Wireless communication device and position estimation method

Abstract

The present invention comprises: a communication unit (11) for acquiring communication quality information including beamforming information of a communication terminal (2) by means of wireless communication with a communication terminal (2); an acquisition unit (12) for acquiring position information of the communication terminal (2); a data generation unit (13) for generating training data using the position information and the beamforming information; a model generation unit (14) for performing machine learning using the training data and constructing a prediction model for position estimation of the communication terminal (2); and a use unit (15) for inputting the beamforming information of the communication terminal (2) acquired by the communication unit (11) to the prediction model during communication between the communication terminal (2) and the communication unit (11), and acquiring the result of position estimation of the communication terminal (2).

Description

Wireless communication device and position estimation method

[0001] The present disclosure relates to a wireless communication device and a position estimation method.

[0002] There is a high demand for estimating the location of communication terminals, and this technology is used in a variety of situations, such as understanding human movement and behavior, managing people and objects in factories, etc. To estimate the location of a communication terminal, methods have been proposed that use communication quality information such as the received signal strength of the communication terminal, or spatial information such as images captured by a camera.

[0003] Non-Patent Document 1 discloses a method for estimating the location of a communication terminal based on beacon information from multiple access points, and Non-Patent Document 2 discloses a method for estimating the location of a communication terminal based on an image captured by a camera and received electromagnetic wave information (RSSI: Received Signal Strength Indication) acquired from the communication terminal.

[0004] YUAN YOU, CHANG WU, "Indoor Positioning System With CellularNetwork Assistance Basedon Received SignalStrength Indication of Beacon," DigitalObject Identifier 10.1109 / ACCESS.2019.2963099Alexandre Alahi,Albert Haque, Li Fei-Fei, "RGB-W: When Vision MeetsWireless," 2015 IEEE International Conference on ComputerVision (ICCV), DOI: 10.1109 / ICCV.2015.376, 7-13 Dec. 2015

[0005] However, in the above-mentioned Non-Patent Documents 1 and 2, in order to predict the position of a communication terminal with high accuracy, it is necessary to install multiple base stations and access points or multiple cameras, which results in a problem of an increase in the scale of the equipment. Also, in the method of using a camera to capture an image of a user carrying a communication terminal, there is a problem in that the position of the communication terminal cannot be estimated if the user moves and enters a blind spot of the camera, or if the user cannot be identified from the captured image.

[0006] The present disclosure has been made in consideration of the above circumstances, and its purpose is to provide a wireless communication device and a position estimation method that are capable of estimating the position of a communication terminal with high accuracy using a simple configuration.

[0007] A wireless communication device of one aspect of the present disclosure includes a communication unit that acquires communication quality information including beamforming information of a communication terminal through wireless communication with the communication terminal, an acquisition unit that acquires location information of the communication terminal, a data generation unit that generates training data using the location information and the beamforming information, a model generation unit that performs machine learning using the training data to construct a predictive model for estimating the location of the communication terminal, and a utilization unit that inputs the beamforming information of the communication terminal acquired by the communication unit into the predictive model during communication between the communication terminal and the communication unit, and acquires a location estimation result for the communication terminal.

[0008] A position estimation method of one aspect of the present disclosure includes a communication unit acquiring communication quality information including beamforming information of a communication terminal through wireless communication with the communication terminal, an acquisition unit acquiring position information of the communication terminal, a data generation unit generating training data using the position information and the beamforming information, a model generation unit performing machine learning using the training data to construct a predictive model for position estimation of the communication terminal, and a utilization unit inputting the beamforming information of the communication terminal acquired by the communication terminal during communication between the communication terminal and the communication unit into the predictive model to obtain a position estimation result for the communication terminal.

[0009] According to the present disclosure, it is possible to estimate the position of a communication terminal with high accuracy using a simple configuration.

