the implementation scheme discussed herein.

[0016] Figure 6A A first mechanism for time prediction of L3 beam-level measurements is illustrated according to the implementation scheme discussed herein.

[0017] Figure 6B A second mechanism for time prediction of L3 beam-level measurements is illustrated according to the implementation scheme discussed herein.

[0018] Figure 6CA third mechanism for time prediction of L3 beam-level measurements is illustrated according to the implementation scheme discussed herein.

[0019] Figure 7 A mechanism for spatial prediction of L3 beam-level measurements according to the implementation scheme discussed herein is illustrated.

[0020] Figure 8 A flowchart illustrating the LTM process between a UE and a base station of the network according to the implementation scheme discussed herein is provided.

[0021] Figure 9 A diagram illustrating the operation of an RSTD-based TA mechanism between a UE, a source cell, and a target cell, according to the implementation scheme discussed herein, is provided.

[0022] Figure 10 A flowchart illustrating a UE-side process for predicting L1 and / or TA measurements, as shown in the embodiments described herein, is provided.

[0023] Figure 11 A flowchart illustrating a two-sided process for predicting L1 and / or TA measurements, as shown in the embodiments described herein, is presented.

[0024] Figure 12 A mechanism for time prediction of L1 measurements according to the implementation scheme discussed herein is illustrated.

[0025] Figure 13 A mechanism for spatial prediction of L1 measurements according to the implementation scheme discussed herein is illustrated.

[0026] Figure 14 The diagram illustrates example scenarios of various TAs used between the UE and each of the first, second, and third cells 1408.

[0027] Figure 15 A mechanism for time prediction of RSTD-based TA measurements is illustrated according to the implementation scheme discussed herein.

[0028] Figure 16 A mechanism for spatial prediction of RSTD-based TA measurements is illustrated according to the implementation scheme discussed herein.

[0029] Figure 17A and Figure 17B A flowchart illustrating the implementation of early target time (TA) prediction in a system comprising a UE, a source base station communicating with the UE on a serving cell, a first target base station having a first target cell, a second target base station having a second target cell, and a server, according to the implementation scheme discussed herein, is illustrated.

[0030] Figure 18A and Figure 18B A flowchart illustrating conditional handover that can be used in some wireless communication systems is presented together.

[0031] Figure 19 A flowchart illustrating the CHO process for using measurement prediction according to the implementation scheme discussed herein, comprising a UE, a source base station communicating with the UE on the serving cell, a first target base station having a first target cell, a second target base station having a second target cell, and a server.

[0032] Figure 20 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0033] Figure 21 An example of a RAN method based on the implementation scheme discussed herein is provided.

[0034] Figure 22 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0035] Figure 23 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0036] Figure 24 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0037] Figure 25 An example of a RAN method based on the implementation scheme discussed herein is provided.

[0038] Figure 26 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0039] Figure 27 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0040] Figure 28 An example of a source base station method for a RAN according to the implementation scheme discussed herein is illustrated.

[0041] Figure 29 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0042] Figure 30 An example of a source base station method for a RAN according to the implementation scheme discussed herein is illustrated.

[0043] Figure 31 A method for a UE based on the implementation scheme discussed herein is illustrated.

[0044] Figure 32An example architecture of a wireless communication system according to the implementation scheme disclosed herein is illustrated.

[0045] Figure 33 A system for performing signaling between a wireless device and a network device according to an embodiment disclosed herein is illustrated. Detailed Implementation

[0046] Various implementations are described for the UE. However, references to the UE are provided for illustrative purposes only. The example implementations can be used with any electronic component capable of establishing a connection to a network and configured with hardware, software, and / or firmware for exchanging information and data with the network. Therefore, the UE as described herein is used to represent any suitable electronic component.

[0047] Frameworks for Artificial Intelligence / Machine Learning in Wireless Communication Systems Figure 1 Example framework 100 for using artificial intelligence (AI) and / or machine learning (ML) in the context of wireless communication systems is illustrated. The discussion in this paper involves the use of AI / ML models (sometimes referred to as "models" in this paper).

[0048] Framework 100 includes data collection functionality 102, model training functionality 104, management functionality 106, inference / prediction functionality 108, and model storage functionality 110.

[0049] As shown in the figure, the data collection function 102 can provide training data 112 to the model training function 104, monitoring data 114 to the management function 106, and / or inference / prediction data 116 to the inference / prediction function 108. The model training function 104 can provide trained / updated model signaling 124 to the model storage function 110. The management function 106 can provide performance feedback / retraining request signaling 122 to the model training function 104, model transfer / delivery request signaling 128 to the model storage function 110, and / or selection / (de)activation / toggle / rollback signaling 120 to the inference / prediction function 108. The inference / prediction function 108 can provide output monitoring signaling 118 to the management function 106. The model storage function 110 can provide model transfer / delivery signaling 126 to the inference / prediction function 108.

[0050] In framework 100, an AI / ML model can be trained at model training function 104 based on training data 112 received from data collection function 102. Once trained, the model can be provided to model storage function 110.

[0051] When the model is to be used, it is provided from the model storage functionality 110 to the inference / prediction functionality 108. The data collection functionality 102 may also provide inference / prediction data 116 (e.g., input data) to the inference / prediction functionality 108. The inference / prediction functionality 108 can then perform inference by applying the inference / prediction data 116 to the model. This inference may be reported to functionalities outside the framework 100 for further use.

[0052] The management function 106 manages the overall operation of the framework 100. Management decisions may be based on monitoring data 114 received at the management function 106 from the data collection function 102 and / or output monitoring signaling 118 received from the inference / prediction function 108. The management function 106 may, for example, provide training data 112 to the model training function 104 to notify the model training function 104 of the performance of the trained model and / or request retraining of the current model. The management function 106 may, for example, provide model transfer / delivery request signaling 128 to the model storage function 110 to control the transfer to the inference / prediction function 108 and the use of the model at the inference / prediction function. The management function 106 may, for example, control the inference / prediction function 108 by selecting / (de)activating / toggling / rollback signaling 120 to indicate the model to be used and / or the method of using the current model, etc.

[0053] Use cases of artificial intelligence / machine learning in wireless communication systems Regarding wireless communication systems, various use cases have been identified for studying useful applications of AI / ML models in categories related to physical layer (PHY) considerations. One such case is investigating the use of AI / ML models in channel state information (CSI) feedback environments with the aim of achieving CSI feedback enhancement. For example, CSI timing prediction using AI / ML models could be considered.

[0054] Another scenario involves beam management considerations. For example, layer 1 (L1) beam temporal / spatial prediction using AI / ML models could be considered.

[0055] Another scenario involves enhancing localization accuracy through the use of AI / ML models.

[0056] Regarding the use of AI / ML within wireless communication systems, various possible levels of UE / base station cooperation exist. For example, in some cases, there may be no cooperation between the UE and the base station regarding the use of ML models. In other cases, there may be signaling-based cooperation between the UE and the base station, but no ML model transfer between the base station and the UE (in which case auxiliary information may be used for ML model selection purposes, for example). In still other cases, there may be signaling-based cooperation between the UE and the base station, which includes the transfer of ML models used between the UE and the base station. At least some of the embodiments discussed herein apply to, for example, signaling-based cooperation scenarios (with or without model transfer).

[0057] Proposals for wireless communication systems may involve situations using AI / ML to enhance mobility. These situations can be divided into various sub-topics. For example, a first such sub-topic could be about AI / ML-based radio resource management (RRM) predictions, such as predictions of future L1 and / or Layer 3 (L3) measurements based on historical measurements. In such cases, the intention might be to reduce UE measurement workload and / or reduce latency in triggering measurement events.

[0058] In another example, another such sub-topic could be about AI / ML-based target cell selection, such as predicting and notifying the network which cell and / or beam and / or when to handover. In such cases, the intention might be to allow the UE not to report all its locally useful observations about the handover to the network, thereby helping the UE stay within given power and / or memory and / or privacy constraints.

[0059] In another example, another sub-topic could be about AI / ML-based fault avoidance, such as predicting and notifying the network of potential future radio link failures (RLF) / handover failures (HOF). In such cases, the intention might be to enable the network to proactively avoid RLFs, rather than reacting only after an RLF has occurred (based on some existing passive mechanisms).

[0060] Other use cases for beneficial applications of AI / ML include, but are not limited to, AI / ML-based UE trajectory prediction, AI / ML-based discontinuous reception (DRX) adaptation, AI / ML-based slicing / QoE mechanisms, and / or AI / ML-based cell reselection mechanisms.

[0061] Therefore, it can be seen that there are multiple proposals to explore AI / ML-based mobility enhancement. This paper discusses details of various implementation schemes for such AI / ML-based mobility enhancement.

[0062] In some implementations described herein, the UE may use an ML model trained at the UE based on its mobility and mobility-related information. In some cases, the UE may inform the network of its predictions of the optimal target cell and / or beam for a regular HO (House of Interest) based on its use of the ML model. In some cases, the UE may inform the network of its predictions of a list of candidate cells for a Conditional Handover (CHO) recommendation / rejection based on its use of the ML model. In some cases, the UE may be able to predict an impending RLF (Restricted Life Failure) based on its use of the ML model and notify the network in advance.

[0063] Example L3 measurement frame Figure 2 An L3 measurement framework 200, which can be used in a wireless communication system according to the embodiments discussed herein, is illustrated. The L3 measurement framework 200 can be, for example, a measurement framework used at a UE in an NR wireless communication system.

[0064] Initially, the results sensed from each of the multiple monitored gNB (base station) beams undergo L1 beam filter 202. This means that, regarding the number... K 1 beam, each beam (from 1 to 1) K The results can first be processed using an L1 filter. As shown in the figure, the specific implementation of the L1 filter can be specific to the UE / UE type / UE manufacturer. Then, as shown in the figure, the per-beam result of the L1 filter enters two different stages: L3 cell-level measurement stage 204 and L3 beam-level measurement stage 206.

[0065] exist Figure 2 The upper right portion illustrates the L3 cell-level measurement phase 204. In the L3 cell-level measurement phase 204, the L1 filtering results for each beam are first combined 208 into a single cell-level value by linear averaging. This cell-level value is then passed to the corresponding L3 filter 210 to generate an output. The output can be reported when it passes certain reporting criteria 212 at the UE. As shown, the parameters used for / during this process can be configured by the network to the UE (e.g., via Radio Resource Control (RRC)).

[0066] L3 beam-level measurement phase 206 Figure 2 The lower right portion is illustrated. In the L3 beam-level measurement phase 206, the L1 filtering result of each beam undergoes a separate L3 beam filter 214, and the UE can then select 216 the qualified beams from these beams (e.g., X Beam, in which X ≤ KThe output is used for this process. As shown in the figure, the parameters used for this process / during this process can be configured by the network to the UE (e.g., via RRC configuration).

[0067] In some wireless communication systems, L3 measurements may be configured to occur on a measurement-object basis, where the measurement object is frequency-based (rather than cell-based). Therefore, in such cases, if L3 beam reporting is configured (e.g., in...), reportConfigNR In the Information Element (IE), the UE will apply the same measurement configuration to all cells using the same frequency (e.g., the UE will generate and transmit L3 beam measurement reports for all cells using the same frequency), even for such cells with poor cell quality.

[0068] Correspondingly, in some wireless communication systems, measurement reporting can be implemented using a significant amount of signaling overhead. For example, even in the case of measurement reporting for a single neighboring cell, up to 3572 bits of overhead can be used because multiple measurement quantities (e.g., Reference Signal Received Power (RSRP) / Reference Signal Received Quality (RSRQ) / Signal-to-Interference-plus-Noise Ratio (SINR)) and multiple reference signal types (e.g., Synchronization Block (SSB) / Channel State Information Reference Signal (CSI-RS)) can be configured for reporting in such cases (and further note that in such cases, up to 64 SSBs / CSI-RSs can be configured for reporting).

[0069] Overview of L3 cell-level and L3 beam-level measurement predictions Therefore, the implementation schemes discussed herein may relate to solutions involving the use of predicted L3 cell-level measurements and / or predicted L3 beam-level measurements. Benefits derived from using such L3 measurement predictions may include, for example, an overall reduction in UE measurement reporting workload corresponding to L3 measurement-related scenarios. For instance, the UE can (e.g., only) report on a number of... N The UE performs L3 measurements on the top cell. Then, the UE can perform predictions (rather than actual measurements) on other cells (and when its prediction is that the cell has entered the top cell). N When dealing with individual cells, you can simply return to perform the actual L3 measurements for those cells.

[0070] Another benefit derived from using L3 measurement prediction is a reduction in the time-to-trigger (TTT) of measurement events, which can, for example, correspondingly reduce the UE's home latency. This can be achieved by configuring measurement events to be triggered based on predicted L3 cell measurements rather than waiting for corresponding actual L3 cell measurements.

[0071] Another benefit derived from using L3 measurement prediction is the reduction in the use of measurement gaps and / or the use of measurement gaps with relatively reduced duration. Instead of implementing measurement gaps to achieve actual L3 measurements for those cells, the UE can perform L3 measurement predictions about one or more cells (e.g., inter-frequency cells). This reduction in the use of measurement gaps allows the UE to utilize more channel resources for data transmission (which can be particularly useful when the UE has pending data for transmission).

[0072] Therefore, this paper discusses various aspects of L3 measurement prediction in its implementation. The first aspect is the overall process between the base station and the UE for using L3 measurement prediction. Another such aspect involves the mechanisms for performing inferences about predicted L3 cell measurements. For example, the use of each of the UE-side and dual-side models to perform L3 cell-level and L3 beam-level measurement predictions is discussed (note that the network-side model allows the UE to report local datasets to the network for model training). Yet another such aspect involves performance monitoring of the ML model's results at the base station and / or UE (as well as other corresponding lifecycle monitoring (LCM) aspects of the ML model). Furthermore, the aspects discussed include auxiliary information that can be passed between the UE and the network / base station in these environments.

[0073] UE-side process for L3 measurement prediction This paper accordingly discusses the details of the generation and use of AI / ML-based L3 measurement predictions for mobility enhancement at the UE. Figure 3 A flowchart 300 illustrates a UE-side procedure for L3 measurement prediction, such as between UE 302 and network 304, according to an embodiment of this document. It should be noted that in some embodiments, the UE-side procedure for L3 measurement prediction may also be combined with the use of a UE server 306, as will be discussed.

[0074] like Figure 3 The UE-side procedure shown for L3 measurement prediction can correspond to, for example, L3 cell-level measurement prediction and / or L3 beam-level measurement prediction.

[0075] Flowchart 300 begins with UE 302 generating a UE capability report 308 and sending it to network 304. In some implementations, the UE capability report 308 may include one or more of the following: whether the UE supports L3 cell-level time measurement prediction; whether the UE supports L3 beam-level time measurement prediction; whether the UE supports L3 cell-level spatial measurement prediction; whether the UE supports L3 beam-level spatial measurement prediction; - the maximum number of historical samples / time slots of a prediction that can be used at the UE / by the UE; the maximum number of prediction samples / time slots that can be used at the UE / by the UE; - the maximum number of parallel predictions that can be used at the UE / by the UE.

[0076] Then, network 304 provides training configuration 310 to UE 302. Training configuration 310 may include one or more of the following: the type of ML model to be trained (e.g., a Long Short-Term Memory (LSTM) ML model type, a Recurrent Neural Network (RNN) ML model type, etc.), the layers to be trained and / or one or more specialized ML models to be used; the window length corresponding to the history of measurements and / or predictions in the time and / or spatial domains to be used with the ML model; and / or the number of parallel predictions that should be provided by UE 302.

[0077] Then, UE 302 performs data collection 312. In some embodiments, this process incorporates the generation / training of an ML model at the UE using the collected data. In some embodiments, the UE provides the collected data to UE server 306, such that offline training 314 (e.g., generation of the ML model) occurs instead at UE server 306, and the UE server then provides the ML model thus generated back to UE 302.

[0078] Then UE 302 transmits notification message 316 to network 304. The content of notification message 316 may notify network 304 which ML models (and in at least some cases, the model identifiers (IDs) corresponding to these ML models) are available at UE 302.

[0079] The content of notification message 316 can notify network 304 of model suitability conditions, which network 304 can use to determine which ML model to use at UE 302. It should be noted that model suitability conditions may include, for example, usage scenario information (e.g., indoor / outdoor), antenna type information, channel type information, UE speed information (e.g., UE travels less than 5 km / h (kmph)), UE altitude information (e.g., corresponding to UE movement at an elevation), etc.

[0080] It should be noted that notification message 316 may be included, for example, as part of a scheduling request (SR), as part of uplink auxiliary information (UAI), in a media access control control element (MAC-CE), or in the transmission of RRC messages (e.g., in...). RRCReconfigurationComplete (In the message or the new supply RRC message) to provide.

[0081] Based on notification message 316, network 304 can determine which ML model to activate at UE 302 and can provide UE 302 with activation message 318 commanding UE 302 to activate the selected ML model. Activation message 318 may be provided, for example, as part of downlink control information (DCI), MAC-CE, or RRC message reception.

[0082] It should be noted that, Figure 3 In the alternative embodiment shown, UE 302 may instead directly inform network 304 which ML model it prefers to use or will use. In such a case, UE 302 may determine its preferred / used model based on information such as UE 302's speed and / or the channel conditions experienced by UE 302.

[0083] Then, UE 302, based on the configuration of network 304, continues to perform inference / prediction 320 for L3 cell-level measurements and / or L3 beam-level measurements (as discussed). Inference / prediction 320 may be based on actual measurements already performed at the UE.

[0084] UE 302 may be / has been configured (e.g., by network 304) to trigger measurement report 322 based on one or more of the actual L3 measurement and / or predicted L3 measurement. Measurement report 322 can be transmitted once the appropriate values ​​of the actual and / or predicted measurements are determined at the UE. Measurement report 322 may report to the base station any / both of the actual and / or predicted L3 cell-level and / or beam-level measurements (which can be done using, for example...). Measurement Results The message is coming.

