Electronic device, method of operating same, and method of operating wireless communication system

Abstract

Disclosed are an electronic device, a method of operating the same, and a method of operating a wireless communication system. The electronic device includes: at least one memory; a communication interface comprising an antenna array configured to form a plurality of candidate beams; and at least one processor operably connected to the communication interface and the at least one memory. The at least one processor is configured to: generate a plurality of pieces of RSRP mode data based on a reference signal received power (RSRP) measured in each of the plurality of candidate beams for a signal received from an external device; estimating an angle of arrival (AoA) distribution for each of the plurality of RSRP pattern data by applying each of the plurality of RSRP pattern data to a neural network trained based on a deep learning algorithm; and performing beam management for wireless communication with the external device based on the estimated AoA distribution.

Description

[0001] This application is based on and claims the benefit of priority from Korean Patent Application No. 10-2023-0102288 filed on August 4, 2023, in the Korean Intellectual Property Office, and Korean Patent Application No. 10-2023-0157686 filed on November 14, 2023, in the Korean Intellectual Property Office, the disclosures of which are incorporated herein in their entirety by reference. Technical Field

[0002] One or more disclosed example embodiments relate to an electronic device and an operating method of the electronic device for neural network-based beam management in a wireless communication system. Background Art

[0003] Recently, as wireless communication technology has developed, high frequency bands in the millimeter wave band are used to ensure faster communication speeds in wireless communication systems (e.g., fifth generation (5G) new radio (NR, or referred to as new air interface) communication systems). Communication via the millimeter wave band achieves high-speed communication by utilizing a wider bandwidth than the frequency band used in the existing wireless communication system (e.g., sub-6GHz band). However, in communication via the millimeter wave band, because the high frequency band is used, the channel propagation loss may be large. The electronic device can reduce the channel propagation loss by using a beamforming technology that uses an antenna array. In a wireless communication system, the electronic device can improve the quality of wireless communication by focusing a signal in a specific direction based on the beamforming technology. During the beam training process, the electronic device (e.g., a terminal) can predetermine a candidate beam to select the best beam that matches the wireless communication environment. In addition, the final beam to be applied to the antenna array can be selected by comparing the measurement results of the signal quality between the beam of an external device (e.g., a base station) and the candidate beam of the electronic device.

[0004] Specifically, during the beam training process, the electronic device (eg, terminal) has physical limitations because the beam width cannot be very small due to space limitations. Therefore, even if the number of candidate beams increases, it is difficult for the electronic device to perform precise beam alignment with the beam of an external device (eg, base station).

[0005] In addition, as the number of candidate beams of an electronic device (e.g., a terminal) increases, the performance of beam alignment between a beam of an external device (e.g., a base station) and a candidate beam of the electronic device may be improved. However, the consumption of communication resources increases in proportion to the increase in the number of candidate beams, and therefore, the overhead of beam training processing may increase. The increase in overhead results in a decrease in performance or a decrease in communication quality in a wireless communication system. Summary of the invention

[0006] One or more disclosed example embodiments provide an electronic device and an operating method of the electronic device, wherein the electronic device is capable of estimating angle of arrival (AoA) distribution of reference signal received power (RSRP) pattern data measured for a plurality of candidate beams of the electronic device using a neural network trained via a deep learning algorithm, and performing beam management using the estimated AoA distribution.

[0007] The technical objectives of the inventive concept are not limited to the above-mentioned technical objectives, and other technical objectives not described herein are clearly understood by those skilled in the art from the following description.

[0008] According to one aspect of the disclosed example embodiment, an electronic device is provided, the electronic device comprising: at least one memory configured to store computer-readable instructions; a communication interface comprising an antenna array configured to form a plurality of candidate beams; and at least one processor operably connected to the communication interface and the at least one memory, wherein the at least one processor is configured to execute the computer-readable instructions to: generate a plurality of RSRP pattern data based on a reference signal received power (RSRP) measured in each of the plurality of candidate beams for a signal received from an external device; estimate an angle of arrival (AoA) distribution for each of the plurality of RSRP pattern data by applying each of the plurality of RSRP pattern data to a neural network trained based on a deep learning algorithm; and perform beam management for wireless communication with the external device based on the estimated AoA distribution.

[0009] According to one aspect of a disclosed example embodiment, there is provided a method for operating an electronic device, the method comprising: generating a plurality of RSRP pattern data based on a reference signal received power (RSRP) measured in each of a plurality of candidate beams for a signal received from an external device; estimating an angle of arrival (AoA) distribution for each of the plurality of RSRP pattern data by applying each of the plurality of RSRP pattern data to a neural network trained based on a deep learning algorithm; and performing beam management for wireless communication with the external device based on the estimated AoA distribution.

[0010] According to one aspect of a disclosed example embodiment, there is provided a method for operating a wireless communication system, the method comprising: forming a plurality of candidate beams for performing wireless communication with an external device; generating an angle of arrival (AoA) estimation model via deep learning by using a plurality of reference signal received power (RSRP) pattern training data for the plurality of candidate beams as input and using an angle of arrival (AoA) distribution corresponding to each of the plurality of RSRP pattern training data as output; generating a plurality of RSRP pattern data by measuring RSRP in each of the plurality of candidate beams for a signal received from the external device; estimating an AoA distribution for each of the plurality of RSRP pattern data by applying each of the plurality of RSRP pattern data to the AoA estimation model; and performing beam management based on the estimated AoA distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Example embodiments will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings.

[0012] Figure 1 An example of operating an electronic device according to one or more embodiments is shown.

[0013] Figure 2 is a block diagram illustrating an electronic device according to one or more embodiments.

[0014] Figure 3A is a diagram illustrating an example of a training operation of a neural network according to one or more embodiments.

[0015] Figure 3B is a diagram illustrating an example of training data for a neural network according to one or more embodiments.

[0016] Figure 3C is a diagram illustrating an example of a training result of a neural network according to one or more embodiments.

[0017] Figure 4A is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0018] Figure 4B is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0019] Figure 5A is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0020] Figure 5B is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0021] Figure 5C is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0022] Figure 6 is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0023] Figure 7 is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0024] Figure 8 is a flowchart illustrating a method of operating an electronic device according to one or more embodiments.

[0025] Fig. 9 is a flowchart illustrating a method of operating an electronic device according to one or more embodiments.

[0026] Fig.10 is a flowchart illustrating a method of operating an electronic device according to one or more embodiments.

[0027] Fig.11 is a schematic block diagram illustrating a wireless communication system according to one or more embodiments. DETAILED DESCRIPTION

[0028] Hereinafter, example embodiments are described in detail with reference to the accompanying drawings. In the example embodiments of the present disclosure, components according to the inventive concept are represented in singular form or plural form, depending on the embodiment presented. However, the singular representation or plural representation is selected for ease of description and is suitable for the context of presentation, and the disclosure is not limited to singular components or plural components, and components represented in plural form may be configured as a single component, and / or components represented in singular form may be configured as multiple components (e.g., one or more neural networks).

[0029] For ease of description, the inventive concept uses some terms and terminologies defined in the 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE) standard or the New Radio (NR, or referred to as New Air Interface) standard. However, the inventive concept is not limited to the above terms and terminologies and can be applied to systems conforming to other standards in the same manner.

[0030] As a non-limiting example, a wireless communication system according to one or more embodiments may include an NR system, a fifth generation (5G) system, an LTE system, a code division multiple access (CDMA) system, a global mobile communication system (GSM) system, a wireless local area network (WLAN) system, and / or any other wireless communication system. Hereinafter, the wireless communication system is described assuming that the wireless communication system is an NR system, an LTE system, or a system capable of supporting communications based on NR and LTE. However, it should be understood that the inventive concept is not limited thereto.

[0031] An electronic device (eg, terminal) according to one or more embodiments may be referred to as, for example, user equipment, mobile station (MS), mobile terminal (MT), user terminal, subscriber station (SS), wireless device, handheld device, etc.

[0032] An electronic device (eg, terminal) according to one or more embodiments may support fourth generation (4G) communication (eg, LTE or LTE-Advanced (LTE-A)) and 5G communication (eg, NR) specified in the 3GPP standard.

[0033] For 4G communication and 5G communication, an electronic device (e.g., a terminal) according to one or more embodiments may support a CDMA-based communication protocol, a wideband CDMA (WCDMA)-based communication protocol, a time division multiple access (TDMA)-based communication protocol, a frequency division multiple access (FDMA)-based communication protocol, an orthogonal frequency division multiplexing (OFDM)-based communication protocol, a cyclic prefix (CP)-OFDM-based communication protocol, a discrete Fourier transform-spread-OFDM (DFT-s-OFDM)-based communication protocol, a non-orthogonal multiple access (NOMA)-based communication protocol, a generalized frequency division multiplexing (GFDM)-based communication protocol, and the like.

[0034] In one or more embodiments, wireless communication may be performed based on any one of 5G communication (e.g., NR) (e.g., FR1 band or FR2 band of 5G communication), 4G communication (e.g., LTE or LTE-A), ultra-wideband (UWB), and wireless fidelity (Wi-Fi) that enables high-speed data transmission between an external device (e.g., a base station) and an electronic device (e.g., a terminal).

[0035] According to one or more embodiments, an external device (e.g., a base station) may represent a fixed station that communicates with an electronic device (e.g., a terminal) and / or other base stations. For example, the external device (e.g., a base station) may be referred to as a node B, an evolved node B (eNB), a sector, a site, a base transceiver system (BTS), an access point (AP), a relay node, a remote radio head (RRH), a radio unit (RU), etc. For example, in a wireless communication system, an external device (e.g., a base station) includes at least one cell, which is the smallest unit of an area for providing communication services to each external device (e.g., a base station). The external device (e.g., a base station) may provide efficient multiple access communications for multiple terminals based on specific frequency resources allocated to each cell. Specifically, in a multiple-input multiple-output (MIMO) system, an electronic device (e.g., a terminal) and an external device (e.g., a base station) may perform wireless communications between the electronic device (e.g., a terminal) and the external device (e.g., a base station) based on multiple component carriers.

[0036] In this specification, reference signal received power (RSRP) pattern data may refer to data including a signal received from an external device (eg, a base station) in an electronic device (eg, Figure 1 The present invention also provides data of RSRP values ​​measured in each of a plurality of candidate beams (100).

