Method and apparatus for model-based rrm measurement for wireless communication system

Abstract

An apparatus and method for radio resource management (RRM) measurement prediction is provided. In one novel aspect, a user equipment (UE) measures a first set of reference signals of a first set of cells and predicts a second set of reference signals of a second set of cells. In one embodiment, the prediction is made in a time domain, a frequency domain, or a spatial domain, or a combination of multiple domains. In one embodiment, the one or more first measurement results are a set of L1 measurement results or a set of L3 measurement results. The one or more predicted results are an L1 cell quality of the second set of cells, an L1 measurement result of the second set of cells, a 3rd layer (L3) cell quality of the second set of reference signals, or an L3 measurement result of the second set of reference signals.

Description

[0001] Cross-references

[0002] This application is filed pursuant to 35 USC §111(a) and enjoys priority under 35 USC §120 and §365(c) for international application PCT / CN2023 / 142377 entitled “Method and Apparatus for AI / ML-based RRM Measurement for Wireless Communication Systems”, filed on December 27, 2023. The entire contents of the foregoing documents are incorporated herein by reference. Technical Field

[0003] The disclosed embodiments generally relate to wireless communication, and more specifically, to model-based wireless resource management (RRM) measurements. Background Technology

[0004] With the rapid development of wireless communication, more efficient processes are needed, such as radio resource management (RRM) measurement, location, and mobility processes in 5G and future 6G. In the legacy networks of the 3rd Generation Partnership Project (3GPP) 5G New Radio (NR), considering the mobility of user equipment (UE), the UE continuously measures and reports the radio resource signal environment quality to maintain service continuity and quality. Traditional handover (HO) design is controlled by a series of L3 processes, including RRM measurement and radio resource control (RRC) reconfiguration, which involve a large amount of signaling and latency. Conditional handover is introduced to allow the UE to make handover decisions based on predefined conditions, thereby reducing signaling overhead and speeding up HO. The Dual Active Protocol Stack (DAPS) aims to allow the UE to communicate with both the source and target cells simultaneously during handover by maintaining two active protocol stacks, minimizing interruptions. To further reduce latency, overhead, and downtime during UE movement, L1 / L2-triggered Mobility (LTM) based on mobility mechanisms has been enhanced, enabling the serving cell to handover via beam management and L1 / L2 signaling. However, with the ever-increasing demands for capacity and efficiency, traditional measurement procedures can no longer keep pace with the exponential growth of mobile networks.

[0005] Therefore, there is a need for devices and mechanisms for model-based RRM measurements in wireless communication systems. Summary of the Invention

[0006] This application provides an apparatus and method for predicting radio resource management (RRM) measurements. In a novel aspect, a user equipment (UE) performs measurements on a first set of reference signals for a first group of cells and predicts a second set of reference signals for a second group of cells. In one embodiment, the prediction is performed in the time domain using one or more historical measurements as model input and one or more future measurements as labels; in the frequency domain using one or more measurements of a first subset of the bandwidth part (BWP) as model input and measurements of the full BWP or a second subset of the BWP as labels; in the spatial domain using one or more measurements of a first group of partial beams as input and measurements of the full beam group or a second group of partial beams as labels; or in a combination of domains, including a combination of time and spatial domains, a combination of time and frequency domains, a combination of spatial and frequency domains, or a combination of time, spatial, and frequency domains. In one embodiment, one or more first measurements are a set of L1 measurements obtained by the UE performing layer-1 (L1) filtering on the corresponding measurements of the first set of reference signals. In one embodiment, the prediction result is the L1 cell quality of the second group of cells, the layer-3 (L3) cell quality of the second group of cells, or the L3 measurement result of the second group of reference signals. In one embodiment, one or more first measurement results are a set of L3 measurement results obtained by the user equipment performing L3 filtering on the corresponding measurement results of the first group of reference signals, and the prediction result is the L3 measurement result of the second group of reference signals.

[0007] This content is not intended to limit the scope of the invention. The invention is defined by the claims. Attached Figure Description

[0008] In the accompanying drawings, the same numbers represent the same components, illustrating embodiments of the present invention.