[0085] It is conceivable that, regarding the UE-side process, UE 302 or network 304 may perform performance monitoring 324 of the ML model (e.g., by comparing predicted measurements with corresponding actual measurements). Based on the results of performance monitoring 324, UE 302 or network 304 may initiate LCM signaling 326 for model switching or model deactivation, which then leads to model switching / deactivation 328 at the UE. In the case of deactivation, UE 302 and network 304 may then fall back to a non-AI / ML-based measurement reporting solution.

[0086] Two-sided process for L3 measurement prediction Figure 4A flowchart 400 illustrates a two-sided process for L3 measurement prediction, such as between UE 402 and network 404, according to an embodiment of this document. It should be noted that in some embodiments, the UE-side process for measurement prediction may also be combined with the use of UE server 406.

[0087] like Figure 3 The two-sided process shown for L3 measurement prediction can correspond to, for example, L3 cell-level prediction and / or L3 beam-level prediction.

[0088] UE capability report 308, training configuration 310, data collection 312, and offline training 314 can all be found in this article. Figure 3 The discussion regarding the UE-side process describes how it occurs.

[0089] Once the ML model exists at UE 402, a model transfer occurs at 408. During the model transfer, UE 402 transmits the ML model to the network at 404. The model transfer signaling can be RRC-based or DRB-based.

[0090] Then, UE 402 may generate actual L3 cell-level and / or beam-level measurements (e.g.) and transmit a measurement report 410 with these actual measurements to network 404.

[0091] Network 404 can then apply these actual measurements along with the previously received ML model to perform L3 cell-level and / or L3 beam-level measurement predictions 412 (as applicable). In some implementations, the detailed prediction method may depend on the specific implementation of network 404.

[0092] Furthermore, performance monitoring 414 may occur at network 404 (e.g., by comparing the actual L3 cell-level and / or beam-level measurements received from UE 402 with the corresponding predicted L3 cell-level and / or beam-level measurements).

[0093] Network 404 can also be configured to trigger ML model retraining performance based on its specific implementation (which may, for example, make the decision at least in part based on the results of performance monitoring 414). As part of triggering this retraining, network 404 may provide the UE with a retraining configuration 416 that commands retraining (and may provide one or more parameters for the UE to analyze / use as part of the ML model retraining process).

[0094] In response to retraining configuration 416, in some implementations, the UE continues to perform data collection 312, offline training 314, and model transfer 408 again, as previously described.

[0095] Example mechanism for L3 cell-level measurement prediction In some implementations, in order to provide a flexible trade-off between UE measurement burden and mobility performance, the UE may be configured (e.g., via RRC signaling) to use one of a variety of alternatives for L3 cell-level measurement prediction by an ML model (which may be referred to as a "measurement prediction model").

[0096] In a first-class implementation for L3 cell-level measurement prediction, the L3 cell-level measurement prediction may correspond to time prediction. For example, the measurement prediction model can be used to predict future L3 cell-level measurements based on the current inputs to the measurement prediction model.

[0097] Figure 5A A first mechanism 502 for time prediction of L3 cell-level measurements according to the implementation scheme discussed herein is illustrated. One or more actual L3 cell-level measurements 504 may be performed at the UE. The measurement prediction model may be configured to receive these one or more actual L3 cell-level measurements 504 as input and, in response, provide one or more predicted L3 cell-level measurements 506 (where each of the one or more predicted L3 cell-level measurements 506 corresponds to a later time).

[0098] Figure 5B A second mechanism 508 for time prediction of L3 cell-level measurements, according to the implementation scheme discussed herein, is illustrated. One or more actual L1 beam-level measurements can be performed. Figure 5B In the example, the UE obtains the first actual L1 beam level measurement 510 of the first beam and the second actual L1 beam level measurement 512 of the second beam.

[0099] The UE then uses a measurement prediction model to predict one or more predicted L1 beam-level measurements. Figure 5B In the example, the UE uses the first actual L1 beam level measurement 510 and the second actual L1 beam level measurement 512 with the measurement prediction model to generate the first predicted L1 beam level measurement 514 and the second predicted L1 beam level measurement 516.

[0100] Then, the UE performs a linear averaging of 518 on one or more predicted beam-level measurements. Regarding Figure 5A For example, the UE performs a linear average on the first predicted L1 beam level measurement 514 and the second predicted L1 beam level measurement 516.

[0101] The result of the linear averaging 518 is then subjected to L3 filtering 520 to generate one or more predicted L3 cell-level measurements 522. L3 filtering 520 may occur based on network-configured (e.g., via RRC signaling) L3 filter coefficients, as shown. In some implementations, the L3 filter coefficients that may be used for L3 filtering are generated by a measurement prediction model (e.g., based on its reception of a first actual L1 beam-level measurement 510 and a second actual L1 beam-level measurement 512).

[0102] Figure 5C A third mechanism 524 for time prediction of L3 cell-level measurements, according to the implementation scheme discussed herein, is illustrated. One or more actual L1 beam-level measurements can be performed. Figure 5C In the example, the UE obtains the first actual L1 beam level measurement 526 of the first beam and the second actual L1 beam level measurement 528 of the second beam.

[0103] Then, the UE performs a linear averaging of 530 on one or more actual L1 beam-level measurements. Figure 5C In the example, the UE performs a linear averaging 530 on the first actual L1 beam level measurement 526 and the second actual L1 beam level measurement 528.

[0104] The result of this linear average (e.g., Figure 5C The derived actual cell-level L3 measurements (532) and the configured (e.g., RRC-configured) set of L3 filter coefficients are applied to the measurement prediction model. The measurement prediction model then uses this information to generate one or more predicted L3 cell-level measurements (534).

[0105] In such implementations, the dwell / effective time that may be used for prediction is also calculated by the measurement prediction model and provided as a result along with one or more predicted L3 cell-level measurements 534.

[0106] Figure 5D A fourth mechanism 536 for time prediction of L3 cell-level measurements, according to the implementation scheme discussed herein, is illustrated. One or more actual L1 beam-level measurements can be performed. Figure 5D In the example, the UE acquires one or more first actual L1 beam level measurements 538 of the first beam and one or more second actual L1 beam level measurements 540 of the second beam.

[0107] One or more actual L1 beam-level measurements (e.g., one or more first actual L1 beam-level measurements 538 and one or more second actual L1 beam-level measurements 540) and L3 coefficients 542 (e.g., L3 coefficients configured by RRC) are provided to a measurement prediction model 544, which uses these terms to generate one or more predicted L3 cell-level measurements 546.

[0108] In such cases, using L3 coefficient 542 to compensate for the fact that L1 measurements (such as one or more first actual L1 beam level measurements 538 and one or more second actual L1 beam level measurements 540) may be less stable than the corresponding L3 measurements (and given the understanding that the UE does not generate more stable actual L3 measurements itself in this case).

[0109] In a second type of implementation for L3 cell-level measurement prediction, L3 cell-level measurement prediction corresponds to spatial prediction. For example, if base station deployment geometry and / or long-term channel time statistics (e.g., correlation) are available, the UE can use a measurement prediction model to infer L3 measurements of neighboring cells based on nearby deployments (e.g., correlation information) and L3 cell measurements of the current cell.

[0110] It should be noted that configuration information (e.g., RRC configuration information) can be provided to instruct the UE to perform L3 cell-level measurement predictions regarding the use of the measurement prediction model. For example, the UE can be configured to generate L3 cell-level measurement predictions for a specific cell (e.g., the current serving cell and / or neighboring cells).

[0111] As another example of using L3 cell-level measurement prediction based on configuration information, a UE can be configured to generate L3 cell-level measurement predictions for a specific frequency.

[0112] As another example of using L3 cell-level measurement prediction based on configuration information, a UE can be configured to generate L3 cell-level measurement predictions based on one or more conditions.

[0113] In the first example using conditions for predicting L3 cell-level measurements, the UE can be configured to generate measurements for at most a number of... N L3 cell-level measurements for each cell, where the UE generates measurements for the number of cells previously known to have the strongest RSRP / RSRQ. M One community ( M < N Actual L3 cell-level measurements, and further generate data for the remaining cells. N – M The predicted L3 cell-level measurements for each cell. The predicted L3 cell-level measurements fall within... M In the case of the highest measurement, the UE will then perform actual L3 cell-level measurements on subsequent cells (and the previous ones). M The last cell in the set is added as an alternative to use the predicted location. N – M gather).

[0114] In the second example using conditions for using predicted L3 cell-level measurements, a configured RSRP / RSRQ / SINR threshold can be compared with the actual or predicted L3 cell measurement. If the actual or predicted L3 cell-level measurement of a cell is less than the threshold, the UE performs L3 cell-level prediction for subsequent cells. In a variation of this scenario, the base station may configure the UE to use separate thresholds for the use of actual and predicted L3 cell-level measurements.

[0115] In the third example of using conditions for predicting L3 cell-level measurements, it's possible that L3 cell-level measurement prediction can be performed even when interference in the cell is strong. For example, prediction can be used when the interference in the cell exceeds a configured interference measurement threshold.

[0116] In the fourth example using the conditions for using predicted L3 cell-level measurements, L3 cell-level measurement prediction can be performed for all or some of the indicated inter-frequency measurements.

[0117] In the fifth example using conditions for predicting L3 cell-level measurements, L3 cell-level measurement prediction can be performed if the measurement uses a measurement gap (e.g., for a measurement of another frequency or another non-overlapping bandwidth portion (BWP)).

[0118] In the sixth example of using conditions for employing predicted L3 cell-level measurements, the use of L3 cell-level measurement predictions may depend on the UE's mobility level. For example, the UE may use L3 cell-level measurement predictions when moving at very low speeds.

[0119] As another example of using L3 cell-level measurement prediction based on configuration information, the configuration information may indicate neighboring cells and / or frequencies that allow the UE to autonomously / independently select to generate actual L3 cell-level measurements or predicted L3 cell-level measurements.

[0120] As another example of using L3 cell-level measurement prediction based on configuration information, the configuration information can instruct the UE to use L3 cell-level measurement prediction for one or more indicated cells.

[0121] As another example of using L3 cell-level measurement prediction based on configuration information, the configuration information can instruct the UE to use L3 cell-level measurement prediction for one or more indicated frequencies.

[0122] Example mechanism for predicting L3 beam level measurements In some implementations, in order to provide a flexible trade-off between UE measurement burden and mobility performance, the UE may be configured (e.g., via RRC signaling) to use one of a variety of alternatives for L3 beam-level measurement prediction by an ML model (which may be referred to as a "measurement prediction model").

[0123] In a first-class implementation for L3-beam level measurement prediction, the L3-beam level measurement prediction may correspond to a time prediction. For example, the measurement prediction model can be used to predict future L3-beam level measurements based on the current inputs to the measurement prediction model.

[0124] Figure 6A A first mechanism 602 for timing prediction of L3 beamlevel measurements according to the embodiments discussed herein is illustrated. One or more actual L3 beamlevel measurements 604 may be performed at the UE. The measurement prediction model may be configured to receive these one or more actual L3 beamlevel measurements 604 and optimized L3 filter coefficients as input, and in response provide one or more predicted L3 beamlevel measurements 606 (each of the one or more predicted L3 beamlevel measurements 606 corresponds to a later time).

[0125] In such implementations, the dwell / effective time that may be used for prediction is also calculated by the measurement prediction model and provided as a result along with one or more predicted L3 beam-level measurements 606.

[0126] Figure 6B A second mechanism 608 for time prediction of L3 beam-level measurements, according to the implementation scheme discussed herein, is illustrated. One or more actual L1 beam-level measurements can be performed. Figure 6B In the example, the UE performs actual L1 beam-level measurements of the first beam 610.

[0127] The UE then uses a measurement prediction model to predict one or more predicted L1 beam-level measurements. Figure 6B In the example, the UE uses the actual L1 beam level measurement 610 with the measurement prediction model to generate the predicted L1 beam level measurement 612.

[0128] Then, an L3 filter 614 is applied to one or more predicted L1 beam-level measurements 612 to generate one or more predicted L3 beam-level measurements 616. The L3 filter 614 can be generated according to the L3 filter coefficients configured by the network (e.g., via RRC signaling), as shown in the figure.

[0129] Figure 6C A third mechanism 618 for time prediction of L3 beam-level measurements is illustrated according to the implementation scheme discussed herein. One or more actual L1 beam-level measurements can be performed. Figure 6C In the example, the UE performs the actual L1 beam level measurement 620 for the first beam.

[0130] One or more actual L1 beam-level measurements 620 and L3 coefficients 622 (e.g., L3 coefficients configured for RRC) are provided to a measurement prediction model 624, which uses these terms to generate one or more predicted L3 beam-level measurements 626.

[0131] In such cases, using the L3 coefficient 622 can compensate for the fact that the actual L1 beam level measurement 620 may not be as stable as the corresponding L3 measurement (and given the understanding that the UE does not generate a more stable actual L3 measurement itself in this case).

[0132] In a second type of implementation for L3 beam-level measurement prediction, L3 beam-level measurement prediction corresponds to spatial prediction. For example, a UE can use its understanding of applicable spatial channel statistics to predict / infer the (predicted) L3 beam-level measurement of a beam based on the actual or predicted L3 beam-level measurement of its neighboring beams.

[0133] Figure 7 A mechanism 700 for spatial prediction of L3 beam-level measurements is illustrated according to the implementation scheme discussed herein. Figure 7 An example of a spatial beam arrangement 702 for each of the first beam (“beam 1”), the second beam (“beam 2”), and the third beam (“beam 3”) is shown.

[0134] It can perform one or more actual L3 beam-level measurements. Figure 7 In the example, the UE performs one or more first actual L3 beam level measurements 704 for the first beam (beam 1) and one or more second actual L3 beam level measurements 706 for the third beam (beam 3).

[0135] One or more actual L3 beamlevel measurements (e.g., one or more first actual L3 beamlevel measurements 704 and one or more second actual L3 beamlevel measurements 706) are provided to a measurement prediction model 708, which uses these terms to generate one or more predicted L3 beamlevel measurements 710 for a second beam (beam 2, as shown, which is a neighboring beam of beams 1 and 3). It should be noted that in some embodiments, the measurement prediction model 708 may use applicable spatial channel statistics to generate one or more predicted L3 beamlevel measurements 710.

[0136] It should be noted that configuration information (e.g., RRC configuration information) can be provided to instruct the UE to perform L3 beam-level measurement predictions regarding the use of the measurement prediction model. For example, the UE can be configured to generate L3 beam-level measurement predictions for a specific beam of a particular cell (e.g., the current serving cell and / or neighboring cells).

[0137] As another example of using L3 beam-level measurement prediction based on configuration information, the UE can be configured to generate L3 beam-level measurement prediction based on one or more conditions.

[0138] In the first example using conditions for predicting L3 beamlevel measurements, the UE can be configured to generate measurements for up to a maximum number of... N L3 beam-level measurements for each beam, where the UE generates measurements for the number of previously known beams with the strongest RSRP / RSRQ. M beams ( M < N Actual L3 beam-level measurements, and further generation of measurements for the remaining beams. N – M The predicted L3 beam level measurement for each beam. The predicted L3 beam level measurement falls within... M In the case of the highest measurement, the UE will then perform actual L3 beam-level measurements on subsequent beams (and the previous ones). M The last beam in the ensemble is added alternatively using the predicted beam. N – M gather).

[0139] In the second example using conditions for using predicted L3 beamlevel measurements, a configured RSRP / RSRQ / SINR threshold can be compared with the actual or predicted L3 beamlevel measurement. If the actual or predicted L3 beamlevel measurement of the cell is less than the threshold, the UE performs L3 beamlevel prediction for the beam. In a variation of this, the base station may configure the UE to use separate thresholds for the use of actual and predicted L3 beamlevel measurements.

[0140] In the third example using conditions for using predicted L3 beam-level measurements, two thresholds can be used. A first configured RSRP threshold can be compared to the actual / predicted L3 cell-level measurements of the cell. If the actual or predicted L3 cell-level measurement of the cell is greater than the threshold, the UE can perform L3 beam-level measurement prediction for one or more beams in that cell. A second configured RSRP / RSRQ threshold can then be compared to the actual or predicted L3 beam-level measurements of the beams in that cell to determine whether beams in subsequent cells use actual or predicted L3 beam-level measurements. In a variation of this scenario, the base station may configure the UE to use separate thresholds for the use of actual and predicted measurements.

[0141] In the fourth example using the conditions for using predicted L3 beam-level measurements, L3 beam-level measurement prediction can be performed for all or some of the indicated inter-frequency measurements.

[0142] In the fifth example using conditions for predicting L3 beam-level measurements, if the measurement uses a measurement gap (e.g., for a measurement at another frequency or another non-overlapping BWP), L3 beam-level measurement prediction can be performed.

[0143] In the sixth example of using conditions for employing predicted L3 beam-level measurements, the use of L3 beam-level measurement prediction may depend on the UE's mobility level. For example, the UE may use L3 beam-level measurement prediction when moving at very low speeds.

[0144] As another example of using L3 beam-level measurement prediction based on configuration information, the configuration information may indicate neighboring cells and / or frequencies that allow the UE to autonomously / independently select to generate actual L3 beam-level measurements or predicted L3 beam-level measurements.

[0145] Implementation scheme for reporting actual and / or predicted L3 measurements The goal is to support periodic and event-triggered L3 measurement reports.