[0037] In this specification, as in training a neural network (e.g., Figure 1 The RSRP pattern training data may include RSRP values ​​measured from a plurality of candidate beams while pre-changing an angle of the electronic device (e.g., an angle of arrival (AoA)) or may include RSRP values ​​of a plurality of candidate beams obtained through a previous experiment.

[0038] In the present specification, as a probability distribution of each AoA class estimated by a neural network (eg, using a deep learning algorithm), the AoA distribution may include a distribution of matching probabilities between RSRP pattern training data and each AoA class.

[0039] In the present specification, the AoA class may represent an AoA in units of a predetermined angle (eg, 5 degrees), as a value to be estimated by a neural network (eg, using a deep learning algorithm).

[0040] In this specification, a dominant AoA class may mean an AoA class having a significantly high matching probability (eg, having the highest matching probability) in an AoA distribution estimated by a neural network (eg, using a deep learning algorithm).

[0041] In this specification, the candidate AoA class may mean: if there is no main AoA class in the AoA distribution estimated by the neural network (for example, using a deep learning algorithm), at least two AoA classes having a matching probability greater than or equal to a threshold in the estimated AoA distribution.

[0042] In the present specification, a neural network may refer to a neural network that is trained to receive RSRP pattern training data of a plurality of candidate beams of an electronic device and estimate an AoA distribution corresponding to the RSRP pattern training data through a deep learning algorithm.

[0043] Figure 1 An example of operating the electronic device 100 according to one or more example embodiments is illustrated.

[0044] Reference Figure 1 , the electronic device 100 may include a processor 110 and a neural network 111. Figure 1 Components of the electronic device 100 (eg, the processor 110 and the neural network 111) are shown, but these components are only some of the components of the electronic device 100. The electronic device 100 may also include various other components. Figure 2 Some specific components of the electronic device 100 are described.

[0045] exist Figure 1 In the present invention, the antenna array (not shown) of the electronic device 100 may form a predetermined number of candidate beams according to the characteristics of the wireless communication system and / or the state of the wireless communication. For example, based on the characteristics of the wireless communication system and / or the state of the wireless communication, the candidate beams of the electronic device 100 may include all beams included in the predefined beam codebook or some beams included in the predefined beam codebook. In order to cover all spatial domains where signals can be received from external devices, the candidate beams of the electronic device 100 (based on the predetermined number of candidate beams) may be arranged not to overlap with each other in space (low resolution characteristics of the RSRP pattern data 10). For example, the candidate beams of the electronic device 100 may include the zeroth candidate beam corresponding to the -60 degree area in the spatial domain, the first candidate beam corresponding to the -40 degree area in the spatial domain, the second candidate beam corresponding to the -20 degree area in the spatial domain, the third candidate beam corresponding to the 0 degree area in the spatial domain, the fourth candidate beam corresponding to the 20 degree area in the spatial domain, the fifth candidate beam corresponding to the 40 degree area in the spatial domain, and the sixth candidate beam corresponding to the 60 degree area in the spatial domain.

[0046] In one or more embodiments, the neural network 111 of the electronic device 100 may receive RSRP pattern data 10 (e.g., with low resolution) and output (or estimate) an AoA distribution 20 (or AoA class) (e.g., with high resolution). That is, the neural network 111 may convert the RSRP pattern data 10 (low resolution) into an AoA distribution 20 (high resolution). The RSRP pattern data 10 as data input to the neural network 111 may include RSRP measured from each of a plurality of candidate beams of the electronic device 100 for a signal received from an external device (e.g., a base station). In addition, the AoA distribution 20 as data output from the neural network 111 may include a distribution of matching probabilities between the RSRP pattern data 10 and each angle in the AoA class. For example, the beam corresponding to the AoA class may cover an angle range of preset units in the spatial domain (e.g., 0 degrees, 10 degrees, 15 degrees, ..., 75 degrees, etc.), and the beam corresponding to the AoA class may be set to be finer than the beam spacing of the candidate beam and / or the beam width of the candidate beam (high resolution characteristics of the AoA distribution 20). For example, the AoA distribution 20 may include a first AoA class corresponding to a 0 degree area in the spatial domain, a first AoA class corresponding to a 5 degree area in the spatial domain, a second AoA class corresponding to a 10 degree area in the spatial domain, ..., a 36th AoA class corresponding to a 180 degree area in the spatial domain. It can be seen that compared to the coverage area of ​​the candidate beam of the electronic device 100 described above (which is a low-resolution beam divided in units of 20 degrees), the coverage area of ​​the beam corresponding to the AoA class is a high-resolution beam divided more finely in units of 5 degrees. Reference will be made to the following. Figure 3A A deep learning based training process for a neural network 111 for outputting (or estimating) an AoA distribution 20 is described.

[0047] In one or more embodiments, the processor 110 may perform beam management based on the output AoA distribution 20 (high resolution). Here, beam management may include a beam selection operation, a beam training operation, and / or a beam recovery operation of the electronic device 100. FIG. 4A to FIG. 7 A beam selection operation and a beam training operation using the AoA distribution 20 (high resolution) of the electronic device 100 are described. For example, the electronic device 100 may select a high-resolution best beam (e.g., an optimal beam in a fine angle range) based on the AoA distribution 20 to ensure optimal wireless communication performance. When the electronic device 100 includes a plurality of antenna arrays, beam management of the electronic device 100 may be performed by each of the plurality of antenna arrays, which will be described below.

[0048] The device and method according to one or more embodiments may estimate the AoA distribution (or AoA class) corresponding to the input RSRP pattern data using a neural network trained via deep learning (without changing the hardware or improving the hardware performance of the electronic device 100), and may select a high-resolution optimal beam for wireless communication with an external device based on the estimated AoA distribution (or AoA class). Therefore, the wireless communication performance between devices may be improved by minimizing the overhead of the beam training process.

[0049] Figure 2 is a block diagram illustrating an electronic device 100 according to one or more example embodiments.

[0050] Reference Figure 2 , the electronic device 100 may include a processor 110 , a communication interface 120 , a sensor 130 , and a memory 140 , wherein the processor 110 includes a neural network 111 .

[0051] In one or more embodiments, the processor 110 may execute programs and / or processes stored in the memory 140 to control the operation of the electronic device 100. In some embodiments, the processor 110 may store program code in the memory 140, which is executed to perform operations of training the neural network 111 according to example embodiments (e.g., Figure 3A ) and operations according to example embodiments that perform beam management based on the trained neural network 111 (e.g., FIG. 4A to FIG. 7 and Figures 8 to 10 ). The processor 110 may access the memory 140 and execute the stored program code. Therefore, the processor 110 may estimate the AoA distribution corresponding to the RSRP pattern data measured from the plurality of candidate beams of the communication interface 120 via the neural network 111, and may control the beam management so that the beam management is performed based on the estimated AoA distribution. Figure 2 It is described in the drawings that one processor 110 and one memory 140 are provided, but example embodiments are not limited thereto, and the electronic device 100 may include two or more numbers of processors 110 and two or more numbers of memories 140 .

[0052] In one or more embodiments, the processor 110 may include a neural network 111. In one embodiment, the neural network 111 may be provided as a logic block formed by logic synthesis, a software block executed by the processor, and / or a combination thereof. In one or more embodiments, the neural network 111 may include a program as a set of multiple instructions executed by the processor 110, and the program may be stored in a memory accessible to the processor 110.

[0053] In one or more embodiments, the neural network 111 may be trained to use the RSRP pattern training data of the plurality of candidate beams as input data, and estimate the AoA distribution corresponding to the RSRP pattern training data via a deep learning algorithm. This will be referred to below. FIG. 3A to FIG. 3C Detailed description.

[0054] In one or more embodiments, the neural network 111 may use a trained deep learning algorithm and estimate the AoA distribution (high resolution) based on RSRP pattern data (low resolution) measured in a plurality of candidate beams for a signal received from an external device. Here, the AoA distribution may include a distribution of matching probabilities between the RSRP pattern data and each AoA class. In addition, the beam corresponding to the AoA class may be set to be finer than the beam spacing of the candidate beams and / or the beam width of the candidate beams (high resolution characteristics of the AoA distribution).

[0055] The neural network 111 may use a multi-layer perception model, a learning algorithm based on gradient descent, a cross entropy function model, etc. as a deep learning algorithm (or deep learning model). However, the deep learning algorithm (or deep learning model) of the neural network 111 according to the embodiment is not limited thereto, and other types of deep learning algorithms (or deep learning models) may be used.

[0056] The processor 110 may perform beam management based on the AoA distribution received from the neural network 111. FIG. 4A to FIG. 7 An example embodiment of a beam management operation of the processor 110 is described. For example, the processor 110 may determine an AoA class corresponding to RSRP pattern data (low resolution), and select a beam corresponding to the determined AoA class from a high-resolution beam codebook as a target beam. The processor 110 may control the communication interface 120, and perform wireless communication with an external device (e.g., a base station) using the target beam.

[0057] In one or more embodiments, if the position of the electronic device 100 and / or the orientation (or referred to as orientation) of the electronic device 100 changes after the target beam is selected, the processor 110 may generate correction information for the target beam based on the sensing data (e.g., data indicating the amount of change in the orientation of the electronic device 100) received from the sensor 130, and then correct the existing target beam based on the generated correction information. The processor 110 may select the corrected existing target beam as the final target beam, and perform wireless communication with an external device (e.g., a base station) using the final target beam.

[0058] The communication interface 120 may include at least one antenna array including a plurality of antennas. In addition, the communication interface 120 may form a plurality of candidate beams using a plurality of antennas (or antenna arrays). The communication interface 120 may send a signal to an external device (e.g., a base station) via a plurality of candidate beams and / or receive a radio frequency (RF) signal from an external device (e.g., a base station) via a plurality of candidate beams. The communication interface 120 may generate an intermediate frequency signal or a baseband signal by down-converting the RF signal received from the external device. The processor 110 may generate a data signal by filtering, decoding and / or digitizing the intermediate frequency signal or the baseband signal. The processor 110 may additionally process the data signal sent to the external device and the data signal received from the external device. The processor 110 may control the communication interface 120 and generate RSRP pattern data (low resolution) including RSRP data measured in each of the plurality of candidate beams for the signal received from the external device.