[0010] FIG. 1 is a block diagram illustrating a configuration of a wireless communication device according to an embodiment. FIG. 2 is a flowchart illustrating a processing procedure of the wireless communication device according to an embodiment. FIG. 3 is an explanatory diagram illustrating an area in which a robot carrying a communication terminal moves. FIG. 4 is an explanatory diagram illustrating errors in the x and y directions when the position of a communication terminal is predicted using RSSI, beamforming information, and RSSI and beamforming information. FIG. 5A is a graph illustrating data in the x direction when the position of a communication terminal is predicted using RSSI. FIG. 5B is a graph illustrating data in the y direction when the position of a communication terminal is predicted using RSSI. FIG. 6A is a graph illustrating data in the x direction when the position of a communication terminal is predicted using beamforming information. FIG. 6B is a graph illustrating data in the y direction when the position of a communication terminal is predicted using beamforming information. FIG. 7A is a graph illustrating data in the x direction when the position of a communication terminal is predicted using RSSI and beamforming information. FIG. 7B is a graph illustrating data in the y direction when the position of a communication terminal is predicted using RSSI and beamforming information. FIG. 8 is a block diagram showing the hardware configuration of this embodiment.

[0011] Hereinafter, an embodiment will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a wireless communication device 1 and its peripheral devices according to an embodiment. As shown in FIG. 1, the wireless communication device 1 is capable of wireless communication with a communication terminal 2. The communication terminal 2 is a mobile device such as a smartphone that can be carried by a user. In order to reduce the influence of surrounding objects, the communication terminal 2 employs beamforming control to improve communication speed and stability, such as in 5G cellular communication and IEEE 802.11ac.

[0012] Beamforming is a technology that improves communication quality by concentrating radio waves at a specific point, and is used in a wide range of fields, including 5G communications and wireless LANs. In particular, it is used in mobile communications in high-frequency bands where radio waves are prone to attenuation, as a technology that amplifies signal strength and enables transmission over long distances.

[0013] The wireless communication device 1 is capable of wireless or wired communication with the sensor 3. The sensor 3 is installed within an area in which the communication terminal moves and acquires location information of the communication terminal. The sensor 3 may be a camera or LiDAR fixed within the area.

[0014] As shown in FIG. 1, the wireless communication device 1 includes a communication unit 11, an acquisition unit 12, a data generation unit 13, a model generation unit 14, and a utilization unit 15.

[0015] The communication unit 11 performs wireless communication with the communication terminal 2. The communication unit 11 acquires communication quality information through communication with the communication terminal 2. The communication quality information refers to the throughput of the communication terminal 2, beamforming information, received signal power, signal-to-noise power ratio, signal-to-interference-plus-noise power ratio, received signal strength indication (RSSI), reference signal received quality (RSRQ), packet error rate, number of arriving bits, bit error rate, number of arriving bits per unit time, modular code index (MCS), number of retransmissions, packet arrival delay time, error correction technology settings, communication terminal user contract information, differential information of these values, indicators calculated from these values ​​using a formula, frequency conditions such as the frequency of the communication terminal 2 and the bandwidth of the resources used, and setting items of the communication terminal 2 that affect these indicators. The communication unit 11 outputs the acquired communication quality information to the data generation unit 13. That is, the communication unit 11 acquires communication quality information including the beamforming information of the communication terminal 2 through wireless communication with the communication terminal 2.

[0016] The acquisition unit 12 acquires location information of the communication terminal 2. The location information can be acquired from, for example, a GPS (Global Positioning System) receiver. The acquisition unit 12 can acquire the location information from an image captured by an externally fixed camera or point cloud information detected by LiDAR. Note that the information acquired by the acquisition unit 12 is used when the data generation unit 13 generates training data (details of which will be described later), but is not used when the utilization unit 15 utilizes the training data.

[0017] The data generation unit 13 generates training data used to build a prediction model that estimates the position of the communication terminal 2, based on the beamforming information and communication quality information other than beamforming (e.g., RSSI) acquired by the communication unit 11. In addition to the beamforming information and communication quality information, the data generation unit 13 may generate training data by combining information on camera images acquired by the acquisition unit 12 and point cloud information collected by LiDAR.

[0018] The data generator 13 may select, from among the communication quality information, information that has a high correlation with the location information of the communication terminal 2. For example, the data generator 13 may extract communication quality information that has a correlation coefficient of a certain value or more with the location information of the communication terminal 2, and generate the training data using the extracted communication quality information in addition to the beamforming information. The data generator 13 may generate training data based only on the beamforming information, without using communication quality information such as RSSI.

[0019] "Training data" refers to data used when executing machine learning to build a prediction model. The training data refers to, for example, data indicating the relationship between information on the position, speed, and direction of the communication terminal 2 and surrounding objects, and the beamforming information of the communication terminal 2 measured at that time and other communication quality information. In other words, the data generator 13 generates training data using the position information and beamforming information of the communication terminal 2.