[0146] In some implementations, for periodic measurement reporting, the UE can be configured by the network to have two periodicities in a single reporting configuration. The first periodicity can be a predicted measurement periodicity, indicating how often the UE should perform and / or report AI / ML-based L3 measurement predictions. The second periodicity can be an actual measurement periodicity, indicating how often the UE should perform and / or report actual L3 measurements. In some such implementations, the predicted measurement periodicity may be less than the actual measurement periodicity, allowing the UE to use L3 measurement predictions most of the time (e.g., to reduce power) while still occasionally reporting actual L3 measurements to provide more accurate updates / information available for model monitoring.

[0147] In some implementations, for event-triggered reporting, the UE can be configured to trigger a measurement reporting event using actual and / or predicted L3 measurements (e.g., events A1-A6 as understood in a 3GPP NR wireless communication system). In the first case, only actual L3 cell-level measurements may trigger a measurement reporting event.

[0148] In the second scenario of event-triggered reporting, a measurement reporting event may be triggered by an actual or predicted L3 measurement. In this second scenario, the UE can also be configured to trigger a measurement report based on a predicted L3 measurement when the confidence level for the predicted measurement is greater than a threshold.

[0149] In either case, once the measurement reporting event is triggered, the UE can report available predicted and / or actual L3 cell-level and / or beam-level measurements as part of the measurement report. Within the measurement report, the UE can indicate which cell and / or beam measurements are predictive measurements and / or which cell and / or beam measurements are actual measurements. In at least some embodiments, the UE can also indicate the reliability probability or confidence level of any predictive measurements when they are included in the measurement report.

[0150] For both periodic measurement reporting and event-triggered measurement reporting, the UE may first report actual cell and / or beam measurements, and then report predicted cell and / or beam measurements in the order of first providing the number of measurements from high to low (e.g., RSRP), and then providing the confidence level from high to low (e.g., corresponding to any predictive measurement).

[0151] In some implementations, the total number of parallel predictions may not exceed the UE's capabilities. In one such implementation, the selection of favorable conditions for multiple parallel predictions may depend on the specific implementation of the base station. In another such implementation, the UE may be allowed to discard measurements in the following order: first, beam-level measurements are discarded; then, any beam-level and / or cell-level measurements with poor radio conditions are discarded; and finally, any predictive beam-level and / or cell measurements corresponding to poor confidence levels are discarded.

[0152] Model monitoring and LCM for L3 measurement prediction Imagine a UE-side process for L3 measurement prediction where model monitoring (e.g., monitoring the performance of the ML model) can be performed at the UE and / or at the base station. In the case of performing model monitoring at the base station, the process may depend on the specific implementation at the base station.

[0153] In cases where model monitoring is performed at the UE, in some such implementations, the model monitoring metric used by the UE may be the error between the predicted L3 cell / beam level measurement and the corresponding actual L3 cell / beam level measurement. For example, the mean square error (MSE) between some predicted L3 cell / beam level measurements and the corresponding actual L3 cell / beam level measurements may be used (and it should be noted that using the MSE metric in this way could be an example of a “confidence level” as discussed herein). For monitoring purposes, the UE may perform both the predicted L3 cell / beam level measurements and the corresponding actual L3 cell / beam level measurements for a small set of cells and / or beams (as applicable).

[0154] It is conceivable that the ML model in use may be switched from time to time (e.g., a different ML model may be selected for use and / or the use of an ML model may be paused or stopped for prediction). This may occur, for example, based on the results of model monitoring, as will be described. The UE may be configured to perform UE-initiated model switching or network-initiated model switching.

[0155] In the case of a UE-initiated handover, the UE can be configured with conditions and UE behavior related to model metrics (e.g., confidence levels). For example, the UE can be configured with an MSE (Mean Sequence Size) threshold of 0.01. When the MSE is greater than 0.01, the UE can be configured to fall back to using the actual measurement.

[0156] In the event of a network-initiated handover, the UE can report model monitoring metrics (e.g., confidence level) and may wait for base station LCM signaling in response (e.g., instructions to stop using the ML model and / or start using a different ML model). The UE can report model monitoring metrics via UAI or MAC-CE.

[0157] The two-sided process for L3 measurement prediction can be performed using model monitoring at the base station. This monitoring can be implemented according to the specific base station requirements.

[0158] Regarding the UE-side and / or dual-side processes for L3 measurement prediction monitored by the base station using ML models, the UE may be configured to provide information to the base station to assist the base station in performing monitoring. This information may include additional temporal information, such as timestamps used for prediction and corresponding timestamps of actual measurements. This information may also / optionally include additional spatial information, such as the UE's actual location, the UE's actual movement orientation, changes in the UE's movement orientation, and / or incremental directions (differences) that can be compared with the prediction.

[0159] Auxiliary information for L3 measurement prediction The auxiliary information used for L3 measurement prediction may include auxiliary information transmitted from the UE to the base station. Additionally, the auxiliary information used for L3 measurement prediction may / optionally include auxiliary information transmitted from the base station to the UE.

[0160] The auxiliary information transmitted by the UE to the base station (e.g., to the network) may include, but is not limited to: the predicted optimal L3 filter coefficients; the predicted optimal measurement report event type; the predicted optimal trigger time (TTT) for MR events; the predicted optimal threshold for measurement report events; one or more suggested cells for actual measurements; one or more suggested beams for actual measurements; and / or suggested T304 timer values.

[0161] Notification messages (e.g., one or more ML models at the network identifier UE) can be used to transmit auxiliary information from the UE to the base station.

[0162] The auxiliary information transmitted from the base station (e.g., the network) to the UE may include, but is not limited to: information about the deployment geometry of nearby base stations; long-term statistics on temporal correlation; and / or long-term statistics on inter-cell correlation and / or inter-beam correlation.

[0163] Downlink (DL) messages can be used to transmit auxiliary information from the base station to the UE. This message can be a MAC-CE or RRC message (e.g., RRCReconfigurationComplete (Message or new RRC message).

[0164] L1-L2 Triggered Mobility Implementation Plan The L1-L2 Triggered Mobility (LTM) procedure can be used in some radio systems, such as NR Release 18 (Rel-18). LTM stands for UE Mobility Mechanism in the system, which is based on L1 measurements at the UE / L1 measurement reports from the UE (rather than, for example, L3 measurements at the UE / L3 measurement reports from the UE).

[0165] Figure 8 A flowchart 800 illustrates the LTM process between UE 802 and base station 804 of the network according to the implementation scheme discussed herein. The LTM process represented by flowchart 800 anticipates an LTM preparation phase 806, an early synchronization phase 808, an LTM execution phase 810, and an LTM completion phase 812.

[0166] During the LTM preparation phase 806, it is anticipated that UE 802 is in RRC connected mode 814, and a measurement report 816 is transmitted to base station 804. Based on the measurement report 816, base station 804 performs LTM candidate preparation 818 (e.g., base station 804 selects one or more cells from the measurement report 816 to configure as LTM candidates). Then, base station 804 transmits LTE candidate configuration information (information about one or more LTM candidates selected by base station 804) to UE 802. RRCReconfiguration Message 820. UE 802 uses RRCReconfigurationComplete Message 822 responded RRCReconfiguration Message 820.

[0167] It should be noted that LTM candidate cell configurations can be added, modified, and / or released by the network via RRC signaling. Each LTM candidate cell configuration can be provided as an incremental configuration relative to a reference configuration.

[0168] The early synchronization phase 808 of the LTM process anticipates that the UE 802 will perform DL / UL synchronization 824 with the candidate cells, preparing for potential mobility to one or more of those candidate cells during the LTM execution phase 810.

[0169] The LTM execution phase 810 of the LTM process anticipates using L1 beam-level measurements (e.g., RSRP / RSRQ) regarding reference signals (e.g., SSB or CSI-RS) on those beams. L1 measurement information is provided by UE 802 to base station 804 in an L1 measurement report 826. Base station 804 can be configured to make an LTM decision 828 based on the information in the L1 measurement report 826. The LTM decision 828 can be a decision to instruct UE 802 on mobility to candidate cells based on the information in the L1 measurement report 826 (as previously stated in...). RRCReconfiguration (Configured for UE 802 in message 820). Corresponding to LTM decision 828, base station 804 transmits cell handover command 830 to UE 802 via MAC-CE. This cell handover command indicates / identifies the selected LTM candidate cell configuration for the selected candidate cell.

[0170] Then, the UE leaves the source cell 832 and applies the identified configuration to the selected candidate / target cells. The UE 802 further initiates a Random Access Channel (RACH) procedure 834 with these cells.

[0171] LTM Completion 836 of LTM Completion Phase 812 corresponds to the end of the LTM process, at which point the UE has completed mobility to the indicated candidate cell.

[0172] The LTM procedure can support candidate target cell TA acquisition via Early Timing Advance (TA) acquisition or RSTD-based TA acquisition. For example, if the TA of the candidate target cell is indicated in the MAC-CE that triggers HO / mobility, RACH-free communication between UE 802 and the candidate cell of the LTM procedure may be allowed / enabled.

[0173] The LTM procedure also enables fault handling. The UE can start an LTM supervisory timer upon receiving a cell handover command. Upon successful LTM cell handover, the UE stops the timer. If the timer expires instead, the UE considers the LTM cell handover a failure and can initiate an RRC connection re-establishment procedure to return to the previous serving cell.

[0174] Implementation plan for timed advance acquisition of target cell In some wireless systems, TA acquisition on a target cell can be achieved through one of a variety of possible TA acquisition mechanisms.

[0175] The first possible TA acquisition mechanism is the network-based TA acquisition mechanism. This network-based TA acquisition mechanism may also be referred to as the "early TA acquisition mechanism" in this paper. Under some of these mechanisms, the network estimates the UE's TA value and maintains those TA values ​​at the UE.

[0176] First, it's important to note that the network ensures that the cell transmission times used by the network's cells are synchronized on the network side. Then, under the early TA acquisition mechanism, the UE transmits preambles to candidate target cells from the contention-free random access (CFRA) resources known to the network side for TA estimation. Based on the reception times of these preambles at each cell (which occur due to the varying distances between the UE and the cells), the network determines (and communicates to / maintains at the UE) one or more TAs that the UE will use for the corresponding cell.

[0177] After transmitting the preamble, the UE can return to its source cell (for example, when the preamble is transmitted for the purpose of network determination of the TA value maintained by the network, it may not be necessary to receive the random access response (RAR) to the preamble).

[0178] Regarding this network-maintained TA value / early TA acquisition situation, the source cell can instruct the UE to perform the early TA acquisition process by triggering the UE to execute the Physical Downlink Control Channel (PDCCH) command sent to the target cell via RACH / preamble.

[0179] The second possible TA acquisition mechanism is the UE-based TA acquisition mechanism. This UE-based TA acquisition mechanism may also be referred to in this paper as the "TA mechanism based on Received Signal Time Difference (RSTD)".

[0180] Figure 9 Figure 900 illustrates the operation of an RSTD-based TA mechanism between UE 902, source cell 904, and target cell 906, according to the implementation scheme discussed herein. First, it should be noted that the network ensures that the source cell transmission time 908 and the target cell transmission time 910 are synchronized on the network side, as shown in the figure. Under the RSTD-based TA mechanism, UE 902 can estimate and maintain the TA for each candidate target cell (such as target cell 906) and report this TA to the network. In such an environment, the UE may derive the TA of target cell 906 based on / by considering both the RSTD between the current serving cell and the target serving cell, and the known TA value of the current serving cell.

[0181] For example, such as Figure 9As shown, UE 902 can determine the RSTD 916 between the source cell reception time 912 of source cell 904 and the target cell reception time 914 of target cell 906. UE 902 can then multiply the RSTD 916 between source cell 904 and target cell 906 by 2 (to take into account uplink (UL) and DL aspects regarding TA usage). UE 902 then adds this value to the known TA value of source cell 904 to obtain the TA value of target cell 906 (note that in some cases, this value can be negative).

[0182] Overview of L1 beam level measurement predictions This paper discusses the details of generating and using AI / ML-based LTM enhancements at the UE. As already discussed, LTM decisions can be based on L1 measurements, which may involve one or more potential considerations. First, L1 measurements may be relatively less stable than their corresponding L3 measurements. Therefore, in some cases, using L1 measurements (compared to L3 measurements) for mobility may lead to frequent cell handover / ping-pong handover (HO). Therefore, the implementation scheme in this paper relates to UE-side AI / ML for L1 measurement prediction, which uses a robust HO decision process that minimizes the likelihood of such problems.

[0183] Another potential consideration is that performing L1 measurements and corresponding reports on multiple candidate cells incurs an additional burden on the UE. Regarding this issue, some implementations in this paper for UE-side and network-side L1 measurement prediction are configured to (relatively) reduce the UE burden for L1 measurement and reporting. For example, in some implementations, more resources can be reserved in candidate LTM cells.

[0184] Furthermore, regarding additional UE efforts in L1 mobility scenarios, for each candidate cell's TA acquisition, in the case of using a network-based / early TA acquisition mechanism, the UE may transmit preambles to one or more target cells, as discussed. Additionally, in the case of a UE-based / RSTD-based TA acquisition mechanism, it may be necessary to perform the measurement and calculation of the target cell's TA and report it to the network.

[0185] Therefore, this paper discusses various aspects of L1 measurement prediction and / or TA prediction. The first aspect is the overall process between the base station and the UE for using L1 measurement and / or TA prediction. Another aspect involves the process for training the ML model to be used for L1 and / or TA prediction. Yet another aspect involves the mechanism for performing inference on L1 measurement prediction and / or RSTD-based mechanism TA prediction. For example, the use of each of the UE-side model and the dual-side model for each of L1 measurement prediction and / or RSTD-based mechanism TA prediction is discussed (and these predictions can be spatial and / or temporal predictions, as discussed in more detail elsewhere). Furthermore, the use of network-side models and the inference performance of early TA acquisition mechanism TA prediction are discussed. In such cases, joint temporal-spatial predictions can be generated. Additional aspects include possible performance monitoring.

[0186] UE-side procedures for L1 measurement and / or TA measurement prediction This paper accordingly discusses the details of the generation and use of AI / ML-based L1 and / or TA measurement predictions for mobility enhancement at the UE. Figure 10 A flowchart 1002 illustrates a UE-side procedure for predicting L1 and / or TA measurements, such as between UE 1004 and network 1006, according to an embodiment of this document. It should be noted that in some embodiments, the UE-side procedure for predicting L1 and / or TA measurements may also be combined with the use of a UE server 1008.

[0187] Flowchart 1002 begins with UE 1004 generating a UE capability report 1010 and sending it to network 1006. In some implementations, the UE capability report 1010 may include one or more of the following: whether the UE supports L1 time measurement prediction; whether the UE supports L1 spatial measurement prediction; whether the UE supports RSTD-based TA prediction in the time domain; whether the UE supports RS-RSTD-based TA prediction in the spatial domain; the maximum number of historical samples / time slots that can be used at the UE; the maximum number of prediction samples / time slots that can be used at the UE; and / or the maximum number of parallel predictions that can be used at the UE.

[0188] Then, network 1006 provides training configuration 1012 to UE 1004. Training configuration 1012 may include one or more of the following: the type of ML model to be trained (e.g., a Long Short-Term Memory (LSTM) ML model type, a Recurrent Neural Network (RNN) ML model type, etc.), the layers to be trained and / or one or more specialized ML models to be used; the window length corresponding to the history of measurements and / or predictions in the time and / or spatial domains to be used with the ML model; and / or the maximum number of parallel predictions that should be provided by UE 1004.

[0189] Then, UE 1004 performs data collection 1014. In some embodiments, this process incorporates generating / training an ML model at UE 1004 using the collected data. In some embodiments, UE 1004 provides the collected data to UE server 1008, such that offline training 1016 (e.g., generation of the ML model) occurs instead at UE server 1008, and the UE server then provides the thus generated ML model back to UE 1004.

[0190] Then UE 1004 transmits notification message 1018 to network 1006. The content of notification message 1018 may notify network 1006 which ML models (and in at least some cases, the model IDs corresponding to these ML models are available at UE 1004).

[0191] The content of notification message 1018 can notify network 1006 of model applicability conditions, which network 1006 can use to determine which ML model to use at UE 1004. It should be noted that model applicability conditions may include, for example, usage scenario information (e.g., indoor / outdoor), antenna type information, channel type information, UE speed information (e.g., UE travels less than 5 km / h (kmph)), UE altitude information (e.g., corresponding to UE movement at an elevation), etc.

[0192] The content of notification message 1018 can notify network 1006 of the UE's preferred model ID.

[0193] It should be noted that notification message 1018 may be, for example, part of an SR, part of a UAI, in a MAC-CE, or in the sending and receiving of RRC messages (e.g., in...). RRCReconfigurationComplete Provided in the message or a newly provided RRC message.

[0194] Based on notification message 1018, network 1006 can determine which ML model to activate at UE 1004 and can provide UE 1004 with activation message 1020 commanding UE 1004 to activate the selected ML model. Activation message 1020 may be provided, for example, as part of DCI, MAC-CE, or RRC message reception.

[0195] Then, UE 1004 continues to perform inference / prediction 1022 for L1 measurements and / or RSTD-based TA mechanism measurements (as applicable), as discussed, based on the configuration of network 1006. Inference / prediction 1022 can be reported to the network in report 1024.

[0196] It is conceivable that, regarding the UE-side process, UE 1004 or network 1006 may perform performance monitoring 1026 of the ML model (e.g., by comparing predicted measurements with corresponding actual measurements). Based on the results of performance monitoring 1026, UE 1004 or network 1006 may initiate LCM signaling 1028 for model switching or model deactivation, which then leads to model switching / deactivation 1030 at the UE. In the case of deactivation, UE 1004 and network 1006 may then fall back to a non-AI / ML-based measurement reporting solution.

[0197] Two-sided process for predicting L1 and / or TA measurements Figure 11 A flowchart 1100 illustrates a two-sided process for predicting L1 and / or TA measurements, such as between UE 1102 and network 1104, according to an embodiment of this document. It should be noted that in some embodiments, the UE-side process for measurement prediction may also be combined with the use of a UE server 1106.