[0059] In addition, the communication interface 120 may receive a data signal from the processor 110. The communication interface 120 may perform encoding, multiplexing, and analog conversion on the received data signal. The communication interface 120 may up-convert the frequency of the intermediate frequency signal or baseband signal output from the processor 110, and transmit the converted intermediate frequency signal or baseband signal as an RF signal to an external device via a plurality of candidate beams.

[0060] The sensor 130 may detect an operating condition (e.g., position, power, temperature, etc.) or an external environmental condition (e.g., a user's state, etc.) of the electronic device 100, and may generate an electrical signal and / or a data value corresponding to the detected condition. The sensor 130 may include, for example, but not limited to, a gesture sensor, a gyroscope sensor, an air pressure sensor, a magnetic sensor, an acceleration sensor, a grip sensor, a proximity sensor, a color sensor, an infrared (IR) sensor, a biometric sensor, a temperature sensor, a humidity sensor, and / or an illumination sensor. For example, when the position of the electronic device 100 or the orientation of the electronic device 100 changes, the sensor 130 may generate sensing data indicating the amount of change in the orientation of the electronic device 100 for each reference axis (e.g., the horizontal axis (x), the vertical axis (y), and the height axis (z)) based on the position of the electronic device 100 before the change (e.g., considered as the origin), and may send the sensing data to the processor 110.

[0061] The memory 140 may include an operating system and may include applications and / or process registers equipped with device drivers, executable libraries, and / or program codes. The operating system and the applications may be provided as software elements and may be implemented by executing codes and / or commands using a processor. For example, the memory 140 may store commands executed to perform operations for training the neural network 111 according to example embodiments (e.g., Figure 3A) and operations according to example embodiments that perform beam management based on the trained neural network 111 (e.g., FIG. 4A to FIG. 7 and Figures 8 to 10 ) program code.

[0062] According to an embodiment, an electronic device and a method of operating the electronic device are provided. The electronic device can estimate the AoA distribution of RSRP pattern data measured for a plurality of candidate beams of the electronic device using a neural network trained by a deep learning algorithm, and perform beam management, thereby minimizing the overhead of beam training.

[0063] An electronic device and a method of operating the electronic device according to example embodiments may perform wireless communication with an external device via a high-resolution target beam selected using an AoA distribution estimated by a neural network, thereby improving wireless communication quality.

[0064] Figure 3A is a diagram illustrating an example of a training operation of the neural network 111 according to one or more example embodiments.

[0065] Reference Figure 3A , when RSRP pattern training data is input to the neural network 111, the neural network 111 may be trained to estimate and output the AoA distribution (or AoA class) corresponding to the RSRP pattern training data using a deep learning algorithm. The RSRP pattern training data may represent RSRP pattern data of a plurality of candidate beams that have been measured or tested in advance to train the neural network 111 (or to be used in the deep learning algorithm of the neural network 111). As a probability distribution of each AoA class estimated by the neural network 111, the AoA distribution may include a distribution of matching probabilities between the RSRP pattern training data and each AoA class. The neural network 111 may be trained to estimate and output the AoA class corresponding to the RSRP pattern training data.

[0066] In the deep learning algorithm (or deep learning model), the ground truth for training the neural network 111 may include a true AoA (e.g., an ideal AoA distribution or an ideal AoA class) as a label value of the RSRP mode training data. For example, the true AoA (e.g., an ideal AoA distribution (or an ideal AoA class) corresponding to the RSRP mode training data) may be labeled in the RSRP mode training data, and the neural network 111 may be trained based on the labeled data. As an example of labeling, a one-hot vector having a dimension corresponding to the number of AoA classes constituting the AoA distribution for the “line-of-sight (LoS) RSRP mode” may be set as a label. The RSRP mode training data may be generated using all or some of the multiple candidate beams of the electronic device 100. In the RSRP mode training data, a field corresponding to an unused beam among the multiple candidate beams may include “0”. A mean square error (MSE) function or a cross entropy function may be used as a loss function for training the neural network 111. The neural network 111 may be trained based on the shuffled RSRP mode training data. In addition to the functions described above, the neural network 111 according to the embodiment may also be trained based on various deep learning algorithms (or deep learning models) and loss functions associated therewith.

[0067] The neural network 111 may be trained to adjust parameters or weights to minimize the difference between the estimated AoA distribution (eg, output values ​​of the neural network 111 ) for the RSRP pattern training data and the pre-labeled ground truth (eg, true AoA).

[0068] Figure 3B is a diagram illustrating an example of training data of the neural network 111 according to one or more example embodiments.

[0069] In detail, Figure 3B Shown as Figure 3A The example of training data for the neural network 111 includes RSRP pattern training data (e.g., first RSRP pattern training data 311 and second RSRP pattern training data 321) and true AoAs (or ground truths) labeled for each RSRP pattern training data (e.g., first true AoAs 315 and second true AoAs 325). The neural network 111 according to one or more embodiments may be trained based on the true AoAs (e.g., ideal AoA distributions or ideal AoA classes) corresponding to the RSRP pattern training data.

[0070] exist Figure 3BIn , it is assumed that the multiple candidate beams of the electronic device 100 may include a zeroth candidate beam corresponding to a -60 degree region in the spatial domain, a first candidate beam corresponding to a -40 degree region in the spatial domain, a second candidate beam corresponding to a -20 degree region in the spatial domain, a third candidate beam corresponding to a 0 degree region in the spatial domain, a fourth candidate beam corresponding to a 20 degree region in the spatial domain, a fifth candidate beam corresponding to a 40 degree region in the spatial domain, and a sixth candidate beam corresponding to a 60 degree region in the spatial domain. In addition, Figure 3B In , the true AoA (eg, ground truth) used to train the neural network 111 may be set to an angle of 5 degrees. Figure 3B In the present invention, the real AoA is assumed to indicate an ideal AoA class corresponding to the RSRP pattern training data. Here, for example, the RSRP pattern training data for each ideal AoA class may include RSRP values ​​measured in a plurality of candidate beams while changing the angle of the electronic device 100 in advance, or may include RSRP values ​​of a plurality of candidate beams obtained through previous experiments.

[0071] Reference Figure 3B , the neural network 111 may be trained to estimate the AoA distribution (or AoA class) based on training data such as row t1, row t2, ..., and row tn (not shown). (For example, Figure 2 The processor 110 in the figure may collect training data (eg, row t1, row t2, ..., row tn, etc.) to train the neural network 111.

[0072] In one or more embodiments, row t1 may include the first RSRP pattern training data 311 and the first true AoA 315 (e.g., 0 degrees) as a label value of the first RSRP pattern training data 311. The first RSRP pattern training data 311 may include RSRP values ​​in the zeroth candidate beam, the first candidate beam, the second candidate beam, the third candidate beam, the fourth candidate beam, the fifth candidate beam, and the sixth candidate beam measured and / or obtained for the first true AoA 315 (e.g., 0 degrees). For example, for the first true AoA 315 (e.g., 0 degrees), the RSRP value of the zeroth candidate beam may be "-96.81", the RSRP value of the first candidate beam may be "-108.89", the RSRP value of the second candidate beam may be "-108.80", the RSRP value of the third candidate beam may be "-106.64", the RSRP value of the fourth candidate beam may be "-104.97", the RSRP value of the fifth candidate beam may be "-100.73", and the RSRP value of the sixth candidate beam may be "-95.16".

[0073] In one or more embodiments, row t2 may include the second RSRP pattern training data 321 and the second true AoA 325 (e.g., 5 degrees) as a label value of the second RSRP pattern training data 321. The second RSRP pattern training data 321 may include RSRP values ​​in the zeroth candidate beam, the first candidate beam, the second candidate beam, the third candidate beam, the fourth candidate beam, the fifth candidate beam, and the sixth candidate beam measured and / or obtained for the second true AoA 325 (e.g., 5 degrees). For example, for the second true AoA 325 (e.g., 5 degrees), the RSRP value of the zeroth candidate beam may be "-97.36", the RSRP value of the first candidate beam may be "-102.84", the RSRP value of the second candidate beam may be "-106.97", the RSRP value of the third candidate beam may be "-107.85", the RSRP value of the fourth candidate beam may be "-110.35", the RSRP value of the fifth candidate beam may be "-100.27", and the RSRP value of the sixth candidate beam may be "-95.88".

[0074] exist Figure 3B In the embodiment, for ease of description, the multiple candidate beams of the electronic device 100 are described as including a total of 7 beams in units of 20 degrees in the spatial domain, but the embodiment is not limited thereto. The multiple candidate beams according to one or more embodiments may include any other number of beams in units of any other angles in the spatial domain depending on the setting.

[0075] exist Figure 3B For ease of description, the real AoA (e.g., Figure 3B The angle in (AoA) is described as being set in units of 5 degrees, but the embodiment is not limited thereto. The actual AoA according to one or more embodiments may be set in units of any other angles depending on the characteristics and / or channel conditions of the wireless communication system.

[0076] The apparatus and method according to one or more embodiments may estimate an AoA class (or AoA distribution) corresponding to a beam that is relatively more finely divided than multiple candidate beams of the electronic device 100 and covers a spatial domain (or physical area) based on the RSRP pattern training data described above and a real AoA marked therewith, and perform precise beam alignment based on the estimated high-resolution AoA class (or AoA distribution), thereby improving the performance and / or quality of wireless communication.

[0077] Figure 3C is a diagram illustrating an example of a result of training the neural network 111 according to one or more embodiments.

[0078] In detail, as Figure 3A An example of the training results of the neural network 111 in, Figure 3CRSRP pattern data (e.g., first RSRP pattern data 351 and second RSRP pattern data 361) and AoA classes (e.g., first AoA class 355 and second AoA class 365) corresponding to the RSRP pattern data estimated by the neural network 111 are shown. The neural network 111 according to one or more embodiments may estimate an AoA distribution or an AoA class corresponding to the RSRP pattern data. Figure 3C In FIG. 1 , for ease of description, the neural network 111 is described as being trained to estimate the AoA class.