[0020] The model generation unit 14 acquires the training data generated by the data generation unit 13 and performs machine learning to construct a prediction model for estimating the position of the communication terminal 2. That is, the model generation unit 14 performs machine learning using the training data to construct a prediction model for estimating the position of the communication terminal 2. The model generation unit 14 provides the constructed prediction model to the utilization unit 15.

[0021] During wireless communication between the communication terminal 2 and the communication unit 11, the utilization unit 15 acquires beamforming information and other communication quality information of the communication terminal 2 acquired by the communication unit 11. The utilization unit 15 inputs the beamforming information and other communication quality information into the prediction model generated by the model generation unit 14. The utilization unit 15 acquires the position estimation result of the communication terminal 2 output from the prediction model.

[0022] During communication between the communication terminal 2 and the communication unit 11, the utilization unit 15 inputs beamforming information of the communication terminal 2 acquired by the communication unit 11 into the prediction model to acquire a position estimation result of the communication terminal 2. When estimating the communication quality of the communication terminal 2 during communication between the communication terminal 2 and the communication unit 11, the utilization unit 15 inputs the beamforming information into the prediction model and acquires a position estimation result without inputting the position information acquired by the acquisition unit 12 into the prediction model.

[0023] The utilization unit 15 can use the position estimation results to monitor the flow of people moving around town, understand people's behavior, manage people and objects in factories, etc. By predicting future position information of the communication terminal 2, the utilization unit 15 can improve the responsiveness and stability of communication control by the communication terminal 2 through more accurate resource control, reflector control, and beamforming control.

[0024] Next, the operation of the wireless communication device 1 configured as described above will be described with reference to the flowchart shown in Fig. 2. First, in step S11 of Fig. 2, the communication unit 11 performs wireless communication with the communication terminal 2 and acquires beamforming information of the communication terminal 2 and communication quality information other than beamforming.

[0025] In step S12, the acquisition unit 12 acquires position information of the communication terminal 2 based on data detected by the sensor 3. The position information may include not only the position of the communication terminal 2 but also the speed and direction. Specifically, the position information of the communication terminal 2 is acquired from an image captured by a camera or point cloud data detected by LiDAR.

[0026] In step S13, the data generator 13 acquires beamforming information and other communication quality information of the communication terminal 2, and further acquires location information of the communication terminal 2, and generates training data based on this information.

[0027] In step S14, the model generation unit 14 performs machine learning based on the training data generated by the data generation unit 13 to construct a prediction model for estimating the position of the communication terminal 2. That is, the training data is generated by combining not only communication quality information indicating signal strength such as the above-mentioned RSSI and RSRP, but also beamforming information that controls radio waves in a specific direction to achieve high-speed and stable communication speeds, and the prediction model is constructed using this training data. The beamforming information may be collected from base stations and surrounding communication terminals. In this way, a prediction model can be constructed using the training data generated using the beamforming information.

[0028] In step S15, the utilization unit 15 acquires communication quality information (e.g., RSSI) during wireless communication between the communication terminal 2 and the communication unit 11, inputs this communication quality information to the prediction model constructed by the model generation unit 14, and acquires a position estimation result for the communication terminal 2. The prediction model constructed by the model generation unit 14 is generated using training data generated using beamforming information, and therefore can estimate with high accuracy the position of the communication terminal 2. Furthermore, because communication quality information from multiple base stations and access points is not required, the position estimation accuracy for the communication terminal 2 can be improved without increasing the size of the device.

[0029] Next, the accuracy of position estimation when the communication terminal 2 is moved will be described with reference to Figures 3, 4, 5A, 5B, 6A, 6B, 7A, and 7B. Figure 3 is an explanatory diagram showing a robot carrying the communication terminal 2 traveling around area D1 in the order of points p1 → p2 → p3 → p4 → p1. The position indicated by the symbol p0 is a 5G base station. The left-right direction of the plane shown in Figure 2 is the x-direction, and the up-down direction is the y-direction.

[0030] RSSI and beamforming information (Beam Index) were collected at 500 ms intervals as communication quality information. The beamforming information here is expressed as an integer value between 0 and 63, and indicates the direction of the beam. A prediction model was prepared that outputs the x and y coordinates of the future position of the communication terminal from the communication quality information over the past second, and machine learning was performed. The prediction model used was "lightGBM."

[0031] The prediction results were evaluated using the absolute mean error. By combining beamforming information with the RSSI of the conventional technology, the accuracy improved by approximately 66% and 78% in the absolute mean error of the x and y coordinates, respectively.