[0009] Figure 1 To illustrate the system diagram, an example wireless network supporting wireless resource management (RRM) measurements based on an artificial intelligence-machine learning (AI-ML) model is shown according to an embodiment of the present invention.

[0010] Figure 2 An exemplary diagram of radio resources based on AI-ML RRM reference signals is shown according to an embodiment of the present invention, wherein the radio resources are defined by spatial domain, time domain, frequency domain, and any combination thereof.

[0011] Figure 3 An exemplary diagram illustrating the overall flow of a high-rise measurement model according to an embodiment of the present invention is shown.

[0012] Figure 4 An exemplary diagram illustrates the overall flow of unfiltered measurement results for AI-ML-based RRM prediction according to an embodiment of the present invention.

[0013] Figure 5 An exemplary diagram illustrating the overall flow of L1 measurement results (after L1 filtering) for AI-ML-based RRM prediction according to an embodiment of the present invention is shown.

[0014] Figure 6 An exemplary diagram illustrating the overall flow of L3 measurement results for AI-ML-based RRM prediction according to an embodiment of the present invention is shown.

[0015] Figure 7 An exemplary diagram illustrating the overall process for predicting cell quality for RRM measurement according to an embodiment of the present invention is shown.

[0016] Figure 8 An exemplary flowchart of an AI-ML-based RRM prediction measurement process is shown according to an embodiment of the present invention.

[0017] Figure 9 An exemplary flowchart illustrating how a user equipment performs RRM measurements based on an AI-ML model according to an embodiment of the present invention is shown. Detailed Implementation

[0018] Reference will now be made in detail to certain embodiments of the invention, examples of which are illustrated in the accompanying drawings.

[0019] Several aspects of a telecommunications system will now be introduced in conjunction with various devices and methods. These devices and methods will be described in detail below and illustrated by various modules, components, circuits, processes, algorithms, etc. (collectively referred to as "elements") shown in the accompanying drawings. These elements can be implemented by electronic hardware, computer software, or any combination of both. Whether these elements are implemented in hardware or software depends on the specific application and design constraints imposed on the overall system.

[0020] Figure 1This is a schematic system diagram illustrating a wireless network supporting Radio Resource Management (RRM) measurements based on an Artificial Intelligence-Machine Learning (AI-ML) model according to embodiments of the present invention. The wireless communication network 100 includes one or more fixed base station infrastructure units, forming a network distributed across a geographical area. The base station unit may also be referred to as an access point, access terminal, base station, Node-B, eNode-B (eNB), gNB, or other terms used in the art. For example, a base station provides services to multiple mobile stations within its service area, such as a cell or a cell sector. In some systems, one or more base stations are connected to a controller to form an access network, which is connected to one or more core networks. gNB 102, gNB 107, and gNB 108 are base stations in the wireless network, and their service areas may overlap or not. gNB 102 is connected to gNB 107 via Xn interface 121. gNB 102 is connected to gNB 108 via Xn interface 122. gNB 107 is connected to gNB 108 via Xn interface 123. Core network (CN) entity 103 is connected to gNBs 102 and 107 via NG interfaces 125 and 126, respectively. CN 103 is connected to gNB 108 via NG interface 127. CN 103 includes core components such as user plane function (UPF) and access and mobility management function (AMF). UE 101 supports AI-ML functionality. For example, UE 101 is connected to gNB 102 via link 111. UE 101 can also establish a mobility link with the neighboring gNB 108 via link 112.

[0021] Figure 1This further demonstrates a simplified interaction process between UE 101 and gNBs 102 and 108. The latest Artificial Intelligence (AI) and Machine Learning (ML) technologies are increasingly being integrated into wireless networks to improve performance, enhance user experience, and reduce complexity and overhead. For example, AI-ML technology has been used to predict Top-K beams in the time or spatial domains, helping to reduce measurement overhead. With increased mobility—i.e., faster / more frequent UE movement or higher cell density—measurement overhead and power consumption increase accordingly. AI / ML is used to predict measurements to reduce unnecessary measurements while providing proactive measurements for handover (HO). For example, in step 181, gNB 102 or 108 provides measurement configuration to UE 101 via RRC configuration signaling. The gNB can request UE 101 to perform in-frequency and out-of-frequency measurements. In step 182, UE 101 performs measurements and evaluates the conditions for a measurement report based on the measurement results. In step 183, the actual measurement results are predicted using an AI-ML model. In one embodiment, the AI-ML model is located on the UE side. In step 184, UE 101 performs measurement prediction and reports the model output to gNB. Subsequently, gNB makes a handover decision and sends a handover command to UE 101.

