Method and apparatus for positioning object by using transformer in wireless communication system

Abstract

According to one embodiment, a method for estimating the position of a first device may comprise the operations of: receiving one or more signals from a second device; estimating at least one of a propagation path length, an angle of departure and an angle of arrival of each of the one or more signals by using the one or more signals as an input of a first transformer-based deep neural network; and acquiring a position estimation value of the second device by using a second transformer-based deep neural network on the basis of the at least one of the propagation path length, the angle of departure and the angle of arrival of each of the one or more signals, and an image of the surrounding environment of the first device.

Description

Method and apparatus for positioning an object using a transformer in a wireless communication system

[0001] The present disclosure relates to a method for estimating the length and angle of a signal propagation path and determining the location of a device by utilizing sequential pilot signals received from a device in a wireless communication system as input to a Transformer-based Deep Neural Network (DNN).

[0002] GPS (global positioning system) satellite signals can be used to obtain location information of a specific device. However, GPS techniques are difficult to use indoors where GPS signals do not reach, and position estimation errors of up to 10m or more can occur due to time synchronization errors.

[0003] In addition, a method utilizing signal exchange between a base station (or access point) and a terminal can be used to obtain location information of a specific device. Since the method using signal exchange between the base station and the terminal utilizes an already established communication system, it offers excellent versatility and can minimize time errors due to the short communication distance. Furthermore, the method using signal exchange between the base station and the terminal has the additional advantage of not requiring additional filters or power consumption. However, when using a location positioning technique based on signal exchange between the base station and the terminal, accurate location estimation may be difficult in environments where a line-of-sight (LoS) propagation path cannot be secured, as the time of arrival (ToA), reference signal received power (RSRP), and angle of arrival (AoA) may change.

[0004] Common location estimation methods utilize trilateration or triangulation, which generally assume that multiple base stations must receive signals via a straight path. Consequently, signals received via a non-line-of-sight (NLoS) path can cause significant errors in location estimation. Additionally, when a terminal moves at a relatively high speed, large errors occur in location estimation, requiring the use of additional filters or sensors to correct for motion.

[0005] The present disclosure proposes a method for determining the location of a moving terminal by utilizing sequential pilot signals received by a base station from a moving terminal as input to a deep neural network to estimate the propagation path length, departure angle, and arrival angle of each signal, and utilizing the estimated signal path length / angle together with an image of the surrounding environment of the base station.

[0006] According to one embodiment, the method for estimating the position of a first device may include the operation of receiving at least one signal from a second device. According to one embodiment, the method for estimating the position of a first device may include the operation of using the at least one signal as an input to a first transformer-based deep neural network to estimate at least one of the propagation path length, angle of departure, and angle of arrival of each of the at least one signal. According to one embodiment, the method for estimating the position of a first device may include the operation of obtaining a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length, angle of departure, angle of arrival of each of the at least one signal, and an image of the surrounding environment of the first device.

[0007] According to one embodiment, the first device may include a transceiver; a memory; and at least one processor. The at least one processor may, by executing instructions stored in the memory: receive at least one signal from the second device and use the at least one signal as an input to a first transformer-based deep neural network to estimate at least one of the propagation path length, angle of departure, and angle of arrival of each of the at least one signal, and obtain a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length, angle of departure, angle of arrival of each of the at least one signal and an image of the surrounding environment of the first device.

[0008] According to one embodiment, a storage medium storing at least one instruction readable by a computer may be implemented. The at least one instruction may cause an electronic device to perform a plurality of operations when executed by at least one processor. The plurality of operations may include an operation of receiving at least one signal from a second device. The plurality of operations may include an operation of estimating at least one of the propagation path length, angle of departure, and angle of arrival of each of the at least one signal by using the at least one signal as an input to a first transformer-based deep neural network. The plurality of operations may include an operation of obtaining a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length, angle of departure, angle of arrival of each of the at least one signal, and an image of the surrounding environment of the first device.

[0009] A method and apparatus according to one embodiment of the present disclosure can enable positioning through position prediction even when a target whose position is to be determined is moving, by utilizing a pilot signal received by a base station from a terminal whose position changes over time as an input to a neural network.

[0010] The method and apparatus according to one embodiment of the present disclosure do not require additional GPS signals, filters, or sensors, and can also take into account the operation time of the deep neural network, thereby minimizing errors caused by the delay time between the location of an object estimated from a signal and the actual location, and can also enable diversification of the model size.

[0011] In relation to the description of the drawings, the same or similar reference numerals may be used for identical or similar components.

[0012] FIG. 1 is a block diagram of an electronic device in a network environment according to one embodiment of the present disclosure.

[0013] FIG. 2 is a diagram showing an example of a DL MU-MIMO environment in a wireless communication system according to one embodiment of the present disclosure.

[0014] FIG. 3 is a diagram showing an example of a DL SU-MIMO environment in a wireless communication system according to one embodiment of the present disclosure.

[0015] FIG. 4 is a drawing illustrating the structure of a wireless communication system according to one embodiment of the present disclosure.

[0016] FIG. 5 is a drawing showing the configuration of an electronic device according to one embodiment of the present disclosure.

[0017] FIG. 6 is a diagram illustrating a method for an electronic device according to one embodiment of the present disclosure to estimate the location of a terminal.

[0018] FIG. 7 is a diagram illustrating the operation of a channel parameter estimation unit according to one embodiment of the present disclosure.

[0019] FIG. 8 is a diagram illustrating the operation of a UE position estimation unit according to one embodiment of the present disclosure.

[0020] FIG. 9 is a flowchart illustrating a method of operation of an electronic device according to one embodiment of the present disclosure.

[0021] FIG. 10 is a diagram illustrating a method for an electronic device according to one embodiment of the present disclosure to estimate the location of a terminal.

[0022] FIG. 11 is a diagram illustrating the operation of a transformer-based position estimation deep neural network according to one embodiment of the present disclosure.

[0023] FIG. 12 is a figure showing the average absolute error of the estimated length / angle according to one embodiment of the present disclosure as a signal-to-noise ratio.

[0024] FIG. 13 is a figure showing the average absolute error of the estimated position according to the signal-to-noise ratio according to one embodiment of the present disclosure.

[0025] Hereinafter, embodiments of the present disclosure are described in detail with reference to the drawings so that those skilled in the art can easily practice them. However, the present disclosure may be embodied in various different forms and is not limited to the embodiments described herein. In relation to the description of the drawings, the same or similar reference numerals may be used for identical or similar components. Furthermore, in the drawings and related descriptions, descriptions of well-known functions and configurations may be omitted for clarity and brevity.

[0026] FIG. 1 is a block diagram of an electronic device (101) in a network environment (100) according to one embodiment.

