Radio positioning system and method

Abstract

To provide a radio positioning system and a method that can easily reduce weight of a neural network used for radio positioning.SOLUTION: The present invention provides a radio positioning system including: a collection unit that, through communication between a plurality of anchor terminals and a mobile terminal installed in a space to carry out positioning of the mobile terminal, collects a plurality of pieces of CIR data and data of a distance between the anchor terminals and the mobile terminal; a preprocessing unit that performs preprocessing on the plurality of pieces of CIR data to generate a plurality of pieces of preprocessing data; a first learning unit that learns a first artificial neural network on the basis of the plurality of pieces of preprocessing data and the distance data; an analysis unit that analyzes the first artificial neural network to select, from the plurality of pieces of preprocessing data, a plurality of pieces of positioning important data used for the positioning of the mobile terminal; and a second learning unit that learns a second artificial neural network on the basis of the plurality of pieces of positioning important data and the distance data.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a wireless positioning system and method thereof, and more particularly to an indoor wireless positioning system and method thereof that can minimize positioning errors while reducing weight by utilizing an artificial neural network (ANN) and explainable artificial intelligence (XAI). [Background technology]

[0002] Radio positioning technology is a technology that estimates the location of a mobile terminal by using the characteristics of radio signals measured through wireless communication. That is, it is a technology that estimates the location of a mobile terminal by processing radio signals sent from multiple anchor terminals that are installed in fixed locations and whose location information is known in advance.

[0003] This wireless positioning technology measures ToF (Time of Flight) to perform ranging. More specifically, it measures time while exchanging messages through wireless communication between anchors and tags, and estimates distance using the speed of light. Since light travels at a speed of about 300,000km / s, a time measurement error of 1ns indicates a distance error of about 30cm. For this reason, accurate time measurement is closely related to the accuracy of distance measurement.

[0004] Conventional wireless positioning technology measures time using Channel Impulse Response (CIR) data collected during wireless communication.

[0005] The CIR data is shown in a graph in Figure 1. Referring to Figure 1, the CIR data shows a signal intensity waveform measured in nanoseconds and is composed of hundreds of peaks.

[0006] Conventional wireless positioning technology uses a first peak algorithm for accurate time measurement. As shown in Figure 1, the first peak algorithm is an algorithm that measures the time of the first peak that exceeds a certain threshold, and is an algorithm devised based on the assumption that the earliest received peak is the peak of a linear signal.

[0007] FIG. 2 is a diagram illustrating an example of a radio signal propagation process in a radio positioning technology.

[0008] Referring to Figure 2, a wireless signal is reflected by surrounding structures and propagates along multiple paths. The CIR has multiple peaks due to signals propagating along multiple paths. Looking at the peaks in Figures 1 and 2, an example of how a signal propagated along multiple paths appears in CIR data can be seen. If the time of peak 2 in Figure 1 is measured, the distance of the path corresponding to peak 2 in Figure 2 will be measured, resulting in a distance error.

[0009] The conventional first peak algorithm has the advantage of being simple, but the disadvantage is that distance measurement errors occur significantly in NLOS (Non-Line-of-Sight) environments where there are many surrounding structures, resulting in multiple paths, and where signal propagation is blocked by obstacles.In particular, since UWB (Ultra-wideband) is often used for indoor positioning, such error problems often occur.

[0010] Recently, research has been presented on technology that uses deep learning technology to learn CIR data and select peaks to solve these problems. This method uses an artificial neural network to learn CIR data and infer the wireless channel propagation environment, thereby selecting the peak of the shortest path.

[0011] Although this deep learning-based wireless positioning technology has significantly improved distance measurement accuracy compared to conventional technologies, it has the problem that it requires a large amount of calculations for the artificial neural network to learn and operate. Generally, devices that use wireless positioning technology are low-power mobile devices that run on batteries, so techniques that use deep learning are difficult to operate in real time. In addition, the causal relationship between the generated learning model and the accuracy results is unclear like a black box, which causes a problem of low reliability. Summary of the Invention [Problem to be solved by the invention]

[0012] An object of the present invention is to provide a wireless positioning system and method that can easily reduce the weight of a neural network used in wireless positioning.

[0013] Another object of the present invention is to provide a wireless positioning system and method that is capable of performing positioning with high accuracy while being lightweight.

[0014] Another object of the present invention is to provide a wireless positioning system and method capable of dramatically reducing costs by minimizing the minimum required computing power of a mobile terminal for positioning.