[0079] exist Figure 3C In the example, it is assumed that the multiple candidate beams of the electronic device 100 may include a zeroth candidate beam corresponding to a -60 degree region in the spatial domain, a first candidate beam corresponding to a -40 degree region in the spatial domain, a second candidate beam corresponding to a -20 degree region in the spatial domain, a third candidate beam corresponding to a 0 degree region in the spatial domain, a fourth candidate beam corresponding to a 20 degree region in the spatial domain, a fifth candidate beam corresponding to a 40 degree region in the spatial domain, and a sixth candidate beam corresponding to a 60 degree region in the spatial domain. In addition, Figure 3C , it is assumed that the AoA class estimated by the neural network 111 includes the AoA class set in units of an angle of 5 degrees. Here, the RSRP pattern data may include RSRP values ​​measured in a plurality of candidate beams of the electronic device 100 for a signal received from an external device.

[0080] Reference Figure 3C , the neural network 111 may estimate the AoA class (or AoA distribution) based on RSRP pattern data (such as those RSRP pattern data included in rows T1, T2, . . . , and Tn (not shown)). The processor 110 may generate RSRP pattern data based on RSRP values ​​measured in each of a plurality of candidate beams for a signal received from an external device (e.g., a base station), and may then transmit (or input) the RSRP pattern data to the neural network 111.

[0081] In one or more embodiments, the neural network 111 may use the first RSRP pattern data 351 in response to the input of the first RSRP pattern data 351. Figure 3AThe deep learning algorithm (or deep learning model) trained in the electronic device 100 may estimate the AoA class corresponding to the first RSRP pattern data 351 (included in row T1). Row T1 may include the first RSRP pattern data 351 and a first AoA class 355 (e.g., 50 degrees) estimated by the neural network 111 to correspond to the first RSRP pattern data 351. Here, the first RSRP pattern data 351 may include RSRP values ​​in the zeroth candidate beam, the first candidate beam, the second candidate beam, the third candidate beam, the fourth candidate beam, the fifth candidate beam, and the sixth candidate beam in the electronic device 100. For example, in the first RSRP pattern data 351, the RSRP value of the zeroth candidate beam may be "-114.81", the RSRP value of the first candidate beam may be "-93.88", the RSRP value of the second candidate beam may be "-90.49", the RSRP value of the third candidate beam may be "-99.22", the RSRP value of the fourth candidate beam may be "-101.16", the RSRP value of the fifth candidate beam may be "-101.89", and the RSRP value of the sixth candidate beam may be "-96.99". The neural network 111 may estimate (or output) the first AoA class 355 (e.g., 50 degrees) as the AoA class corresponding to the first RSRP pattern data 351 described above.

[0082] In one or more embodiments, the neural network 111 may use the second RSRP pattern data 361 in response to input. Figure 3AThe deep learning algorithm (or deep learning model) trained in the electronic device 100 may be used to estimate the AoA class corresponding to the second RSRP pattern data 361 (included in row T2). Row T2 may include the second RSRP pattern data 361 and a second AoA class 365 (e.g., 75 degrees) estimated by the neural network 111 to correspond to the second RSRP pattern data 361. Here, the second RSRP pattern data 361 may include RSRP values ​​in the zeroth candidate beam, the first candidate beam, the second candidate beam, the third candidate beam, the fourth candidate beam, the fifth candidate beam, and the sixth candidate beam in the electronic device 100. For example, in the second RSRP pattern data 361, the RSRP value of the zeroth candidate beam may be "-100.07", the RSRP value of the first candidate beam may be "-107.60", the RSRP value of the second candidate beam may be "-102.50", the RSRP value of the third candidate beam may be "-104.92", the RSRP value of the fourth candidate beam may be "-100.06", the RSRP value of the fifth candidate beam may be "-98.10", and the RSRP value of the sixth candidate beam may be "-100.55". The neural network 111 may estimate (or output) the second AoA class 365 (e.g., 75 degrees) as the AoA class corresponding to the second RSRP pattern data 361 described above.

[0083] exist Figure 3C In the embodiment, for ease of description, the multiple candidate beams of the electronic device 100 are described as including a total of 7 beams in units of 20 degrees in the spatial domain, but the embodiment is not limited thereto. The multiple candidate beams according to one or more embodiments may include any other number of beams in units of any other angles in the spatial domain depending on the setting.

[0084] exist Figure 3C In the embodiment, for ease of description, the neural network 111 is described as estimating the AoA class in response to the input of RSRP pattern data, but the embodiment is not limited thereto. The neural network 111 according to one or more embodiments may be used in the same manner as described above with reference to FIG. 3A to FIG. 3B The AoA distribution is estimated in response to the input of RSRP pattern data in a manner identical or similar to that described.

[0085] exist Figure 3C In the case of the estimated AoA class (e.g., Figure 3C The angle in (i) is described as being set in units of 5 degrees, but the embodiment is not limited thereto. The estimated AoA class according to one or more embodiments may be set in units of any other angles (e.g., 1 degree, 2 degrees, etc.) depending on the characteristics and / or channel conditions of the wireless communication system.

[0086] According to the embodiment, a device and a method of operating the device are provided. The device can minimize the overhead of beam training by using the neural network 111 trained based on the RSRP pattern training data described above and the real AoA labeled therewith, and can perform accurate beam alignment by estimating high-resolution AoA classes (or AoA distributions). FIG. 4A to FIG. 7 In the following, one or more embodiments of beam management based on the AoA class (or AoA distribution) estimated by the neural network 111 are described.

[0087] Figure 4A is a diagram illustrating an example of an operation of beam management of the electronic device 100 according to one or more example embodiments.

[0088] Reference Figure 4A , the electronic device 100 may be based on (for example, Figure 2 The AoA distribution estimated by the neural network 111 of the processor 110 in the embodiment of the present invention is used to perform beam management using the main AoA class. Figure 4A When, in view of the above reference Figures 1 to 3C Description, and reference Figures 1 to 3C Descriptions that are duplicate of those given may be omitted.

[0089] exist Figure 4A In the present invention, the RSRP pattern data may refer to data including an RSRP value measured in each of a plurality of candidate beams of the electronic device 100 for a signal received from an external device (eg, a base station).

[0090] In one or more embodiments, the electronic device 100 (eg, Figure 2 The processor 110 may be configured to generate a beam code based on a predefined low-resolution beam codebook (eg, see Figure 4B The processor 110 ( Figure 2 ) The generated RSRP pattern data may be sent or input to the neural network 111.

[0091] In one or more embodiments, in response to input of RSRP pattern data (eg, Figure 2 The neural network 111 of the processor 110 may be Figure 3A The deep learning algorithm (or deep learning model) trained in the process is used to output the main AoA class corresponding to the RSRP pattern data. For example, the processor 110 ( Figure 2 ) can be obtained by applying the input RSRP pattern data to Figure 3AThe processor 110 may estimate the AoA distribution using a deep learning algorithm (or deep learning model) trained in the AoA distribution, and may output the AoA class with the highest matching probability in the estimated AoA distribution as the main AoA class. Here, if there is one AoA class with a significantly high matching probability (for example, the difference between the two highest matching probabilities is a threshold or more) in the estimated AoA distribution, then the one AoA class may represent the main AoA class. For example, if there are at least two AoA classes with matching probabilities higher than a threshold in the estimated AoA distribution, the processor 110 ( Figure 2 ) Executable according to FIG. 5A to FIG. 5C and Figure 6 As a probability distribution for each AoA class estimated by the neural network 111, the AoA distribution may include a distribution of matching probabilities between the RSRP pattern training data and each AoA class (e.g., the sum of the matching probabilities for all AoA classes may be "1").

[0092] In one or more embodiments, the processor 110 (e.g., Figure 2 ) can control the communication interface 120 ( Figure 2 ) Select a high-resolution beam codebook (e.g., Figure 4B A beam corresponding to the main AoA class in 450 of the electronic device 100 is used as a target beam, and wireless communication between an external device (for example, a base station) and the electronic device 100 can be performed using the target beam.

[0093] Figure 4B is a diagram illustrating an example of an operation of beam management of the electronic device 100 according to one or more example embodiments.

[0094] In detail, Figure 4B Show reference Figure 4A Examples and references to the low-resolution beam set 410 of the described low-resolution beam codebook Figure 4A An example of a high-resolution beam set 450 of a high-resolution beam codebook is described.

[0095] Reference Figure 4B , as a plurality of candidate beams used by the electronic device 100 when generating RSRP pattern data input to the neural network 111, the low-resolution beam set 410 represents beams belonging to the low-resolution beam codebook. For example, the processor 110 (e.g., Figure 2) may use all or some beams in a predefined low-resolution beam codebook to generate RSRP pattern data. The low-resolution beam set 410 may have a wider spacing between beams than the high-resolution beam set 450, and may have a larger beam width than the high-resolution beam set 450. For example, the low-resolution beam set 410 may include a first candidate beam 411-1, a second candidate beam 411-2, ..., an mth candidate beam 411-m as multiple candidate beams of the electronic device 100.

[0096] The high-resolution beam set 450 may represent a beam belonging to a high-resolution beam codebook as a beam corresponding to each AoA class of the AoA distribution output from the neural network 111. The high-resolution beam set 450 may have a narrower beam spacing than the low-resolution beam set 410, and may have a smaller beam width (finer) than the low-resolution beam set 410. For example, the high-resolution beam set 450 may include a first candidate beam 451-1, a second candidate beam 451-2, ..., and an nth candidate beam 451-n.

[0097] Reference Figure 4A , the processor 110 (eg, Figure 2 ) may select a beam corresponding to the main AoA class in the high-resolution beam set 450 as a target beam, and may control the antenna array of the communication interface 120 so that wireless communication is performed based on the target beam.

[0098] The electronic device and the method of operating the electronic device according to one or more embodiments may minimize the overhead of beam training by performing beam training based on the low-resolution beam set 410, and may perform wireless communication according to the AoA class estimated based on the high-resolution beam set 450. Therefore, accurate beam alignment may be performed, and wireless communication quality may be improved.

[0099] Figure 5A is a diagram illustrating an example of an operation of beam management of the electronic device 100 according to one or more example embodiments.

[0100] Reference Figure 5A , the electronic device 100 may be based on the processor 110 (eg, Figure 2 ) and uses the candidate AoA classes to perform beam management. Figure 5A When, in view of the above reference Figures 1 to 4B Description, and reference Figures 1 to 4B Descriptions that are duplicate of those given may be omitted.

[0101] exist Figure 5AIn the present invention, the RSRP pattern data may refer to data including an RSRP value measured in each of a plurality of candidate beams of the electronic device 100 for a signal received from an external device (eg, a base station).