[0198] UE capability report 1010, training configuration 1012, data collection 1014, and offline training 1016 can all be found in this article. Figure 10 The discussion regarding the UE-side process describes how it occurs.

[0199] Once the ML model exists at UE 1102, a model transfer 1108 occurs. During the model transfer, UE 1102 transmits the ML model to network 1104. The model transfer signaling can be RRC-based or DRB-based.

[0200] Then, UE 1102 can generate actual L1 measurements and / or actual RSTD-based TA measurements, and transmit a measurement report 1110 with these actual measurements to network 1104.

[0201] Network 1104 can then apply these actual L1 measurements / actual RSTD-based TA measurements together with the previously received ML model to perform L1 measurement / or RSTD TA measurement prediction 1112 (as applicable). In some implementations, the detailed prediction method may depend on the specific implementation of network 1104.

[0202] Furthermore, performance monitoring 1114 may occur at network 1104 (e.g., by comparing the actual L1 measurement and / or the actual RSTD-based TA measurement received from UE 1102 with the corresponding predicted L1 measurement and / or the predicted RSTD-based TA measurement).

[0203] Network 1104 can also be configured to trigger ML model retraining performance based on its specific implementation (which may make the decision, for example, at least in part, based on the results of performance monitoring 1114). As part of triggering this retraining, network 1104 may provide the UE with a retraining configuration 1116 that commands retraining (and may provide one or more parameters for the UE to analyze / use as part of the ML model retraining process).

[0204] In response to the retraining configuration 1116, in some implementations, the UE continues to perform data collection 1014, offline training 1016, and model transfer 1108 again, as previously described.

[0205] Example mechanism for L1 measurement prediction In some implementations, the UE may be configured (e.g., via RRC signaling) for one of several alternatives to L1 measurement prediction performed by an ML model (which may be referred to as a “measurement prediction model”). It should be noted that L1 measurement prediction as discussed herein may correspond to beam-level measurement (and this may not be explicitly mentioned in the various subsequent implementations).

[0206] In a first-class implementation for L1 measurement prediction, the L1 measurement prediction may correspond to a time prediction. For example, the measurement prediction model can be used to predict future L1 measurements based on the current inputs to the measurement prediction model.

[0207] Figure 12 A mechanism 1200 for time prediction of L1 measurements according to the implementation discussed herein is illustrated. One or more actual L1 measurements 1202 (e.g., SSB and / or CSI-RS measurements) may be performed at the UE. The measurement prediction model may be configured to receive these one or more actual L1 measurements 1202 as input and, in response, provide one or more predicted L1 measurements 1204 (each of the one or more predicted L1 measurements 1204 corresponding to a later time). Each predicted L1 measurement in the predicted L1 measurements 1204 may be used, for example, for SSB or CSI-RS.

[0208] In a second type of implementation for L1 measurement prediction, L1 measurement prediction corresponds to spatial prediction. For example, the UE can use its understanding of applicable spatial channel statistics to predict / infer the (predicted) L1 measurement of a beam based on the actual or predicted L1 measurements of its neighboring beams. In such an example, the UE can predict the L1 measurement of a beam based on the actual L1 measurements of its neighboring beams.

[0209] Figure 13A mechanism 1300 for spatial prediction of L1 measurements according to the implementation scheme discussed herein is illustrated. Figure 13 An example of a spatial beaming arrangement 1302 for each of the first beam (“beam 1”), the second beam (“beam 2”), and the third beam (“beam 3”) is shown.

[0210] One or more actual L1 measurements can be performed. Figure 13 In the example, the UE performs one or more first actual L1 measurements 1304 for the first beam (beam 1) and one or more second actual L1 measurements 1306 for the third beam (beam 3).

[0211] One or more actual L1 measurements (e.g., one or more first actual L1 measurements 1304 and one or more second actual L1 measurements 1306) are provided to a measurement prediction model 1308, which uses these terms to generate one or more predicted L1 measurements 1310 for a second beam (beam 2, as shown, which is a neighboring beam of beams 1 and 3). It should be noted that in some embodiments, the measurement prediction model 1308 may use applicable spatial channel statistics to generate one or more predicted L1 measurements 1310.

[0212] In some implementations, the predicted dwell / validity time can be provided for both the temporal and / or spatial predictions of L1 measurements. Furthermore, in some implementations, the confidence level of the prediction can be provided for both the temporal and / or spatial predictions of L1 measurements.

[0213] In some implementations, the temporal and spatial predictions of L1 measurements may be configured to be performed simultaneously in a two-dimensional space.

[0214] It should be noted that configuration information (e.g., RRC configuration information) can be provided to instruct the UE to perform L1 measurement predictions regarding the use of the measurement prediction model. For example, the UE can be configured to generate L1 measurement predictions for a specific beam of a particular cell (e.g., the current serving cell and / or neighboring cells).

[0215] As another example of using L1 measurement prediction based on configuration information, the UE can be configured to generate L1 measurement prediction based on one or more conditions.

[0216] In the first example using conditions for predicting L1 measurements, the UE can be configured to generate measurements for at most a number of... N L1 measurements for each beam, where the UE generates measurements for the number of previously known beams with the strongest RSRP / RSRQ. M beams ( M < N The actual L1 measurement of ) and further generate for the remainingN – M The predicted L1 measurement for each beam. The predicted L1 measurement falls within... M In the case of the highest measurement, the UE will then perform actual L1 measurements on subsequent beams (and the previous ones). M The last beam in the ensemble is added alternatively using the predicted beam. N – M gather).

[0217] In the second example using conditions for using predicted L1 measurements, a configured RSRP / RSRQ / SINR threshold can be compared with the actual or predicted L1 measurement. If the actual or predicted L1 measurement of the cell is less than the threshold, the UE performs L1 measurement prediction for the beam. In a variation of this, the base station may configure the UE to use separate thresholds for the use of actual and predicted L1 measurements.

[0218] In the third example using conditions for predicting L1 measurements, a configured interference measurement threshold can be used. In this case, the UE can use L1 measurement prediction if the interference intensity in the beam is higher than the threshold.

[0219] In the fourth example using conditions for using predicted L1 measurements, two thresholds can be used. A first configured RSRP threshold can be compared to the actual / predicted L3 cell-level measurement of the cell. If the actual or predicted L3 cell-level measurement of the cell is greater than the threshold, the UE can perform L1 (beam-level) measurement prediction for one or more beams in that cell. Then, a second configured RSRP / RSRQ threshold can be compared to the actual or predicted L1 measurement of the beams in that cell to determine whether beams in subsequent cells use actual L1 measurements or predicted L1 beam-level measurements.

[0220] In the fifth example using the conditions for using predicted L1 measurements, L1 measurement prediction can be performed for measurements across all frequencies.

[0221] In the sixth example using L1 measurements for prediction, L1 measurement prediction can be performed if the measurement uses a measurement gap (e.g., for a measurement of another frequency or another non-overlapping BWP).

[0222] As another example of using L1 measurement prediction based on configuration information, the configuration information may indicate neighboring cells and / or frequencies that allow the UE to autonomously / independently select to generate actual L1 measurements or predicted L1 measurements.

[0223] Implementation scheme for reporting actual and / or predicted L1 measurements Imagine supporting periodic and event-triggered L1 measurement reports.

[0224] In some implementations, for periodic L1 measurement reporting, the UE can be configured by the network to have two periodicities in a single reporting configuration. The first periodicity can be a predicted measurement periodicity, indicating how often the UE should perform and / or report an AI / ML-based L1 measurement prediction. The second periodicity can be an actual measurement periodicity, indicating how often the UE should perform and / or report an actual L1 measurement. In some such implementations, the predicted measurement periodicity may be less than the actual measurement periodicity, allowing the UE to use L1 measurement predictions most of the time (e.g., to reduce power) while still occasionally reporting actual L1 measurements to provide more accurate updates / information available for model monitoring.

[0225] In some implementations, for event-triggered reporting, the UE may be configured to trigger a measurement reporting event using actual and / or predicted L1 measurements. Examples of such events may include a first event in which the actual or predicted L1 measurement is less than a threshold, a second event in which a first actual or predicted L1 measurement at the serving cell is less than a first threshold and a second actual or predicted L1 measurement at a neighboring cell is greater than a second threshold, and / or a third event in which the first actual or predicted L1 measurement at the neighboring cell is better than / greater than the second actual or predicted L1 measurement at the serving cell by more than / greater than a threshold.

[0226] The UE may be configured to determine whether a predicted L1 measurement can trigger a measurement reporting event (e.g., the opposite of triggering a measurement reporting event using only the actual L1 measurement). In such cases where a predicted L1 measurement is available, the UE may also be configured to trigger a measurement report based on the predicted L1 measurement (e.g., only if the confidence level for the predicted L1 measurement is greater than a threshold).

[0227] It should be noted that when analyzing L1 measurements of neighboring cells, the UE can be configured to determine whether the predicted L1 measurements are only for neighboring cells (and not, for example, the serving cell).

[0228] In these cases, the UE could also be configured to use, for example, one or more of events A1-A6, as can be understood in the NR wireless communication system regarding any predicted L3 cell-level measurement (e.g., if the confidence level for the predicted L3 cell-level measurement is greater than the corresponding threshold).

[0229] In either case, once the measurement reporting event is triggered, the UE can report the available predicted and / or actual L1 measurements as part of the measurement report.

[0230] In some implementations, for both periodic and event-triggered reporting, the UE may first report the actual L1 measurement and then report the predicted L1 measurement. In the first case, the order in which such L1 measurements are reported may then be based on the number of measurements (RSRP / RSRQ) (e.g., from high to low).

[0231] In the second case, the order in which such L1 measurements are reported can be based on the confidence level for any predictive measurement (e.g., from high to low).

[0232] In the third case, the order of reporting such L1 measurements could be to report the SSB measurement first, followed by the CSI-RS measurement.

[0233] It should be noted that any combination of the above three scenarios is possible. For example, the UE can be configured to report the SSB first, then the highest RSRP first, and finally the CSI-RS (e.g., in the case where more than one SSB has the same RSRP).

[0234] Within the measurement report, the UE indicates which L1 measurements are predictive measurements and / or which L1 measurements are actual measurements. In at least some embodiments, the UE may also indicate the reliability probability or confidence level of any predictive measurement if it includes predictive measurements in the measurement report.

[0235] Regarding the use of L1 measurement predictions, the total number of parallel predictions in the measurement report may be within the UE's capabilities. In some cases, the base station configures the UE to perform multiple predictions within that UE's capabilities (e.g., it may be selected to perform measurements corresponding to previously reported high channel conditions).

[0236] When a UE discards one or more measurement predictions to remain within its capabilities, the UE may discard predictions corresponding to poor channel conditions; discard predictions corresponding to poor confidence levels; and / or discard predictions based on CSI-RS. It should also be noted that any combination of these criteria can be used for discarding purposes.

[0237] Example mechanism for RSTD-based TA measurement prediction Figure 14 Figure 1400 illustrates example scenarios of various TAs (Transmission Actions) between UE 1402 and each of the first cell 1404, the second cell 1406, and the third cell 1408. For example... Figure 14As shown, each of the first cell 1404, the second cell 1406, and the third cell 1408 may be located at a different position relative to the UE 1402. Therefore, the first TA 1410 for communication between the UE 1402 and the first cell 1404, the second TA 1412 for communication between the UE 1402 and the second cell 1406, and the third TA 1414 for communication between the UE 1402 and the third cell 1408 may all be independent and / or different from each other.

[0238] In some implementations, the UE can be configured (e.g., via RRC signaling) for one of a variety of alternatives to RSTD-based TA measurement predictions performed by an ML model (which may be referred to as a "measurement prediction model").

[0239] In a first-class implementation for RSTD-based TA measurement prediction, the RSTD-based TA measurement prediction may correspond to time prediction. For example, the measurement prediction model can be used to predict future RSTD-based TA measurements based on the current inputs to the measurement prediction model.

[0240] Figure 15 A mechanism 1500 for time prediction of RSTD-based TA measurements according to the implementation scheme discussed herein is illustrated. One or more actual RSTD-based TA measurements 1502 can be performed at the UE. The measurement prediction model can be configured to receive these one or more actual RSTD-based TA measurements 1502 as input and, in response, provide one or more predicted RSTD-based TA measurements 1504 (where each predicted RSTD-based TA in the one or more predicted RSTD-based TA measurements 1504 corresponds to a later time).

[0241] In a second type of implementation for RSTD-based TA measurement prediction, RSTD-based TA measurement prediction corresponds to spatial prediction. For example, a UE can use its understanding of applicable spatial channel statistics to predict / infer the (predicted) RSTD-based TA measurement of a cell based on the actual or predicted RSTD-based TA measurements of its neighboring cells. For example, a UE can predict the RSTD-based TA measurement of a cell based on the actual RSTD-based TA measurements of its neighboring cells.

[0242] For spatial prediction, the UE can predict the TA of a candidate cell based on the actual RSTD-based TA measurement of another candidate cell. In such cases, the network can provide the UE with various information (e.g., base station deployment geometry) as auxiliary information.

[0243] Figure 16A mechanism 1600 for spatial prediction of RSTD-based TA measurements is illustrated according to the implementation scheme discussed herein. Figure 16 Corresponding to the first cell (e.g., Figure 14 The first community 1404), the second community (for example, Figure 14 The second community 1406) and the third community (e.g., Figure 14 The spatial arrangement of each of the three communities (1408).

[0244] One or more actual RSTD-based TA measurements can be performed. Figure 16 In the example, the UE performs one or more first actual RSTD-based TA measurements 1602 in the first cell and one or more second actual RSTD-based TA measurements 1604 in the third cell.

[0245] One or more actual RSTD-based TA beamlevel measurements (e.g., one or more first actual RSTD-based TA measurements 1602 (e.g., for the first cell 1404) and one or more second actual RSTD-based TA measurements 1604 (e.g., for the third cell 1408) are provided to a measurement prediction model 1606, which uses these terms to generate one or more predicted RSTD-based TA beamlevel measurements 1608 for the second cell (e.g., the second cell 1406). It should be noted that in some embodiments, the measurement prediction model 1606 may use applicable spatial channel statistics to generate one or more predicted RSTD-based TA beamlevel measurements 1608.

[0246] In some implementations, the predicted dwell / validity time can be provided for both temporal and / or spatial predictions of RSTD-based TA measurements. Furthermore, in some implementations, the confidence level of the prediction can be provided for both temporal and / or spatial predictions of RSTD-based TA measurements.

[0247] In some implementations, temporal and spatial predictions of TA measurements based on RSTD may be configured to be performed simultaneously in a two-dimensional space.

[0248] Implementation scheme for reporting actual and / or predicted RSTD-based TA measurements Imagine supporting periodic and event-triggered RSTD-based TA measurement reports.

[0249] In some implementations for periodic RSTD-based TA measurement reporting, the UE can be configured by the network to have two periodicities in a single reporting configuration. The first periodicity can be a predictive measurement periodicity, indicating how often the UE should perform and / or report an AI / ML-based RSTD-based TA measurement prediction. The second periodicity can be an actual measurement periodicity, indicating how often the UE should perform and / or report an actual RSTD-based TA measurement.

[0250] For an event-triggered reporting implementation, the UE may be configured with a new TA-related event (e.g., "Event-4"). This event may occur when the change in the RSTD-based TA of a candidate cell is greater than a threshold compared to the last reported instance of that value. The network can be configured to allow the UE to determine whether actual RSTD-based TA measurements and / or predicted RSTD-based TA measurements can trigger the event. Additionally or alternatively, the UE may be configured to trigger the event only when the confidence level of the predicted RSTD-based measurement is greater than a threshold.

[0251] In either case, once the measurement report event is triggered, the UE can report the available predicted and / or actual RSTD-based TA measurements as part of the measurement report.

[0252] In some implementations, for both periodic and event-triggered reporting, the UE may first report actual RSTD-based TA measurements, and then report predicted RSTD-based TA measurements. In the first case, the order in which these RSTD-based TA measurements are reported may then be based on the number of cell-level L3 measurements (RSRP / RSRQ / SINR) from high to low.

[0253] In the second case, the order in which such RSTD-based TA measurements are reported can be based on the confidence level for any predictive measurement (e.g., from high to low).

[0254] It should be noted that any combination of the two scenarios described above is possible. For example, the UE could be configured to first report based on the number of cell-level L3 measurements, and then secondarily based on the confidence level of any predictive RSTD-based TA measurements.

[0255] Within the measurement report, the UE may indicate which RSTD-based TA measurements are predictive measurements and / or which RSTD-based TA measurements are actual measurements. In at least some embodiments, the UE may also indicate the reliability probability or confidence level of any predictive measurement when it includes predictive measurements in the measurement report.

[0256] Regarding the use of RSTD-based TA measurement predictions, the total number of parallel predictions in the measurement report may be within the UE's capabilities. In some cases, any additional RSTD-based TA measurements / predictions may be discarded based on, for example, the order of measurement reports used for RSTD-based TA measurement predictions (e.g., as discussed herein).

[0257] Model monitoring and LCM for L1 measurement prediction and / or RSTD-based TA measurement prediction Imagine a UE-side procedure for L1 measurement prediction and / or RSTD-based TA measurement prediction, in which model monitoring (e.g., monitoring the performance of the ML model) can be performed at the UE and / or at the base station. In the case of performing model monitoring at the base station, the procedure may depend on the specific implementation at the base station.

[0258] In cases where model monitoring is performed at the UE, in some such implementations, the model monitoring metric used by the UE may be the error between the predicted L1 measurement / RSTD-based TA measurement and the corresponding actual L1 measurement / RSTD-based TA measurement. For example, the MSE between some predicted L1 measurement / RSTD-based TA measurement and the corresponding actual L1 measurement / RSTD-based TA measurement may be used (and it should be noted that using the MSE metric in this way could be an example of a "confidence level" as discussed herein). For monitoring purposes, the UE may perform both the predicted L1 measurement / RSTD-based TA measurement and the corresponding actual L1 measurement / RSTD-based TA measurement for a small set of cells and / or beams (as applicable).