[0032] FIG. 4 is an explanatory diagram showing the error between the position to which the communication terminal 2 actually moved and the position of the communication terminal 2 estimated using the prediction model constructed by the model generation unit 14.

[0033] 4 shows the result of constructing a prediction model using training data generated using RSSI and estimating the position of the communication terminal 2 using this prediction model. In other words, it shows the result of estimating the position of the communication terminal using a prediction model constructed without using beamforming information.

[0034] The "beamforming information" shown in Figure 4 shows the results of constructing a prediction model using training data generated using the beamforming information of communication terminal 2 and estimating the position of communication terminal 2 using this prediction model.

[0035] The "RSSI, beamforming information" shown in Figure 4 shows the results of constructing a prediction model using training data generated using the RSSI and beamforming information of communication terminal 2, and estimating the position of communication terminal 2 using this prediction model.

[0036] As can be seen from Figure 4, when "RSSI" is used, there is a large error between the actual measured values ​​and the estimated values ​​for "0 seconds ahead (present)," "1 second ahead," and "3 seconds ahead." When "beamforming information" is used, the error is reduced compared to when "RSSI" is used. Furthermore, when both "RSSI and beamforming information" are used, the error is further reduced compared to when "RSSI" is used. In other words, it can be seen that by using a prediction model constructed using training data generated using "beamforming information," the position of the communication terminal 2 can be estimated with extremely high accuracy.

[0037] Fig. 5A is a graph showing data in the x direction when the position of communication terminal 2 is predicted using "RSSI" shown in Fig. 4, with curve s1 showing observed values ​​and curve s2 showing predicted values ​​0 seconds later. Fig. 5B is a graph showing data in the y direction when the position of communication terminal 2 is predicted using "RSSI" shown in Fig. 4, with curve s3 showing observed values ​​and curve s4 showing predicted values ​​0 seconds later. As can be seen from each of curves s1 to s4 shown in Figs. 5A and 5B, a large error occurs between the observed values ​​and the predicted values.

[0038] Fig. 6A is a graph showing data in the x direction when the position of communication terminal 2 is predicted using the "beamforming information" shown in Fig. 4, with curve s11 showing observed values ​​and curve s12 showing predicted values ​​0 seconds later. Fig. 6B is a graph showing data in the y direction when the position of communication terminal 2 is predicted using the "beamforming information" shown in Fig. 4, with curve s13 showing observed values ​​and curve s14 showing predicted values ​​0 seconds later. As can be seen from curves s11 to s14 shown in Figs. 6A and 6B, the error between the observed values ​​and predicted values ​​is small, and it can be seen that the predicted values ​​are approaching the observed values.

[0039] Fig. 7A is a graph showing data in the x direction when the position of communication terminal 2 is predicted using the "RSSI, beamforming information" shown in Fig. 4, with curve s21 showing observed values ​​and curve s22 showing predicted values ​​0 seconds later. Fig. 7B is a graph showing data in the y direction when the position of communication terminal 2 is predicted using the "RSSI, beamforming information" shown in Fig. 4, with curve s23 showing observed values ​​and curve s24 showing predicted values ​​0 seconds later. As can be seen from curves s21 to s24 shown in Figs. 7A and 7B, the error between the observed values ​​and predicted values ​​is smaller, and it can be seen that the predicted values ​​are getting closer to the observed values.

[0040] 3 shows an example of estimating the x and y coordinates of the communication terminal 2 when the communication terminal 2 moves within the area D1, but it is also possible to perform position estimation based on other conditions, such as estimating the position on an image captured by a camera installed within the area D1. Furthermore, it is also possible to make predictions by combining the position estimation result of the communication terminal 2 with spatial information such as image information from the camera or point cloud information obtained from LiDAR.

[0041] As such, the wireless communication device 1 of this embodiment includes a communication unit 11 that acquires communication quality information including beamforming information of the communication terminal 2 through wireless communication with the communication terminal 2, an acquisition unit 12 that acquires position information of the communication terminal 2, a data generation unit 13 that generates training data using the position information and beamforming information, a model generation unit 14 that performs machine learning using the training data and constructs a prediction model for position estimation of the communication terminal 2, and a utilization unit 15 that inputs the beamforming information of the communication terminal 2 acquired by the communication unit 11 into the prediction model during communication between the communication terminal 2 and the communication unit 11, and acquires the position estimation result of the communication terminal 2.