[0027] Referring to FIG. 1, in a network environment (100), an electronic device (101) may communicate with an electronic device (102) through a first network (198) (e.g., a short-range wireless communication network) or with at least one of an electronic device (104) or a server (108) through a second network (199) (e.g., a long-range wireless communication network). According to one embodiment, the electronic device (101) may communicate with the electronic device (104) through a server (108). According to one embodiment, the electronic device (101) may include a processor (120), memory (130), input module (150), sound output module (155), display module (160), audio module (170), sensor module (176), interface (177), connection terminal (178), haptic module (179), camera module (180), power management module (188), battery (189), communication module (190), subscriber identification module (196), or antenna module (197). In some embodiments, at least one of these components (e.g., connection terminal (178)) may be omitted from the electronic device (101), or one or more other components may be added. In some embodiments, some of these components (e.g., sensor module (176), camera module (180), or antenna module (197)) may be integrated into a single component (e.g., display module (160)).

[0028] The processor (120) can control at least one other component (e.g., hardware or software component) of the electronic device (101) connected to the processor (120) by executing software (e.g., program (140)), for example, and can perform various data processing or operations. According to one embodiment, as at least part of the data processing or operations, the processor (120) can store commands or data received from other components (e.g., sensor module (176) or communication module (190)) in volatile memory (132), process the commands or data stored in volatile memory (132), and store the resulting data in non-volatile memory (134). According to one embodiment, the processor (120) may include a main processor (121) (e.g., central processing unit or application processor) or an auxiliary processor (123) that can operate independently or together with it (e.g., graphics processing unit, neural processing unit (NPU), image signal processor, sensor hub processor, or communication processor). For example, if the electronic device (101) includes a main processor (121) and an auxiliary processor (123), the auxiliary processor (123) may be configured to use less power than the main processor (121) or to be specialized for a designated function. The auxiliary processor (123) may be implemented separately from the main processor (121) or as part thereof.

[0029] The auxiliary processor (123) may control at least some of the functions or states associated with at least one component of the electronic device (101) (e.g., display module (160), sensor module (176), or communication module (190)) on behalf of the main processor (121) while the main processor (121) is in an inactive (e.g., sleep) state, or together with the main processor (121) while the main processor (121) is in an active (e.g., application execution) state. According to one embodiment, the auxiliary processor (123) (e.g., image signal processor or communication processor) may be implemented as part of another functionally related component (e.g., camera module (180) or communication module (190)). According to one embodiment, the auxiliary processor (123) (e.g., neural network processing unit) may include a hardware structure specialized for processing an artificial intelligence model. The artificial intelligence model may be generated through machine learning. Such learning may be performed, for example, on the electronic device (101) itself where the artificial intelligence model is executed, or through a separate server (e.g., server (108)). The learning algorithm may include, for example, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but is not limited to the examples described above. The artificial intelligence model may include a plurality of artificial neural network layers.An artificial neural network may be a deep neural network (DNN), a convolutional neural network (CNN), a recurrent neural network (RNN), a restricted Boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), a deep Q-network, or a combination of two or more of the above, but is not limited to the examples described above. In addition to the hardware structure, the artificial intelligence model may include a software structure, either additionally or substantially.

[0030] The memory (130) can store various data used by at least one component of the electronic device (101) (e.g., processor (120) or sensor module (176)). The data may include, for example, input data or output data for software (e.g., program (140)) and related commands. The memory (130) may include volatile memory (132) or non-volatile memory (134).

[0031] The program (140) may be stored as software in memory (130) and may include, for example, an operating system (142), middleware (144), or an application (146).

[0032] The input module (150) can receive commands or data to be used for a component of the electronic device (101) (e.g., processor (120)) from outside the electronic device (101) (e.g., user). The input module (150) may include, for example, a microphone, a mouse, a keyboard, a key (e.g., a button), or a digital pen (e.g., a stylus pen).

[0033] The sound output module (155) can output a sound signal to the outside of the electronic device (101). The sound output module (155) may include, for example, a speaker or a receiver. The speaker may be used for general purposes, such as multimedia playback or recording playback. The receiver may be used to receive incoming calls. According to one embodiment, the receiver may be implemented separately from the speaker or as part thereof.

[0034] The display module (160) can visually provide information to an external (e.g., user) of the electronic device (101). The display module (160) may include, for example, a display, a holographic device, or a projector and a control circuit for controlling said device. According to one embodiment, the display module (160) may include a touch sensor configured to detect a touch, or a pressure sensor configured to measure the intensity of the force generated by said touch.

[0035] The audio module (170) can convert sound into an electrical signal or, conversely, convert an electrical signal into sound. According to one embodiment, the audio module (170) can acquire sound through the input module (150) or output sound through the sound output module (155) or an external electronic device (e.g., electronic device (102)) (e.g., speaker or headphones) connected directly or wirelessly to the electronic device (101).

[0036] The sensor module (176) can detect the operating state of the electronic device (101) (e.g., power or temperature) or the external environmental state (e.g., user state) and generate an electrical signal or data value corresponding to the detected state. According to one embodiment, the sensor module (176) may include, for example, a gesture sensor, a gyroscope sensor, a barometric pressure sensor, a magnetic sensor, an accelerometer sensor, a grip sensor, a proximity sensor, a color sensor, an IR (infrared) sensor, a biosensor, a temperature sensor, a humidity sensor, or an illuminance sensor.

[0037] The interface (177) may support one or more specified protocols that can be used for the electronic device (101) to be connected directly or wirelessly to an external electronic device (e.g., electronic device (102)). According to one embodiment, the interface (177) may include, for example, a high definition multimedia interface (HDMI), a universal serial bus (USB) interface, an SD card interface, or an audio interface.

[0038] The connection terminal (178) may include a connector through which the electronic device (101) can be physically connected to an external electronic device (e.g., electronic device (102)). According to one embodiment, the connection terminal (178) may include, for example, an HDMI connector, a USB connector, an SD card connector, or an audio connector (e.g., a headphone connector).

[0039] The haptic module (179) can convert an electrical signal into a mechanical stimulus (e.g., vibration or movement) or an electrical stimulus that the user can perceive through tactile or kinesthetic senses. According to one embodiment, the haptic module (179) may include, for example, a motor, a piezoelectric element, or an electric stimulation device.

[0040] The camera module (180) can capture still images and video. According to one embodiment, the camera module (180) may include one or more lenses, image sensors, image signal processors, or flashes.

[0041] The power management module (188) can manage the power supplied to the electronic device (101). According to one embodiment, the power management module (188) can be implemented, for example, as at least part of a power management integrated circuit (PMIC).

[0042] The battery (189) can supply power to at least one component of the electronic device (101). According to one embodiment, the battery (189) may include, for example, a non-rechargeable primary battery, a rechargeable secondary battery, or a fuel cell.