[0259] It is conceivable that the ML model in use may be switched from time to time (e.g., a different ML model may be selected for use and / or the use of an ML model may be paused or stopped for prediction). This may occur, for example, based on the results of model monitoring, as will be described. The UE may be configured to perform UE-initiated model switching or network-initiated model switching.

[0260] In the case of a UE-initiated handover, the UE can be configured with conditions and UE behavior related to model metrics (e.g., confidence levels). For example, the UE can be configured with an MSE (Mean Sequence Size) threshold of 0.01. When the MSE is greater than 0.01, the UE can be configured to fall back to the normal measurement.

[0261] In the event of a network-initiated handover, the UE can report model monitoring metrics (e.g., confidence level) and may wait for base station LCM signaling in response (e.g., instructions to stop using the ML model and / or start using a different ML model). The UE can report model monitoring metrics via UAI or MAC-CE.

[0262] It should be noted that the network may provide the UE with auxiliary information about model monitoring occurring on the UE side, which the UE can use. This auxiliary information may include any one or more of the following: the geometry of nearby base station deployments; statistics on temporal correlations; and / or long-term statistics on inter-cell or inter-beam correlations.

[0263] For the two-sided process used for L1 measurement prediction and / or RSTD-based TA measurement prediction, model monitoring can be performed on the base station side, and it may depend on the specific implementation of the gNB.

[0264] Regarding the UE-side and / or dual-side processes for L1 measurement prediction and / or RSTD-based TA measurement prediction monitored by the base station using ML models, the UE may be configured to provide information to the base station to assist the base station in performing monitoring. This information may include additional temporal information, such as timestamps used for prediction and / or timestamps corresponding to actual measurements. This information may also / optionally include additional spatial information, such as the UE's actual location, the UE's actual movement orientation, changes in the UE's movement orientation, and / or incremental directions (differences) that can be compared with the prediction.

[0265] Base station-side processes for early TA prediction Figure 17A and Figure 17B A flowchart 1700 illustrating the implementation of early TA prediction in a system comprising a UE 1702, a source base station 1704 communicating with the UE on the serving cell, a first target base station (“Target Base Station 1”) having a first target cell 1706, a second target base station (“Target Base Station 2”) having a second target cell 1708, and a server 1710, is illustrated below. It should be noted that a base station-side procedure may be appropriate regarding the use of the early TA mechanism, as TA is maintained at the base station side.

[0266] Flowchart 1700 begins with data collection and offline model training 1712. Then, UE 1702 transmits actual and / or predicted L3 cell-level and / or beam / level measurements 1714 to source base station 1704. These measurements can indicate to source base station 1704 that the first target cell 1706 and the second target cell 1708 are appropriate target cells.

[0267] As shown in the figure, the source base station 1704 continues to perform LTM candidate preparation 1716 with the first target cell 1706 and the second target cell 1708, so that the first target cell 1706 and the second target cell 1708 are ready to take action according to LTM.

[0268] Then, the source base station 1704 transmits a configuration message 1718 to the UE 1702, indicating candidate configurations for the first target cell 1706 and the second target cell 1708 (e.g., RRCReconfiguration (Message). Furthermore, the source base station 1704 may include an applicability condition request corresponding to the applicability conditions used to select the ML model to be used by the base station. For example, a configuration message 1718 may indicate that a UE speed threshold will be used to select a suitable ML model.

[0269] The UE may provide a configuration response 1720 to the source base station 1704 in response to receiving a configuration message 1718. The configuration response 1720 may include feedback on the applicability conditions indicated in the configuration message 1718 in the form of an applicability condition response, which informs the source base station 1704 of the value of the applicability condition. Based on the value of the applicability condition received in the configuration response 1720, the source base station 1704 may (e.g., from multiple such models that may exist at the source base station 1704) identify the ML model used for early TA prediction. It should be noted that it is envisioned that the UE may provide updated feedback (e.g., updated values ​​of the applicability conditions) to the source base station 1704 at any time via the UAI.

[0270] UE 1702 then transmits preamble 1722 to each of the first target cell 1706 and the second target cell 1708. As shown, UE 1702 transmits the first preamble to the first target cell 1706 at a first time ("T1"), the second preamble to the second target cell 1708 at a second time ("T2"), the third preamble to the first target cell 1706 at a third time ("T3"), and the fourth preamble to the second target cell 1708 at a fourth time ("T4").

[0271] After receiving the preamble 1722, the first target base station 1706 of the first target cell and the second target base station 1708 of the second target cell each transmit TA information 1724 to the source base station 1704. The TA information 1724 may include the collected data corresponding to the preamble 1722 and can be transmitted to the source base station 1704 through an inter-node signaling process.

[0272] TA information 1724 may include the UE ID of UE 1702.

[0273] The TA information 1724 used for transmitting to the target cell may also include a TA value corresponding to the transmission by UE 1702 to the target cell. As discussed herein, this TA value may have already been determined at the network using the timing corresponding to one or more preambles transmitted by UE 1702 to the target cell.

[0274] The TA information 1724 for transmitting the target cell may also include one or more timestamps corresponding to the preamble used to generate the reported TA value for transmitting the target cell (e.g., such as...). Figure 17B (As shown in TA information 1724).

[0275] It should be noted that TA information 1724 can generally be considered to include additional instances of similar communications between the first target cell 1706 and the second target cell 1708, in addition to those explicitly illustrated (e.g., it may include unillustrated information about a previous unillustrated preamble that predates 1722).

[0276] The source base station 1704 may then receive one or more actual or predicted L1 measurements 1726 from the UE 1702. Based on these actual or predicted L1 measurements 1726, the source base station 1704 may make a HO decision 1728, in which it selects one of the first target cell 1706 and the second target cell 1708 where the UE should perform the handover.

[0277] After making the HO decision 1728, the source base station 1704 can apply the TA information 1724 to the selected ML model to generate a predicted early TA 1730. In some examples, the predicted early TA 1730 is a joint temporal-spatial prediction.

[0278] The predicted early TA 1730 can then be included in a cell handover command 1732 (e.g., a MAC-CE cell handover command), which is transmitted to UE 1702 to cause the UE to perform LTM for the selected target cell (one of the first target cell 1706 and the second target cell 1708). As part of this LTM, the UE uses the predicted early TA 1730 received from the source base station 1704 to adjust its transmission timing with respect to the selected target cell.

[0279] It should be noted that in such implementations, model monitoring can be performed on the network side and may depend on the specific network implementation.

[0280] Implementation plan for fault handling using predicted L1 measurements In some implementations, if LTM execution has failed (e.g., caused / determined by the expiration of an LTM supervisor timer), the UE may consider (potentially) performing the cell selection process using predicted L1 measurements. For example, if the configured candidate target cells become suitable, the UE may select a target cell via the following priority ranking rules. First, the UE may select from cells with actual L1 measurements (e.g., L1 RSRP / RSRQ) from high to low. Then, the UE may select cells following a confidence level that only has predicted L1 measurements.

[0281] It should be noted that these priority ordering rules are given as examples rather than as restrictions. It is conceivable that the priority ordering rules used in this case may vary depending on the specific implementation of the UE.

[0282] Conditional handover method CHO (Crossover Event) is a feature used to improve mobility robustness. In a CHO, a UE can be configured with a handover command and associated CHO conditions to be monitored (sometimes alternatively referred to as "event conditions," "trigger conditions," or "conditions"). When the associated condition becomes true, the UE can execute the stored handover command. Event conditions may include, for example, when a neighboring cell becomes better than a specific cell (SpCell) by a certain offset (e.g., event condition A3), or when SpCell becomes worse than a first threshold and a neighboring cell becomes better than a second threshold (e.g., event condition A5). SpCell is the primary serving cell of a primary cell group (MCG) or secondary cell group (SCG), and the offset can be positive or negative. When more than one candidate target cell meets the condition, the cell for which the UE executes the HO may be determined depending on the specific UE implementation. In some wireless communication systems (e.g., 3GPP Release 17 wireless communication systems), new location- and time-related trigger conditions can be defined to help enhance CHO for non-terrestrial networks (NTNs).

[0283] Figure 18A and Figure 18B A flowchart 1800 illustrating conditional handover that can be used in some wireless communication systems is provided. Flowchart 1800 illustrates a wireless communication system including a UE 1802, a source gNB 1804, a target gNB 1806, other potential target gNBs 1808, an Access and Mobility Management Function (AMF) 1810, and one or more User Plane Functions (UFP) 1812. It can be seen that flowchart 1800 corresponds to the situation within the AMF / UPF. It should be noted that in other implementations, the source gNB 1804, the target gNB 1806, and other potential target gNBs 1808 may each (e.g., independently) be a base station type other than a gNB.

[0284] like Figure 18AAs illustrated, flowchart 1800 begins with the handover preparation phase 1814. Currently, as shown, user data 1816 is transmitted between UE 1802 and source gNB 1804, and between source gNB 1804 and UFP 1812. AMF 1810 provides mobility control information 1818 to source gNB 1804. Then, during measurement control and reporting 1820, source gNB 1804 configures measurements at UE 1802, and UE 1802 performs the measurements and reports the measurement results to source gNB 1804. Based on the receipt of the measurement report, source gNB 1804 makes a CHO decision 1822. Based on CHO decision 1822, source gNB 1804 transmits handover request 1824 to other gNBs (in flowchart 1800, target gNB 1806, which will eventually be selected as the target of handover, and other potential target gNB 1808 are both represented as receiving handover request 1824).

[0285] Other gNBs (e.g., target gNB 1806 and other potential target gNBs 1808) each perform admission control 1826 and respond to source gNB 1804 with a handover request confirmation 1828, including the configuration of any CHO candidate cells at that gNB.

[0286] Figure 18B Continuing from the previous article about Figure 18A The flowchart discussed is 1800. Source gNB 1804 transmits configurations for CHO candidate cells (CHO configuration for candidate cells) to UE 1802. RRCReconfiguration Message 1830. UE1802 transmits to source gNB 1804 RRCReconfigurationComplete Message 1832.

[0287] Then, flowchart 1800 proceeds to the handover execution phase 1834. UE 1802 evaluates the CHO condition 1836. Furthermore, in some implementations (e.g., in the case of using early data forwarding), target gNB 1806 transmits an early status transfer message 1838 to other potential target gNBs 1808.

[0288] Then, UE 1802 leaves the old cell 1840 and synchronizes to the new cell (e.g., on target gNB 1806). As part of this process, the UE performs an evaluation of the conditions on the candidate cells and determines that the new cell (on target gNB 1806) meets these conditions and is therefore to be handed over to that cell. The configuration for the new cell is then applied at the UE.

[0289] Furthermore, user data 1842 is transmitted between UFP 1812 and target gNB 1806 and / or other potential target gNB 1808 via source gNB 1804. Once UE 1802 becomes associated with a new cell on source gNB 1804 (and UE 1802 can transmit accompanying data to target gNB 1806), RRCReconfigurationComplete (News) The handover of CHO was completed in 1844.

[0290] Then, flowchart 1800 proceeds to the handover completion phase 1846. First, the target gNB 1806 transmits a handover success message 1848 to the source gNB 1804. Next, the source gNB 1804 transmits a sequence number (SN) status transfer 1850 to the target gNB 1806. User data 1852 is transmitted between UFP 1812 and the target gNB 1806 via the source gNB 1804. Finally, the source gNB 1804 may transmit a handover cancellation message 1854 to the target gNB 1806 and / or other potential target gNBs 1808.

[0291] Overview of measurement predictions used in the CHO method The details of generating and using AI / ML-based CHO enhancements at the UE will now be discussed. In some cases, using AI / ML-based CHO enhancements may result in a reduction of reserved radio resources for candidate target cells based on predicted L3 measurements. In such cases, the UE may suggest changes to unfavorable CHO conditions, such as candidate target cells that are unlikely to meet CHO conditions within the relevant timeframe, CHO event types, and / or applicable thresholds.

[0292] In some cases, using AI / ML-based CHO enhancements can lead to a reduction in the time required to initiate CHO execution. For example, CHO can be executed when the predicted L3 measurement meets the CHO condition (e.g., instead of waiting for the actual L3 measurement to meet the CHO condition).

[0293] In some cases, using AI / ML-based CHO enhancements can make the UE's target cell selection more robust. Depending on the specific UE implementation, the target cell selection may vary depending on the existing / defined CHO procedures, provided that more than one target cell meets the applicable CHO conditions. Therefore, in some cases, measurement predictions can be used to further select among such cells.

[0294] This paper discusses various aspects of the overall process between the base station and the UE regarding the AI / ML-enhanced CHO mechanism. Furthermore, it discusses new CHO configurations that can be used in such environments. Further, it discusses UE behavior regarding CHO condition evaluation in such environments. Still further, it discusses what can be provided from the UE to the source cell / source base station (e.g., via...) in such environments. RRCReconfigurationComplete Messages or UAI (User AI) auxiliary information. Finally, various aspects of UE behavior used for fault handling in such environments are discussed.

[0295] The process of using measurement to predict CHO Figure 19 A flowchart 1900 illustrates a CHO process for using a UE 1902, a source base station 1904 communicating with the UE on a serving cell, a first target base station (“Target Base Station 1”) having a first target cell 1906, a second target base station (“Target Base Station 2”) having a second target cell 1908, and a server 1910, according to the implementation scheme discussed herein.

[0296] As shown in the figure, UE 1902, source base station 1904, and server 1910 work together to generate one or more L3 cell-level and / or beam-level measurement predictions 1912 corresponding to one or more of the first target cell 1906 and / or the second target cell 1908. These are ultimately reported by UE 1902 to source base station 1904. This can occur as discussed elsewhere in this document.

[0297] Then, the source base station 1904 may perform a first CHO preparation 1914 with respect to the first target cell 1906, as shown in the figure. The first CHO preparation 1914 may be based on the prediction of the L3 cell / beam level measurement of the first target cell 1906 reported by the UE (and in some cases, this may include analysis of any corresponding confidence levels of these predicted L3 measurements, as may also be provided by the UE 1702).

[0298] As shown in the figure, the source base station 1904 can also perform a similar second CHO preparation 1916 with respect to the second target cell 1908 (and for similar reasons).

[0299] In some cases, the specific metrics and procedures for selecting cells for CHO preparation, as described, may vary depending on the specific implementation of the source cell.

[0300] The UE source base station 1904 then provides the CHO configuration 1918 to the UE 1902 (e.g., in...). RRCReconfiguration(As shown in the message). As illustrated, in some environments, CHO configuration 1918 can be understood as a CHO command. CHO configuration 1918 provides a list of candidate target cells, which identifies each of the first target cell 1906 and the second target cell 1908 as a candidate target cell.

[0301] Furthermore, CHO configuration 1918 may include one or more CHO events corresponding to one or more CHO conditions (e.g., relevant thresholds for measurement), which will be evaluated with respect to candidate target cells to determine whether an HO should be performed on a particular candidate target cell. As an example, for some 3GPP wireless communication systems, these CHO events may include events A3, A4, and / or A5.

[0302] For each measurement-based CHO event in a configuration, CHO configuration 1918 can provide an indication of whether the predicted L3 measurement can be used to evaluate the corresponding CHO condition.

[0303] In addition, for each configuration of measurement-based CHO events, CHO configuration 1918 can provide a confidence threshold (in... Figure 19 The term "th" is used to evaluate CHO conditions using predicted L3 measurements where the use of predicted L3 measurements is permitted.

[0304] The CHO configuration 1918 is envisioned to include multiple sets of CHO conditions and criteria for selecting a specific CHO condition from these conditions for use. For example, the use of various such CHO conditions could be used to indicate a UE mobility speed threshold.

[0305] In some cases, CHO configuration 1918 may also include a priority value for each candidate target cell.

[0306] In response to receiving CHO configuration 1918, UE 1902 provides CHO configuration response 1920 to source base station 1904 (e.g., RRCReconfigurationComplete (Message). The UE can include various information in the CHO configuration response 1920.

[0307] In some cases, the CHO configuration response 1920 may include a proposed change to the list of target cells used by the source base station 1904. This can help reduce any mismatch between the L3 measurement reports predicted by the UE 1902 and the currently applicable channel observations (e.g., corresponding to CHO configuration 1918), which may be outdated.

[0308] In some cases, CHO configuration response 1920 may include updated L3 cell-level / beam-level measurement predictions. This helps reduce any mismatches between the L3 measurement reports predicted by UE 1902 and currently applicable channel observations (e.g., corresponding to CHO configuration 1918), which may be outdated.

[0309] In some cases, the CHO configuration response 1920 may include a prediction error metric for one of the predicted L3 measurements (e.g., the MSE between the predicted L3 measurement and its actual L3 measurement). In some cases, the CHO configuration response 1920 may include suggested changes to CHO configuration information, including but not limited to changes to the CHO event type; changes to the CHO conditions / thresholds used to assess whether a CHO event has occurred; changes to the TTT value; and / or changes to the suggested priority value for the cell.

[0310] It should be noted that although CHO configuration response 1920 has been cited as using RRCReconfigurationComplete The message, but also envisions that the information found therein could be carried alternatively / additionally in MAC-CE and / or a new / some other UL RRC message.

[0311] It should be noted that in at least some cases, the content of CHO configuration response 1920 may depend on the content of CHO configuration 1918.

[0312] As shown in the figure, source base station 1904 may optionally provide UE 1902 with an update 1922 on the CHO configuration (e.g., based on information received in the CHO configuration response 1920). The mechanism for such an update may vary depending on the specific implementation of source base station 1904. As shown in the figure, update 1922 may be available in... RRCReconfiguration The message is transmitted, and the UE may further provide the source base station 1904 with a response 1924 to update 1922 (e.g., RRCReconfigurationComplete information).

