Artificial intelligence / machine learning for radio resource management inter-frequency and cell level spatial prediction

Abstract

Systems and methods for artificial intelligence (AI) / machine learning (ML) for radio resource management (RRM) inter-frequency and cell level spatial prediction are discussed herein. A UE connects to a serving cell of a first plurality of cells configured with a first carrier frequency and receives, an AI / ML model trained by the serving cell to determine a prediction on a second carrier frequency from measurements based on the first carrier frequency. Then, the UE measures reference signals transmitted by the first plurality of cells on the first carrier frequency to generate cell feature information and processes the cell feature information, using the AI / ML model at the UE, to determine the prediction on the second carrier frequency.

Description

ARTIFICIAL INTELLIGENCE / MACHINE LEARNING FOR RADIO RESOURCEMANAGEMENT INTER-FREQUENCY AND CELL LEVEL SPATIAL PREDICTIONTECHNICAL FIELD

[0001] This application relates generally to wireless communication systems, including systems with beam prediction.BACKGROUND

[0002] Wireless mobile communication technology uses various standards and protocols to transmit data between a base station and a wireless communication device. Wireless communication system standards and protocols can include, for example. 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE) (e g., 4G), 3GPP New Radio (NR) (e.g., 5G), and Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard for Wireless Local Area Networks (WLAN) (commonly known to industry groups as Wi-Fi®).

[0003] As contemplated by the 3GPP, different wireless communication systems' standards and protocols can use various radio access networks (RANs) for communicating between a base station of the RAN (which may also sometimes be referred to generally as a RAN node, a network node, or simply a node) and a wireless communication device known as a user equipment (UE). 3GPP RANs can include, for example, Global System for Mobile communications (GSM), Enhanced Data Rates for 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 may use one or more radio access technologies (RATs) to perform communication between the base station and the UE. For example, the GERAN implements GSM and / or EDGE RAT, the UTRAN implements Universal Mobile Telecommunication System (UMTS) RAT or other 3GPP RAT, the E-UTRAN implements LTE RAT (sometimes simply referred to as LTE). and NG-RAN implements NR RAT (sometimes referred to herein as 5G RAT, 5G NR RAT. or simply NR). In certain deployments, the E-UTRAN may also implement NR RAT. In certain deployments, NG-RAN may also implement LTE RAT.14899-4873-7899,1 P70146WO1

[0005] A base station used by a RAN may correspond to that RAN. One example of an E-UTRAN base station is an Evolved Universal Terrestrial Radio Access Network (E- UTRAN) Node B (also commonly denoted as evolved Node B, enhanced Node B, eNodeB, or eNB). One example of an NG-RAN base station is a next generation Node B (also sometimes referred to as a g Node B or gNB).

[0006] A RAN provides its communication services with external entities through its connection to a core network (CN). For example, E-UTRAN may utilize an Evolved Packet Core (EPC) while NG-RAN may utilize a 5G Core Network (5GC).BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0007] To easily identify the discussion of any particular element or act, the most significant digit or digits in a reference number refer to the figure number in which that element is first introduced.

[0008] FIG. 1 illustrates an example of a measurement procedure for inter-frequency based prediction.

[0009] FIG. 2A illustrates an example of inter-frequency based beam prediction, according to embodiments herein.

[0010] FIG. 2B illustrates an example AI / ML model used in the inter-frequency based beam prediction, according to embodiments herein.

[0011] FIG. 3 A illustrates a second example of inter-frequency based beam prediction, according to embodiments herein.

[0012] FIG. 3B illustrates a second example AI / ML model used in the inter-frequency based beam prediction, according to embodiments herein.

[0013] FIG. 4A illustrates a third example of inter-frequency based beam prediction, according to embodiments herein.

[0014] FIG. 4B illustrates a third example AI / ML model used in the inter-frequency based beam prediction, according to embodiments herein.

[0015] FIG. 5 is a diagram illustrating training of an AI / ML model configured to generate predicted receive signal characteristics of Tx-Rx beam pairs, according to embodiments herein.

[0016] FIG. 6 illustrates an example of a 3D domain of reduced measurements, according to embodiments herein.24899-4873-7899,1 P70146WO1

[0017] FIG. 7 illustrates an example AI / ML classifier model used to perform interfrequency based beam prediction, according to embodiments herein.

[0018] FIG. 8 illustrates a method for a UE, according to embodiments herein.

[0019] FIG. 9 illustrates a method for a cell configured with a first carrier frequency, according to embodiments herein.

[0020] FIG. 10 illustrates an example architecture of a wireless communication system, according to embodiments disclosed herein.

[0021] FIG. 11 illustrates a sy stem for performing signaling between a wireless device and a network device, according to embodiments disclosed herein.DETAILED DESCRIPTION

[0022] Various embodiments are described with regard to a UE. However, reference to a UE is merely provided for illustrative purposes. The example embodiments may be utilized with any electronic component that may establish a connection to a network and is configured with the hardware, software, and / or firmware to exchange information and data with the network. Therefore, the UE as described herein is used to represent any appropriate electronic component.

[0023] In some wireless communication systems, for artificial intelligence (AI) / machine learning (ML) based mobility, enhancements in radio resource control (RRC) connected modes over an air interface follows an existing mobility framework (i.e., a handover decision is made at the network). Mobility use cases may use a standalone NR primary cell (PCell) change. Further, a UE-sided and a network-sided AI / ML model may both be used.

[0024] Various objectives for mobility may be considered. For example, testability and interoperability of AI / ML based mobility, as well as the impacts on radio resource management (RRM) requirements and performance, may be evaluated and considered.

[0025] Additionally, various cases for AI / ML based mobility have been considered and may be further specified. For example, in a first case, RRM measurement prediction may be considered as a first round of performance evaluation. In a second case, measurement event predictions may be considered. However they may be deprioritized. In a third case, radio link failure (RLF) and / or handover (HO) prediction may be considered. For example, the RLF prediction may be further specified based on initial considerations and simulation assumptions. The HO prediction may be deprioritized.34899-4873-7899,1 P70146WO1

[0026] In some wireless communication mechanisms, various protentional benefits and gains of AI / ML aided mobility for network triggered Layer 3 (L3)-based handover have been studied and evaluated. Such benefits may include various aspects such as AI / ML based RRM measurement and event prediction, cell level measurement prediction including intra and inter-frequency (using a UE sided or a network sided model), intercell beam-level measurement prediction for L3 mobility (using a UE sided or a network sided model), HO failure / RLF prediction (using a UE sided model), and measurement events prediction (using a UE sided model).

[0027] Additionally, the needs and / or benefits of UE assistance information for the network side model may be further considered. For example, the evaluation of the AI / ML aided mobility benefits may consider HO performance key performance indicators (KPIs) (e.g., ping-pong HO, handover failure (HOF) / RLF, time of stay, handover interruption, prediction accuracy and measurement reduction), etc.) and complexity tradeoffs. Note that a simulation assumption and methodology can include, for example, which are discussed in 3GPP technical report (TR) 38.901, 3GPP TR 38.843 and 3GPP TR 36.839. Potential AI / ML mobility specific enhancements may be based on a Rel-19 AI / ML air interface general framework (e.g., life cycle management (LCM), performance monitoring).

[0028] In some wireless communication systems, potential specification impacts of AI / ML aided mobility may be considered. For example, the testability, interoperability', and impacts on RRM requirements and performance may be evaluated. Note that such considerations may be defined by pre-allocated time units (TUs). Further, note that a two-sided model might not be included in such considerations.