[0102] In one or more embodiments, the processor 110 (eg, Figure 2 ) can be based on a predefined low-resolution beam codebook (see, for example, Figure 4B The processor 110 ( Figure 2 ) The generated RSRP pattern data may be sent or input to the neural network 111.

[0103] In one or more embodiments, in response to input of RSRP pattern data, processor 110 (eg, Figure 2 ) of the neural network 111 can be Figure 3A The deep learning algorithm (or deep learning model) trained in the process is used to output the AoA distribution corresponding to the RSRP pattern data. For example, the processor 110 ( Figure 2 ) can apply the input RSRP pattern data to Figure 3A A deep learning algorithm (or deep learning model) trained in the algorithm may output an estimated AoA distribution. For example, when there are at least two AoA classes with a matching probability greater than or equal to a threshold in the estimated AoA distribution (e.g., when there are no AoA classes in the estimated AoA distribution Figure 4A When the main AoA class is determined by the neural network 111, the neural network 111 may output an estimated AoA distribution. As a probability distribution of each AoA class estimated by the neural network 111, the AoA distribution may include a distribution of matching probabilities between the RSRP pattern training data and each AoA class (for example, the sum of matching probabilities of all AoA classes may be "1").

[0104] In one or more embodiments, the processor 110 (e.g., Figure 2 ) may select an AoA class having a matching probability greater than or equal to a threshold in the AoA distribution as a candidate AoA class, and perform beam training on the beam corresponding to the candidate AoA class (e.g., see Figure 6 ). In the AoA distribution, the number of candidate AoA classes can be determined according to a predetermined threshold. Figure 6 Describe the operation of beam training. Processor 110 ( Figure 2 ) The target beam can be selected based on the results of beam training.

[0105] In one or more embodiments, the processor 110 (e.g., Figure 2) An AoA class having a matching probability greater than or equal to a predetermined threshold in the AoA distribution may be selected as a candidate AoA class, and a beam generated by combining beams corresponding to the candidate AoA class may be selected as a target beam. In the AoA distribution, the number of candidate AoA classes may be determined according to a predetermined threshold.

[0106] In one or more embodiments, the processor 110 (e.g., Figure 2 ) can use the selected target beam to perform wireless communication with an external device (e.g., a base station).

[0107] Figure 5B is a diagram illustrating an example of an operation of beam management of the electronic device 100 according to one or more example embodiments.

[0108] In detail, Figure 5B Show Figure 5A The AoA distribution of includes two candidate AoA classes. Figure 5B , there are a total of two AoA classes (including a first candidate AoA class and a second candidate AoA class) having a matching probability greater than or equal to a threshold in the AoA distribution, and it is assumed that beam gains having the same value are allocated to the beam 511 corresponding to the first candidate AoA class and the beam 513 corresponding to the second candidate AoA class.

[0109] Reference Figure 5B In one or more embodiments, the processor 110 (eg, Figure 2 ) A first candidate AoA class and a second candidate AoA class whose matching probability with the input RSRP pattern data is greater than or equal to a threshold value may be selected from the AoA distribution output from the neural network 111, a beam 511 corresponding to the first candidate AoA class and a beam 513 corresponding to the second candidate AoA class may be selected from the high-resolution beam codebook, and beam training may be performed on the selected beams. Figure 6 One or more embodiments of beam training are described. Processor 110 ( Figure 2 ) may select a target beam for performing wireless communication with an external device (eg, a base station) based on a result of beam training.

[0110] In one or more embodiments, the processor 110 (e.g., Figure 2 ) A first candidate AoA class and a second candidate AoA class whose matching probability with the input RSRP pattern data is greater than or equal to a threshold value can be selected from the AoA distribution output from the neural network 111, and "a beam formed by combining the beam 511 corresponding to the first candidate AoA class and the beam 513 corresponding to the second candidate AoA class in the high-resolution beam codebook" can be set as a target beam.

[0111] exist Figure 5B , the beam gain of the beam 511 corresponding to the first candidate AoA class and the beam gain of the beam 513 corresponding to the second candidate AoA class having the same value are described as being allocated, but example embodiments are not limited thereto. The beam gain of the beam 511 corresponding to the first candidate AoA class and the beam gain of the beam 513 corresponding to the second candidate AoA class according to one or more embodiments may be differentially allocated (e.g., have different values) according to the orientation of each AoA (e.g., the orientation corresponding to each AoA class) depending on the matching probability of the AoA class corresponding to each beam in the AoA distribution.

[0112] Figure 5C is a diagram illustrating an example of an operation of beam management of the electronic device 100 according to one or more embodiments.

[0113] In detail, Figure 5C Show Figure 5A The AoA distribution of includes three candidate AoA classes. Figure 5C , there are a total of three AoA classes (including a first candidate AoA class, a second candidate AoA class, and a third candidate AoA class) having a matching probability greater than or equal to a threshold in the AoA distribution, and it is assumed that beam gains having the same value are assigned to beam 521 corresponding to the first candidate AoA class, beam 522 corresponding to the second candidate AoA class, and beam 523 corresponding to the third candidate AoA class.

[0114] Reference Figure 5C In one or more embodiments, the processor 110 (eg, Figure 2 ) A first candidate AoA class, a second candidate AoA class, and a third candidate AoA class whose matching probability with the input RSRP pattern data is greater than or equal to a threshold value may be selected from the AoA distribution output from the neural network 111, a beam 521 corresponding to the first candidate AoA class, a beam 522 corresponding to the second candidate AoA class, and a beam 523 corresponding to the third candidate AoA class may be selected from the high-resolution beam codebook, and beam training may be performed on the selected beams. Figure 6 One or more embodiments of beam training are described. Processor 110 ( Figure 2 ) may select a target beam for performing wireless communication with an external device (eg, a base station) based on a result of beam training.

[0115] In one or more embodiments, the processor 110 (e.g., Figure 2) A first candidate AoA class, a second candidate AoA class, and a third candidate AoA class whose matching probability with the input RSRP pattern data is greater than or equal to a threshold value may be selected from the AoA distribution output from the neural network 111, and “a beam formed by combining beam 521 corresponding to the first candidate AoA class, beam 522 corresponding to the second candidate AoA class, and beam 523 corresponding to the third candidate AoA class in the high-resolution beam codebook” may be set as a target beam.

[0116] exist Figure 5C , the beam gain of the beam 521 corresponding to the first candidate AoA class, the beam gain of the beam 522 corresponding to the second candidate AoA class, and the beam gain of the beam 523 corresponding to the third candidate AoA class having the same value are described as being allocated, but example embodiments are not limited thereto. The beam gain of the beam 521 corresponding to the first candidate AoA class, the beam gain of the beam 522 corresponding to the second candidate AoA class, and the beam gain of the beam 523 corresponding to the third candidate AoA class according to one or more embodiments may be differentially allocated (e.g., having different values) in the orientation of each AoA (e.g., the orientation corresponding to the AoA class) depending on the matching probability of the AoA class corresponding to each beam in the AoA distribution.

[0117] Figure 6 is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0118] Figure 6 Can be shown as reference FIG. 5A to FIG. 5C 1 is a diagram of one or more embodiments of beam training operations of the electronic device 100 described. Figure 6 The embodiment of the beam training of the electronic device 100 in the embodiment can also be applied to the above-described FIG. 5A to FIG. 5C The description of beam training in the diagrams other than Figure 6 When, in view of the above reference Figures 1 to 5C Description, and reference Figures 1 to 5C Descriptions that are duplicate of those given may be omitted.

[0119] exist Figure 6, it is assumed that a beam (e.g., UE candidate beam) (high resolution) corresponding to a candidate AoA class (e.g., candidate AoA #0, candidate AoA #1, . . . , and candidate AoA #K-1) belongs to a zeroth candidate beam (low resolution) among a plurality of candidate beams of the electronic device 100 in the spatial domain. For example, when the coverage area of ​​the zeroth candidate beam in the spatial domain includes a first angular area (e.g., 0 degrees to 20 degrees (low resolution)), the coverage area of ​​the beam corresponding to candidate AoA #0 may include a partial area of ​​the first angular area (e.g., 5 degrees (high resolution)).

[0120] Reference Figure 6 , a processor of the electronic device 100 (eg, Figure 2 110 ) may select a total of K candidate AoA classes having a matching probability greater than or equal to a threshold based on the AoA distribution output from the neural network 111 , and may perform beam training on the selected candidate AoA classes (the total of K AoA classes).

[0121] In one or more embodiments, the processor 110 (e.g., Figure 2 ) may measure at least one channel indicator of a channel based on a beam corresponding to a candidate AoA class (e.g., candidate AoA #0, candidate AoA #1, ..., and candidate AoA #K-1). Each candidate AoA class (e.g., candidate AoA #0, candidate AoA #1, ..., and candidate AoA #K-1) may perform beam alignment with a beam (e.g., a gNB beam) of an external device (e.g., a base station). Here, as a beam codebook belonging to a high-resolution beam codebook (e.g., Figure 4B 450), the beams corresponding to the candidate AoA classes (eg, candidate AoA #0, candidate AoA #1, . . . , and candidate AoA #K-1) are comparable to the multiple candidate beams (eg, Figure 4B 410) more accurately performs beam alignment with the external device. The processor 110 ( Figure 2 ) may perform beam training for each of the beams corresponding to the candidate AoA classes (eg, candidate AoA #0, candidate AoA #1, . . . , and candidate AoA #K-1). For example, the processor 110 ( Figure 2 ) may perform a first beam training 600-1 between a beam (e.g., a gNB beam) of an external device (e.g., a base station) and a beam corresponding to candidate AoA #0, perform a second beam training 600-2 between a beam (e.g., a gNB beam) of an external device (e.g., a base station) and a beam corresponding to candidate AoA #1, ..., and perform a K-th beam training 600-k between a beam (e.g., a gNB beam) of an external device (e.g., a base station) and a beam corresponding to candidate AoA #K-1. The processor 110 ( Figure 2 ) may measure at least one channel indicator indicating a channel state between a beam (e.g., a gNB beam) of an external device (e.g., a base station) and a beam corresponding to a candidate AoA class (e.g., candidate AoA #0, candidate AoA #1, ..., and candidate AoA #K-1). The at least one channel indicator may include, for example, at least one of RSRP, a signal-to-noise ratio (SNR), and a reference signal received quality (RSRQ) of signals respectively passing through the channel.