[0022] Figure 1 A simplified block diagram of the RAN node / base station and mobile device / UE supporting RRM measurements based on AI-ML models is further illustrated. The gNB / RAN node has an antenna 156 for transmitting and receiving radio signals. An RF transceiver circuit 153 connected to the antenna 156 receives RF signals from the antenna 156, converts them into baseband signals, and sends them to the processor 152. The RF transceiver 153 also converts the baseband signals received by the processor 152 into RF signals and sends them to the antenna 156. The processor 152 processes the received baseband signals and invokes different functional modules to implement functions within the gNB. A memory 151 stores program instructions and data 155 for controlling the operation of the base station / gNB. The base station / gNB also includes a set of control modules 157 that perform functional tasks for communicating with the mobile station. These control modules can be implemented via circuitry, software, firmware, or a combination thereof.

[0023] Figure 1It also includes a simplified block diagram of the UE, such as UE 101. The UE may also be referred to as a mobile station, mobile terminal, mobile phone, smartphone, wearable device, IoT device, tablet, laptop, or other terms used in the technology. The UE performs functions such as RRM measurement based on an AI-ML model. The UE interacts with the gNB via an air interface. The UE has an antenna 166 for transmitting and receiving radio signals. An RF transceiver circuit 163 connected to the antenna receives RF signals from the antenna 166, converts them into baseband signals, and sends them to the processor 162. The RF transceiver 163 also converts the baseband signals received by the processor 162 into RF signals and sends them to the antenna 166. The processor 162 processes the received baseband signals and calls different functional modules to implement the functions in UE 101. Memory 161 stores program instructions and data 165 for controlling the operation of UE 101. Antenna 166 transmits uplink transmissions to antenna 156 of the base station / gNB and receives downlink transmissions.

[0024] The UE also includes a set of control modules that perform functional tasks. These control modules can be implemented through circuitry, software, firmware, or a combination thereof. Measurement module 191 performs a first measurement on a first set of reference signals (RS) for a first group of cells. Prediction module 192, based on the first measurement result, uses one or more artificial intelligence-machine learning (AI-ML) models to predict the second set of RS for a second group of cells, wherein the prediction is performed over one or more RS radio resource domains, including one or a combination of spatial, temporal, and frequency domains. Filtering module 193, based on the prediction result, determines the cell quality of the first group of cells, the second group of cells, or one or more of the first and second groups of cells. Reporting module 194 triggers one or more measurement reports based on one or more reporting conditions, wherein at least one reporting condition considers the first measurement result, the prediction result, or both the first measurement and the prediction result, and at least one or more measurement reports include the first measurement result, the prediction result, or both the first measurement and the prediction result.

[0025] Figure 2This invention illustrates radio resources for Radio Resource Management (RRM) reference signals based on an Artificial Intelligence-Machine Learning (AI-ML) model, defined by the spatial domain, time domain, frequency domain, and any combination of these domains, in an embodiment of the invention. In one embodiment 200, a User Equipment (UE) performs AI-ML model training. In step 250, the UE measures a first set of reference signals for a first group of cells and obtains a first measurement result 251. In step 260, the UE measures a second set of reference signals for a second group of cells during the training phase and obtains a second measurement result 261. In one embodiment, the UE uses the measurement result 251 of the first set of reference signals for the first group of cells as model input and the measurement result 261 of the second set of reference signals for the second group of cells as labels for model training. The UE generates or updates the AI-ML model 270 using the model input 271 and the labels 272. The UE may have one or more AI-ML models.

[0026] The radio resources of the reference signal are defined by the spatial domain, time domain, frequency domain, and any combination of these domains, including combinations of the time and spatial domains, combinations of the time and frequency domains, combinations of the spatial and frequency domains, and combinations of the time, spatial, and frequency domains. AI-ML model training and measurement prediction based on AI-ML models can be performed in different reference signal domains.