[0029] In some wireless systems, three alternative methods for measurement reduction have been identified and may be further considered. In a first alternative, spatial prediction may be used wherein the UE does not measure every cell and / or beam. In a second alternative, temporal prediction is used wherein the UE may skip some of the measurement time steps (same cell as for temporal prediction). In a third alternative, inter-frequency prediction may be used wherein a set of measured or predicted cells or beams includes cells or beams from different frequency layers. Inter-frequency prediction may be utilized to reduce the number of inter-frequency measurements that a UE may need to perform. As a result, this reduction can also decrease the number of measurement gaps that the UE may be configured to use. It should be understood that in44899-4873-7899,1 P70146WO1current systems, the temporal prediction, spatial prediction, and inter-frequency prediction are done independently of each other.

[0030] However, in certain embodiments disclosed herein, temporal, spatial, and / or frequency based measurements and predictions may be combined. For example, spatial prediction of a cell may include using a first cell’s partial transmit (Tx) beam measurement results as a reduced set (set B) to predict all of the first cell’s Tx beams (setA). Cell prediction may use the first cell’s measurement results (cell level measurement results can be based on Tx beam averaging or based on best Tx beam) as set B, for example, to predict the first cell’s reference signal received power (RSRP) in a future prediction window using temporal prediction, to predict a second cell on a second carrier frequency (F2) (i.e., a collocated inter-frequency cell) using spatial / frequency predictions, to predict a second cell on first carrier frequency (Fl) (i.e., an intrafrequency non-collocated cell) using spatial prediction, or to predict a second cell on F2 (i.e., an inter-frequency non-collocated cell) using spatial / frequency prediction. As used herein, “Fl” may refer to a first frequency, a first carrier frequency, a serving carrier frequency, or a frequency layer. Similarly, “F2” may refer to a second frequency, a second carrier frequency, a secondary carrier frequency, or a frequency layer that is different than the frequency layer Fl.

[0031] In some embodiments, an AI / ML model may use Tx beam measurement results of a subset of Tx beams of a first plurality of cells at a first carrier frequency (Fl) to predict future Tx beams of the first plurality of cells using cell spatial and temporal prediction. In other embodiments, an AI / ML model may use Tx beam measurement of a first plurality of cells at a first carrier frequency (Fl) to predict future RSRP measurements for the first plurality of cells using cell temporal prediction. In other embodiments, an AI / ML model may use reduced Tx beam measurement results (i.e.. setB) of a first plurality of cells to predict the Tx beams of a second plurality of cells using intra-frequency non-collocated cell spatial prediction. In other embodiments, an AI / ML model may use cell measurement results for a first plurality of cells to predict RSRP values for a second plurality of cells using intra-frequency non-collocated cell prediction. In other embodiments, an AI / ML model may use Tx beam measurement results (set B) for a first plurality7of cells in Fl to predict Tx beams for the first plurality of cells in F2 using inter-frequency collocated cell spatial prediction. In other embodiments, an AI / ML model may use cell measurement results of a first plurality of54899-4873-7899,1 P70146WO1cells to predict RSRP values for the first plurality of cells using inter-frequency collocated cell prediction. In other embodiments, an AI / ML model may use Tx beam measurement results (set B) of a first plurality of cells to predict Tx beams of a second plurality of cells using inter-frequency non-collocated cell spatial prediction. In other embodiments, an AI / ML model may use cell measurement results of a first plurality of cells to predict RSRP values of a second plurality of cells using inter-frequency noncollocation.

[0032] In some wireless communication systems, for inter-frequency based prediction, it may be beneficial to reduce the measurement overhead across spatial and frequency domains. In order to determine the inter-frequency RSRP values, a UE may perform measurements across different sites (i.e., cells) at a same frequency layer (intrafrequency measurements), across different frequency layers (inter-frequency measurements), and different Tx / Rx beams.

[0033] For example, FIG. 1 illustrates an example of a measurement procedure for inter-frequency based prediction. A UE (not shown) may measure Tx beams from a plurality of sites 102a, 102b, 102c at a first frequency 108 (Freq A), a plurality of sites 104a, 104b, 104c at a second frequency 110 (Freq B), a plurality of sites 106a. 106b, 106c at a third frequency 112 (Freq C), etc. The sites at the different frequency layers may be non-collocated. Thus, the amount of overhead (i.e., resources, time, and power used for measurements) and of collected measurement data can be significant. In certain embodiments, the UE or the network may use an AI / ML model to predict RSRP measurements across different frequencies layers to generate a spatial / frequency domain estimation 114. For example, RSRP fingerprinting can be used to predict the RSRP values for all sites at one frequency layer from the measurements at one or more other frequency layers. In this way, measurement gaps can be significantly reduced (i.e.. synchronization signal block measurement timing configuration (SMTC) periodicity is increased).

[0034] In some cases, however, denser networks and an increasing number of frequency bands may make it challenging to efficiently assign an optimal serving cell and frequency carrier to a UE. This is particularly difficult to achieve without requiring the UE to perform an excessive number of inter-frequency measurements or to frequently report the quality of alternative (secondary ) carrier cells. Inter-frequency measurements may require UEs to reconfigure one of their receiver chains to tune into a different64899-4873-7899,1 P70146WO1carrier than that of the serving cell. If the UE has only a single receiver chain, it will be unable to maintain communication with the serving cell during these measurements, which may lead to a degradation in service quality. Additionally, inter-frequency measurements are power-intensive and should be minimized to preserve battery life.

[0035] Thus, in some embodiments, an AI / ML model may be used to predict interfrequency measurements by analyzing patterns in historical (previous) channel state information (CSI) data, user mobility trends, and other relevant network data. For example, if an AI / ML model can predict when a UE is likely to transition from one frequency layer to another, the AI / ML model may proactively initiate a measurement or handover, which may provide a seamless transition and prevent the signal quality from deteriorating.

[0036] Additionally, predicting inter-frequency measurements may require analyzing how users transition between different frequency layers as they move across various coverage areas. In some embodiments, by understanding these mobility patterns, the AI / ML model may anticipate when a UE is likely to switch to a new frequency layer and / or carrier frequency, which may optimize handover and reduce unnecessary measurements.

[0037] Predicting a strongest cell of a secondary carrier (F2) from a serving carrier (Fl)

[0038] Certain embodiments disclosed herein introduce a procedure for predicting a best or strongest cell on a secondary carrier frequency (F2), using only measurements from the serving carrier configured with a serving carrier frequency (Fl). As a result, such embodiments reduce or eliminate the need for direct inter-frequency measurements on the secondary carrier frequency, which optimizes the measurement and prediction process while maintaining accuracy in predicting the best / strongest cell. By allowing the serving cell to predict the strongest cell on a secondary' carrier for a UE, without requiring additional inter-frequency measurements from the UE, measurement overhead may be reduced while maintaining accurate cell selection on the secondary carrier.