[0043] The communication module (190) can support the establishment of a direct (e.g., wired) communication channel or a wireless communication channel between an electronic device (101) and an external electronic device (e.g., electronic device (102), electronic device (104), or server (108)), and the performance of communication through the established communication channel. The communication module (190) may include one or more communication processors that operate independently of the processor (120) (e.g., application processor) and support direct (e.g., wired) communication or wireless communication. According to one embodiment, the communication module (190) may include a wireless communication module (192) (e.g., cellular communication module, short-range wireless communication module, or GNSS (global navigation satellite system) communication module) or a wired communication module (194) (e.g., LAN (local area network) communication module, or power line communication module). The corresponding communication module among these communication modules can communicate with an external electronic device (104) through a first network (198) (e.g., a short-range communication network such as Bluetooth, WiFi (wireless fidelity) direct, or IrDA (infrared data association)) or a second network (199) (e.g., a legacy cellular network, a 5G network, a next-generation communication network, the Internet, or a computer network (e.g., a LAN or WAN)). These various types of communication modules may be integrated into a single component (e.g., a single chip) or implemented as multiple separate components (e.g., multiple chips). The wireless communication module (192) can identify or authenticate the electronic device (101) within a communication network such as the first network (198) or the second network (199) using subscriber information (e.g., International Mobile Subscriber Identifier (IMSI)) stored in the subscriber identification module (196).

[0044] The wireless communication module (192) can support 5G networks and next-generation communication technologies following 4G networks, for example, new radio access technology. NR access technology can support high-speed transmission of high-capacity data (enhanced mobile broadband (eMBB)), minimization of terminal power and connection of multiple terminals (massive machine type communications (mMTC)), or high reliability and low latency (ultra-reliable and low-latency communications (URLLC)). The wireless communication module (192) can support a high-frequency band (e.g., mmWave band) to achieve a high data transmission rate, for example. The wireless communication module (192) can support various technologies for securing performance in the high-frequency band, such as beamforming, massive MIMO (multiple-input and multiple-output), full-dimensional MIMO (FD-MIMO), array antenna, analog beam-forming, or large-scale antenna. The wireless communication module (192) can support various requirements specified in the electronic device (101), external electronic device (e.g., electronic device (104)), or network system (e.g., second network (199)). According to one embodiment, the wireless communication module (192) can support a Peak data rate (e.g., 20 Gbps or more) for realizing eMBB, loss coverage (e.g., 164 dB or less) for realizing mMTC, or U-plane latency (e.g., downlink (DL) and uplink (UL) each 0.5 ms or less, or round trip 1 ms or less) for realizing URLLC.

[0045] An antenna module (197) can transmit a signal or power to or from an external source (e.g., an external electronic device). According to one embodiment, the antenna module (197) may include an antenna comprising a radiator made of a conductor or a conductive pattern formed on a substrate (e.g., a PCB). According to one embodiment, the antenna module (197) may include a plurality of antennas (e.g., an array antenna). In this case, at least one antenna suitable for a communication method used in a communication network, such as a first network (198) or a second network (199), may be selected from the plurality of antennas, for example, by a communication module (190). A signal or power may be transmitted or received between the communication module (190) and an external electronic device through the selected at least one antenna. According to some embodiments, in addition to the radiator, other components (e.g., a radio frequency integrated circuit (RFIC)) may be additionally formed as part of the antenna module (197).

[0046] According to various embodiments, the antenna module (197) may form a mmWave antenna module. According to one embodiment, the mmWave antenna module may include a printed circuit board, an RFIC disposed on or adjacent to a first surface (e.g., bottom surface) of the printed circuit board and capable of supporting a specified high frequency band (e.g., mmWave band), and a plurality of antennas (e.g., array antennas) disposed on or adjacent to a second surface (e.g., top surface or side surface) of the printed circuit board and capable of transmitting or receiving a signal of the specified high frequency band.

[0047] At least some of the above components can be connected to each other via a communication method between peripheral devices (e.g., bus, GPIO (general purpose input and output), SPI (serial peripheral interface), or MIPI (mobile industry processor interface)) and exchange signals (e.g., commands or data) with each other.

[0048] According to one embodiment, commands or data may be transmitted or received between the electronic device (101) and an external electronic device (104) through a server (108) connected to a second network (199). Each of the external electronic devices (102, or 104) may be the same or different type of device as the electronic device (101). According to one embodiment, all or part of the operations performed on the electronic device (101) may be performed on one or more of the external electronic devices (102, 104, or 108). For example, if the electronic device (101) needs to perform a function or service automatically or in response to a request from a user or another device, the electronic device (101) may request one or more external electronic devices to perform at least part of the function or service instead of performing the function or service itself or additionally. One or more external electronic devices that receive the above request may execute at least part of the requested function or service, or additional function or service related to the request, and transmit the result of the execution to the electronic device (101). The electronic device (101) may provide the result as is or additionally processed as at least part of the response to the request. For this purpose, for example, cloud computing, distributed computing, mobile edge computing (MEC), or client-server computing technology may be used. The electronic device (101) may provide ultra-low latency services using, for example, distributed computing or mobile edge computing. In another embodiment, the external electronic device (104) may include an Internet of Things (IoT) device. The server (108) may be an intelligent server using machine learning and / or neural networks. According to one embodiment, the external electronic device (104) or the server (108) may be included within a second network (199).The electronic device (101) can be applied to intelligent services (e.g., smart home, smart city, smart car, or healthcare) based on 5G communication technology and IoT-related technology.

[0049] A method and apparatus according to one embodiment of the present disclosure utilize a received signal in a multiple-input-multiple-output (MIMO) communication system. The present disclosure relates to a method and apparatus in which a first device (e.g., a base station) estimates distance and / or angle information of a second device (e.g., a terminal (UE: user equipment)) using a deep learning technique based on a signal received through a wireless communication system, and estimates the location of the second device (e.g., a terminal) using a deep learning network.

[0050] FIG. 2 is a diagram showing an example of a MU (multi-user)-MIMO environment in a wireless communication system according to one embodiment of the present disclosure.

[0051] Referring to FIG. 2, a base station (210), and M r Each of the UEs (220, 230, 240) can be implemented as the electronic device (101) of FIG. 1. In the downlink of a wireless communication system, M t A base station (210) using multiple antennas and M using at least one receiving antenna r A DL MU (multi-user)-MIMO environment including 220, 230, 240 UEs can be implemented. In addition, M in the uplink of the wireless communication system r A base station (210) using multiple antennas and M using at least one transmitting antenna r A UL MU-MIMO environment including 220, 230, 240 UEs can be implemented.

[0052] FIG. 3 is a diagram illustrating an example of a SU-MIMO environment in a wireless communication system according to an embodiment of the present disclosure.

[0053] Referring to FIG. 3, the base station (310) and the UE (350) can each be implemented as the electronic device (101) of FIG. 1. In the downlink of a wireless communication system, M t Base station (310) using multiple antennas and M r A DL SU (single user)-MIMO environment including a UE (350) using multiple receiving antennas can be implemented. Additionally, in the uplink of the wireless communication system, M t Base station (310) using multiple antennas and M r A UL SU-MIMO environment including a UE (350) using multiple transmitting antennas can be implemented.