[0027] In one embodiment 210, the UE performs measurement prediction using one or more AI-ML models in the time domain. For example, the first set of reference signals (model input) of the first group of cells and the second set of reference signals (model labels) of the second group of cells are identical in both the frequency and spatial domains. During the training phase, the UE performs measurements from TN 211 to T 212 and a second measurement from T+1 to T+M 213. The AI-ML-based RRM model uses historical measurement results [TN, T] as model input and future measurement results [T+1, T+M] as labels. The UE possesses one or more trained AI-ML models. For example, the UE measures the first set of reference signals of the first group of cells at time [TN, T] to predict the result for the second set of reference signals of the second group of cells at time [T, T+M].

[0028] In one embodiment 220, the UE performs measurement prediction using one or more AI-ML models in the spatial domain. For example, the first set of reference signals for a first group of cells (e.g., cell A 221) and the second set of reference signals for a second group of cells (e.g., cell B 222) are identical in both the frequency and time domains. Solid lines represent the measured beam set, and dashed lines represent the predicted beam set. During the training phase, the AI-ML-based RRM model uses the measurement results of a subset of beams as model input and the full beam results as labels. In another embodiment, during the training phase, the AI-ML-based RRM model uses the measurement results of a subset of beams as model input and the results of another subset of beams as labels. The UE possesses one or more trained AI-ML models. For example, the UE measures the first set of reference signals for a first group of cells on the subset of beams represented by dashed lines to predict the results for the second set of reference signals for a second group of cells on the full beam set represented by solid lines or another subset of beams.

[0029] In one embodiment 230, the UE performs measurement prediction using one or more AI-ML models in the frequency domain. For example, the first set of reference signals of the first group of cells and the second set of reference signals of the second group of cells are identical in both the spatial and temporal domains. Solid lines represent the measured bandwidth portion BWP / frequency 231, and dashed lines represent the predicted full BWP / other BWP / frequency 232. The AI-ML-based RRM model uses the measurement results of the portion of BWP / frequency as model input and the results of the full BWP / frequency as labels. In another embodiment, the AI-ML-based RRM model uses the measurement results of the portion of BWP / frequency as model input and the results of another portion of BWP / frequency as labels. In one embodiment, the first set of reference signals and the second set of reference signals are at different frequencies. In another embodiment, the first set of reference signals and the second set of reference signals are at the same frequency.

[0030] In other embodiments, a combination of domains is employed. In one embodiment, the first set of reference signals for the first group of cells and the second set of reference signals for the second group of cells are any combination of these domains. In one embodiment, the first set of reference signals and the second set of reference signals are identical in the frequency domain, and the UE uses one or more AI-ML models to perform measurement prediction in the time-space domain, combining 210 and 220, using historical measurement results [TN, T] of a portion of the beam as model input, and using future measurement results [T+1, T+M] of the full beam or another portion of the beam as labels. In another embodiment, the first set of reference signals for the first group of cells and the second set of reference signals for the second group of cells are identical in the spatial domain, and the UE uses one or more AI-ML models to perform measurement prediction in the time-frequency domain, combining 210 and 230, using historical measurement results [TN, T] of a portion of the BWP / frequency as model input, and using future measurement results [T+1, T+M] of the full BWP / other BWP / frequency as labels. In another embodiment, the first set of reference signals of the first group of cells and the second set of reference signals of the second group of cells are identical in the time domain. The UE uses one or more AI-ML models to perform measurement prediction in the spatial-frequency domain. Combining 220 and 230, the UE uses the measurement results of a portion of the beam on a portion of the BWP / frequency as the model input and uses the results of the full beam or another portion of the beam on the full BWP / other BWP / frequency as the label. In one embodiment, the UE uses one or more AI-ML models to perform measurement prediction in the time-spatial-frequency domain. Combining 210, 220 and 230, the UE uses the historical measurement results [TN, T] of a portion of the beam on a portion of the BWP / frequency as the model input and uses the future measurement results [T+1, T+M] of the full beam or another portion of the beam on the full BWP / other BWP / frequency as the label.