[0039] In certain such embodiments, each macro cell on a primary' carrier trains a supervised AI / ML model using the strongest micro / macro cell identifier (ID) as the output class. The AI / ML model may be trained to predict which secondary carrier cell (at F2) will likely provide the strongest signal, based solely on measured data from the serving carrier frequency (Fl). UEs connect to a primary' carrier cell (e.g., a macro cell) and perform intra-frequency measurements on this carrier. In addition, the UEs may74899-4873-7899,1 P70146WO1carry out inter-frequency measurements on the secondary carrier to obtain ground truth data, which is used to validate the accuracy of the predicted strongest secondary carrier cell. The measurement dataset may take the form of Dn={xi,Vi}, for index i=l,—, n, where Xi G Rd, where d is the number of features and output ground truth y; is in a finite set. The output of the AI / ML model y represents, for a particular UE, the identifier (ID) of the strongest cell on the target carrier. Thus, the AI / ML model may be understood as a multiclass classification AI / ML model and the AI / ML model may be used to find, for each cell, a predicted mapping between the measurements and the strongest cell on the secondary7carrier frequency.

[0040] In some embodiments, various features may be used as an input for the AI / ML model such as RSRP values, precoder choices, and a precoder matrix indicator (PMI). The RSRP values may be understood as a UE-based feature in which the UEs within the network are assumed to send RSRP measurement reports. These reports include Layer 3 (L3) measurements of the RSRP values from the serving cell to provide information about signal strength for network optimization and handover decisions. The macro cells are assumed to have multiple antenna ports with many precoder choices for single layer transmissions that may be used as an input by the AI / ML model. Additionally, the PMI indicates the precoder selection for the UE's serving cell and may be considered a network-based feature. This information may help the network optimize the beamforming strategy to improve signal quality and overall communication performance for the UE. Alternatively, e.g., for frequency range 2 (FR2), a best Tx beam index may serve as a feature inputted into the AI / ML model. Further, each precoder choice corresponds to directing energy in a specific direction, which in open areas provides a clear indication of the UE's angular position relative to the base station. In some instances, a learning algorithm, y = F(x) of the AI / ML model may output an estimate of the true unknown function.

[0041] FIG. 2A illustrates an example of inter-frequency based beam prediction, according to embodiments herein. In the illustrated example, a UE is at a location 202 with respect to center macro cells, surrounding macro cells, and micro cells within an environment. Although not shown, the environment may be an urban environment with building, streets, vehicles, etc., or a rural environment with trees, hills, etc. In this example, the center macro cells transmit beams at frequency layer Fl and the micro cells transmit beams at frequency layer F2.84899-4873-7899,1 P70146WO1

[0042] The network provides an AI / ML model 208 (see FIG. 2B) to the UE configured to predict a strongest cell on frequency layer F2 using only measurements on frequency layer Fl. The AI / ML model 208 may be a multi-class classifier model. For inference, based on the measurements, the UE determines features (e.g., RSRP values, PMI, Tx beam index, etc.) across a plurality of cells 204a, 204b, 204c, 204d, 204e, 204f at the frequency layer Fl. Based on the determined features provided as inputs to the AI / ML model 208, the AI / ML model 208 predicts a best cell ID (e.g.. PceiiiD for cell 206) at frequency layer F2. In other embodiments, the AI / ML model 208 may output a top-K best cell IDs (Pceiim) at frequency layer F2. Thus, in the illustrated example, the UE may skip inter-frequency measurements at frequency layer F2 for all other cells (at F2) to reduce measurement overhead.

[0043] Predicting a strongest carrier in inter-frequency measurements

[0044] Certain embodiments disclosed herein provide a procedure for predicting a strongest carrier frequency in inter-frequency measurements, using only measurements from the serving carrier configured with a serving carrier frequency (Fl). FIG. 3 A illustrates a second example of inter-frequency based beam prediction, according to embodiments herein. In the illustrated example, a UE is at a location 302 with respect to center macro cells, surrounding macro cells, and micro cells within an environment. Similar to the example shown in FIG. 2A, the environment may be an urban environment with building, streets, vehicles, etc., or a rural environment with trees, hills, etc. In this example, the center macro cells transmit beams at frequency layer Fl and the other cells transmit beams at frequency layers F2,..., FN.

[0045] The network provides an AI / ML model 306 (see FIG. 3B) to the UE configured to predict a strongest frequency layer using only measurements on frequency layer FL The AI / ML model 306 may be a classifier model. For inference, based on the measurements, the UE determines features (e.g., RSRP values, PMI, Tx beam index, etc.) across a plurality' of cells 304a, 304b, 304c, 304d, 304e, 304f at the frequency layer FL Based on the determined features provided as inputs to the AI / ML model 306, the AI / ML model 306 predicts a best frequency layer ID (e.g.. PFreqLayeriD for Fbest). In other embodiments, the AI / ML model 306 may output a top-K best frequency layer IDs (PFreqLayerin). Thus, in the illustrated example, the UE may skip the inter-frequency measurements at frequency layers that are not in the best or strongest frequency layer (Fj £ Fbest) to reduce measurement overhead.94899-4873-7899,1 P70146WO1

[0046] Predicting a strongest carrier / cell ID in multi-frequency multiple cell measurements

[0047] Certain embodiments disclosed herein provide a procedure for predicting a strongest carrier frequency and a strongest cell in multiple-frequency multiple cell operation. FIG. 4A illustrates a third example of inter-frequency based beam prediction, according to embodiments herein. In the illustrated example, a UE is at a location 402 with respect to center macro cells, surrounding macro cells, and micro cells within an environment. Similar to the example shown in FIG. 2A, the environment may be an urban environment with building, streets, vehicles, etc., or a rural environment with trees, hills, etc. In this example, the center macro cells transmit beams at frequency layer Fl and the other cells transmit beams at frequency layers F2,..., FN.

[0048] The network provides an AI / ML model 412 (see FIG. 4B) to the UE configured to predict a strongest frequency layer and a strongest cell on the strongest frequency layer. The AI / ML model 412 may be a multi-class classifier model. For inference, based on the measurements, the UE determines features (e.g., RSRP values, PMI, Tx beam index, etc.) across selected cells and selected frequency layers. For example, the UE measures a plurality of selected cells 404a, 404b. 404c, 404d, 404e at the frequency layer Fl, a plurality of selected cells 406a, 406b at the frequency layer F2, a plurality of selected cells 408a, 408b at a frequency layer (subsequently determined by the model to be the best frequency layer Fbest), a plurality of cells 410a, 410b at the frequency layer FN.

[0049] Based on the determined features provided as inputs to the AI / ML model 412, the AI / ML model 412 predicts a best cell ID (e.g., Cell^stcorresponding to cell 414) and a best frequency layer Fbest. In other embodiments, the AI / ML model 412 may output a top-K best cell ID and best frequency layer (e.g., {Cell^st, Fbest}). Thus, in the illustrated example, the UE may skip the inter-frequency measurements at frequencylayers that are not in the best or strongest frequency layer (Fj g Fbest) and best cell level measurements (CelliD £ Cellj^st) to reduce measurement overhead.

[0050] FIG. 5 is a diagram illustrating training 500 of an AI / ML model 502 configured to generate predicted receive signal characteristics of Tx-Rx beam pairs, according to embodiments herein.

[0051] Embodiments herein relate to the use of mapped images that represent signal measurements for a given characteristic (e.g.. actual measurements, predicted104899-4873-7899,1 P70146WO1measurements) and their corresponding to codebook el ements / b earns (e.g., beams of a corresponding Tx codebook, beams of a corresponding Rx codebook) in a graphical way. For example, a set of measurements may be represented as a two dimensional image with one axis representing Tx codebook beams and another axis representing Rx codebook beams, and where each element of the mapped image corresponding to one Tx-Rx point is shaded in a way that represents the measurement for that Tx-Rx beam pair.