[0122] In one or more embodiments, the processor 110 (e.g., Figure 2 ) may select a target beam from beams corresponding to the candidate AoA classes (e.g., candidate AoA #0, candidate AoA #1, ..., and candidate AoA #K-1) based on a comparison result of the channel indicator. For example, the processor 110 (e.g., Figure 2 ) may select a target beam based on a beam corresponding to a channel indicator having a better quality than other channel indicators. The processor 110 ( Figure 2 ) can use the selected target beam to perform wireless communication with an external device (e.g., a base station).

[0123] Figure 7 is a diagram illustrating an example of operation of beam management of an electronic device according to one or more embodiments.

[0124] Figure 7 It may be shown that when the position of the electronic device 100 or the orientation of the electronic device 100 changes, the processor 110 (for example, Figure 2 ) is a diagram of the beam management operation to correct the target beam based on the primary AoA class. Figure 7 When, in view of the above reference Figures 1 to 6 Description, and reference Figures 1 to 6 Descriptions that are duplicate of those given may be omitted.

[0125] Assumptions Figure 7 The spatial coordinate axes (eg, the X axis (horizontal axis), the Y axis (vertical axis), and the Z axis (height axis)) in FIG. 1 are reference axes for indicating the position of the electronic device 100 before the change, and the origin 710 indicates the position of the electronic device 100 before the change. In addition, assuming that Figure 7 The spatial coordinate axes (eg, X1 axis, Y1 axis, and Z1 axis) in are reference axes for indicating the change amounts of the corresponding orientations after the position of the electronic device 100 is changed.

[0126] Reference Figure 7, the processor 110 of the electronic device 100 (eg, Figure 2 ) may detect a change in the orientation of the electronic device 100 (or a change in the position of the electronic device 100) based on sensing data sensed from at least one sensor (eg, an accelerometer, a gyroscope, and a magnetometer) in the electronic device 100. For example, in Figure 7 In (a), by at least one sensor of the electronic device 100, the amount of change in the orientation of the electronic device 100 on the X-axis and the Y-axis can be sensed as "α degrees" counterclockwise about the Z-axis (Z=Z1, which means no change with respect to the Z-axis). Figure 7 In (b), the change in orientation of the electronic device 100 on the Z axis and the X axis can be sensed as "β degrees" counterclockwise about the Y axis (Y=Y2, which means no change with respect to the Y axis) by the sensor of the electronic device 100. Figure 7 In (c), by the sensor of the electronic device 100, the change amount of the orientation of the electronic device 100 on the Y axis and the Z axis can be sensed as "γ degrees" counterclockwise about the X axis (X=X3, which means no change with respect to the X axis).

[0127] In one or more embodiments, the processor 110 (e.g., Figure 2 ) may calculate the position coordinates of the electronic device 100 before the change based on the sensor information (eg, ). For example, the processor may calculate the position coordinates of the electronic device 100 before the change based on the following equation 1: .

[0128] [Equation 1] .

[0129] here, x represents the position coordinates of the electronic device 100 before the change on the X axis, y represents the position coordinates of the electronic device 100 before the change with respect to the Y axis, and z ∫ represents the position coordinates of the electronic device 100 before the change with respect to the Z axis. θ represents the azimuth angle of the primary AoA class, and represents the elevation angle of the main AoA class. Equation 1 re-expresses the azimuth angle of the main AoA class for the X-axis, Y-axis, and Z-axis ( θ ) and pitch angle ( ).

[0130] In one or more embodiments, the processor 110 (e.g., Figure 2 ) can calculate the position coordinates of the electronic device 100 after being changed based on the sensing data (eg, α, β, γ) (eg, ). For example, the processor may calculate the coordinates of the position of the electronic device 100 after the change based on the following equation 2: .

[0131] [Equation 2] .

[0132] here, x' represents the position coordinates of the electronic device 100 after the change on the X axis, y' represents the position coordinates of the electronic device 100 after the change with respect to the Y axis, and z' ∫ represents the position coordinates of the electronic device 100 after the change with respect to the Z axis. α represents the change in the orientation of the electronic device 100 on the X-axis and the Y-axis (Z=Z1), β represents the amount of change in the orientation of the electronic device 100 in the Z axis and the X axis (Y=Y2), and γ Indicates the amount of change in the orientation of the electronic device 100 on the Y axis and the Z axis (X=X3).

[0133] In one or more embodiments, the processor 110 (e.g., Figure 2 ) can select and change the position coordinates of the electronic device 100 (for example, ) as the final target beam. For example, processor 110 ( Figure 2 ) may be based on the position coordinates of the electronic device 100 before the change (eg, ) and the position coordinates of the electronic device 100 after the change (for example, ) to generate correction information about the current target beam, and the current target beam may be corrected based on the correction information. Processor 110 ( Figure 2 ) can select the corrected target beam as the final target beam and perform wireless communication with an external device (e.g., a base station).

[0134] Figure 8 is a flowchart illustrating a method of operating the electronic device 100 according to one or more example embodiments.

[0135] Reference Figure 8 , the beam management method using the AoA distribution (or AoA class) estimated based on the neural network 111 of the electronic device 100 may include operations S100 to S120. Figure 8 When, in view of the above reference Figures 1 to 7 Description, and reference Figures 1 to 7 Descriptions that are duplicated with those given above may be omitted. Figure 8 The electronic device 100 and the neural network 111 described correspond to the reference Figures 1 to 7An electronic device 100 and a neural network 111 are described.

[0136] In operation S100, the electronic device 100 may generate RSRP pattern data based on RSRP measured in each of a plurality of candidate beams for a signal received from an external device. Here, the RSRP pattern data may refer to data including RSRP values ​​measured in each of a plurality of candidate beams of the electronic device 100 for a signal received from an external device (e.g., a base station).

[0137] In operation S110, the electronic device 100 may estimate the AoA distribution (or AoA class) for the RSRP pattern data by applying the RSRP pattern data to the neural network 111 based on the deep learning algorithm. The neural network 111 may be trained to estimate the AoA distribution (or AoA class) corresponding to the RSRP pattern training data of the plurality of candidate beams via the deep learning algorithm using the RSRP pattern training data as input data. The RSRP pattern training data may represent RSRP pattern data for the plurality of candidate beams, which are measured in advance or experimentally obtained in order to train the neural network 111. As a probability distribution of each AoA class estimated by the neural network 111, the AoA distribution may include a distribution of matching probabilities between the RSRP pattern training data and each AoA class.

[0138] In operation S120, the electronic device 100 may perform beam management based on the estimated AoA distribution. Fig. 9 and Fig.10 An operation method of performing beam management of the electronic device 100 is described.

[0139] Fig. 9 is a flowchart illustrating a method of operating the electronic device 100 according to one or more example embodiments.

[0140] Fig. 9 Involved Figure 8 The beam management method using the AoA distribution (or AoA class) estimated based on the neural network 111 of the electronic device 100 may include operations S210 to S260. Fig. 9 When, in view of the above reference Figures 1 to 8 Description, and reference Figures 1 to 8 Descriptions that are duplicated with those given above may be omitted. Fig. 9 The electronic device 100 and the neural network 111 described correspond to the reference Figures 1 to 8 An electronic device 100 and a neural network 111 are described.

[0141] In operation S210, the electronic device 100 may identify whether a main AoA class exists. Here, the main AoA class may refer to an AoA class having a significantly high matching probability (e.g., a matching probability with RSRP pattern data (input)) among the AoA classes in the AoA distribution estimated by the neural network 111. When the main AoA class exists, the electronic device 100 may perform operation S220, and when the main AoA class does not exist, the electronic device 100 may perform operation S230.

[0142] In operation S220, the electronic device 100 may select a beam corresponding to the main AoA class as the target beam. For example, when there is a main AoA class, the electronic device 100 may select a beam corresponding to the main AoA class as the target beam among beams in the high-resolution beam codebook.

[0143] In operation S230, the electronic device 100 may select at least two candidate AoA classes. For example, when there are at least two AoA classes having a matching probability greater than or equal to a threshold in the AoA distribution estimated by the neural network 111 (e.g., when there is no primary AoA class), the electronic device 100 may perform beam management based on the at least two candidate AoA classes (e.g., see FIG. 5A to FIG. 5C and Figure 6 ). The candidate AoA class may represent an AoA class having a matching probability greater than or equal to a predetermined threshold in the AoA distribution estimated by the neural network 111.

[0144] In operation S240, the electronic device 100 may perform beam training based on the beam corresponding to the candidate AoA class. For example, the electronic device 100 may select a beam corresponding to the candidate AoA class from among the beams in the high-resolution beam codebook, and may measure at least one channel indicator of the channel based on the beam corresponding to the candidate AoA class. The at least one channel indicator may include at least one of RSRP, SNR, and RSRQ of a signal passing through the channel.

[0145] In operation S250, the electronic device 100 may select a target beam based on the result of the beam training. For example, the electronic device 100 may select a target beam based on at least one channel indicator of channels respectively corresponding to beams corresponding to the candidate AoA class (eg, comparison between channel indicators).

[0146] Although not shown, the electronic device 100 may select a beam generated by combining beams corresponding to the candidate AoA classes as a target beam.

[0147] In operation S260 , the electronic device 100 may perform wireless communication between the external device and the electronic device 100 using the target beam.

[0148] Fig.10 is a flowchart illustrating a method of operating the electronic device 100 according to one or more example embodiments.

[0149] Fig.10 Involved Fig. 9 The beam management method using the AoA distribution (or AoA class) estimated based on the neural network 111 of the electronic device 100 may include operations S310 to S350. Fig.10 When, in view of the above reference Figures 1 to 9 Description, and reference Figures 1 to 9 Given those descriptions, duplicate descriptions may be omitted. Fig.10 The electronic device 100 and the neural network 111 described may correspond to the reference Figures 1 to 9 An electronic device 100 and a neural network 111 are described.

[0150] In operation S310, the electronic device 100 may identify whether the position of the electronic device 100 (or the orientation of the electronic device 100) has changed. For example, the electronic device 100 may detect a change in the position of the electronic device 100 or a change in the orientation of the electronic device 100 based on sensing data from at least one sensor in the electronic device 100 (e.g., an accelerometer, a gyroscope, a magnetometer).