[0031] Figure 3An exemplary illustration of the overall flow of a high-layer measurement model according to an embodiment of the present invention is shown. This exemplary measurement flow includes beam-level and cell-level filtering. K beams (pairs), such as 301, 302, and 303, correspond to measurements on SSB or CSI-RS resources, configured for L3 mobility by the gNB and detected by the User Equipment (UE) at L1. Point 311 represents the measurement within the physical layer (beam-specific sample). Layer 1 (L1) filtering 310 performs internal L1 filtering on the input measured at point 311, the specific implementation of which is determined by the UE. Point 313 represents the L1-filtered measurement reported to Layer 3 (L3) by L1. Beam combining 320 combines the beam-specific measurements to obtain cell quality at output 321. L3 filtering 330 for cell quality filters the measurement provided at point 321. Point 331 is the L3-filtered cell quality measurement. At 340, the UE evaluates the reporting conditions based on 331 and one or more optional other cell quality parameters 332. Measurement report message 341 is sent via the radio interface. L3 beam filtering 350 filters the measurement provided at point 313 and outputs 351. Beam selection 360 for beam reporting selects X measurements 361 from the L3 beam-filtered measurements 351. The behavior of beam combining / selection 320, L3 filter 330, reporting condition evaluation 340, L3 beam filtering 350, and beam selection 360 for beam reporting is standardized, and the configuration of the L3 filter is provided by RRC signaling at 325, 335, 345, 355, and 365, respectively. In one embodiment, the Radio Resource Management (RRM) measurement procedure based on an Artificial Intelligence-Machine Learning (AI-ML) model uses one or more AI-ML models to predict one or more beam measurements and / or cell quality. In one embodiment, in step 371, the UE performs a first measurement on a first set of reference signals RS for a first set of cells. In step 372, the UE, based on one or more first measurement results of the first measurement, uses one or more AI-ML models 370 to predict one or more prediction results for the second group of RSs of the second group of cells. The AI-ML models can be cell-specific or cluster-specific. In one embodiment, the prediction is performed over one or more RS radio resource domains, including spatial, time, and frequency domains, as shown in 380. The prediction can be performed in the time domain, spatial domain, frequency domain, a combination of time and spatial domains, a combination of time and frequency domains, a combination of spatial and frequency domains, or a combination of time, spatial, and frequency domains.

[0032] Figure 4An exemplary illustration of the overall flow of unfiltered measurement results for AI-ML-based Radio Resource Management (RRM) prediction according to an embodiment of the present invention is shown. In one embodiment, using the measurement results (unfiltered) of a first set of reference signals as model input, the UE first predicts the measurement results of a second set of reference signals, and then performs L1 and L3 filtering on the measurement results of the first and second sets of reference signals, respectively. The UE receives gNB signal 481. The UE predicts beam measurement results 482 using one or more AI-ML models 470. In step 410, the UE performs L1 filtering on the prediction results and outputs a filtered signal 411. In step 420, the UE performs beam combining / selection and outputs 421. In step 430, the UE performs L3 filtering on cell quality based on output 421 and outputs cell quality 431. In step 430, the UE obtains cell quality 441 based on cell quality 431 and one or more other cell quality indicators 432. In one embodiment, the first set of reference signals and the second set of reference signals are defined by the spatial domain, time domain, frequency domain, and any combination thereof. The behavior of beam combining / selection 420, L3 filter 430, reporting condition evaluation 440, L3 beam filtering 450, and beam selection for beam reporting 460 is standardized, and the configuration of the L3 filter is provided by RRC signaling at 425, 435, 445, 455, and 465, respectively.