[0052] It further is anticipated that, assuming that such mapped images are available, an AI / ML model 502 may leverage image processing functionality corresponding to such mapped images to generate predicted receive signal characteristics of Tx-Rx beam pairs. A mechanism for training the Al / ML model 502 for such functionality will now be described.

[0053] The AI / ML model 502 may include a convolutional neural network (CNN) image processor for image super-resolution (SR). The SR functionality of the AI / ML model 502 may be used to predict a super-resolved image from a reduced image that is input to the AI / ML model 502. The CNN used by the AI / ML model 502 for this SR functionality may include a plain feed-forward network which has convolutional layers and may utilize residual learning. Accordingly, a training dataset for the CNN of the AI / ML model 502 may include low resolution images as “before” images and the corresponding high resolution images as “after” images.

[0054] Preliminarily, a full mapped image 504 (denoted T) is provided. The full mapped image 504 includes image data that represents measurements for a full Tx-Rx beam sweep (the full mapped image 504 includes a measurement for each base station Tx beam using each UE Rx beam) and one or more frequency layers 522. The full mapped image 504 is provided to the AI / ML model 502 as ground truth data 514.

[0055] The full mapped image 504 then undergoes downsampling 506 in order to generate the downsampled mapped image 508 (with the procedure downsampling 506 represented in FIG. 5 as / / ?). This downsampled mapped image 508 represents some proper subset of the data included in the full mapped image 504. In the example illustrated in FIG. 5, the downsampled mapped image 508 represents the case where one quarter of the samples of the full mapped image 504 are included in the downsampled mapped image 508 by omitting data corresponding to even-numbered Tx beams and even-numbered Rx beams. Note that this particular mechanism for the downsampled mapped image 508 is given by way of example, and that the use of other possible114899-4873-7899,1 P70146WO1mechanisms and / or methods for performing the downsampling 506 from the full mapped image 504 are contemplated. The downsampled mapped image 508 may also include downsampled frequency layers 524.

[0056] The downsampled mapped image 508 then undergoes mapping reduction 510 to generate a reduced mapped image 512 (denoted X). As illustrated, the sampled data from the downsampled mapped image 508 is formed into the downsampled mapped image 508 by removing locations for the blank / non-sampled data from the downsampled mapped image 508 and applying appropriate corresponding downsampled codebook indices. The downsampled mapped image 508 is accordingly understood to be of lower resolution than the full mapped image 504, as illustrated.

[0057] The reduced mapped image 512 is then provided to the AI / ML model 502 as input data 516 that corresponds to the ground truth data 514.

[0058] Note that many such instances of ground truth data 514 and corresponding input data 516 as described here may be provided to the AI / ML model 502 for training purposes.

[0059] The AI / ML model 502 then performs deep learning 518 that iterates over these instances of the ground truth data 514 and its corresponding input data 516 such that it is trained to predict a predicted full mapped image 520 (denoted Y ) based on a reduced mapped image 512 (e.g., the AI / ML model 502 leams a particular mechanism for performing an upsampling JR'1with respect to an inputted X). Once trained, the AI / ML model 502 is useful to generate a predicted full mapped image 520 using a reduced mapped image 512 without having previously trained on a corresponding full mapped image 504 (without having previously seen the corresponding T).

[0060] Note that the procedure as described in relation to FIG. 5 assumes the use of degradation-free images.

[0061] In some embodiments, it may be desirable to find the top-K Tx-Rx beam pairs from the interpolated image map [r, t]1:JV= g(Y). In such cases, an optimum Tx-Rx beam may be denoted by [r*, t*] = g(Y), from the generated labels Y. Simulations for computing the probability of having [r*, t*] G [r, t]1:JVmay be applied.

[0062] As has been described, assuming the use of clean mapped images, the training dataset consists of low resolution images as “before” images and the corresponding high124899-4873-7899,1 P70146WO1resolution images as “after’ image. However, real-world in-field measurements cannot be assumed to be clean (because they are affected by degradation due to in-field distortive transmission factors). Types of degradation may include, for example, additive white Gaussian noise (AWGN) that may exist within a real-world measurement, antenna spacing mismatches between training data and real-world data, beam width mismatches between training data and real-world data, etc.

[0063] Accordingly, as an extension to the training and use case for the AI / ML models described in relation to FIG. 5, it will be beneficial to develop these AI / ML models in such a way that they can handle various different degradation types.

[0064] In some cases, an AI / ML model that performs image restoration that accounts for degradation may be trained in a way similar to that discussed in relation to FIG. 5 but with a further improvement to the input image dataset during the training stage. For example, the inputted image data could include both "before” images (no degradation) and one or more “after” images (corresponding to various the types of degradation as applied to the “before” image).

[0065] The resulting AI / ML model would obtain the ability- to predict a cleaned image from the different ty pes of degraded images represented in the training data without having previously seen the non-degraded version of the image. Such image restoration processors may be classified as blind image restoration processors. Such blind image restoration processors may thus be understood to restore images by implicitly or explicitly estimating a degradation model and corresponding parameters.

[0066] It is noted that while such learning-based blind image restoration processors have high performance, they are very sensitive to the perturbation of the implicit degradation model they use. For instance, such learning-based blind-restoration processors will easily fail to properly restore a sufficiently accurate image in cases where the applicable real-world / in-field degradation factors are even slightly different from those implicitly assumed by the trained degradation model based on the training data.

[0067] Beam level to cell level spatial / frequency RSRP prediction

[0068] FIG. 6 illustrates an example of a three-dimensional (3D) domain of reduced measurements, according to embodiments herein.

[0069] In some embodiments, the UE may perform reduced measurements (i.e.. measure a subset of Tx beams) across a 3D domain of reduced measurements to achieve a subset of beam measurements from a plurality- of measurements. The 3D domain of134899-4873-7899,1 P70146WO1reduced measurements includes measurements at a Tx beam level domain, measurements at a frequency layer domain, and measurements at a site (i.e., cell) spatial domain.

[0070] For example, the performed reduced measurements may be understood as measurements performed on a subsets of beams 612 from various beams of a Tx-Rx beam pair map 620 of a larger overall set (e.g., all) of the Tx-Rx beam pairs between the Tx beams 602 of the network and the Rx beams of the UE. The Tx-Rx beam pair map 620 may correspond to reference signals transmitted from various sites 604 (i.e., cells) at corresponding frequency layers. That is, the reduced measurements may be performed at various frequency layers (e.g., first frequency layer 606, second frequency layer 608, third frequency layer 610). The Tx-Rx beam pair map 620 may be in terms of three dimensions. The X dimension 616 and the Y dimension 614 respectively correspond to horizontal and vertical indexes that together identify applicable Tx beams 602. The Z dimension 618 then a frequency layer corresponding to the Tx beams 602 measured. It should be understood that the reduced measurements are performed on a subset of beams of the overall set of the Tx-Rx beam pairs per site (i.e., cell), per frequency layer.

[0071] FIG. 7 illustrates an example AI / ML classifier model used to perform interfrequency based beam prediction, according to embodiments herein.