[0054] The embodiments of the present disclosure can be applied to a MIMO environment in which the number of transmitting / receiving antennas of a base station and the number of transmitting / receiving antennas of a terminal are implemented in various ways.

[0055] FIG. 4 is a drawing illustrating the structure of a wireless communication system according to one embodiment of the present disclosure.

[0056] FIG. 4 illustrates an example in which a terminal moves and changes the connected base station in a wireless communication system to which embodiments of the present disclosure are applied. Base stations (420, 430) may be connected to some of the surrounding base stations, and base stations (420, 430) may be connected to a mobile communication core network (CN: Core Network) (440), such as an EPC (Evolved Packet Core), a 5GC (5G Core Network), or a 6G network.

[0057] The radio access technology of the base stations (420, 430) may be LTE, NR, Wi-Fi, 6G, etc., and is not limited to one example. For example, the base stations (420, 430) may be mobile communication base stations unrelated to the radio access technology.

[0058] A terminal (410) can be connected to a base station to receive mobile communication services, and as the terminal (410) moves, the connected base station may change, and the terminal (410) can receive mobile communication services without interruption through a handover (HO; Handover, or handoff) procedure. In one example of FIG. 4, the terminal (410) is connected to a base station (420), and then disconnects from the base station (420) through a handover and connects to a new base station (430). A location estimation control device (440) can control multiple base stations to estimate the location of the terminal (410).

[0059] According to one embodiment, a base station (420) or a base station (430) can estimate at least one of the propagation path length, departure angle, and arrival angle of each signal by utilizing sequential pilot signals received from a moving terminal (410) as input to a deep neural network. The base station (420) or a base station (430) can determine the location of the moving terminal (410) by utilizing the estimated signal path length and / or angle together with an image of the surrounding environment of the base station (e.g., a plan view including the location of surrounding obstacles).

[0060] According to one embodiment, the location estimation control device (440) can estimate at least one of the propagation path length, departure angle, and arrival angle of each signal by utilizing sequential pilot signals received from a moving terminal (410) by a base station (420) or a base station (430) as input to a deep neural network. The location estimation control device (440) can determine the location of the moving terminal (410) by utilizing the estimated signal path length and / or angle together with an image of the surrounding environment of the base station (e.g., a plan view including the location of surrounding obstacles).

[0061] FIG. 5 is a drawing showing the configuration of an electronic device according to one embodiment of the present disclosure.

[0062] Referring to FIGS. 4 and 5, the electronic device (500) may be implemented as any one of a base station (420), a base station (430), and a location estimation control device (440). The electronic device (500) may be identical or substantially identical to the electronic device (101) of FIG. 1. The electronic device (500) may include a processor (120), a memory (130), and a communication module (190). The descriptions of the processor (120), memory (130), and communication module (190) described above in FIG. 1 may be applied in the same way.

[0063] Referring to FIG. 5, the processor (120) may include a channel parameter estimation unit (510) including a first deep neural network (515) and a UE location estimation unit (520) including a second deep neural network (525). The channel parameter estimation unit (510) may estimate the length and / or angle of a signal path, and the UE location estimation unit (520) may estimate the location of a terminal. The first deep neural network (515) may include at least one of a convolutional neural network (CNN), a long short-term memory (LSTM), and a transformer. The second deep neural network (525) may include at least one of a CNN, an LSTM, and a transformer.

[0064] According to one embodiment, the channel parameter estimation unit (510) may receive sequentially received uplink signals as input to a first deep neural network (515) (e.g., a transformer-based deep neural network). The channel parameter estimation unit (510) may use the first deep neural network (515) (e.g., a transformer-based deep neural network) to estimate the propagation path length, angle of departure, and angle of arrival of the signal used as input to the UE location estimation unit (520).

[0065] According to one embodiment, the channel parameter estimation unit (510) can estimate at least one of the azimuth angle and elevation angle of the signal path. The received signal is a superposition of signals propagated through multiple paths, and has at least one of the signal propagation time, departure angle, and arrival angle of each path as a parameter. According to one embodiment, the channel parameter estimation unit (510) can estimate at least one of the propagation time, departure angle, and arrival angle of the superimposed signal. According to one embodiment, the channel parameter estimation unit (510) may include a function that derives an output from an input through learning using big data. Since the angle, which is the output of the channel parameter estimation unit (510), belongs to a continuous domain, a regression model can be constructed using the mean squared error as a loss function for learning. According to one embodiment, the first deep neural network (515) can be replaced with a signal path parameter estimation technique of an existing communication system as needed (e.g., compressed sensing; CS).

[0066] According to one embodiment, the UE location estimation unit (520) can estimate the location of a terminal by using a second deep neural network (525) (e.g., a neural network with a transformer structure) and using the lengths and / or angles of signal paths estimated by the channel parameter estimation unit (510) and a plan view of the surrounding environment of the base station as inputs to the network.

[0067] According to one embodiment, when both the length and angle information of the signal propagation paths are given, if only the location where the signal is reflected is known, the possible paths can be limited to one by combining the information, and thus the location of the terminal can also be limited. If a plan of the area around the base station is available, the location of the reflection point can be known by combining it with the length and / or angle information of the signal propagation path, and this can be used as an input to the UE location estimation unit (520). Since the terminal location, which is the output of the UE location estimation unit (520), belongs to the continuous domain, a regression model can be constructed that utilizes mean squared error as the loss function.

[0068] FIG. 6 is a diagram illustrating a method for an electronic device according to one embodiment of the present disclosure to estimate the location of a terminal.

[0069] Referring to FIG. 6, the electronic device (600) comprises a channel parameter estimation unit (610) including at least one deep neural network, a UE position estimation unit (620) including at least one deep neural network, and N R It may include antennas (630). The electronic device (600) may be implemented as any one of the base station (420), base station (430), and position estimation control device (440) of FIG. 4. The electronic device (600) may be identical or substantially identical to the electronic device (101) of FIG. 1.

[0070] The electronic device (600) is N R At least one RF signal (through the antennas (630)) Can receive ). RF signal( ) can be determined based on mathematical formula 1.

[0071] [Mathematical Formula 1]

[0072]

[0073] Here represents the time step of the received signal entering as input, S (=1:S) represents the subcarrier of the received signal entering as input, and n t,s can mean a noise vector. Channel H t,s It can be constructed as in mathematical formula 2.

[0074] [Mathematical Formula 2]

[0075]

[0076] Here, N path represents the number of channel paths, and represents the channel gain of the i-th path, and and represent the i-th path azimuth and elevation angle on the receiver side, respectively, and can represent the i-th path azimuth and elevation angle on the transmitter side, respectively. Also, a R and a T is a steering vector defined as in mathematical equation 3.