[0313] Then, UE 1902 continues to CHO evaluation cycle 1926, in which the CHO conditions configured by CHO are actively monitored regarding the actual L3 cell-level and / or beam-level measurements and / or predicted L3 cell-level and / or beam-level measurements generated from time to time at UE 1902.

[0314] It should be noted that during the CHO evaluation cycle 1926, the UE may transmit a UAI message 1928 to the source base station 1904 to perform one or more of the following: a proposed change to the candidate target cell list; a proposed target cell for performing an unconditional (e.g., regular) HO; and / or a proposed change to the priority value of one or more cells.

[0315] As shown in the figure, the CHO evaluation cycle 1926 ends when UE 1902 determines that the CHO conditions for an applicable CHO event with applicable CHO configuration have been met. Regarding the implementation using predicted L3 measurements, it can be assumed that these CHO conditions are met under various possible circumstances.

[0316] In the first case, the CHO condition is considered met when the actual L3 measurement satisfies the CHO condition, or when the predicted L3 measurement satisfies the CHO condition and has a corresponding confidence level greater than the threshold.

[0317] In the second case, the CHO condition is considered met when both the actual L3 measurement and the predicted L3 measurement satisfy the CHO condition, and when the predicted L3 measurement has a corresponding confidence level greater than the threshold.

[0318] In the third case, the CHO condition is considered met when the predicted L3 measurement satisfies the CHO condition and has a corresponding confidence level greater than the threshold (e.g., without referencing the actual L3 measurement in one way or another).

[0319] It should be noted that the additional case corresponds to the CHO condition to be satisfied when the actual L3 measurement satisfies the CHO condition (e.g., not referencing the predicted L3 measurement in one way or another).

[0320] The UE behavior regarding which of these conditions must be met for the CHO condition to be considered can be configured by the source base station 1904 at the UE 1902 (e.g., via RRC signaling).

[0321] If more than one target cell (e.g., each of the first target cell 1906 and the second target cell 1908) satisfies the CHO condition, the UE may select one of the target cells based on the configured priority values ​​of these target cells. Regarding this mechanism, it should be noted that the source base station 1904 may use DL signaling (e.g., MAC-CE or DCI) to cause a change in the priority of the target cell.

[0322] Flowchart 1900 illustrates the scenario where UE 1902 determines that the second target cell 1908 meets the CHO conditions for the applicable CHO event of CHO configuration 1918 and therefore performs a HO to it. As shown, UE 1902 performs RACH procedure 1930 with the second target cell 1908 to attach to the second target cell 1908. Once the HO to the second target cell 1908 is completed, UE 1902 sends a CHO completion message 1932 to the second target cell 1908 (e.g., ...). RRCReconfigurationComplete The CHO completion message (via the second target cell 1908) indicates to the network that the CHO has occurred.

[0323] Implementation plan for LTM fault handling of the CHO process using predictive measurements If CHO execution fails (e.g., due to a RACH failure with the selected target cell), the UE may perform the cell selection process by considering the information provided in CHO Configuration 1918. For example, if the configured cell is suitable, the UE may select the target cell via the following priority ranking rules. First, the UE may follow any configured priority value (if configured). Then, the UE may select the cell based on whether the relevant CHO conditions satisfy the actual measurements of the cell. Finally, the UE may select the cell based on the confidence level (for predicted measurements).

[0324] It should be noted that these priority ordering rules are given as examples rather than as restrictions. It is conceivable that the priority ordering rules used in this case may vary depending on the specific implementation of the UE.

[0325] Figure 20 A method 2000 for a UE according to the embodiments discussed herein is illustrated. Method 2000 includes transmitting to the network 2002 a notification message identifying one or more measurement prediction models at the UE. Method 2000 also includes receiving from the network 2004 an activation message for a first measurement prediction model available for use among the one or more measurement prediction models at the UE. Method 2000 further includes generating 2006 one or more actual measurements of one or more reference signals received at the UE from a cell of the network. Method 2000 further includes generating 2008 one or more predicted measurements based on the one or more actual measurements using the first measurement prediction model. Method 2000 further includes transmitting to the network 2010 a first measurement report including one or more predicted measurements.

[0326] In some implementations of method 2000, one or more predicted measurements include one or more predicted L3 cell-level measurements.

[0327] In some such implementations, one or more actual measurements include one or more actual L3 cell-level measurements for one or more reference signals received at the UE; and one or more predicted L3 cell-level measurements are generated by providing one or more actual L3 cell-level measurements to the first measurement prediction model using a first measurement prediction model.

[0328] In some such implementations, one or more actual measurements include one or more actual L1 beam-level measurements of one or more reference signals received at the UE; and one or more predicted L3 cell-level measurements are generated using a first measurement prediction model by: providing one or more actual L1 beam-level measurements to the first measurement prediction model to generate one or more predicted L1 beam-level measurements; and performing linear averaging and L3 filtering on the one or more actual L1 beam-level measurements and the one or more predicted L1 beam-level measurements. In some of these cases, the L3 filter coefficients used for L3 filtering are generated by the measurement prediction model.

[0329] In some such implementations, one or more actual measurements include one or more actual L1 beam-level measurements of one or more reference signals received at the UE; and one or more predicted L3 cell-level measurements are generated using a first measurement prediction model by: performing a linear average on one or more actual L1 beam-level measurements; and providing the first measurement prediction model with one or more actual linear average results and a configured set of L3 filter coefficients.

[0330] In some such implementations, one or more actual measurements include one or more actual L1 beam-level measurements of one or more reference signals received at the UE; and one or more predicted L3 cell-level measurements are generated using a first measurement prediction model by providing one or more actual L1 beam-level measurements and a set of L3 filter coefficients configured to the first measurement prediction model.

[0331] In some such implementations, one or more predicted L3 cell-level measurements are for the cell’s neighboring cells; and the generation of one or more predicted L3 cell-level measurements using a first measurement prediction model is also based on the correlation information of the neighboring cells.

[0332] In some such implementations, method 2000 further includes receiving configuration information identifying frequencies from the network, and one or more of the predicted L3 cell-level measurements are frequency-specific.

[0333] In some such implementations, method 2000 further includes receiving configuration information from the network that identifies conditions for generating one or more predicted L3 cell-level measurements, wherein the one or more predicted L3 cell-level measurements are generated in response to a determination at the UE that the conditions have been met.

[0334] In some such implementations, method 2000 further includes receiving configuration information identifying a cell from the network, and wherein the UE selects to generate one or more predicted L3 cell-level measurements based on the cell identification in the configuration information.

[0335] In some embodiments of method 2000, one or more predicted measurements include one or more predicted L3 beam-level measurements.

[0336] In some such implementations, one or more actual measurements include one or more actual L3 beam-level measurements of one or more reference signals received at the UE; and one or more predicted L3 beam-level measurements are generated by providing one or more actual L3 beam-level measurements to the first measurement prediction model using a first measurement prediction model.

[0337] In some such implementations, one or more actual measurements include one or more actual L1 beam-level measurements of one or more reference signals received at the UE; and one or more predicted L3 beam-level measurements are generated using a first measurement prediction model by: providing one or more actual L1 beam-level measurements to the first measurement prediction model to generate one or more predicted L1 beam-level measurements; and performing L3 filtering on the one or more actual L1 beam-level measurements and the one or more predicted L1 beam-level measurements.

[0338] In some such implementations, one or more actual measurements include one or more actual L1 beam-level measurements of one or more reference signals received at the UE; and one or more predicted L3 beam-level measurements are generated using a first measurement prediction model by providing one or more actual L1 beam-level measurements to the first measurement prediction model.

[0339] In some such implementations, one or more reference signals are received on one or more beams; the one or more predicted L3 beam-level measurements are for neighboring beams of the one or more beams that are not part of the one or more beams; and the generation of one or more predicted L3 beam-level measurements using a first measurement prediction model is also based on statistical information of the channel between the UE and the cell.

[0340] In some such implementations, method 2000 further includes receiving configuration information identifying the beam from the network, and one or more of the predicted L3 beam-level measurements are beam-specific.

[0341] In some such implementations, method 2000 further includes receiving configuration information from the network that identifies conditions for generating one or more predicted L3 beam-level measurements, wherein the one or more predicted L3 beam-level measurements are generated in response to a determination at the UE that the conditions have been met.

[0342] In some such implementations, method 2000 further includes receiving configuration information identifying a beam from the network, and wherein the UE selects to include predicted L3 cell-level measurements for the beam in one or more predicted L3 beam-level measurements based on the beam identifier in the configuration information.

[0343] In some embodiments of method 2000, one or more predicted measurements include one or more predicted L1 beam-level measurements.

[0344] In some such implementations, one or more actual measurements include one or more actual L1 beam-level measurements for one or more reference signals received at the UE; and one or more predicted L1 beam-level measurements are generated by providing one or more actual L1 beam-level measurements to the first measurement prediction model using a first measurement prediction model.

[0345] In some such implementations, one or more reference signals are received on one or more beams; the one or more predicted L1 beam-level measurements are for neighboring beams of the one or more beams that are not part of the one or more beams.

[0346] In some such implementations, method 2000 further includes receiving configuration information identifying the beam from the network, and one or more of the predicted L1 beam-level measurements are beam-specific.

[0347] In some such implementations, method 2000 further includes receiving configuration information from the network that identifies conditions for generating one or more predicted L1 beam-level measurements, wherein the one or more predicted L1 beam-level measurements are generated in response to determining at the UE that the conditions have been met.

[0348] In some such implementations, method 2000 further includes receiving configuration information identifying a beam from the network, and wherein the UE selects to include predicted L1 cell-level measurements for the beam in one or more predicted L1 beam-level measurements based on the beam identifier in the configuration information.

[0349] In some implementations, method 2000 further includes receiving configuration information from the network indicating predicted measurement periodicity and actual measurement periodicity; and transmitting a second measurement report to the UE according to the actual measurement periodicity, including one or more actual measurements; wherein the first measurement report, including the one or more predicted measurements, is transmitted according to the predicted measurement periodicity.

[0350] In some embodiments of method 2000, the transmission of a first measurement report to the network, which includes one or more predicted measurements, is triggered by one or more predicted measurements.

[0351] In some embodiments of method 2000, the transmission of a first measurement report to the network, which includes one or more predicted measurements, is triggered by actual measurements of one or more reference signals.

[0352] In some embodiments of method 2000, the first measurement report also includes an indication that one or more predicted measurements are predictive measurements.

[0353] In some embodiments of method 2000, the first measurement report also includes one or more actual measurements of one or more reference signals.

[0354] In some embodiments of method 2000, the notification message further includes one or more of the following: the predicted optimal L3 filter coefficients; the predicted optimal measurement report event type; the predicted optimal TTT for the measurement report event; the predicted optimal threshold for the measurement report event; the recommended cell for the actual measurement; the recommended beam for the first actual measurement in one or more actual measurements; and the recommended T304 timer value.

[0355] In some implementations, method 2000 further includes transmitting auxiliary information to the network, which includes one or more of the following: the deployment geometry of nearby base stations; a first statistic corresponding to temporal correlation; a second statistic corresponding to inter-cell correlation; and a third statistic corresponding to inter-beam correlation.

[0356] Figure 21 Method 2100 of a RAN according to the implementation discussed herein is illustrated. Method 2100 includes receiving 2102 a measurement prediction model from a UE. Method 2100 also includes transmitting 2104 one or more reference signals to the UE. Method 2100 also includes receiving 2106 actual measurements of the one or more reference signals from the UE. Method 2100 also includes generating 2108 one or more predicted measurements based on the actual measurements of the one or more reference signals using the measurement prediction model; wherein the measurement prediction model is one of the following: an L3 cell-level measurement prediction model; an L3 beam-level measurement prediction model; and an L1 beam-level measurement prediction model.

[0357] In some implementations, method 2100 also includes transmitting configuration information to the UE for use by the UE to retrain the measurement prediction model.

[0358] In some implementations, method 2100 also includes receiving a timestamp from the UE to generate the actual measurement.

[0359] In some implementations, method 2100 further includes receiving the UE's location and the UE's movement orientation from the UE.

[0360] In some implementations, method 2100 also includes receiving a change in the UE's mobility orientation from the UE.

[0361] Figure 22 A method 2200 for a UE according to the implementation discussed herein is illustrated. Method 2200 includes generating 2202 one or more predictions based on a first reference signal received at the UE from a cell in the network using a measurement prediction model. Method 2200 also includes generating 2204 one or more actual measurements corresponding to the one or more predictions by measuring a second reference signal received at the UE from the cell. 2200 further includes calculating 2206 a confidence level using the one or more predicted measurements and the one or more actual measurements. Method 2200 also includes reporting 2208 the confidence level to the network.

[0362] In some implementations of method 2200, calculating the confidence level includes determining the MSE between the predicted measurement and the actual measurement.

[0363] In some implementations, method 2200 also includes receiving an instruction from the network to stop using the measurement prediction model.

[0364] In some implementations, method 2200 further includes reporting to the network a first timestamp of the generated predicted measurement and a second timestamp of the generated actual measurement.

[0365] In some implementations, method 2200 also includes reporting the UE's location and UE's movement orientation to the network.

[0366] Figure 23 Method 2300 of a UE according to the implementation discussed herein is illustrated. Method 2300 includes generating 2302 one or more predictions based on a first reference signal received at the UE from a cell in the network using a measurement prediction model. Method 2300 also includes generating 2304 one or more actual measurements corresponding to the one or more predictions by measuring a second reference signal received at the UE from the cell. Method 2300 further includes calculating 2306 a confidence level using the one or more predicted measurements and the one or more actual measurements. Method 2300 also includes stopping 2308 the use of the measurement prediction model based on the confidence level.

[0367] In some implementations of method 2300, calculating the confidence level includes determining the MSE between the predicted measurement and the actual measurement.

[0368] Figure 24Method 2400 of a UE according to the implementation scheme discussed herein is illustrated. Method 2400 includes transmitting to the network 2402 a notification message identifying one or more RSTD-based TA prediction models at the UE. Method 2400 also includes receiving from the network 2404 an activation message identifying a first RSTD-based TA prediction model available for use among the one or more RSTD-based TA prediction models at the UE. 2400 further includes generating 2406 one or more actual RSTD-based TA measurements based on one or more reference signals received at the UE from one or more target cells of the network and the TA value of the serving cell of the network. Method 2400 further includes generating 2408 one or more predicted RSTD-based TA measurements of the first target cell based on one or more actual RSTD-based TA measurements of the one or more target cells using the first RSTD-based TA prediction model. Method 2400 also includes transmitting to the network 2410 an RSTD-based TA prediction report including one or more predicted RSTD-based TA measurements.

[0369] In some implementations of method 2400, one or more reference signals are reference signals of the first target cell.

[0370] In some embodiments of method 2400, one or more reference signals are reference signals not transmitted by the first target cell.

[0371] In some implementations of method 2400, the RSTD-based TA prediction report also includes the validity period of one or more predicted RSTD-based TA measurements for the first target cell.

[0372] In some implementations of method 2400, the RSTD-based TA prediction report also includes confidence levels of one or more predicted RSTD-based TA measurements for the first target cell.

[0373] In some implementations, method 2400 further includes receiving configuration information from the network indicating the predicted RSTD-based TA measurement periodicity and the actual RSTD-based TA measurement periodicity; and transmitting to the UE an actual RSTD-based TA report including one or more actual RSTD-based TA measurements of a first target cell according to the actual RSTD-based TA measurement periodicity; including that the RSTD-based TA prediction report of one or more predicted RSTD-based TA measurements of the first target cell is transmitted according to the predicted RSTD-based TA measurement periodicity.

[0374] In some embodiments of method 2400, transmitting an RSTD-based TA prediction report to the network, which includes one or more predicted RSTD-based TA measurements, is triggered by the UE determining that a first predicted RSTD-based TA measurement in one or more predicted RSTD-based TA measurements of the first target cell differs from a previously predicted RSTD-based TA measurement of the first target cell by at least a threshold.

[0375] In some embodiments of method 2400, the RSTD-based TA prediction report also includes an indication that one or more predicted RSTD-based TA measurements are predictive RSTD-based TA measurements.

[0376] In some implementations of method 2400, the RSTD-based TA prediction report also includes one or more actual measurements of one or more reference signals.

[0377] In some implementations of method 2400, one or more predicted RSTD-based TA measurements are first ranked in an RSTD-based TA prediction report based on the Layer 3 (L3) measurement of the corresponding cell and then based on the confidence level associated with one or more predicted RSTD-based TA measurements.

[0378] Figure 25 Method 2500 of a RAN according to the implementation discussed herein is illustrated. Method 2500 includes receiving 2502 an RSTD-based TA prediction model from a UE. Method 2500 also includes transmitting 2504 one or more reference signals from one or more target cells to the UE. Method 2500 also includes receiving 2506 one or more actual RSTD-based TA measurements of the one or more reference signals from the UE. Method 2500 also includes generating 2508 one or more predicted RSTD-based TA measurements of a first target cell based on one or more actual RSTD-based TA measurements of the one or more target cells using the RSTD-based TA prediction model.

[0379] In some implementations, method 2500 also includes transmitting configuration information to the UE for use by the UE to retrain the measurement prediction model.

[0380] In some implementations, method 2500 further includes receiving from the UE a timestamp for generating one or more actual RSTD-based TA measurements.

[0381] In some implementations, method 2500 further includes receiving the UE's location and the UE's movement orientation from the UE.

[0382] In some implementations, method 2500 also includes receiving a change in the UE's mobility orientation from the UE.