[0072] In some embodiments, the AI / ML classifier model may be trained across the 3D domain of reduced measurements including, for example, Tx beam level measurements 708, cell level spatial measurements 704, and frequency level measurements 706 (as discussed herein). The measurements can be saved across all the domains in, for example, a multiple domain set B beams 702 (where multiple domain set A corresponds to a full set). The AI / ML model may perform 3D-convolution padding 710 on the multiple domain set B beams 702 (e.g., with a padding value of 0) and the result may be passed to a convolutional layer 712 of the AI / ML model. The output from the convolutional layer 712 may be flattened 714 and passed to a first dense layer 716, a second dense layer 718, and an output layer 720 that may output 722 a predicted best cell ID, best / strongest frequency layer, and / or a best Tx beam ID.

[0073] In some cases, the AI / ML model may be understood as a CNN classifier network that may process the reduced set B of measurements to predict / output top-K beam indices that include a predicted best cell ID, a best / strongest frequency layer and / or a best Tx beam ID.144899-4873-7899,1 P70146WO1

[0074] FIG. 8 illustrates a method 800 for a UE. according to embodiments herein. The illustrated method 800 includes connecting 802 to a serving cell of a first plurality of cells configured with a first carrier frequency. The method 800 further includes receiving 804 an AI / ML model trained by the serving cell to determine a prediction on a second carrier frequency from measurements based on the first carrier frequency. The method 800 further includes measuring 806 reference signals transmitted by the first plurality of cells on the first carrier frequency to generate cell feature information. The method 800 further includes processing 808 the cell feature information, using the AI / ML model at the UE, to determine the prediction on the second carrier frequency.

[0075] In some embodiments of the method 800, the cell feature information comprises one or more of RSRP values, a PMI, and a best Tx beam index.

[0076] In some embodiments of the method 800, the AI / ML model comprises a multiclass classifier model, and wherein the prediction comprises a best cell ID among a second plurality of cells at the second carrier frequency. Some such embodiments further comprise reporting, to the serving cell, the best cell ID on the second carrier frequency.

[0077] In some embodiments of the method 800, the AI / ML model comprises a classifier model, and wherein the prediction indicates that the second carrier frequency corresponds to a strongest frequency layer among a plurality' of frequency layers. Some such embodiments further comprise reporting, to the serving cell, the strongest frequency layer.

[0078] In some embodiments of the method 800, the first plurality' of cells comprises a first downsampled subset of cells on the first carrier frequency, and wherein the method 800 further comprises: measuring the reference signals transmitted by a second downsampled subset of cells on a plurality of carrier frequencies, in addition to the first carrier frequency, to generate the cell feature information. In some such embodiments, the AI / ML model comprises a multi-class classifier model, and wherein the prediction indicates that the second carrier frequency corresponds to a strongest frequency layer among a plurality of frequency layers and a best cell ID at the second carrier frequency. Some such embodiments further comprise reporting, to the serving cell, the strongest frequency layer and the best cell ID at the second carrier frequency.

[0079] In some embodiments of the method 800, the AI / ML model is trained across a three dimensional domain of reduced measurements including Tx beam level measurements, cell level spatial measurements, and frequency layer measurements, and154899-4873-7899,1 P70146WO1wherein measuring the reference signals further comprises: for each of a plurality of frequency layers, measuring subsets of Tx beams of a plurality of Tx beams across a plurality of cells including at least the first plurality of cells at the first carrier frequency and a second plurality' of cells at the second carrier frequency. In some such embodiments, the AI / ML model comprises a CNN configured to generate an output comprising at least one of a best cell ID, a best frequency layer, and a best Tx beam ID. Some such embodiments further comprise reporting the output to the serving cell.

[0080] FIG. 9 illustrates a method 900 for a cell configured with a first carrier frequency, according to embodiments herein. The illustrated method 900 includes transmitting 902, to a UE, reference signals on the first carrier frequency. The method 900 further includes receiving 904, from the UE, cell feature information, corresponding to the reference signals measured by the UE. The method 900 further includes processing 906 the cell feature information, using an AI / ML model trained to predict a second carrier frequency from measurements based on the first carrier frequency, to generate a predicted signal on the second carrier frequency. The method 900 further includes transmitting 908, to the UE, information corresponding to the predicted signal on the second carrier frequency.

[0081] In some embodiments of the method 900, the cell feature information comprises one or more of RSRP values, a PMI, and a best Tx beam index.

[0082] In some embodiments of the method 900, the AI / ML model comprises a multiclass classifier model, and wherein the prediction comprises a best cell ID among a second plurality of cells at the second carrier frequency. Some such embodiments further comprise reporting, to the UE, the best cell ID on the second carrier frequency.

[0083] In some embodiments of the method 900, the AI / ML model comprises a classifier model, and wherein the prediction indicates that the second carrier frequency corresponds to a strongest frequency layer among a plurality of frequency layers. Some such embodiments further comprise reporting, to the UE, the strongest frequency layer.

[0084] In some embodiments of the method 900, the AI / ML model is trained across a three dimensional domain of reduced measurements including Tx beam level measurements, cell level spatial measurements, and frequency layer measurements. In some such embodiments, the AI / ML model comprises a CNN configured to generate an output comprising at least one of a best cell ID, a best frequency layer, and a best Tx beam ID. Some such embodiments further comprise reporting the output to the UE.164899-4873-7899,1 P70146WO1

[0085] FIG. 10 illustrates an example architecture of a wireless communication system 1000, according to embodiments disclosed herein. The following description is provided for an example wireless communication system 1000 that operates in conjunction with the LTE system standards and / or 5G or NR system standards as provided by 3GPP technical specifications.

[0086] As shown by FIG. 10, the wireless communication system 1000 includes UE 1002 and UE 1004 (although any number of UEs may be used). In this example, the UE 1002 and the UE 1004 are illustrated as smartphones (e.g., handheld touchscreen mobile computing devices connectable to one or more cellular networks), but may also comprise any mobile or non-mobile computing device configured for wireless communication.

[0087] The UE 1002 and UE 1004 may be configured to communicatively couple with a RAN 1006. In embodiments, the RAN 1006 may be NG-RAN, E-UTRAN, etc. The UE 1002 and UE 1004 utilize connections (or channels) (shown as connection 1008 and connection 1010, respectively) with the RAN 1006, each of which comprises a physical communications interface. The RAN 1006 can include one or more base stations (such as base station 1012 and base station 1014) that enable the connection 1008 and connection 1010.

[0088] In this example, the connection 1008 and connection 1010 are air interfaces to enable such communicative coupling, and may be consistent with RAT(s) used by the RAN 1006, such as. for example, an UTE and / or NR.

[0089] In some embodiments, the UE 1002 and UE 1004 may also directly exchange communication data via a sidelink interface 1016. The UE 1004 is shown to be configured to access an access point (shown as AP 1018) via connection 1020. By way of example, the connection 1020 can comprise a local wireless connection, such as a connection consistent with any IEEE 802.11 protocol, wherein the AP 1018 may comprise a Wi-Fi® router. In this example, the AP 1018 may be connected to another network (for example, the Internet) without going through a CN 1024.

[0090] In embodiments, the UE 1002 and UE 1004 can be configured to communicate using orthogonal frequency division multiplexing (OFDM) communication signals with each other or with the base station 1012 and / or the base station 1014 over a multicarrier communication channel in accordance with various communication techniques, such as, but not limited to, an orthogonal frequency division multiple access (OFDMA) communication technique (e.g., for downlink communications) or a single carrier174899-4873-7899,1 P70146WO1frequency division multiple access (SC-FDMA) communication technique (e.g., for uplink and ProSe or sidelink communications), although the scope of the embodiments is not limited in this respect. The OFDM signals can comprise a plurality of orthogonal subcarriers.