[0042] In this embodiment, training data is generated using beamforming information acquired through communication between the communication terminal 2 and the communication unit 11, and this training data is used to construct a prediction model for estimating the position of the communication terminal 2. This makes it possible to perform highly accurate position estimation without requiring communication quality information from multiple base stations and access points, and it is possible to improve the position estimation accuracy of the communication terminal 2 without increasing the size of the device.

[0043] This embodiment utilizes the characteristics of beamforming adopted in 5G cellular communication, IEEE 802.11ac, or later communication methods, and therefore, it is not necessary to acquire communication quality information such as received signal strength indicator (RSSI) from multiple access points or base stations, and it is possible to estimate the location information of the communication terminal 2 with high accuracy based on communication quality information acquired from a single access point or base station.

[0044] In this embodiment, when estimating the position of the communication terminal 2, the position of the communication terminal 2 is estimated using beamforming information detected by the communication unit 11. Therefore, it is possible to estimate the position information of the communication terminal 2 with high accuracy without using spatial information detected by a sensor 3 such as a camera or LiDAR. Therefore, it is possible to acquire a highly accurate position estimation result of the communication terminal 2 without being affected by a blind spot occurring in the detection range of the sensor 3, which reduces the detection accuracy.

[0045] In this embodiment, in an indoor factory where GPS information cannot be obtained, it is possible to grasp the movements of workers wearing VR (Virtual Reality) goggles and use this information for behavioral analysis. Also, at indoor and outdoor work sites, by monitoring the movements of workers in dangerous places such as high places and high-temperature areas where it is difficult to install cameras, this can be useful for preventing accidents during work.

[0046] The wireless communication device 1 of the present embodiment described above can be, for example, a general-purpose computer system including a CPU (Central Processing Unit, processor) 901, a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), a communication device 904, an input device 905, and an output device 906, as shown in Fig. 8. The memory 902 and the storage 903 are storage devices. In this computer system, the CPU 901 executes a predetermined program loaded on the memory 902, thereby realizing each function of the wireless communication device 1.

[0047] The wireless communication device 1 may be implemented by one computer or by multiple computers, and may also be a virtual machine implemented on a computer.

[0048] The program for the wireless communication device 1 can be stored in a computer-readable recording medium such as an HDD, an SSD, a Universal Serial Bus (USB) memory, a Compact Disc (CD), or a Digital Versatile Disc (DVD), or can be distributed via a network. The computer-readable recording medium is, for example, a non-transitory recording medium.

[0049] The present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the present disclosure.

[0050] REFERENCE SIGNS LIST 1 wireless communication device 2 communication terminal 3 sensor 11 communication unit 12 acquisition unit 13 data generation unit 14 model generation unit 15 utilization unit

Claims

1. A wireless communication device comprising: a communication unit that acquires communication quality information including beamforming information of a communication terminal through wireless communication with the communication terminal; an acquisition unit that acquires location information of the communication terminal; a data generation unit that generates training data using the location information and the beamforming information; a model generation unit that performs machine learning using the training data to construct a predictive model for estimating the location of the communication terminal; and a utilization unit that inputs the beamforming information of the communication terminal acquired by the communication unit into the predictive model during communication between the communication terminal and the communication unit, and acquires a location estimation result for the communication terminal.

2. The wireless communication device described in claim 1, wherein the data generation unit extracts communication quality information from the communication quality information that has a correlation coefficient of a certain value or more with the location information of the communication terminal, and generates the training data using the extracted communication quality information in addition to the beamforming information.

3. A wireless communication device as described in claim 1 or 2, wherein when estimating the communication quality of the communication terminal during communication between the communication terminal and the communication unit, the utilization unit inputs the beamforming information into the prediction model and acquires the position estimation result without inputting the position information acquired by the acquisition unit into the prediction model.

4. A position estimation method in which a communication unit acquires communication quality information including beamforming information of a communication terminal through wireless communication with the communication terminal, an acquisition unit acquires location information of the communication terminal, a data generation unit generates training data using the location information and the beamforming information, a model generation unit performs machine learning using the training data to construct a predictive model for position estimation of the communication terminal, and a utilization unit inputs the beamforming information of the communication terminal acquired by the communication terminal during communication between the communication terminal and the communication unit into the predictive model to obtain a position estimation result of the communication terminal.

Images