[0077] [Mathematical Formula 3]

[0078]

[0079] Here and represent the number of receiver antennas (the product of horizontal and vertical) and the number of transmitter antennas, respectively. When using a deep neural network, the signal received at the receiver Since is a complex number, z that actually enters as input to the neural network t,s It uses the real and imaginary parts combined in order. Therefore, z t,s Is It consists of repetitions of, where y t,s Is It is an element of

[0080] The channel parameter estimation unit (610) may use a transformer structure. According to one embodiment, in the deep neural network of the channel parameter estimation unit (610), zt,s Divided into matrices by t=1:T and s=1:S It can be used as input. The input passes through a single linear layer and is embedded. This becomes the transformation process of a single vector, as shown in Equation 4.

[0081] [Mathematical Formula 4]

[0082]

[0083] Here are the weights and deviations of the linear layer for embedding in the channel parameter estimation unit (610), respectively. Input Z' is a positional embedding This is added Position embeddings can be used to distinguish signals with different time steps and subcarriers, allowing deep neural networks to recognize the order of sequential inputs and consider terminal movements based on sequence information. Z0 can be used to pass through multiple transformer blocks and perform various operations sequentially.

[0084] The first operation performed in each transformer block of the channel parameter estimation unit (610) is multi-head attention. The channel parameter estimation unit (610) can perform a residual connection by adding the result of the multi-head attention operation and the input prior to the multi-head attention.

[0085] The channel parameter estimation unit (610) can perform layer normalization after residual linking. After layer normalization, the channel parameter estimation unit (610) can pass through one fully connected layer, perform residual linking by adding the input before passing through the fully connected layer, and perform the layer normalization process again.

[0086] In the channel parameter estimation unit (610), a series of processes from multi-head attention to layer normalization after the linear layer (multi-head attention, residual linking, layer normalization, linear layer, residual linking, layer normalization) can be formed as a single transformer block, and a total of N1 transformer blocks can be configured. The final output of the channel parameter estimation unit (610) is the output that has passed through the last transformer block. The signal path length / angle at time step t=T can be estimated by passing the first row vector through a linear layer.

[0087] According to one embodiment, the channel parameter estimation unit (610) can estimate the signal path length / angle as in Equation 5.

[0088] [Mathematical Formula 5]

[0089]

[0090] Here W f,1 Wa b f,1 are the weights and deviations of the linear layer for estimating the length / angle of the signal propagation path in the channel parameter estimation unit (610), respectively. The deep neural network of the channel parameter estimation unit (610) obtains N from the input Z. path Estimated distance / angle of the path that is received first and has the strongest (dominant) strength among the paths and the components of the actual signal propagation path Learning can be done in a direction that reduces the difference. To this end, the loss function is the mean squared error (MSE), which is the difference between the estimate and the propagation path length / angle, and the loss function J1 can be defined as Equation 6.

[0091] [Mathematical Formula 6]

[0092]

[0093] Here, N bis the size of the mini-batch used in one training iteration. When training the deep neural network of the channel parameter estimation unit (610), the weights and deviations of the deep neural network can be optimized in a direction that minimizes the loss function. At this time, various backpropagation algorithms (e.g., Adam optimization, adagrad optimization, RMSprop optimization) can be used as algorithms to optimize the weights and deviations of the deep neural network.

[0094] The UE location estimation unit (620) can estimate the location of a terminal by utilizing the signal / propagation path estimate derived from the channel parameter estimation unit (610) and a plan view around the electronic device (600) (e.g., base station) as input. The UE location estimation unit (620) can utilize sequential estimates up to time step t (=1:T) as input. Length / angle input estimated by the channel parameter estimation unit (610). is a matrix separated by time step This can be. This is part of the input to the deep neural network of the UE position estimation unit (620) through a single linear layer. It becomes. Embedded input vector of length / angle information It can be expressed as in mathematical formula 7.

[0095] [Mathematical Formula 7]

[0096]

[0097] A plan view of the surroundings of an electronic device (600) (e.g., a base station), which is an input other than distance / angle, can be used as an input to the deep neural network of the UE location estimation unit (620) after undergoing a patch embedding process through a convolutional neural network (CNN). The plan view of the surroundings of the electronic device (600) (e.g., a base station) used by the UE location estimation unit (620) may be a plan view in RGB format or a plan view representing the height of buildings or obstacles in image form. A plan view representing the height of buildings or obstacles in image form may be a grayscale image. Patch embedding may refer to a process of dividing the plan view into multiple areas (patches) and embedding the image corresponding to each area into a vector through a CNN.

[0098] Plan view image X of the area around the electronic device (600) (e.g., base station) im is N pe It is divided into patches, and each patch is E patch After multiplying by and passing through the flatten function, it can be converted into an embedding vector. Subsequently, the embedded length / angle vector and the embedded patch image vector can be combined. The resulting combined matrix is ​​a position embedding matrix E of the same dimension. pos,2 It is added, and X0, the result of this series of processes, is equal to mathematical formula 8.

[0099] [Mathematical Formula 8]

[0100]

[0101] Position embedding, as in the channel parameter estimation unit (610), allows the deep neural network to recognize the order of sequential inputs and facilitate the identification of the terminal's mobility. X0 passes through multiple transformer blocks, similar to the operation of the channel parameter estimation unit (610). The UE position estimation unit (620) passes through a total of N2 transformer blocks and then produces a final output. The first row vector in (e.g., of Fig. 8) Time step considering model operation time by passing ) through a linear layer The location of the terminal at can be estimated.

[0102] According to one embodiment, the UE location estimation unit (620) can estimate the location of the terminal as in Equation 9.

[0103] [Mathematical Formula 9]

[0104]

[0105] Here, W f,2 wa b f,2 are the weights and deviations of the linear layer for estimating the terminal's location in the UE location estimation unit (620), respectively. The deep neural network of the UE location estimation unit (620) is the terminal's location predicted during the training phase. and the correct answer location The mean squared error is used as the loss function. The network parameters can be optimized using the backpropagation algorithm through the value of the loss function. The loss function J2 of the deep neural network of the UE location estimation unit (620) is defined as Equation 10.

[0106] [Mathematical Formula 10]

[0107]

[0108] FIG. 7 is a diagram illustrating the operation of a channel parameter estimation unit according to one embodiment of the present disclosure.

[0109] The channel parameter estimation unit (610) of FIG. 7 may be identical or substantially identical to the channel parameter estimation unit (610) of FIG. 6.

[0110] The channel parameter estimation unit (610) corresponds to Z= of the processed received signals. Perform input embedding (710) with the input and It can generate. The channel parameter estimation unit (610) can generate through Z' and positional encoding (720). It can generate.

[0111] The channel parameter estimation unit (610) includes a transformer-based deep neural network, and the transformer-based deep neural network may include a multi-head attention module (730), a layer normalization module (740), and a feed-forward network module (750). According to one embodiment, the channel parameter estimation unit (610) may repeatedly perform the entire operation performed on the multi-head attention module (730), the layer normalization module (740), and the feed-forward network module (750).

[0112] The multi-head attention operation performed in the multi-head attention module (730) is as follows. When matrix A is the input to the multi-head attention module (730), a total of N h The k-th head among the heads is formed by multiplying the input matrix A by the query weight matrix, key weight matrix, and value weight matrix, respectively, to obtain the query Q k , Key K k , Value V k can calculate.