[0091] In some embodiments, all or parts of the base station 1012 or base station 1014 may be implemented as one or more software entities running on server computers as part of a virtual network. In addition, or in other embodiments, the base station 1012 or base station 1014 may be configured to communicate with one another via interface 1022. In embodiments where the wireless communication system 1000 is an LTE system (e.g.. when the CN 1024 is an EPC), the interface 1022 may be an X2 interface. The X2 interface may be defined between two or more base stations (e.g., two or more eNBs and the like) that connect to an EPC, and / or between two eNBs connecting to the EPC. In embodiments where the wireless communication system 1000 is an NR system (e.g.. when CN 1024 is a 5GC), the interface 1022 may be an Xn interface. The Xn interface is defined between two or more base stations (e.g., two or more gNBs and the like) that connect to 5GC, between a base station 1012 (e.g., a gNB) connecting to 5GC and an eNB, and / or between two eNBs connecting to 5GC (e.g., CN 1024).

[0092] The RAN 1006 is shown to be communicatively coupled to the CN 1024. The CN 1024 may comprise one or more network elements 1026, which are configured to offer various data and telecommunications services to customers / subscribers (e.g., users of UE 1002 and UE 1004) who are connected to the CN 1024 via the RAN 1006. The components of the CN 1024 may be implemented in one physical device or separate physical devices including components to read and execute instructions from a machine- readable or computer-readable medium (e g., a non-transitory machine-readable storage medium).

[0093] In embodiments, the CN 1024 may be an EPC, and the RAN 1006 may be connected with the CN 1024 via an SI interface 1028. In embodiments, the SI interface 1028 may be split into two parts, an SI user plane (Sl-U) interface, which carries traffic data between the base station 1012 or base station 1014 and a serving gateway (S-GW). and the SI -MME interface, which is a signaling interface between the base station 1012 or base station 1014 and mobility management entities (MMEs).

[0094] In embodiments, the CN 1024 may be a 5GC, and the RAN 1006 may be connected with the CN 1024 via an NG interface 1028. In embodiments, the NG184899-4873-7899,1 P70146WO1interface 1028 may be split into two parts, an NG user plane (NG-U) interface, which carries traffic data between the base station 1012 or base station 1014 and a user plane function (UPF), and the SI control plane (NG-C) interface, which is a signaling interface between the base station 1012 or base station 1014 and access and mobility management functions (AMFs).

[0095] Generally, an application server 1030 may be an element offering applications that use internet protocol (IP) bearer resources with the CN 1024 (e.g., packet switched data services). The application server 1030 can also be configured to support one or more communication services (e g., VoIP sessions, group communication sessions, etc.) for the UE 1002 and UE 1004 via the CN 1024. The application server 1030 may communicate with the CN 1024 through an IP communications interface 1032.

[0096] FIG. 11 illustrates a system 1100 for performing signaling 1134 between a wireless device 1102 and a network device 1118, according to embodiments disclosed herein. The system 1100 may be a portion of a wireless communications system as herein described. The wireless device 1102 may be, for example, a UE of a wireless communication system. The network device 1118 may be, for example, a base station (e.g.. an eNB or a gNB) of a wireless communication system.

[0097] The wireless device 1102 may include one or more processor(s) 1104. The processor(s) 1104 may execute instructions such that various operations of the wireless device 1102 are performed, as described herein. The processor(s) 1104 may include one or more baseband processors implemented using, for example, a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a controller, a field programmable gate array (FPGA) device, another hardware device, a firmware device, or any combination thereof configured to perform the operations described herein.

[0098] The wireless device 1102 may include a memory' 1106. The memory 1106 may be a non-transitory computer-readable storage medium that stores instructions 1108 (which may include, for example, the instructions being executed by the processor(s) 1104). The instructions 1108 may also be referred to as program code or a computer program. The memory 1106 may also store data used by, and results computed by, the processor(s) 1104.

[0099] The wireless device 1102 may include one or more transceiver(s) 1110 that may include radio frequency (RF) transmitter circuitry and / or receiver circuitry that use the194899-4873-7899,1 P70146WO1antenna(s) 1112 of the wireless device 1102 to facilitate signaling (e.g.. the signaling 1134) to and / or from the wireless device 1102 with other devices (e.g., the network device 1118) according to corresponding RATs.

[0100] The wireless device 1102 may include one or more antenna(s) 1112 (e.g., one, two, four, or more). For embodiments with multiple antenna(s) 1112, the wireless device 1102 may leverage the spatial diversity of such multiple antenna(s) 1112 to send 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 the multiple antennas used at each of a transmitting device and a receiving device that enable this aspect). MIMO transmissions by the wireless device 1102 may be accomplished according to precoding (or digital beamforming) that is applied at the wireless device 1102 that multiplexes the data streams across the antenna(s) 1112 according to known or assumed channel characteristics such that each data stream is received with an appropriate signal strength relative to other streams and at a desired location in the spatial domain (e.g., the location of a receiver associated with that data stream). Certain embodiments may use single user MIMO (SU-MIMO) methods (where the data streams are all directed to a single receiver) and / or multi user MIMO (MU-MIMO) methods (where individual data streams may be directed to individual (different) receivers in different locations in the spatial domain).

[0101] In certain embodiments having multiple antennas, the wireless device 1102 may implement analog beamforming techniques, whereby phases of the signals sent by the antenna(s) 1112 are relatively adjusted such that the (joint) transmission of the antenna(s) 1112 can be directed (this is sometimes referred to as beam steering).

[0102] The wireless device 1102 may include one or more interface(s) 1114. The interface(s) 1114 may be used to provide input to or output from the wireless device 1102. For example, a wireless device 1102 that is a UE may include interface(s) 1114 such as microphones, speakers, a touchscreen, buttons, and the like in order to allow for input and / or output to the UE by a user of the UE. Other interfaces of such a UE may be made up of transmitters, receivers, and other circuitry (e.g., other than the transceiver(s) 1110 / antenna(s) 1112 already described) that allow- for communication between the UE and other devices and may operate according to known protocols (e.g., Wi-Fi®, Bluetooth®, and the like).204899-4873-7899,1 P70146WO1

[0103] The wireless device 1102 may include an inter-frequency based prediction module 1116. The inter- frequency based prediction module 1116 may be implemented via hardware, software, or combinations thereof. For example, the inter-frequency based prediction module 1116 may be implemented as a processor, circuit, and / or instructions 1108 stored in the memory 1106 and executed by the processor(s) 1104. In some examples, the inter-frequency based prediction module 1116 may be integrated within the processor(s) 1104 and / or the transceiver(s) 1110. For example, the inter-frequency based prediction module 1116 may be implemented by a combination of software components (e.g., executed by a DSP or a general processor) and hardware components (e.g., logic gates and circuitry) within the processor(s) 1104 or the transceiver(s) 1110.