[0033] Figure 5An exemplary illustration of the overall flow of L1 measurement results (after L1 filtering) for AI-ML-based Radio Resource Management (RRM) prediction according to an embodiment of the present invention is shown. In one embodiment, using the L1-filtered measurement results of a first set of reference signals as model input, the UE first predicts the measurement results of a second set of reference signals, and then performs L3 filtering on the measurement results of the first and second sets of reference signals respectively. In step 510, the UE performs L1 filtering on a set of reference signals 511 (e.g., 501, 502, and 503) and outputs the filtered result 512. The UE uses one or more AI-ML models 570 to predict the measurement results of the second set of reference signals and outputs predicted beam-1 581, predicted beam-2 582, and predicted beam-3 583. In step 520, the UE performs beam combining / selection based on the predicted beam 526 and outputs 521. In step 530, the UE performs L3 filtering on the cell quality based on 521 and outputs the cell quality 531. In step 540, the UE obtains cell quality based on 531 and one or more optional other cell quality 532. A measurement report message 541 is sent via the radio interface. L3 beam filtering 550 filters the measurements provided by the predicted beam 526 and outputs 551. Beam selection 560 for beam reporting selects X measurements 561 from the L3 beam-filtered measurements 551. The behavior of beam combining / selection 520, L3 filter 530, reporting condition evaluation 540, L3 beam filtering 550, and beam selection 560 for beam reporting are standardized, and the configuration of the L3 filter is provided by RRC signaling at 525, 535, 545, 555, and 565, respectively. In one embodiment, the first set of reference signals and the second set of reference signals are defined by the spatial domain, time domain, frequency domain, and any combination thereof.

[0034] Figure 6A schematic diagram illustrating the overall flow of L3 measurement results for Radio Resource Management (RRM) prediction based on Artificial Intelligence-Machine Learning (AI-ML) is provided, according to an embodiment of the present invention. In one embodiment, the User Equipment (UE) uses the measurement results of a first set of reference signals after L3 filtering as model input to first predict the measurement results of a second set of reference signals. In another embodiment, the UE first performs L3 filtering on the measurement results of the first set of reference signals, then performs measurement prediction on the second set of reference signals, followed by beam combining / selection to derive cell quality. In step 610, the UE performs L1 filtering on gNB reference signals 611 (such as beams 601, 602, and 603) and outputs the filtering result 612. In step 650, the UE performs L3 beam filtering. The filtering result is used to predict L3 beams through one or more AI-ML models 670, and outputs predicted beam 1 681, predicted beam 2 682, and predicted beam 3 683. The predicted beam 626 is used in step 620 to perform beam combining and selection, outputting result 631. In step 640, based on 631 and one or more optional other cell quality parameters 632, the UE determines the cell quality. A measurement report message 641 is transmitted via the radio interface. Beam selection 660 for beam reporting selects X measurement results 661 from the predicted beam set 627. The behaviors of beam combining / selection 620, reporting condition evaluation 640, L3 beam filtering 650, and beam selection 660 for beam reporting are standardized, and the configuration of the L3 filter is provided via RRC signaling at 625, 635, 655, and 665, respectively. In one embodiment, the first set of reference signals and the second set of reference signals are defined in the spatial domain, time domain, frequency domain, and any combination thereof.

[0035] Figure 7A schematic diagram of the overall process for predicting cell quality for RRM measurements is shown, according to an embodiment of the present invention. In one embodiment 700, the UE measures a first group of cells and predicts the measurement results for a second group of cells. In one embodiment, the first group of cells is the same as the predicted second group of cells. In another embodiment, the first group of cells is different from the predicted second group of cells. In one embodiment 701, the measurement results used for prediction are L1 measurement results (after L1 filtering). In one embodiment 702, the measurement results used for prediction are L3 measurement results (after L3 cell filtering). In step 710, the UE performs L1 filtering on a reference signal set 711 and outputs a filtered signal set B 712. In step 700, the UE performs cell quality prediction based on one or more AI-ML models. In step 720, the UE performs beam combining / selection based on 712. In one embodiment 701, the UE predicts cell quality 781 using one or more AI-ML models 771. In step 735, the UE performs L3 filtering on the cell quality and outputs the predicted cell quality C 731. In another embodiment 702, the UE predicts cell quality 782 based on L3 cell quality filtering using one or more AI-ML models 772 and outputs predicted cell quality C 731. In step 740, based on 731 and one or more optional other cell quality 732, the UE derives the cell quality. A measurement report message 741 is sent via the radio interface. L3 beam filtering 750 performs L3 filtering on the L1-filtered beam 712 and outputs 751. Beam selection 760 for beam reporting selects X measurement results 761 based on 751. The behaviors of beam combining / selection 720, L3 filtering for cell quality 730, reporting condition assessment 740, L3 beam filtering 750, and beam selection for beam reporting 760 are standardized, and the configuration of the L3 filter is provided via RRC signaling at 725, 735, 755, and 765, respectively. In one embodiment, the reference signals of the first group of cells and the reference signals of the second group of cells are defined in the spatial domain, time domain, frequency domain, and any combination thereof.