[0113] Query Q in the multi-head attention module (730) k , Key K k , using the dimension (d) of the embedded vector N h Correlation Matrix of the k-th head among the heads (C k ) can be calculated. According to one embodiment, the correlation matrix (C) of the k-th head k ) can be calculated as in mathematical formula 11.

[0114] [Mathematical Formula 11]

[0115]

[0116] In the multi-head attention module (730), the correlation matrix (C) k ), Value V k Using N h The attention score (S) of the k-th head among the heads k ) can be calculated. According to one embodiment, the attention score (S) of the k-th head k ) can be calculated as in mathematical formula 12.

[0117] [Mathematical Formula 12]

[0118]

[0119] In the multi-head attention module (730), N h The attention scores calculated from each head among the heads are combined to form the final result of multi-head attention ( It can be calculated as ).

[0120] The result of the operation of the multi-head attention module (730) can be added to the input prior to the multi-head attention module (730) to perform residual connection. After residual connection, layer normalization can be performed in the layer normalization module (740). After layer normalization, it can pass through a fully connected layer, perform residual connection by adding to the input prior to passing through the fully connected layer, and perform the layer normalization process again.

[0121] In the feed-forward network module (750), the length / angle of the signal path at time step t=T (based on the layer normalization result by the layer normalization module (740) ) can be estimated. According to one embodiment, a deep neural network estimates N from input Z. pathEstimated distance / angle of the path that is received first and has the strongest (dominant) strength among the paths and the components of the actual signal propagation path Learning can be done in a way that reduces the difference.

[0122] FIG. 8 is a diagram illustrating the operation of a UE position estimation unit according to one embodiment of the present disclosure.

[0123] The UE position estimation unit (620) of FIG. 8 may be identical or substantially identical to the UE position estimation unit (620) of FIG. 6.

[0124] The UE position estimation unit (620) provides plan view image data (X) of the area surrounding the electronic device (600). im ) can be used as an input value for a deep neural network. For example, the UE position estimation unit (620) can use a planar image (X) of the area surrounding the electronic device (600). im ) can be used as input to a deep neural network through a patch embedding process (810) using a convolutional neural network, CNN.

[0125] Plan view (X) of the area surrounding the electronic device (600) used in the UE position estimation unit (620) im ) may be a floor plan in RGB format, as in a photograph, or a floor plan showing the height of an object (e.g., a building or obstacle) in image form. A floor plan showing the height of an object in image form may be a grayscale image.

[0126] The UE location estimation unit (620) is a plan view (X) of the area surrounding the electronic device (600). im Patch embedding (810) can be performed to generate patch image vectors by dividing the area into multiple zones and embedding the image corresponding to each zone into a vector through a CNN. At this time, the multiple patches may include information about the height of the buildings. Plan view image (X) of the area surrounding the electronic device (600).im ) can be divided into Npe patches, and the patches can be converted into embedding vectors by multiplying by the patch.

[0127] The UE location estimation unit (620) estimates the distance and angle values ​​for the terminal (that provided the received signal) estimated by the channel parameter estimation unit (610). Combined and position embedding (820) can be performed using the results of ) and patch embedding (810). The UE position estimation unit (620) can combine the embedded distance and angle vectors and the embedded patch image vectors. According to one embodiment, when combining the embedded distance and angle vectors and the embedded patch image vectors, the UE position estimation unit (620) can convert each embedded vector into a column vector and form multiple converted column vectors into a matrix. Subsequently, the combined matrix can be combined with a position embedding matrix pos having the same dimension to derive X0 as a result. According to one embodiment, X0 can be defined by Equation 13 as follows.

[0128] [Mathematical Formula 13]

[0129]

[0130] The UE location estimation unit (620) includes a transformer block (TF block) (830) that implements a transformer-based deep neural network, and the transformer block (830) may include a multi-head attention module (831), a first layer normalization module (833), a feed-forward network module (835), and a second layer normalization module (837).

[0131] The transformer block (830) recognizes the order of sequential inputs and facilitates the identification of the terminal's mobility therefrom. X0 can pass through multiple transformer blocks. After passing through a total of N2 transformer blocks, the final output The first row vector in ( A time step that takes into account the model operation time by passing ) through the linear layer (840). The location of the terminal at can be estimated ( ).

[0132] FIG. 9 is a flowchart illustrating a method of operation of an electronic device according to one embodiment of the present disclosure.

[0133] Referring to FIG. 9, in operation 910, the first device may receive at least one signal from the second device. The first device may be a base station or a location estimation control device, and the second device may be user equipment (UE). According to one embodiment, the at least one signal may include at least one of information regarding the number of channel paths, channel gain, the azimuth angle and elevation angle of the receiver in each channel, the azimuth angle and elevation angle of the transmitter in each channel, the number of antennas of the receiver, and the number of antennas of the transmitter.

[0134] In operation 920, the first device can use the at least one signal as an input to a first transformer-based deep neural network to estimate at least one of the propagation path length, angle of departure, and angle of arrival of each of the at least one signal.

[0135] In operation 930, the first device can obtain a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length of each of the at least one signal, the departure angle, the arrival angle, and an image of the surrounding environment of the first device.

[0136] According to one embodiment, each of the first transformer-based deep neural network and the second transformer-based deep neural network may include at least one of a multi-head attention module, a layer normalization module, and a feed-forward network module.

[0137] According to one embodiment, the image of the surrounding environment may be a plan view in the form of RGB (red-green-blue) or a plan view showing the height of an object in a grayscale form.

[0138] According to one embodiment, the first transformer-based deep neural network may be configured as a regression model that uses mean squared error as a loss function for learning.

[0139] According to one embodiment, the second transformer-based deep neural network may be configured as a regression model that uses mean squared error as a loss function for learning.

[0140] According to one embodiment, the second transformer-based deep neural network may be configured as a model using a loss function to which at least one output value of the first transformer-based deep neural network and at least one weight value capable of setting a range for the at least one output value are applied.

[0141] According to one embodiment, the first device can determine the location where the at least one signal is reflected by using the propagation path length of each of the at least one signal, the departure angle, the arrival angle, and an image of the surrounding environment of the first device. According to one embodiment, the first device can estimate the location of the second device based on the location where the at least one signal is reflected.

[0142] According to one embodiment, a storage medium storing at least one instruction readable by a computer may be implemented. The at least one instruction may cause an electronic device to perform a plurality of operations when executed by at least one processor. The plurality of operations may include: an operation of receiving at least one signal from a second device; an operation of using the at least one signal as an input to a first transformer-based deep neural network to estimate at least one of the propagation path length, angle of departure, and angle of arrival of each of the at least one signal; and an operation of obtaining a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length, angle of departure, angle of arrival of each of the at least one signal, and an image of the surrounding environment of the first device.