[0104] The inter-frequency based prediction module 1116 may be used for various aspects of the present disclosure, for example, aspects of FIG. 1, FIG. 2A, FIG. 2B, FIG. 3A, FIG. 3B, FIG. 4A, FIG. 4B, FIG. 5, FIG. 6. and FIG. 7. The inter-frequency based prediction module 1116 is configured to cause the wireless device 1102 to connect to a serving cell of a first plurality of cells configured with a first carrier frequency. The inter-frequency based prediction module 1116 is further configured to cause the wireless device 1102 receive, an AI / ML model trained by the network device 1118 to determine a prediction on a second carrier frequency from measurements based on the first carrier frequency. The inter-frequency based prediction module 1116 is further configured to cause the wireless device 1102 to measure reference signals transmitted by the first plurality of network devices 1118 on the first carrier frequency to generate cell feature information. The inter-frequency based prediction module 1116 is further configured to cause the wireless device 1102 to process the cell feature information, using the AI / ML model at the wireless device 1102, to determine the prediction on the second carrier frequency.

[0105] The network device 1118 may include one or more processor(s) 1120. The processor(s) 1120 may execute instructions such that various operations of the network device 1118 are performed, as described herein. The processor(s) 1120 may include one or more baseband processors implemented using, for example, a CPU, a DSP, an ASIC, a controller, an FPGA device, another hardware device, a firmware device, or any combination thereof configured to perform the operations described herein.

[0106] The network device 1118 may include a memory 1122. The memory 1122 may be a non-transitory computer-readable storage medium that stores instructions 1124214899-4873-7899,1 P70146WO1(which may include, for example, the instructions being executed by the processor(s) 1120). The instructions 1124 may also be referred to as program code or a computer program. The memory 1122 may also store data used by, and results computed by, the processor(s) 1120.

[0107] The network device 1118 may include one or more transceiver(s) 1126 that may include RF transmitter circuitry and / or receiver circuitry that use the antenna(s) 1128 of the network device 1118 to facilitate signaling (e.g., the signaling 1134) to and / or from the network device 1118 with other devices (e.g., the wireless device 1102) according to corresponding RATs.

[0108] The network device 1118 may include one or more antenna(s) 1128 (e.g., one, two, four, or more). In embodiments having multiple antenna(s) 1128, the network device 1118 may perform MIMO, digital beamforming, analog beamforming, beam steering, etc., as has been described.

[0109] The network device 1118 may include one or more interface(s) 1130. The interface(s) 1130 may be used to provide input to or output from the network device 1118. For example, a network device 1118 that is a base station may include interface(s) 1130 made up of transmitters, receivers, and other circuitry (e.g., other than the transceiver(s) 1126 / antenna(s) 1128 already described) that enables the base station to communicate with other equipment in a core network, and / or that enables the base station to communicate with external networks, computers, databases, and the like for purposes of operations, administration, and maintenance of the base station or other equipment operably connected thereto.

[0110] The network device 1118 may include an inter-frequency based prediction module 1132. The inter-frequency based prediction module 1132 may be implemented via hardware, softw are, or combinations thereof. For example, the inter-frequency based prediction module 1132 may be implemented as a processor, circuit, and / or instructions 1124 stored in the memory 1122 and executed by the processor(s) 1120. In some examples, the inter-frequency based prediction module 1132 may be integrated within the processor(s) 1120 and / or the trans ceiver(s) 1126. For example, the inter-frequency based prediction module 1132 may be implemented by a combination of software components (e.g., executed by a DSP or a general processor) and hardware components (e.g.. logic gates and circuitry) within the processor(s) 1120 or the transceiver(s) 1126.224899-4873-7899,1 P70146WO1

[0111] The inter-frequency based prediction module 1132 may be used for various aspects of the present disclosure, for example, aspects of FIG. 1, FIG. 2A, FIG. 2B, FIG. 3A, FIG. 3B, FIG. 4A, FIG. 4B, FIG. 5, FIG. 6, and FIG. 8. The inter-frequency based prediction module 1132 is configured to cause the network device 1118 configured with a first carrier frequency to transmit, to a wireless device 1102, reference signals on the first carrier frequency. The inter-frequency based prediction module 1132 is further configured to cause the network device 1118 to receive, from the wireless device 1102, cell feature information, corresponding to the reference signals measured by the wireless device 1102. The inter-frequency based prediction module 1132 is further configured to cause the network device 1118 to process the cell feature information, using an AI / ML model trained to predict a second carrier frequency from measurements based on the first carrier frequency, to generate a predicted signal on the second carrier frequency. The inter-frequency based prediction module 1132 is further configured to cause the network device 1118 to transmit, to the wireless device 1102, information corresponding to the predicted signal on the second carrier frequency.

[0112] Embodiments contemplated herein include an apparatus comprising means to perform one or more elements of the method 800. This apparatus may be, for example, an apparatus of a UE (such as a wireless device 1102 that is a UE. as described herein).

[0113] Embodiments contemplated herein include one or more non-transitory computer-readable media comprising instructions to cause an electronic device, upon execution of the instructions by one or more processors of the electronic device, to perform one or more elements of the method 800. This non-transitory computer-readable media may be, for example, a memory of a UE (such as a memory 1106 of a wireless device 1102 that is a UE, as described herein).

[0114] Embodiments contemplated herein include an apparatus comprising logic, modules, or circuitry to perform one or more elements of the method 800. This apparatus may be, for example, an apparatus of a UE (such as a wireless device 1102 that is a UE, as described herein).

[0115] Embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of the method 800. This apparatus may be. for example, an apparatus of a UE (such as a wireless device 1102 that is a UE, as described herein).234899-4873-7899,1 P70146WO1

[0116] Embodiments contemplated herein include a signal as described in or related to one or more elements of the method 800.

[0117] Embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution of the program by a processor is to cause the processor to carry out one or more elements of the method 800. The processor may be a processor of a UE (such as a processor(s) 1104 of a wireless device 1102 that is a UE, as described herein). These instructions may be, for example, located in the processor and / or on a memory of the UE (such as a memory 1106 of a wireless device 1102 that is a UE, as described herein).

[0118] Embodiments contemplated herein include an apparatus comprising means to perform one or more elements of the method 900. This apparatus may be, for example, an apparatus of a base station (such as a network device 1118 that is a base station, as described herein).

[0119] Embodiments contemplated herein include one or more non-transitory computer-readable media comprising instructions to cause an electronic device, upon execution of the instructions by one or more processors of the electronic device, to perform one or more elements of the method 900. This non-transitory computer-readable media may be, for example, a memory' of a base station (such as a memory 1122 of a network device 1118 that is a base station, as described herein).

[0120] Embodiments contemplated herein include an apparatus comprising logic, modules, or circuitry' to perform one or more elements of the method 900. This apparatus may be, for example, an apparatus of a base station (such as a network device 1118 that is a base station, as described herein).

[0121] Embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of the method 900. This apparatus may be, for example, an apparatus of a base station (such as a network device 1118 that is a base station, as described herein).

[0122] Embodiments contemplated herein include a signal as described in or related to one or more elements of the method 900.

[0123] Embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution of the program by a244899-4873-7899,1 P70146WO1processing element is to cause the processing element to carry out one or more elements of the method 900. The processor may be a processor of a base station (such as a processor(s) 1120 of a network device 1118 that is a base station, as described herein). These instructions may be, for example, located in the processor and / or on a memory of the base station (such as a memory 1122 of a network device 1118 that is a base station, as described herein).