[0151] In operation S320, the electronic device 100 may calculate the position coordinates of the electronic device 100 before the change based on the sensor information (eg, see Figure 7 ). The AoA class information corresponding to the existing target beam may represent the main AoA class information at the position coordinates of the electronic device 100 before the change. For example, the electronic device 100 may use the reference Figure 7 The described [Equation 1] calculates the position coordinates of the electronic device 100 before the change based on the sensor information (for example, referring to Figure 7 Description ).

[0152] In operation S330, the electronic device 100 may generate a signal based on the sensing data (eg, Figure 7 α, β and γ in ), the position coordinates of the electronic device 100 after the change are calculated according to the position coordinates of the electronic device 100 before the change (for example, see Figure 7 ). For example, the electronic device 100 may use a reference Figure 7 The described [Equation 2] is based on the position coordinates of the electronic device 100 before the change (for example, referring to Figure 7 Description ) to calculate the position coordinates of the electronic device 100 after the change (for example, referring to Figure 7 Description ).

[0153] In operation S340 , the electronic device 100 may generate correction information about an existing target beam (eg, a target beam at the position coordinates of the electronic device 100 before the change) based on the position coordinates of the electronic device 100 before the change and the position coordinates of the electronic device 100 after the change.

[0154] In operation S350, the electronic device 100 may select a beam generated by correcting an existing target beam based on the correction information as a final target beam. For example, the electronic device 100 may correct the existing target beam based on the correction information and select the corrected target beam as the final target beam. The electronic device 100 may perform wireless communication with an external device (e.g., a base station) based on the final target beam.

[0155] Fig.11 is a schematic block diagram illustrating a wireless communication system 1100 according to one or more example embodiments.

[0156] Fig.11 The electronic device 1101 may correspond to the reference Figures 1 to 10 The electronic device 100 described, Fig.11 The processor 1120 may correspond to the reference Figures 1 to 10 The processor 110 described, Fig.11 The memory 1130 may correspond to Figure 2 The memory 140, Fig.11 The sensor module 1176 may correspond to Figure 2 The sensor 130, and Fig.11 The communication module 1190 may correspond to Figure 2 The communication interface 120 is provided.

[0157] Reference Fig.11In the wireless communication system 1100, the electronic device 1101 may communicate with another electronic device 1102 via a first network 1198 (e.g., a short-range wireless communication network, etc.), or communicate with another electronic device 1104 and / or a server 1108 via a second network 1199 (e.g., a long-range wireless communication network, etc.). The electronic device 1101 may communicate with another electronic device 1104 via the server 1108. The electronic device 1101 may include a processor 1120, a memory 1130, an input device 1150, an audio output device 1155, a display device 1160, an audio module 1170, a sensor module 1176, an interface 1177, a connection terminal 1178, a haptic module 1179, a camera module 1180, a power management module 1188, a battery 1189, a communication module 1190, a user identification module 1196, and / or an antenna module 1197. In the electronic device 1101, some of these components (such as the display device 1160, for example) may be omitted, or other components may be added. Some of these components may be provided as a single integrated circuit. For example, the sensor module 1176 (fingerprint sensor, iris sensor, illumination sensor, etc.) may be provided by being embedded in the display device 1160 (e.g., display, etc.).

[0158] The processor 1120 may execute software (e.g., program 1140, etc.) to control one or more other components (e.g., hardware, software components, etc.) of the electronic device 1101 connected to the processor 1120, and may perform various data processing or calculations. As part of the data processing or calculation, the processor 1120 may load commands and / or data received from other components (e.g., sensor module 1176, communication module 1190, etc.) into the volatile memory 1132, process the commands and / or data stored in the volatile memory 1132, and store the resulting data in the non-volatile memory 1134. The processor 1120 may include a main processor 1121 (e.g., central processing unit, application processor, etc.) and an auxiliary processor 1123 (e.g., graphics processor, image signal processor, sensor hub processor, communication processor, etc.) that can operate independently or together with the main processor 1121. The auxiliary processor 1123 may use less power than the main processor 1121 and perform dedicated functions.

[0159] The auxiliary processor 1123 may control functions and / or states related to some components of the electronic device 1101 (e.g., display device 1160, sensor module 1176, communication module 1190, etc.) on behalf of the main processor 1121 when the main processor 1121 is in an inactive state (e.g., a sleep state), or control functions and / or states related to some components of the electronic device 1101 (e.g., display device 1160, sensor module 1176, communication module 1190, etc.) together with the main processor 1121 when the main processor 1121 is in an active state (e.g., an application execution state). The auxiliary processor 1123 (e.g., an image signal processor, a communication processor, etc.) may also be provided as part of other function-related components (e.g., a camera module 1180, a communication module 1190, etc.).

[0160] In one or more embodiments, for a signal received from an external device (e.g., a base station), the processor 1120 may estimate an AoA distribution (or AoA class) corresponding to RSRP pattern data by applying RSRP pattern data measured from a plurality of candidate beams of the electronic device 1101 to a neural network of the processor 1120.

[0161] In one or more embodiments, the processor 1120 may perform various beam managements based on the estimated AoA distribution (or AoA class). For example, the processor 1120 may select a beam corresponding to a main AoA class in the estimated AoA distribution as a target beam, and may perform wireless communication with an external device (e.g., a base station). For example, the processor 1120 may select beams corresponding to at least two candidate AoA classes in the estimated AoA distribution, perform beam training on the beams corresponding to the candidate AoA classes, select a target beam among the beams corresponding to the candidate AoA classes based on a result of the beam training, and perform wireless communication with an external device (e.g., a base station). For example, when the position / orientation of the electronic device 1101 changes after selecting a target beam, the processor 1120 can correct the existing target beam by using the sensing data of the sensor module 1176 (e.g., the amount of change in the orientation of the electronic device 1101, etc.) and the AoA class information corresponding to the existing target beam (e.g., the main AoA class information at the position of the electronic device 1101 before the change), and can perform wireless communication with an external device (e.g., a base station) by selecting the corrected beam as the final target beam.

[0162] The memory 1130 may store various data required by the components of the electronic device 1101 (e.g., the processor 1120, the sensor module 1176, etc.). The data may include, for example, input data and / or output data for software (e.g., such as the program 1140) and commands related thereto. The memory 1130 may include a volatile memory 1132 and / or a nonvolatile memory 1134. The nonvolatile memory 1132 may include an internal memory 1136 fixedly installed in the electronic device 1101 and an external memory 1138 detachable from the electronic device 1101.

[0163] The program 1140 may be stored as software in the memory 1130 , and include an operating system 1142 , middleware 1144 , and / or applications 1146 .

[0164] The input device 1150 may receive commands and / or data to be used in components (e.g., processor 1120, etc.) of the electronic device 1101 from outside the electronic device 1101 (e.g., a user, etc.). The input device 1150 may include a microphone, a mouse, a keyboard, and / or a digital pen (e.g., a stylus, etc.).

[0165] The audio output device 1155 may output an acoustic signal to the outside of the electronic device 1101. The audio output device 1155 may include a speaker and / or a receiver. The speaker may be used for general purposes (such as multimedia playback or recording playback), and the receiver may be used to receive incoming calls. The receiver may be integrated as part of the speaker, or may be provided as a separate independent device.

[0166] The display device 1160 may visually provide information to the outside of the electronic device 1101. The display device 1160 may include a display, a holographic device or a projector, and a control circuit for controlling these devices. The display device 1160 may include a touch circuit system configured to detect a touch and / or a sensor circuit system (e.g., a pressure sensor, etc.) configured to measure the strength of a force generated by a touch.

[0167] The audio module 1170 can convert sound into an electrical signal, or vice versa, convert an electrical signal into sound. The audio module 1170 can obtain sound via the input device 1150, or can output sound via the audio output device 1155 and / or via a speaker and / or earphone on another electronic device (e.g., electronic device 1102, etc.) directly or wirelessly connected to the electronic device 1101.

[0168] The sensor module 1176 may detect operating conditions (e.g., power, temperature, etc.) or external environmental conditions (e.g., the state of the user, etc.) of the electronic device 1101, and may generate an electrical signal and / or a data value corresponding to the sensed condition. The sensor module 1176 may include a gesture sensor, a gyroscope sensor, an air pressure sensor, a magnetic sensor, an acceleration sensor, a grip sensor, a proximity sensor, a color sensor, an IR sensor, a biometric sensor, a temperature sensor, a humidity sensor, and / or an illumination sensor. For example, the sensor module 1176 may sense a change in the position of the electronic device 1101 or a change in the orientation of the electronic device 1101.

[0169] The interface 1177 may support one or more designated protocols that may be used to directly or wirelessly connect the electronic device 1101 to another electronic device (e.g., the electronic device 1102, etc.). The interface 1177 may include a high-definition multimedia interface (HDMI), a universal serial bus (USB) interface, a secure digital (SD) card interface, and / or an audio interface.

[0170] The connection terminal 1178 may include a connector that physically connects the electronic device 1101 to another electronic device (eg, the electronic device 1102, etc.). The connection terminal 1178 may include an HDMI connector, a USB connector, an SD card connector, and / or an audio connector (eg, a headphone connector, etc.).

[0171] The haptic module 1179 may convert the electric signal into mechanical stimulation (eg, vibration, movement, etc.) or electric stimulation sensed by the user through tactile or kinesthetic sense. The haptic module 1179 may include a motor, a piezoelectric element, and / or an electric stimulation device.

[0172] The camera module 1180 can capture still images and moving images. The camera module 1180 may include an image acquisition device, and may also include an additional lens assembly, an image signal processor, and / or a flash. The lens assembly of the camera module 1180 may collect light emitted from an object as a target of image capture.

[0173] The power management module 1188 may manage power supplied to the electronic device 1101. The power management module 1188 may be provided as a part of a power management integrated circuit (PMIC).

[0174] The battery 1189 may supply power to the components of the electronic device 1101. The battery 1189 may include a non-rechargeable primary battery, a rechargeable secondary battery, and / or a fuel cell.