[0383] Figure 26 Method 2600 of a UE according to the implementation scheme discussed herein is illustrated. Method 2600 includes generating 2602 one or more predicted RSTD-based TA measurements based on a first reference signal received at the UE from one or more target cells of the network using an RSTDTA prediction model. Method 2600 also includes generating 2604 one or more actual RSTD-based TA measurements corresponding to the one or more predicted RSTD-based TA measurements by measuring a second reference signal received at the UE from one or more target cells. Method 2600 further includes calculating 2606 a confidence level using the one or more predicted RSTD-based TA measurements and the one or more actual RSTD-based TA measurements. Method 2600 also includes reporting 2608 the confidence level to the network.

[0384] In some implementations of method 2600, calculating the confidence level includes determining the MSE between the predicted RSTD-based TA measurement and the actual RSTD-based TA measurement.

[0385] In some implementations, method 2600 also includes receiving an instruction from the network to stop using the RSTD-based TA prediction model.

[0386] In some implementations, method 2600 further includes reporting to the network a first timestamp of generating the predicted RSTD-based TA measurement and a second timestamp of generating the actual RSTD-based TA measurement.

[0387] In some implementations, method 2600 also includes reporting the UE's location and UE's movement orientation to the network.

[0388] In some implementations, method 2600 also includes receiving a change in the UE's mobility orientation from the UE.

[0389] Figure 27 Method 2700 of a UE according to the implementation discussed herein is illustrated. 2700 includes generating 2702 one or more predicted RSTD-based TA measurements based on a first reference signal received at the UE from one or more target cells of the network using an RSTD TA prediction model. Method 2700 also includes generating 2704 one or more actual RSTD-based TA measurements corresponding to the one or more predicted RSTD-based TA measurements by measuring a second reference signal received at the UE from one or more target cells. Method 2700 further includes calculating 2706 a confidence level using the one or more predicted RSTD-based TA measurements and the one or more actual RSTD-based TA measurements. 2700 also includes stopping the use of the RSTD-based TA prediction model based on the confidence level.

[0390] In some implementations of method 2700, calculating the confidence level includes determining the MSE between the predicted RSTD-based TA measurement and the actual RSTD-based TA measurement.

[0391] Figure 28 A method 2800 for a source base station of a RAN according to an embodiment discussed herein is illustrated. Method 2800 includes receiving 2802 TA information corresponding to communication between a UE and one or more target cells from one or more target cells corresponding to one or more target base stations. Method 2800 also includes receiving 2804 L1 measurements corresponding to one or more target cells from the UE. Method 2800 further includes selecting 2806 a first target cell among the one or more target cells for handover to the UE based on the L1 measurements. Method 2800 further includes generating 2808 an early TA corresponding to a predicted TA for the UE and the first target cell based on the TA information using an early TA prediction model. Method 2800 further includes transmitting 2810 a MAC-CE commanding the UE to perform handover to the first target cell to the UE, wherein the MAC-CE includes the predicted early TA corresponding to the UE and the first target cell.

[0392] In some embodiments, method 2800 further includes transmitting to the UE an applicability condition request corresponding to an applicability condition for selecting an early TA prediction model from one or more early TA prediction models at the RAN; receiving from the UE an applicability condition response indicating a value of the applicability condition; and selecting an early TA prediction model from one or more early TA prediction models based on the value of the applicability condition. In some such embodiments, the applicability condition includes a threshold for the UE's speed, the applicability condition request includes a request for a value for the UE's speed, and the applicability response includes a value for the UE's speed.

[0393] In some implementations of method 2800, the TA information includes a TA value and a timestamp corresponding to the TA value.

[0394] Figure 29 A method 2900 for a UE according to the implementation discussed herein is illustrated. Method 2900 includes determining, 2902, that an LTM handover to a first target cell has failed based on the expiration of the LTM supervisor time. Method 2900 also includes performing cell selection, 2904, to a second target cell, which is identified by first evaluating one or more actual L1 measurements from high to low, and then evaluating one or more predicted L1 measurements in order of one or more confidence levels corresponding to one or more predicted L1 measurements.

[0395] Figure 30A method 3000 for a source base station of a RAN according to an embodiment discussed herein is illustrated. Method 3000 includes receiving from a UE 3002 a first or more predicted L3 measurements corresponding to a first target cell of a first target base station. Method 3000 also includes performing a first CHO preparation for the first target cell with the first target base station based on the predicted L3 measurements corresponding to the first target cell 3004. Method 3000 further includes transmitting to the UE 3006 a CHO configuration including a first condition for performing a first handover to the first target cell and a first indication of whether a second or more predicted L3 measurements corresponding to the first target cell can be used to evaluate the first condition. Method 3000 further includes receiving from the UE 3008 a CHO configuration response in response to the CHO configuration.

[0396] In some implementations of method 3000, the CHO configuration includes a confidence level threshold for evaluating the first condition using a second or more predicted L3 measurement.

[0397] In some implementations, method 3000 further includes receiving from the UE a third or more predicted Layer 3 (L3) measurement of a second target cell corresponding to a second target base station; and performing a second CHO preparation for the second target cell with the second target base station based on the third or more predicted L3 measurement corresponding to the second target cell; wherein the CHO configuration further includes a second condition for performing a second handover to the second target cell and a second indication of whether a fourth or more predicted L3 measurement corresponding to the second target cell can be used to evaluate the second condition.

[0398] In some implementations of method 3000, the CHO configuration also includes a second condition for performing a first handover to a first target cell and a second indication of whether a second or more predicted L3 measurements can be used to evaluate the second condition.

[0399] In some implementations of method 3000, the CHO configuration also includes a priority value for the first target cell.

[0400] In some implementations of method 3000, the CHO configuration response includes a proposed change to the list of target cells used by the RAN.

[0401] In some implementations of method 3000, the CHO configuration response includes an update to one or more predicted L3 measurements corresponding to the first target cell.

[0402] In some implementations of method 3000, the CHO configuration response includes a prediction error metric for the first or more predicted L3 measurements.

[0403] In some embodiments of method 3000, the CHO configuration response includes a suggested change to the CHO configuration. In some such embodiments, the suggested change to the CHO configuration includes one or more of a suggested change to the CHO event type, a suggested change to the threshold of the CHO event, and a suggested change to TTT.

[0404] In some implementations of method 3000, the CHO configuration response includes a suggested priority value change for the priority value of the first target cell.

[0405] In some implementations, method 3000 further includes transmitting an update to the CHO configuration to the UE based on information received from the UE in the CHO configuration response.

[0406] In some implementations, method 3000 also includes receiving from the UE a UAI message that includes a proposed change to the list of target cells used by the RAN.

[0407] In some implementations, method 3000 further includes receiving from the UE a UAI message that includes a proposed target cell for performing unconditional handover.

[0408] In some implementations, method 3000 further includes receiving from the UE a UAI message that includes a proposed priority value change for a priority value of a first target cell.

[0409] Figure 31 A method 3100 for a UE according to the implementation discussed herein is illustrated. Method 3100 includes receiving from a source base station of the network 3102 a first CHO configuration including a first condition for performing a first handover to a first target cell of a first target base station and a first indication of whether the first condition can be evaluated using one or more predicted L3 measurements corresponding to the first target cell. Method 3100 further includes transmitting a CHO configuration response to the network 3104 in response to the first CHO configuration. Method 3100 further includes evaluating 3106 that the first condition of the first CHO configuration has been met. Method 3100 further includes initiating 3108 a first handover to the first target cell in response to the evaluation that the first condition of the first CHO configuration has been met.

[0410] In some implementations, method 3100 further includes generating a second or more predicted L3 measurement corresponding to the first target cell; and transmitting the second or more predicted L3 measurement corresponding to the first target cell to the network before receiving the first CHO configuration from the network.

[0411] In some implementations of method 3100, the CHO configuration response includes an update to a second or more predicted L3 measurement corresponding to the first target cell.

[0412] In some implementations of method 3100, the CHO configuration response includes a prediction error metric for a second or more predicted L3 measurement.

[0413] In some implementations of method 3100, the first CHO configuration includes a confidence level threshold for evaluating the first condition using one or more first predicted L3 measurements.

[0414] In some embodiments of method 3100, the first CHO configuration further includes a second condition for performing a second handover to the second target cell and a second indication of whether the second condition can be evaluated using a second or more predicted L3 measurement corresponding to the second target cell.

[0415] In some embodiments of method 3100, the first CHO configuration further includes a second condition for performing a first handover to a first target cell and a second indication of whether the second condition can be evaluated using one or more of the first predicted L3 measurements.

[0416] In some implementations of method 3100, the first CHO configuration also includes a priority value for the first target cell.

[0417] In some implementations of method 3100, the CHO configuration response includes a suggested change to the list of target cells used by the network.

[0418] In some embodiments of method 3100, the CHO configuration response includes a suggested change to the first CHO configuration. In some such embodiments, the suggested change to the CHO configuration includes one or more of a suggested change to the CHO event type, a suggested change to the threshold of the CHO event, and a suggested change to TTT.

[0419] In some implementations of method 3100, the CHO configuration response includes a suggested priority value change for the priority value of the first target cell.

[0420] In some implementations, the method 3100 also includes receiving an update to the configuration of the first CHO from the network.

[0421] In some implementations, method 3100 also includes transmitting a UAI message to the network that includes a proposed change to the list of target cells used by the RAN.

[0422] In some implementations, method 3100 also includes transmitting to the network a UAI message including a proposed target cell for performing an unconditional handover.

[0423] In some implementations, method 3100 further includes transmitting a UAI message to the network that includes a proposed priority value change for a priority value of the first target cell.

[0424] In some implementations of method 3100, evaluating a first condition that has been met by the first CHO configuration includes at least one of the following: determining one or more actual L3 measurement satisfaction conditions for the first target cell; and determining one or more predicted L3 measurement satisfaction conditions.

[0425] In some implementations of method 3100, evaluating the first condition that the first CHO configuration has been met includes each of the following: determining one or more actual L3 measurement satisfaction conditions for the first target cell; and determining one or more predicted L3 measurement satisfaction conditions.

[0426] In some implementations of method 3100, in response to a comparison between the first priority value of the first target cell and the second priority value of the second target cell of the second target base station that has met the second condition of the second CHO configuration, a first handover to the first target cell is further initiated.

[0427] In some implementations, method 3100 further includes determining that a first handover to the first target cell has failed; and performing cell selection to the second target cell based on a priority value configured for the second target cell at the second target base station.

[0428] In some implementations, method 3100 further includes determining that a first handover to a first target cell has failed; and performing cell selection to a second target cell based on determining that a second CHO configuration for a second handover to a second target cell for a second target base station is satisfied.

[0429] In some implementations, method 3100 further includes determining that a first handover to a first target cell has failed; and performing cell selection to a second target cell based on a confidence level of a second or more predicted L3 measurements corresponding to a second target cell at a second target base station.

[0430] Figure 32 An example architecture of a wireless communication system 3200 according to the embodiments disclosed herein is illustrated. The following description is provided for an example wireless communication system 3200 operating in conjunction with LTE system standards and / or 5G or NR system standards provided by 3GPP technical specifications.

[0431] like Figure 32As shown, the wireless communication system 3200 includes UE 3202 and UE 3204 (but any number of UEs may be used). In this example, UE 3202 and UE 3204 are exemplified as smartphones (e.g., handheld touchscreen mobile computing devices capable of connecting to one or more cellular networks), but may also include any mobile or non-mobile computing device configured for wireless communication.

[0432] UE 3202 and UE 3204 can be configured to communicatively couple with RAN 3206. In an implementation, RAN 3206 can be NG-RAN, E-UTRAN, etc. UE 3202 and UE 3204 utilize connections (or channels) with RAN 3206 (shown as connection 3208 and connection 3210, respectively), each connection including a physical communication interface. RAN 3206 may include one or more base stations (such as base station 3212 and base station 3214) implementing connection 3208 and connection 3210.

[0433] In this example, Connection 3208 and Connection 3210 are air interfaces that enable this type of communication coupling and can conform to the RAT used by RAN 3206, such as LTE and / or NR, for example.

[0434] In some implementations, UE 3202 and UE 3204 may also exchange communication data directly via sidelink interface 3216. UE 3204 is shown configured to access an access point (shown as AP 3218) via connection 3220. As an example, connection 3220 may include a local wireless connection, such as a connection conforming to any IEEE 802.11 protocol, wherein AP 3218 may include Wi-Fi. ® Router. In this example, AP 3218 can connect to another network (e.g., the Internet) without using CN 3224.

[0435] In the implementation, UE 3202 and UE 3204 may be configured to communicate with each other or with base station 3212 and / or base station 3214 on a multi-carrier communication channel using orthogonal frequency division multiplexing (OFDM) communication signals according to various communication technologies, such as but not limited to orthogonal frequency division multiple access (OFDMA) communication technology (e.g., for downlink communication) or single-carrier frequency division multiple access (SC-FDMA) communication technology (e.g., for uplink and ProSe or sidelink communication), but the scope of the implementation is not limited in this respect. The OFDM signal may include multiple orthogonal subcarriers.

[0436] In some implementations, all or some of the base stations in base station 3212 or base station 3214 may be implemented as one or more software entities running on a server computer as part of a virtual network. Furthermore, or in other implementations, base station 3212 or base station 3214 may be configured to communicate with each other via interface 3222. In implementations where the wireless communication system 3200 is an LTE system (e.g., when CN 3224 is an EPC), interface 3222 may be an X2 interface. This X2 interface may be defined between two or more base stations (e.g., two or more eNBs, etc.) connected to the EPC and / or between two eNBs connected to the EPC. In implementations where the wireless communication system 3200 is an NR system (e.g., when CN 3224 is a 5GC), interface 3222 may be an Xn interface. The Xn interface is defined between two or more base stations (e.g., two or more gNBs, etc.) connected to the 5GC, between a base station 3212 (e.g., a gNB) connected to the 5GC and an eNB, and / or between two eNBs connected to the 5GC (e.g., CN 3224).

[0437] RAN 3206 is shown communicatively coupled to CN 3224. CN 3224 may include one or more network elements 3226 configured to provide various data and telecommunications services to customers / subscribers (e.g., users of UE 3202 and UE 3204) connected to CN 3224 via RAN 3206. Components of CN 3224 may be implemented in a single physical device or a separate physical device including components for reading and executing instructions from machine-readable or computer-readable media (e.g., non-transitory machine-readable storage media).

[0438] In the implementation scheme, CN 3224 can be an EPC, and RAN 3206 can be connected to CN 3224 via S1 interface 3228. In the implementation scheme, S1 interface 3228 can be divided into two parts: an S1 user plane (S1-U) interface, which carries service data between base station 3212 or base station 3214 and the serving gateway (S-GW); and an S1-MME interface, which is the signaling interface between base station 3212 or base station 3214 and the mobility management entity (MME).

[0439] In the implementation scheme, CN 3224 may be a 5GC, and RAN 3206 may be connected to CN 3224 via NG interface 3228. In the implementation scheme, NG interface 3228 may be divided into two parts: an NG user plane (NG-U) interface, which carries service data between base station 3212 or base station 3214 and user plane function (UPF); and an S1 control plane (NG-C) interface, which is the signaling interface between base station 3212 or base station 3214 and access and mobility management function (AMF).

[0440] Generally, application server 3230 can be an element that provides Internet Protocol (IP) bearer resources (e.g., packet-switched data services) for use with CN 3224. Application server 3230 can also be configured to support one or more communication services (e.g., VoIP sessions, group communication sessions, etc.) for UE 3202 and UE 3204 via CN 3224. Application server 3230 can communicate with CN 3224 via IP communication interface 3232.

[0441] Figure 33 A system 3300 for performing signaling 3334 between a wireless device 3302 and a network device 3318 according to an embodiment disclosed herein is illustrated. System 3300 may be part of a wireless communication system as described herein. Wireless device 3302 may be, for example, a UE (User Equipment) of a wireless communication system. Network device 3318 may be, for example, a base station (e.g., an eNB or gNB) of a wireless communication system.

[0442] Wireless device 3302 may include one or more processors 3304. Processor 3304 is executable instructions that cause various operations of wireless device 3302 to be performed as described herein. Processor 3304 may include one or more baseband processors, which are implemented using, for example, a central processing unit (CPU), digital signal processor (DSP), application-specific integrated circuit (ASIC), controller, field-programmable gate array (FPGA) device, another hardware device, firmware device, or any combination thereof configured to perform the operations described herein.

[0443] Wireless device 3302 may include memory 3306. Memory 3306 may be a non-transitory computer-readable storage medium that stores instructions 3308, which may include, for example, instructions executed by processor 3304. Instructions 3308 may also be referred to as program code or a computer program. Memory 3306 may also store data used by processor 3304 and results calculated by the processor.

[0444] Wireless device 3302 may include one or more transceivers 3310, which may include radio frequency (RF) transmitter circuitry and / or receiver circuitry, which use antenna 3312 of wireless device 3302 to facilitate signaling (e.g., signaling 3334) to and / or from wireless device 3302 and other devices (e.g., network device 3318) according to the corresponding RAT.

[0445] Wireless device 3302 may include one or more antennas 3312 (e.g., one, two, four or more). In embodiments with multiple antennas 3312, wireless device 3302 may utilize spatial diversity of such multiple antennas 3312 to transmit and / or receive multiple different data streams on the same time and frequency resources. This behavior may be referred to as, for example, multiple-input multiple-output (MIMO) behavior (referring to multiple antennas used at each of the transmitting and receiving devices to implement this aspect). MIMO transmission by wireless device 3302 may be achieved according to pre-decoding (or digital beamforming) applied at wireless device 3302, which multiplexes data streams across antennas 3312 based on known or assumed channel characteristics, such that each data stream is received with appropriate signal strength relative to the others at a desired location in the spatial domain (e.g., the location of the receiver associated with that data stream). Some implementations may use a single-user MIMO (SU-MIMO) approach (where all data streams are directed to a single receiver) and / or a multi-user MIMO (MU-MIMO) approach (where individual data streams may be directed to individual (different) receivers at different locations in the airspace).