[0143] FIG. 10 is a diagram illustrating a method for an electronic device according to one embodiment of the present disclosure to estimate the location of a terminal.

[0144] Referring to FIG. 10, the electronic device (1000) comprises a UE position estimation unit (1010) including at least one deep neural network, and N R It may include antennas. The electronic device (1000) may be implemented as any one of the base station (420), base station (430), and position estimation control device (440) of FIG. 4. The electronic device (1000) may be identical or substantially identical to the electronic device (101) of FIG. 1.

[0145] The UE position estimation unit (1010) has at least one RF signal (y t,s It may include a deep neural network capable of directly estimating the location of a terminal from the signal. The deep neural network can use sequentially received signals and a plan view of the surrounding environment of the base station as network inputs and estimate the location of the terminal through a transformer structure. The deep neural network can approximate a complex function that derives an output from an input through learning with a large amount of data. According to one embodiment, the model of the deep neural network may be composed of a regression model that utilizes mean squared error as a loss function.

[0146] The electronic device (1000) is N R At least one RF signal (y) through the antennas t,s Can receive ). RF signal(y t,s ) can be determined based on mathematical formula 14.

[0147] [Mathematical Formula 14]

[0148]

[0149] Here, t (=1:T) represents the time step of the received input signal, s (=1:S) represents the subcarrier of the received input signal, and nt,s can mean a noise vector.

[0150] The UE location estimation unit (1010) is N R At least one RF signal (y) received through antennas t,s The location of the terminal can be estimated by using a map image around the electronic device (1000) (e.g., a base station) as input. The map image may be a plan view in RGB format or a plan view showing the height of buildings or obstacles in image form. A plan view showing the height of buildings or obstacles in image form may be a grayscale image. The UE location estimation unit (1010) takes into account the model operation time and time step The terminal's location at It can be estimated.

[0151] FIG. 11 is a diagram illustrating the operation of a transformer-based position estimation deep neural network according to one embodiment of the present disclosure.

[0152] The transformer-based position estimation deep neural network of FIG. 11 may be included in an electronic device (e.g., the electronic device (1000) of FIG. 10). Referring to FIG. 11, the transformer-based position estimation deep neural network may include at least one of a positional embedding module (1110), a self-attention transformer (1120) that supports self-attention operations, a fully-connected network (1130, 1170), an input embedding module (1140), a cross-attention transformer (1150, 1160) that supports cross-attention operations, and a feed-forward network (1180).

[0153] The self-attention transformer (1120) may include a self-attention module (1121) that performs self-attention operations and a feed-forward network (1123). The cross-attention transformer (1150, 1160) may include a cross-attention module (1151) that performs cross-attention operations and a feed-forward network (1153). The transformer-based location estimation deep neural network receives a received signal (y t,s The location of the terminal can be derived directly from ). The environmental planar map around the base station is X, formed by adding positional embedding after patch embedding. img It can be. The corresponding input passes through several transformer blocks and then through a linear layer to Z img It becomes.

[0154] The operation of multi-head self-attention among the operations of the Transformer block is as follows. Let matrix A be the input for multi-head attention, and the total N h By multiplying the matrix A received as input to the k-th head among the heads by the query weight matrix, key weight matrix, and value weight matrix respectively, query Q k , Key K k , Value V k ... can be calculated. Then, the attention score can be calculated using the query, key, and value. The attention score S of the k-th head. k It can be calculated as in mathematical formula 15.

[0155] [Mathematical Formula 15]

[0156]

[0157] Here, d is the dimension of the embedded vector. The attention scores calculated at each head can be combined to produce the final result.

[0158] The Transformer-based location estimation deep neural network uses the received signal y t,s Receive and, identically It can be converted into and used as input. The input passes through a single linear layer Z'= It can be embedded as.

[0159] Input Z' is position embedding E pos,3 This is added This can be. The corresponding input is X img Used together with, the channel parameter estimate passes through multiple transformer blocks and then through a single linear layer and estimated distance d to the obstacle int,t This can be derived as an intermediate result for time step 1:T.

[0160] Transformer-based location estimation deep neural networks are intermediate results The final result is obtained by passing through a network consisting of multiple linear layers and activation functions using... It can derive.

[0161] The operation of multi-head cross attention within the Transformer block is as follows. Given matrices A and B as input, a total of N h At the k-th head among the heads, matrix A is multiplied by the query weight matrix to obtain query Q k Calculate and multiply matrix B by the key-value weight matrix to obtain key K k , Value V k Calculate . After that, the attention score of the k-th head is the same as self-attention This can be calculated.

[0162] According to one embodiment, matrix X for the signal sig This A, matrix X for the image img This may apply.

[0163] The loss function J3 of the deep neural network is defined as Equation 16.

[0164] [Mathematical Formula 16]

[0165]

[0166]

[0167] Here is the weight. The corresponding loss function serves to impose constraints on the network's intermediate results, thereby guiding the model to operate in the desired direction after training.

[0168] FIG. 12 is a figure showing the average absolute error of the estimated length / angle according to one embodiment of the present disclosure as a signal-to-noise ratio.

[0169] In FIGS. 12 (a) to (e), it is assumed that the base station is located in the center of a simulated urban environment of, for example, 140 m * 140 m, and the terminal moves in a straight line, accelerates, decelerates, and rotates along a main road. In both the channel parameter estimation unit (610 in FIG. 6) and the UE location estimation unit (620 in FIG. 6), the activation function may use a Gaussian error rectified linear unit (GELU) defined as Equation 14.

[0170] [Mathematical Formula 14]

[0171]

[0172] Figures 12 (a) to (e) show the root mean squared error (RMSE) of the signal propagation path length / angle estimation of the channel parameter estimation unit (610 in Figure 6) according to the signal-to-noise ratio (SNR).

[0173] FCN refers to a length / angle estimation technique based on a Fully Connected Network as a control group (e.g., set to 8 hidden layers and a dimension of 1,000). Transformer is the technique proposed in this disclosure, set to an embedding dimension of 256 and 4 transformer blocks. Both of these techniques use a single input of t=1, s=1. Transformer-seq is a small model using an embedding dimension of 128 and 2 transformer blocks; while smaller in size, it uses sequential inputs of t=1:3, s=1:6.

[0174] FIG. 12(a) is about distance error, FIG. 12(b) is about AoA (Azimuth) error, FIG. 12(c) is about AoA (Elevation) error, FIG. 12(d) is about AoD (Azimuth) error, and FIG. 12(e) is about AoD (Elevation) error.

[0175] FIG. 13 is a figure showing the average absolute error of the estimated position according to the signal-to-noise ratio according to one embodiment of the present disclosure.

[0176] Figure 13 shows the root mean square error of the position estimation of the UE position estimation unit (620 in Figure 6) according to the signal-to-noise ratio.