[0175] The communication module 1190 may support the establishment of a direct (e.g., wired) communication channel and / or a wireless communication channel between the electronic device 1101 and other electronic devices (e.g., electronic device 1102, electronic device 1104, server 1108, etc.), and communication via the established communication channel. The communication module 1190 may operate independently of the processor 1120 (e.g., an application processor, etc.), and include one or more communication processors that support direct communication and / or wireless communication. The communication module 1190 may include a wireless communication module 1192 (e.g., a cellular communication module, a short-range wireless communication module, a global navigation satellite system (GNSS) communication module, etc.) and / or a wired communication module 1194 (e.g., a local area network (LAN) communication module, a power line communication module, etc.). Among these communication modules, the corresponding communication module can communicate with other electronic devices via a first network 1198 (e.g., a short-range communication network (such as Bluetooth, WiFi Direct, and Infrared Data Association (IrDA))) or a second network 1199 (e.g., a telecommunication network (such as a cellular network, the Internet, and a computer network (e.g., LAN, WAN, etc.))). These other types of communication modules may be integrated into a single component (e.g., a single chip, etc.), or may be provided as multiple separate components (e.g., multiple chips). The wireless communication module 1192 may identify and authenticate the electronic device 1101 within a communication network (such as the first network 1198 and / or the second network 1199) using user information (e.g., an International Mobile Subscriber Identifier (IMSI), etc.) stored in the user identification module 1196. For example, the communication module 1190 may perform wireless communication with an external device (e.g., a base station) according to a target beam selected based on the AoA distribution estimated by the neural network via the processor 1120.

[0176] The antenna module 1197 may send signals and / or power to the outside (e.g., other electronic devices, etc.), or receive signals and / or power from the outside (e.g., other electronic devices, etc.). The antenna may include a radiator, which includes a conductive pattern formed on a substrate (e.g., a printed circuit board (PCB), etc.). The antenna module 1197 may include one or more antennas. If multiple antennas are provided, an antenna suitable for a communication method used in a communication network (such as the first network 1198 and / or the second network 1199) among the multiple antennas may be selected by the communication module 1190. Signals and / or power may be sent or received between the communication module 1190 and other electronic devices via the selected antenna. In addition to the antenna, other components (e.g., a radio frequency integrated circuit (RFIC), etc.) may be provided as part of the antenna module 1197.

[0177] Some components may be connected to each other via a communication method between peripheral devices (bus, general purpose input and output (GPIO), serial peripheral interface (SPI), mobile industry processor interface (MIPI), etc.) and may exchange signals (eg, commands, data, etc.).

[0178] Commands or data may be sent or received between the electronic device 1101 and the external electronic device 1104 via a server 1108 connected to the second network 1199. The other electronic devices 1102 and 1104 may be of the same or different type as the electronic device 1101. All or some of the operations performed on the electronic device 1101 may be performed for one or more of the other electronic devices 1102 and 1104 and the server 1108. For example, when the electronic device 1101 needs to perform certain functions or services, the electronic device 1101 may request one or more other electronic devices to perform part or all of the functions or services, rather than performing the functions or services autonomously. The one or more other electronic devices that receive the request may perform additional functions or services related to the request, and the execution results may be sent to the electronic device 1101. To this end, cloud computing, distributed computing and / or client-server computing technologies may be used.

[0179] While the inventive concepts have been particularly shown and described with reference to example embodiments thereof, it will be understood that various changes in form and details may be made therein without departing from the spirit and scope of the appended claims and their equivalents.

Claims

1. An electronic device, comprising: at least one memory configured to store computer-readable instructions; A communication interface including an antenna array configured to form a plurality of candidate beams; as well as at least one processor operably connected to the communication interface and the at least one memory, wherein the at least one processor is configured to execute the computer readable instructions to: generating a plurality of reference signal received power pattern data based on a reference signal received power measured in each of the plurality of candidate beams for a signal received from an external device; estimating an arrival angle distribution for each of the plurality of reference signal received power pattern data by applying each of the plurality of reference signal received power pattern data to a neural network trained based on a deep learning algorithm; and Based on the estimated angle-of-arrival distribution, beam management for wireless communication with an external device is performed.

2. The electronic device according to claim 1, wherein: The neural network is trained via a deep learning algorithm to estimate an arrival angle distribution corresponding to each of the plurality of reference signal received power pattern training data for the plurality of candidate beams by using the plurality of reference signal received power pattern training data for the plurality of candidate beams as input data.

3. The electronic device according to claim 1, wherein: The arrival angle distribution output from the neural network includes a distribution of matching probabilities between each of the plurality of reference signal reception power pattern data and each arrival angle class.

4. The electronic device according to claim 1, wherein: In performing the beam management process, the at least one processor is configured to execute the computer readable instructions to: Identify whether there is a main arrival angle class in the arrival angle distribution, and the main arrival angle class has the highest matching probability with the corresponding reference signal received power pattern data; Based on identifying the presence of the main angle of arrival class, selecting a beam corresponding to the main angle of arrival class as a target beam; and Wireless communication with an external device is performed using a target beam.

5. The electronic device according to claim 1, wherein: In performing the beam management process, the at least one processor is configured to execute the computer readable instructions to: selecting at least two candidate angle-of-arrival classes based on the angle-of-arrival distribution; performing beam training based on beams corresponding to the at least two candidate angle of arrival classes, and selecting a target beam based on a result of the beam training; and Wireless communication with an external device is performed using a target beam.

6. The electronic device according to claim 5, wherein: In the arrival angle distribution, each of the at least two candidate arrival angle classes has a matching probability with the corresponding reference signal received power pattern data that is greater than or equal to a predetermined threshold.

7. The electronic device according to claim 5, wherein: In the process of selecting a target beam based on the result of beam training, the at least one processor is configured to execute the computer readable instructions to: measuring at least one channel indicator of a channel based on beams corresponding to the at least two candidate angle of arrival classes; and selecting a target beam based on the at least one channel indicator, The at least one channel indicator includes at least one of a reference signal received power, a signal-to-noise ratio, and a reference signal received quality of a signal passing through the channel.

8. The electronic device according to claim 5, wherein: The at least one processor is further configured to execute the computer-readable instructions to select, as the target beam, a beam generated by combining beams corresponding to the at least two candidate angle-of-arrival classes.

9. The electronic device according to any one of claims 1 to 8, further comprising: at least one sensor, Wherein, the at least one processor is further configured to execute the computer readable instructions to: selecting a target beam among the plurality of candidate beams based on the estimated angle-of-arrival distribution; receiving, from the at least one sensor, sensed data related to a change in at least one of a position and an orientation of the electronic device; obtaining, based on the angle-of-arrival information corresponding to the target beam, first position coordinates of the electronic device before the change of the electronic device; obtaining, based on the sensing data and the first position coordinates, second position coordinates of the electronic device after the change of the electronic device; generating correction information based on the first position coordinates and the second position coordinates; and A beam generated by correcting the target beam based on the correction information is selected as a final target beam.

10. A method for operating an electronic device, the method comprising: generating a plurality of reference signal received power pattern data based on a reference signal received power measured in each of a plurality of candidate beams for a signal received from an external device; estimating an arrival angle distribution for each of the plurality of reference signal received power pattern data by applying each of the plurality of reference signal received power pattern data to a neural network trained based on a deep learning algorithm; as well as Based on the estimated angle-of-arrival distribution, beam management for wireless communication with an external device is performed.

11. The method according to claim 10, wherein: The neural network is trained via a deep learning algorithm to estimate an arrival angle distribution corresponding to each of the plurality of reference signal received power pattern training data for the plurality of candidate beams by using the plurality of reference signal received power pattern training data for the plurality of candidate beams as input data.

12. The method according to claim 10, wherein: The arrival angle distribution output from the neural network includes a distribution of matching probabilities between each of the plurality of reference signal reception power pattern data and each arrival angle class.

13. The method according to claim 10, wherein: The steps of performing the beam management include: Identify whether there is a main arrival angle class in the arrival angle distribution, and the main arrival angle class has the highest matching probability with the corresponding reference signal received power pattern data; Based on identifying the presence of the main angle of arrival class, selecting a beam corresponding to the main angle of arrival class as a target beam; and Wireless communication with an external device is performed using a target beam.

14. The method according to claim 10, wherein: The steps of performing the beam management include: selecting at least two candidate angle-of-arrival classes based on the angle-of-arrival distribution; performing beam training based on beams corresponding to the at least two candidate angle of arrival classes, and selecting a target beam based on a result of the beam training; and Wireless communication with an external device is performed using a target beam.

15. The method according to claim 14, wherein: In the arrival angle distribution, each of the at least two candidate arrival angle classes has a matching probability with the corresponding reference signal received power pattern data that is greater than or equal to a predetermined threshold.

16. The method according to claim 14, wherein: The steps to select the target beam include: at least one channel indicator of a channel measured based on beams corresponding to the at least two candidate angle of arrival classes; and selecting a target beam based on the at least one channel indicator, The at least one channel indicator includes at least one of a reference signal received power, a signal-to-noise ratio, and a reference signal received quality of a signal passing through the channel.

17. The method according to claim 14, wherein: The step of selecting the target beam includes selecting a beam generated by combining beams corresponding to the at least two candidate arrival angle classes as the target beam.

18. The method according to any one of claims 10 to 17, further comprising: selecting a target beam among the plurality of candidate beams based on the estimated angle-of-arrival distribution; receiving sensory data related to a change in at least one of a position and an orientation of the electronic device; obtaining, based on the angle-of-arrival information corresponding to the target beam, first position coordinates of the electronic device before the change of the electronic device; obtaining, based on the sensing data and the first position coordinates, second position coordinates of the electronic device after the change of the electronic device; generating correction information based on the first position coordinate and the second position coordinate; as well as A beam generated by correcting the target beam based on the correction information is selected as a final target beam.

19. A method of operating a wireless communication system, the method comprising: forming a plurality of candidate beams for performing wireless communication with an external device; generating an angle of arrival estimation model via deep learning by using a plurality of reference signal received power pattern training data for the plurality of candidate beams as input and using an angle of arrival distribution corresponding to each of the plurality of reference signal received power pattern training data as output; generating a plurality of reference signal received power pattern data based on a reference signal received power measured in each of the plurality of candidate beams for a signal received from an external device; estimating an angle of arrival distribution for each of the plurality of reference signal received power pattern data by applying each of the plurality of reference signal received power pattern data to an angle of arrival estimation model; as well as Based on the estimated angle-of-arrival distribution, beam management is performed.

20. The method according to claim 19, wherein: The arrival angle distribution output from the arrival angle estimation model includes a distribution of matching probabilities between each of the plurality of reference signal reception power pattern data and each arrival angle class.

Images