[0036] Figure 8A schematic diagram of the AI-ML-based RRM prediction measurement process is shown, according to an embodiment of the present invention. UE 801 is connected to the core network (CN network, abbreviated as CN) 802. In step 811, the UE performs measurement and model input construction. In step 812, the UE performs RRM measurement prediction. In one embodiment, for temporal beam quality prediction, the UE performs a full beam scan during the observation time window and stops beam scanning during the prediction time window. In one embodiment, for spatial beam quality prediction, the UE performs a partial beam scan. In one embodiment, for spatiotemporal beam quality prediction, the UE performs a partial beam scan during the observation time window. In step 813, the UE evaluates reporting conditions based on the predicted measurement results. In one embodiment, the UE considers the actual measured measurement results, the predicted measurement results, or both when evaluating the reporting conditions. In one embodiment, the reporting conditions are based on network configuration. In step 820, the UE sends a measurement report to the network based on the predicted measurement results. In one embodiment, the UE reports the actual measured measurement results, the predicted measurement results, or both. In one embodiment, the measurement report is configured by the network. In one embodiment, one or more AI-ML models are cell-specific, meaning one AI-ML model corresponds to one cell. In another embodiment, one or more AI-ML models are cluster-specific, meaning one AI-ML model corresponds to a group of cells. In one embodiment, the measured reference signal quality (RS) can be the signal-to-noise ratio (SNR), reference signal received power (RSRP), or received signal strength indicator (RSSI).

[0037] Figure 9 A flowchart illustrating a UE performing RRM measurements based on an AI-ML model, according to an embodiment of the present invention, is shown. In step 901, the UE performs one or more first measurements on a first set of reference signals (RS) for a first group of cells. In step 902, based on the one or more first measurement results, the UE uses one or more AI-ML models to predict one or more prediction results for a second set of RSs for a second group of cells, wherein the prediction is performed over one or more RS radio resource domains, the radio resource domains including one or a combination of spatial, temporal, and frequency domains. In step 903, based on the one or more prediction results, the UE determines the cell quality of the first group of cells, the second group of cells, or one or more cells from the first and second groups of cells.

[0038] Although the present invention has been described in conjunction with certain specific embodiments for educational purposes, the invention is not limited thereto. Therefore, various modifications, adaptations, and combinations of features to the described embodiments can be made without departing from the scope defined by the claims of the present invention.

Claims

1. A method for a user equipment to use an artificial intelligence-machine learning model in a wireless network, comprising: Perform a first measurement on a first group of reference signals for a first group of cells; Based on one or more first measurement results of the first measurement, one or more artificial intelligence-machine learning models are used to predict one or more prediction results of a second group of reference signals of a second group of cells, wherein the prediction is performed on one or more reference signal radio resource domains, the radio resource domains including one or a combination of a spatial domain, a time domain and a frequency domain. as well as Based on the one or more prediction results, determine the quality of one or more cells in the first group of cells, the second group of cells, or the first group of cells and the second group of cells.

2. The method as described in claim 1, wherein, At least one of the one or more artificial intelligence-machine learning models is trained by the user equipment using a model, using one or more first measurement results as model input, and using one or more second measurement results as labels, wherein the one or more second measurement results are obtained by the user equipment performing a second measurement on a second set of reference signals for the second set of cells.

3. The method as described in claim 1, wherein, The prediction is made using one or more historical measurements as model inputs and one or more future measurements as labels in the time domain.

4. The method of claim 1, wherein, The prediction is made in the frequency domain using one or more measurements of a first subset of the bandwidth portion as model input, and one or more measurements of a full bandwidth portion or a second subset of the bandwidth portion as labels.

5. The method of claim 1, wherein, The prediction is made in the spatial domain using one or more measurements from a first set of partial beams as input and one or more measurements from a full beam group or a second set of partial beams as labels.

6. The method of claim 1, wherein, The prediction is made on a combination of the time domain and the spatial domain, a combination of the time domain and the frequency domain, a combination of the spatial domain and the frequency domain, or a combination of the time domain, the spatial domain, and the frequency domain.