[0446] In some implementations with multiple antennas, wireless device 3302 can implement analog beamforming technology, whereby the phase of the signal transmitted by antenna 3312 is relatively adjusted so that the (joint) transmission of antenna 3312 can be directed (this is sometimes referred to as beam control).

[0447] Wireless device 3302 may include one or more interfaces 3314. Interfaces 3314 can be used to provide input to or from wireless device 3302. For example, wireless device 3302 as a UE may include interfaces 3314, such as microphones, speakers, touchscreens, and buttons, to allow users of the UE to make inputs and / or outputs to the UE. Other interfaces of such a UE may consist of transmitters, receivers, and other circuitry that allow communication between the UE and other devices (e.g., in addition to the transceiver 3310 / antenna 3312 already described), and may be based on known protocols (e.g., Wi-Fi). ® and Bluetooth ® (etc.) to perform the operation.

[0448] Wireless device 3302 may include prediction module 3316. Prediction module 3316 may be implemented via hardware, software, or a combination thereof. For example, prediction module 3316 may be implemented as a processor, circuitry, and / or instructions 3308 stored in memory 3306 and executed by processor 3304. In some examples, prediction module 3316 may be integrated within processor 3304 and / or transceiver 3310. For example, prediction module 3316 may be implemented via a combination of software components (e.g., executed by a DSP or general-purpose processor) and hardware components (e.g., logic gates and circuitry) within processor 3304 or transceiver 3310.

[0449] The prediction module 3316 can be used in various aspects of this disclosure, for example, Figures 1 to 19 All aspects. The prediction module 3316 can be configured to enable the wireless device 3302 to perform UE-based functionalities corresponding to L3 beam-level measurement prediction, L1 measurement prediction, TA prediction, and / or the use of CHO, as discussed herein.

[0450] Network device 3318 may include one or more processors 3320. Processor 3320 is executable instructions that cause various operations of network device 3318 to be performed as described herein. Processor 3320 may include one or more baseband processors, which are implemented using, for example, a CPU, DSP, ASIC, controller, FPGA device, another hardware device, firmware device, or any combination thereof configured to perform the operations described herein.

[0451] Network device 3318 may include memory 3322. Memory 3322 may be a non-transitory computer-readable storage medium that stores instructions 3324, which may include, for example, instructions executed by processor 3320. Instructions 3324 may also be referred to as program code or a computer program. Memory 3322 may also store data used by processor 3320 and results calculated by the processor.

[0452] Network device 3318 may include one or more transceivers 3326, which may include RF transmitter circuitry and / or receiver circuitry that uses the antenna 3328 of network device 3318 to facilitate signaling (e.g., signaling 3334) to and / or from network device 3318 and other devices (e.g., wireless device 3302) in accordance with the corresponding RAT.

[0453] Network device 3318 may include one or more antennas 3328 (e.g., one, two, four or more). In embodiments having multiple antennas 3328, network device 3318 may perform MIMO, digital beamforming, analog beamforming, beam control, etc., as described.

[0454] Network device 3318 may include one or more interfaces 3330. Interfaces 3330 can be used to provide input to or output to network device 3318. For example, network device 3318 as a base station may include interfaces 3330 consisting of transmitters, receivers, and other circuitry (e.g., in addition to the transceiver 3326 / antenna 3328 already described), which enable the base station to communicate with other equipment in the core network and / or enable the base station to communicate with external networks, computers, databases, etc., for the purpose of performing operations, management, and maintenance of the base station or other equipment operatively connected to the base station.

[0455] Network device 3318 may include prediction module 3332. Prediction module 3332 may be implemented via hardware, software, or a combination thereof. For example, prediction module 3332 may be implemented as a processor, circuitry, and / or instructions 3324 stored in memory 3322 and executed by processor 3320. In some examples, prediction module 3332 may be integrated within processor 3320 and / or transceiver 3326. For example, prediction module 3332 may be implemented via a combination of software components (e.g., executed by a DSP or general-purpose processor) and hardware components (e.g., logic gates and circuitry) within processor 3320 or transceiver 3326.

[0456] The prediction module 3332 can be used in various aspects of this disclosure, for example, Figures 1 to 19 All aspects. The prediction module 3332 can be configured to enable the network device 3318 to perform base station-based functionalities corresponding to L3 beam-level measurement prediction, L1 measurement prediction, TA prediction, and / or CHO usage, as discussed herein.

[0457] The embodiments contemplated herein include an apparatus comprising components for performing one or more elements of any one or more of methods 2000, 2200, 2300, 2400, 2600, 2700, 2900, and / or 3100. The apparatus may be, for example, a UE (such as wireless device 3302 as a UE, as described herein).

[0458] The embodiments contemplated herein include one or more non-transitory computer-readable media comprising instructions to cause the electronic device to perform one or more elements of any one or more of methods 2000, 2200, 2300, 2400, 2600, 2700, 2900, and / or 3100 when executed by one or more processors of the electronic device. The non-transitory computer-readable medium may, for example, be a memory of the UE (such as memory 3306 of a wireless device 3302 serving as a UE, as described herein).

[0459] The embodiments contemplated herein include an apparatus comprising logic components, modules, or circuitry for performing one or more elements of any one or more of methods 2000, 2200, 2300, 2400, 2600, 2700, 2900, and / or 3100. The apparatus may be, for example, a UE (such as wireless device 3302 as a UE, as described herein).

[0460] The embodiments contemplated herein include an apparatus comprising: one or more processors; and one or more computer-readable media including instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of any one or more of methods 2000, 2200, 2300, 2400, 2600, 2700, 2900, and / or 3100. The apparatus may be, for example, a UE (such as wireless device 3302 as a UE, as described herein).

[0461] The implementation schemes envisioned herein include signals described or associated with one or more elements of any one or more methods such as method 2000, method 2200, method 2300, method 2400, method 2600, method 2700, method 2900 and / or method 3100.

[0462] The embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution of the program by a processor will cause the processor to perform one or more elements of any one or more of methods 2000, 2200, 2300, 2400, 2600, 2700, 2900, and / or 3100. The processor may be a processor of the UE (such as processor 3304 as a wireless device 3302 of the UE, as described herein). These instructions may, for example, be located in the processor and / or in the memory of the UE (such as memory 3306 as a wireless device 3302 of the UE, as described herein).

[0463] The embodiments contemplated herein include an apparatus comprising components for performing one or more elements of any one or more of methods 2100, 2500, 2800, and / or 3000. The apparatus may be, for example, a base station (such as network device 3318 as a base station, as described herein).

[0464] The embodiments contemplated herein include one or more non-transitory computer-readable media comprising instructions to cause the electronic device to perform one or more elements of any one or more methods of method 2100, method 2500, method 2800, and / or method 3000 when executed by one or more processors of the electronic device. The non-transitory computer-readable medium may be, for example, the memory of a base station (such as memory 3322 of a network device 3318 serving as a base station, as described herein).

[0465] The embodiments contemplated herein include an apparatus comprising logic components, modules, or circuitry for performing one or more elements of any one or more of methods 2100, 2500, 2800, and / or 3000. The apparatus may be, for example, an apparatus for a base station (such as network device 3318 as a base station, as described herein).

[0466] The embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media including instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of any one or more methods of method 2100, method 2500, method 2800, and / or method 3000. The apparatus may be, for example, an apparatus for a base station (such as network device 3318 as a base station, as described herein).

[0467] The implementation scheme envisioned herein includes a signal as described in or associated with one or more elements of any one or more methods of method 2100, method 2500, method 2800, and / or method 3000.

[0468] The embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution by a processing element causes the processing element to perform one or more elements of any one or more of methods 2100, 2500, 2800, and / or 3000. The processor may be a processor of a base station (such as processor 3320 of network device 3318 as a base station, as described herein). These instructions may, for example, be located in the processor and / or in the memory of the base station (such as memory 3322 of network device 3318 as a base station, as described herein).

[0469] For one or more embodiments, at least one of the components illustrated in one or more of the foregoing figures may be configured to perform one or more operations, techniques, processes, and / or methods as described herein. For example, a baseband processor as described herein in conjunction with one or more of the foregoing figures may be configured to operate according to one or more of the examples illustrated herein. Similarly, circuitry associated with a UE, base station, network element, etc., as described above in conjunction with one or more of the foregoing figures may be configured to operate according to one or more of the examples illustrated herein.

[0470] Unless otherwise expressly stated, any of the embodiments described above may be combined with any other embodiment (or combination of embodiments). The foregoing description of one or more specific embodiments provides illustrative and descriptive information, but is not intended to be exhaustive or to limit the scope of the embodiments to the precise form disclosed. In light of the teachings above, modifications and variations are possible, or modifications and variations may be derived from practice with various embodiments.

[0471] Implementations and specific embodiments of the systems and methods described herein may include various operations embodied in machine-executable instructions to be executed by a computer system. The computer system may include one or more general-purpose or special-purpose computers (or other electronic devices). The computer system may include hardware components, including specific logical parts for performing the operations; or may include a combination of hardware, software, and / or firmware.

[0472] It should be recognized that the systems described herein include descriptions of specific implementations. These implementations may be combined into a single system, partially integrated into other systems, divided into multiple systems, or otherwise partitioned or combined. Furthermore, it is conceivable to use parameters, attributes, aspects, etc., of one implementation in one implementation. For clarity, these parameters, attributes, aspects, etc., are described only in one or more implementations, and it should be recognized that, unless expressly stated herein, these parameters, attributes, aspects, etc., may be combined with or substituted for parameters, attributes, aspects, etc., of another implementation.

[0473] As is widely recognized, the use of personally identifiable information should comply with privacy policies and practices that are generally accepted to meet or exceed industry or governmental requirements for protecting user privacy. Specifically, personally identifiable information data should be managed and processed to minimize the risk of unintentional or unauthorized access or use, and the nature of authorized use should be clearly explained to users.

[0474] Although the foregoing has been described in considerable detail for clarity, it will be apparent that certain changes and modifications can be made without departing from the principles of the invention. It should be noted that there are many alternative ways to implement both the processes and apparatus described herein. Therefore, embodiments of the invention should be considered illustrative rather than restrictive, and this description is not limited to the details given herein, but can be modified within the scope of the appended claims and their equivalents.

Claims

1. A method for a user equipment (UE), the method comprising: Transmit to the network a notification message identifying one or more timing advance (TA) prediction models based on reference signal time difference (RSTD) at the UE; Receive from the network an activation message for the first RSTD-based TA prediction model available for use among the one or more RSTD-based TA prediction models at the UE; One or more actual RSTD-based TA measurements are generated based on one or more reference signals received at the UE from one or more target cells of the network and the TA value of the serving cell of the network. The first RSTD-based TA prediction model is used to generate one or more predicted RSTD-based TA measurements for the first target cell based on the one or more actual RSTD-based TA measurements of the one or more target cells. as well as Transmit an RSTD-based TA prediction report to the network, including one or more predicted RSTD-based TA measurements.

2. The method according to claim 1, wherein the one or more reference signals are reference signals of the first target cell.

3. The method according to claim 1, wherein the one or more reference signals are reference signals not transmitted by the first target cell.

4. The method of claim 1, wherein the RSTD-based TA prediction report further includes the validity period of the one or more predicted RSTD-based TA measurements of the first target cell.

5. The method of claim 1, wherein the RSTD-based TA prediction report further includes the confidence level of the one or more predicted RSTD-based TA measurements of the first target cell.

6. The method according to claim 1, further comprising: Receive configuration information from the network indicating the predicted RSTD-based TA measurement periodicity and the actual RSTD-based TA measurement periodicity; as well as Based on the actual RSTD-based TA measurement, the system periodically transmits an actual RSTD-based TA report, including one or more actual RSTD-based TA measurements of the first target cell, to the UE. This includes the RSTD-based TA prediction report of one or more predicted RSTD-based TA measurements of the first target cell, which is periodically transmitted based on the predicted RSTD-based TA measurements.

7. The method of claim 1, wherein transmitting the RSTD-based TA prediction report, which includes the one or more predicted RSTD-based TA measurements, to the network is triggered by the UE determining that a first predicted RSTD-based TA measurement in the one or more predicted RSTD-based TA measurements of the first target cell differs from a previously predicted RSTD-based TA measurement of the first target cell by at least a threshold.

8. The method of claim 1, wherein the RSTD-based TA prediction report further includes an indication that the one or more predicted RSTD-based TA measurements are predictive RSTD-based TA measurements.

9. The method of claim 1, wherein the RSTD-based TA prediction report further includes the one or more actual measurements of the one or more reference signals.

10. The method of claim 1, wherein the one or more predicted RSTD-based TA measurements are first ranked in the RSTD-based TA prediction report based on the Layer 3 (L3) measurement of the corresponding cell and then based on the confidence level associated with the one or more predicted RSTD-based TA measurements.

11. A method for a radio access network (RAN), the method comprising: Reference signal reception time difference (RSTD) timing advance (TA) prediction model from user equipment (UE); Transmit one or more reference signals from one or more target cells to the UE; One or more actual RSTD-based TA measurements received from the UE of the one or more reference signals; as well as The RSTD-based TA prediction model is used to generate one or more predicted RSTD-based TA measurements for the first target cell based on the one or more actual RSTD-based TA measurements of the one or more target cells.

12. The method according to claim 11, further comprising: The configuration information to be used by the UE to retrain the measurement prediction model is transmitted to the UE.

13. The method according to claim 11, further comprising: The UE receives the timestamps from which the one or more actual RSTD-based TA measurements are generated.

14. The method according to claim 11, further comprising: Receive the UE's location and the UE's movement orientation from the UE.

15. The method according to claim 11, further comprising: The UE receives the change in the UE's movement orientation.

16. A method for a user equipment (UE), the method comprising: The Reference Signal Time Difference (RSTD) Timing Advance (TA) prediction model generates one or more predicted RSTD-based TA measurements based on a first reference signal received at the UE from one or more target cells in the network. One or more actual RSTD-based TA measurements corresponding to the one or more predicted RSTD-based TA measurements are generated by measuring a second reference signal received at the UE from the one or more target cells; The confidence level is calculated using the one or more predicted RSTD-based TA measurements and the one or more actual RSTD-based TA measurements; as well as Report the stated confidence level to the network.

17. The method of claim 16, wherein calculating the confidence level includes determining the mean square error (MSE) between the predicted RSTD-based TA measurement and the actual RSTD-based TA measurement.

18. The method according to claim 16, further comprising: Receive an instruction from the network to stop using the RSTD-based TA prediction model.

19. The method according to claim 16, further comprising: The network is reported a first timestamp for generating the predicted RSTD-based TA measurement and a second timestamp for generating the actual RSTD-based TA measurement.

20. The method of claim 16, further comprising: The location of the UE and the UE's movement orientation are reported to the network.

21. The method according to claim 16, further comprising: The UE receives the change in the UE's movement orientation.

22. A method for a user equipment (UE), the method comprising: The Reference Signal Time Difference (RSTD) Timing Advance (TA) prediction model generates one or more predicted RSTD-based TA measurements based on a first reference signal received at the UE from one or more target cells in the network. One or more actual RSTD-based TA measurements corresponding to the one or more predicted RSTD-based TA measurements are generated by measuring a second reference signal received at the UE from the one or more target cells; The confidence level is calculated using the one or more predicted RSTD-based TA measurements and the one or more actual RSTD-based TA measurements; as well as The use of the RSTD-based TA prediction model will be discontinued based on the stated confidence level.

23. The method of claim 22, wherein calculating the confidence level includes determining the mean square error (MSE) between the predicted RSTD-based TA measurement and the actual RSTD-based TA measurement.

24. A method for a source base station in a radio access network (RAN), the method comprising: Receive timing advance (TA) information corresponding to communication between a user equipment (UE) and the one or more target cells corresponding to one or more target base stations; Receive L1 measurements corresponding to the one or more target cells from the UE; Based on the L1 measurement, a first target cell among the one or more target cells is selected for the handover of the UE; The early TA prediction model is used to generate an early TA corresponding to the UE and the first target cell based on the TA information; as well as A Media Access Control (MAC-CE) element is transmitted to the UE, instructing the UE to perform a handover to the first target cell, wherein the MAC-CE includes the predicted early TA corresponding to the UE and the first target cell.

25. The method according to claim 24, further comprising: Transmit to the UE an applicability condition request corresponding to the applicability conditions for selecting the early TA prediction model from one or more early TA prediction models at the RAN. Receive from the UE an applicability condition response indicating the value of the applicability condition; as well as The early TA prediction model is selected from the one or more early TA prediction models based on the value of the applicability condition.

26. The method of claim 25, wherein the applicability condition includes a threshold for the speed of the UE, the applicability condition request includes a request for a value for the speed of the UE, and the applicability response includes the value for the speed of the UE.

27. The method of claim 24, wherein the TA information includes a TA value and a timestamp corresponding to the TA value.

28. A method for a user equipment (UE), the method comprising: The expiration of the mobility (LTM) supervisor timer triggered by Layer 1 (L1) / Layer 2 (L2) determines that the LTM handover to the first target cell has failed; and Cell selection is performed on the second target cell, which is identified in the following way: First, evaluate one or more actual L1 measurements from high to low; and Then, the one or more predicted L1 measurements are evaluated in order of one or more confidence levels corresponding to the one or more predicted L1 measurements.

29. An apparatus comprising components for performing the method according to any one of claims 1 to 28.

30. A computer-readable medium comprising instructions to cause the electronic device to perform the method according to any one of claims 1 to 28 when the instructions are executed by one or more processors of the electronic device.

31. An apparatus comprising a logic component, module, or circuit for performing the method according to any one of claims 1 to 28.

Images