[0124] For one or more embodiments, at least one of the components set forth in one or more of the preceding figures may be configured to perform one or more operations, techniques, processes, and / or methods as set forth herein. For example, a baseband processor as described herein in connection with one or more of the preceding figures may be configured to operate in accordance with one or more of the examples set forth herein. For another example, circuitry associated with a UE, base station, network element, etc. as described above in connection with one or more of the preceding figures may be configured to operate in accordance with one or more of the examples set forth herein.

[0125] Any of the above described embodiments may be combined with any other embodiment (or combination of embodiments), unless explicitly stated otherwise. The foregoing description of one or more implementations provides illustration and description, but is not intended to be exhaustive or to limit the scope of embodiments to the precise form disclosed. Modifications and variations are possible in light of the above teachings or may be acquired from practice of various embodiments.

[0126] Embodiments and implementations of the systems and methods described herein may include various operations, which may be embodied in machine-executable instructions to be executed by a computer system. A computer system may include one or more general-purpose or special-purpose computers (or other electronic devices). The computer system may include hardware components that include specific logic for performing the operations or may include a combination of hardware, software, and / or firmware.

[0127] It should be recognized that the systems described herein include descriptions of specific embodiments. These embodiments can be combined into single systems, partially combined into other systems, split into multiple systems or divided or combined in other ways. In addition, it is contemplated that parameters, attributes, aspects, etc. of one embodiment can be used in another embodiment. The parameters, attributes, aspects,254899-4873-7899,1 P70146WO1etc. are merely described in one or more embodiments for clarity, and it is recognized that the parameters, attributes, aspects, etc. can be combined with or substituted for parameters, attributes, aspects, etc. of another embodiment unless specifically disclaimed herein.

[0128] It is well understood that the use of personally identifiable information should follow privacy policies and practices that are generally recognized as meeting or exceeding industry or governmental requirements for maintaining the privacy of users. In particular, personally identifiable information data should be managed and handled so as to minimize risks of unintentional or unauthorized access or use, and the nature of authorized use should be clearly indicated to users.

[0129] Although the foregoing has been described in some detail for purposes of clarity, it will be apparent that certain changes and modifications may be made without departing from the principles thereof. It should be noted that there are many alternative ways of implementing both the processes and apparatuses described herein. Accordingly, the present embodiments are to be considered illustrative and not restrictive, and the description is not to be limited to the details given herein, but may be modified within the scope and equivalents of the appended claims.264899-4873-7899,1 P70146WO1

Claims

CLAIMS1. A method for a user equipment (UE), comprising: connecting to a serving cell of a first plurality of cells configured with a first carrier frequency; receiving, an artificial intelligent (AI) / machine learning (ML) model trained by the serving cell to determine a prediction on a second carrier frequency from measurements based on the first carrier frequency; measuring reference signals transmitted by the first plurality of cells on the first carrier frequency to generate cell feature information; and processing the cell feature information, using the AI / ML model at the UE, to determine the prediction on the second carrier frequency.

2. The method of claim 1, wherein the cell feature information comprises one or more of reference signal received power (RSRP) values, a precoding matrix indicator (PMI). and a best transmit (Tx) beam index.

3. The method of claim 1, wherein the AI / ML model comprises a multi-class classifier model, and wherein the prediction comprises a best cell identifier (ID) among a second plurality of cells at the second carrier frequency.

4. The method of claim 3, further comprising reporting, to the serving cell, the best cell ID on the second carrier frequency.

5. The method of claim 1, wherein the AI / ML model comprises a classifier model, and wherein the prediction indicates that the second carrier frequency corresponds to a strongest frequency layer among a plurality of frequency layers.

6. The method of claim 5, further comprising reporting, to the serving cell, the strongest frequency layer.

7. The method of claim 1 , wherein the first plurality of cells comprises a first downsampled subset of cells on the first carrier frequency, and wherein the method further comprises:274899-4873-7899,1 P70146WO1measuring the reference signals transmitted by a second downsampled subset of cells on a plurality of carrier frequencies, in addition to the first carrier frequency, to generate the cell feature information.

8. The method of claim 7, wherein the AI / ML model comprises a multi-class classifier model, and wherein the prediction indicates that the second carrier frequency corresponds to a strongest frequency layer among a plurality of frequency layers and a best cell identifier (ID) at the second carrier frequency.

9. The method of claim 8, further comprising reporting, to the serving cell, the strongest frequency layer and the best cell ID at the second carrier frequency.

10. The method of claim 1, wherein the AI / ML model is trained across a three dimensional domain of reduced measurements including transmit (Tx) beam level measurements, cell level spatial measurements, and frequency layer measurements, and wherein measuring the reference signals further comprises: for each of a plurality7of frequency layers, measuring subsets of Tx beams of a plurality of Tx beams across a plurality of cells including at least the first plurality of cells at the first carrier frequency and a second plurality of cells at the second carrier frequency.

11. The method of claim 10, wherein the AI / ML model comprises a convolutional neural network (CNN) configured to generate an output comprising at least one of a best cell identifier (ID), a best frequency layer, and a best Tx beam ID.

12. The method of claim 11, further comprising reporting the output to the serving cell.

13. A method for a cell configured with a first carrier frequency, the method comprising: transmitting, to a user equipment (UE), reference signals on the first carrier frequency; receiving, from the UE. cell feature information, corresponding to the reference signals measured by the UE; processing the cell feature information, using an artificial intelligence (Al) / machine learning (ML) model trained to predict a second carrier frequency from measurements based on the first carrier frequency, to generate a predicted signal on the second carrier frequency; and284899-4873-7899,1 P70146WO1transmitting, to the UE, information corresponding to the predicted signal on the second carrier frequency.

14. The method of claim 13, wherein the cell feature information comprises one or more of reference signal received power (RSRP) values, a precoding matrix indicator (PMI), and a best transmit (Tx) beam index.

15. The method of claim 13, wherein the AI / ML model comprises a multi-class classifier model, and wherein the prediction comprises a best cell identifier (ID) among a second plurality of cells at the second carrier frequency.

16. The method of claim 15, further comprising reporting, to the UE, the best cell ID on the second carrier frequency.

17. The method of claim 13, wherein the AI / ML model comprises a classifier model, and wherein the prediction indicates that the second carrier frequency corresponds to a strongest frequency layer among a plurality of frequency layers.

18. The method of claim 17, further comprising reporting, to the UE, the strongest frequency layer.

19. The method of claim 13, wherein the AI / ML model is trained across a three dimensional domain of reduced measurements including transmit (Tx) beam level measurements, cell level spatial measurements, and frequency layer measurements.

20. The method of claim 19, wherein the AI / ML model comprises a convolutional neural network (CNN) configured to generate an output comprising at least one of a best cell identifier (ID), a best frequency layer, and a best Tx beam ID.

21. The method of claim 20, further comprising reporting the output to the UE.

22. An apparatus comprising means to perform the method of any of claim 1 to claim 21.

23. A computer-readable media comprising instructions to cause an electronic device, upon execution of the instructions by one or more processors of the electronic device, to perform the method of any of claim 1 to claim 21.294899-4873-7899,1 P70146WO124. An apparatus comprising logic, modules, or circuitry to perform the method of any of claim 1 to claim 21.

25. A baseband processor for a user equipment (UE) that is configured to cause the UE to perform one or more elements of any one of claim 1 to claim 12.

26. A baseband processor for a base station of a cell that is configured to cause the base station to perform one or more elements of any one of claim 13 to claim 21.304899-4873-7899,1 P70146WO1

Images