[0177] FCN, Transformer, and Transformer-seq are the results obtained when using, for example, an embedding dimension of 288 and 6 Transformer blocks, the length / angle derived from the FCN, Transformer, and Transformer-seq models of the channel parameter estimation unit (610 in Fig. 6) as input. Transformer-seq (small) is the result of the small model using, for example, an embedding dimension of 144 and 3 Transformer blocks, the length / angle derived from the Transformer-seq model of the channel parameter estimation unit (610 in Fig. 6) as input.

[0178] The embodiments of this document and the terms used therein are not intended to limit the technical features described in this document to specific embodiments, and should be understood to include various modifications, equivalents, or substitutions of said embodiments. In connection with the description of the drawings, similar reference numerals may be used for similar or related components. The singular form of a noun corresponding to an item may include one or more of said items unless the relevant context clearly indicates otherwise. In this document, phrases such as "A or B," "at least one of A and B," "at least one of A or B," "A, B or C," "at least one of A, B and C," and "at least one of A, B, or C" each may include any one of the items listed together in the corresponding phrase, or all possible combinations thereof. Terms such as "first," "second," or "first" or "second" may be used simply to distinguish said components from other said components and do not limit said components in any other aspect (e.g., importance or order). Where any (e.g., 1st) component is referred to as “coupled” or “connected” to another (e.g., 2nd) component, with or without the terms “functionally” or “communicationly,” it means that said any component may be connected to said other component directly (e.g., via a wire), wirelessly, or through a third component.

[0179] The term “module” as used in the embodiments of this document may include a unit implemented in hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit, for example. A module may be a component formed integrally, or a minimum unit of said component or a part thereof that performs one or more functions. For example, according to one embodiment, a module may be implemented in the form of an application-specific integrated circuit (ASIC).

[0180] One embodiment of the present document may be implemented as software (e.g., program (140)) comprising one or more instructions stored in a storage medium (e.g., internal memory (136) or external memory (138)) readable by a machine (e.g., electronic device (101)). For example, a processor (e.g., processor (120)) of the machine (e.g., electronic device (101)) may call at least one of the one or more instructions stored in the storage medium and execute it. This enables the machine to be operated to perform at least one function according to the at least one called instruction. The one or more instructions may include code generated by a compiler or code that can be executed by an interpreter. The storage medium readable by the machine may be provided in the form of a non-transitory storage medium. Here, 'non-temporary' simply means that the storage medium is a tangible device and does not contain a signal (e.g., electromagnetic waves), and the term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily.

[0181] According to one embodiment, the method according to the embodiments disclosed herein may be provided by being included in a computer program product. The computer program product may be traded between a seller and a buyer as a product. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)), or distributed online (e.g., download or upload) through an application store (e.g., Play Store™) or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily created on a device-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0182] According to one embodiment, each component (e.g., module or program) of the components described above may include a singular or multiple entities, and some of the multiple entities may be separated and placed in other components. According to one embodiment, one or more of the components or operations of the aforementioned components may be omitted, or one or more other components or operations may be added. Generally or additionally, multiple components (e.g., module or program) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each of the multiple components in the same or similar manner as those performed by the corresponding component among the multiple components prior to integration. According to one embodiment, operations performed by the module, program, or other components may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, omitted, or one or more other operations may be added.

Claims

1. A method for estimating the position of a first device in a wireless communication system, The operation of receiving at least one signal from a second device; The operation of using the above at least one signal as an input to a first transformer-based deep neural network to estimate at least one of the propagation path length, angle of departure, and angle of arrival of each of the above at least one signal; and A method characterized by including the operation of obtaining a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length of each of the at least one signal, the departure angle, the arrival angle, and an image of the surrounding environment of the first device.

2. In Paragraph 1, The first device above is a base station or a location estimation control device, and A method characterized in that the above-mentioned second device is a UE (user equipment).

3. In paragraph 1, each of the first transformer-based deep neural network and the second transformer-based deep neural network is, A method characterized by including at least one of a multi-head attention module, a layer normalization module, and a feed-forward network module.

4. A method according to claim 1, characterized in that the image of the surrounding environment is a plan view in the form of RGB (red-green-blue) or a plan view showing the height of an object in a grayscale form.

5. In paragraph 1, the at least one signal is, A method characterized by including at least one of the following: the number of channel paths, channel gain, the azimuth angle and elevation angle of a receiver in each channel, the azimuth angle and elevation angle of a transmitter in each channel, the number of receiver antennas, and the number of transmitter antennas.

6. A method according to claim 1, characterized in that the first transformer-based deep neural network is composed of a regression model that uses mean squared error as a loss function for learning.

7. A method according to claim 1, characterized in that the second transformer-based deep neural network is composed of a regression model that uses mean squared error as a loss function for learning.

8. In paragraph 1, the second transformer-based deep neural network is A method characterized by comprising a model that uses a loss function to which at least one output value of the first transformer-based deep neural network and at least one weight value capable of setting a range for the at least one output value are applied.

9. In Paragraph 1, An operation to determine the position where the at least one signal is reflected using the propagation path length, the departure angle, the arrival angle, and an image of the surrounding environment of the first device for each of the at least one signal; and A method characterized by further including an operation of estimating the position of the second device based on the position where at least one signal is reflected.

10. In a first device of a wireless communication system, transceiver; Memory; and It includes at least one processor, The above at least one processor executes instructions stored in the memory: Receive at least one signal from a second device, and Using the above at least one signal as an input to a first transformer-based deep neural network, at least one of the propagation path length, angle of departure, and angle of arrival of each of the above at least one signal is estimated, and A device characterized by obtaining a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length of each of the at least one signal, the departure angle, the arrival angle, and an image of the surrounding environment of the first device.

11. In Paragraph 10, The first device above is a base station or a location estimation control device, and The above second device is characterized as being a UE (user equipment).

12. In paragraph 10, each of the first transformer-based deep neural network and the second transformer-based deep neural network is, A device characterized by including at least one of a multi-head attention module, a layer normalization module, and a feed-forward network module.

13. A device according to claim 10, characterized in that the image of the surrounding environment is a plan view in the form of RGB (red-green-blue) or a plan view showing the height of an object in a grayscale form.

14. In paragraph 10, the above at least one signal is, A device characterized by including at least one of the following: the number of channel paths, channel gain, the azimuth angle and elevation angle of a receiver in each channel, the azimuth angle and elevation angle of a transmitter in each channel, the number of antennas of a receiver, and the number of antennas of a transmitter.

15. In a storage medium storing at least one instruction readable by a computer, The above at least one instruction causes an electronic device to perform a plurality of operations when executed by at least one processor, and The above multiple operations are, The operation of receiving at least one signal from a second device; The operation of using the above at least one signal as an input to a first transformer-based deep neural network to estimate at least one of the propagation path length, angle of departure, and angle of arrival of each of the above at least one signal; and A storage medium characterized by including an operation of obtaining a position estimation value of the second device using a second transformer-based deep neural network based on at least one of the propagation path length of each of the at least one signal, the departure angle, the arrival angle, and an image of the surrounding environment of the first device.

Images