7. The method of claim 1, wherein, The one or more first measurement results are a set of first-level measurement results obtained by the user equipment through first-level filtering of the corresponding measurement results of the first set of reference signals.

8. The method of claim 7, wherein, The one or more prediction results are a Layer 1 cell quality of the second group of cells, a Layer 3 cell quality of the second group of cells, or one or more Layer 3 measurements of the second group of reference signals.

9. The method of claim 1, wherein, The one or more first measurement results are a set of third-level measurement results obtained by the user equipment through third-level filtering of the corresponding measurement results of the first set of reference signals, and the one or more prediction results are third-level measurement results of the second set of reference signals.

10. The method of claim 1, wherein, The first group of cells consists of serving cells or neighboring cells that operate on the same frequency or different frequencies, and the second group of cells consists of serving cells or neighboring cells that operate on the same frequency or different frequencies.

11. The method of claim 1, further comprising: One or more measurement reports are triggered based on one or more reporting conditions.

12. The method of claim 11, wherein, At least one of the one or more reporting conditions considers the one or more first measurement results, or the one or more prediction results, or the one or more first measurement results and the one or more prediction results.

13. The method of claim 11, wherein, At least one or more measurement reports include the one or more first measurement results, or the one or more prediction results, or the one or more first measurement results and the one or more prediction results.

14. The method of claim 1, wherein, At least one of the one or more artificial intelligence-machine learning models is cell-specific or cluster-specific.

15. A user equipment that uses an artificial intelligence-machine learning model in a wireless network, comprising: A transceiver used to transmit and receive radio frequency signals in this wireless network; A measurement module is used to perform a first measurement on a first set of reference signals of a first set of cells; A prediction module is configured to predict one or more prediction results of a second group of reference signals for a second group of cells based on one or more first measurement results of the first measurement, using one or more artificial intelligence-machine learning models, wherein the prediction is performed over one or more reference signal radio resource domains, the radio resource domains including one or a combination of a spatial domain, a time domain, and a frequency domain; and A quality module is used to determine the quality of one or more cells, including the first group of cells, the second group of cells, or the first group of cells and the second group of cells, based on one or more prediction results.

16. The user equipment as claimed in claim 15, wherein, At least one of the one or more artificial intelligence-machine learning models is trained for the user equipment using a model, using one or more first measurement results as model input and using one or more second measurement results as labels, wherein the one or more second measurement results are obtained by the user equipment performing a second measurement on the second set of reference signals for the second set of cells.

17. The user equipment as claimed in claim 15, wherein, The prediction is made using one or more historical measurements as model inputs and one or more future measurements as labels in the time domain. In this frequency domain, one or more measurements of a first subset of the bandwidth portion are used as model inputs, and one or more measurements of a full bandwidth portion or a second subset of the bandwidth portion are used as labels; In this spatial domain, one or more measurements from a first partial beamset are used as input, and one or more measurements from a full beamset or a second partial beamset are used as labels; or Performed on a combination of the time domain and the spatial domain, a combination of the time domain and the frequency domain, a combination of the spatial domain and the frequency domain, or a combination of the time domain, the spatial domain, and the frequency domain.

18. The user equipment as claimed in claim 15, wherein, The one or more first measurement results are a set of first-layer measurement results obtained by the user equipment through first-layer filtering of the corresponding measurement results of the first set of reference signals, and wherein the one or more prediction results are a first-layer cell quality of the second set of cells, a third-layer cell quality of the second set of cells, or one or more third-layer measurement results of the second set of reference signals.

19. The user equipment as claimed in claim 15, wherein, The one or more first measurement results are a set of third-layer measurement results obtained by the user equipment through third-layer filtering of the corresponding measurement results of the first set of reference signals, and the one or more prediction results are the third-layer measurement results of the second set of reference signals.

20. The user equipment of claim 15, further comprising: A reporting module that triggers one or more measurement reports based on one or more reporting conditions, wherein at least one of the one or more reporting conditions considers one or more first measurement results, one or more prediction results, or one or more first measurement results and one or more prediction results, and wherein at least one or more measurement reports include one or more first measurement results, one or more prediction results, or one or more first measurement results and one or more prediction results.

Images