[0015] The technical problems to be achieved by the present invention are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those having ordinary skill in the art to which the present invention pertains from the following description. [Means for solving the problem]

[0016] In order to solve the above problem, the present invention provides a wireless positioning system including: a collection unit that collects a plurality of CIR data and distance data between the anchor terminal and the mobile terminal by communicating with a plurality of anchor terminals installed in a space where positioning of the mobile terminal is to be performed; a preprocessing unit that preprocesses the plurality of CIR data to generate a plurality of preprocessed data; a first learning unit that trains a first artificial neural network based on the plurality of preprocessed data and the distance data; an analysis unit that analyzes the first artificial neural network and selects a plurality of positioning important data used for positioning of the mobile terminal from the plurality of preprocessed data; and a second learning unit that trains a second artificial neural network based on the plurality of positioning important data and the distance data.

[0017] Here, the plurality of pre-processed data may comprise a plurality of pixels in the form of a two-dimensional image indicating the relationship between the plurality of peaks of the CIR data.

[0018] The analysis unit may also remove connections between arbitrary neural network nodes from the first artificial neural network to construct a plurality of third artificial neural networks having accuracy levels equal to or greater than a reference value.

[0019] In addition, the analysis unit may input a plurality of pre-processed data to a plurality of third artificial neural networks, and select pre-processed data that output the same distance data from a reference number or more of the third artificial neural networks among the plurality of third artificial neural networks as selected data.

[0020] The analysis unit can also evaluate the importance of each of a plurality of pixels of the preprocessed data, and select a plurality of important positioning data consisting of a plurality of pixels whose importance is equal to or greater than a reference value.

[0021] Moreover, the wireless positioning system of the present invention may further include a third learning unit that trains the autoencoder based on the plurality of preprocessed data and the positioning important data.

[0022] Here, the autoencoder can output positioning-important data when preprocessed data is input.

[0023] In addition, the location information of the mobile terminal can be calculated using the location information of the anchor terminal and the communication distance between the mobile terminal and the anchor terminal.

[0024] The mobile terminal may also include a pre-processing unit, an auto-encoder, and a second artificial neural network to perform positioning of the mobile terminal.

[0025] In addition, the mobile terminal can preprocess the CIR data through a preprocessing unit to generate preprocessed data, input the preprocessed data to an autoencoder to output positioning important data, and input the positioning important data to a second artificial neural network to perform positioning of the mobile terminal.

[0026] The present invention also provides a wireless positioning method, including: a step of a mobile terminal communicating with a plurality of anchor terminals installed in a space where positioning of the mobile terminal is to be performed, and collecting a plurality of CIR data and distance data between the anchor terminals and the mobile terminal; a step of pre-processing the plurality of CIR data to generate a plurality of pre-processed data; a step of training a first artificial neural network based on the plurality of pre-processed data and the distance data; a step of analyzing the first artificial neural network to select a plurality of positioning important data used for positioning of the mobile terminal from the plurality of pre-processed data; a step of training an auto-encoder based on the plurality of pre-processed data and the positioning important data; and a step of training a second artificial neural network based on the plurality of positioning important data and the distance data.

[0027] In addition, the wireless positioning method of the present invention may further include, after the step of training the second artificial neural network, a step of the mobile terminal preprocessing the CIR data through a preprocessing unit to generate preprocessed data, a step of the mobile terminal inputting the preprocessed data to an autoencoder to output positioning important data, and a step of inputting the positioning important data to the second artificial neural network to perform positioning of the mobile terminal. Effect of the Invention

[0028] According to the present invention, it is possible to analyze a complex neural network using explainable artificial intelligence (XAI) technology and extract up to several tens of key positioning characteristic data, and it is possible to easily reduce the weight of the neural network using the extracted positioning characteristic data.

[0029] In addition, the present invention provides a method for performing positioning that is significantly lighter than conventional methods while showing higher accuracy than conventional complex machine learning-based positioning technologies, and has the effect of enabling quick and convenient search for key positioning characteristic data in new environments.

[0030] Furthermore, according to the present invention, the minimum required computing power of a mobile terminal for positioning can be minimized, so that a dramatic cost reduction effect can be expected.

[0031] The effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those having ordinary skill in the art from the following description. [Brief description of the drawings]

[0032] [Figure 1] 1 is a graph illustrating CIR data. [Diagram 2] 1 is a diagram illustrating an example of a radio signal propagation process in a radio positioning technology; [Diagram 3] 1 is an overall configuration diagram of a wireless positioning system according to an embodiment of the present invention. [Figure 4] 1 is a block diagram of a wireless positioning system according to an embodiment of the present invention. [Diagram 5] FIG. 2 is a block diagram of a server of the wireless positioning system according to an embodiment of the present invention. [Figure 6] 1 is a diagram illustrating the structure of a first artificial neural network according to an embodiment of the present invention. [Figure 7] 4 is a diagram illustrating the structure of a second artificial neural network according to an embodiment of the present invention. [Figure 8] 2 is a block diagram of a mobile terminal in a wireless positioning system according to an embodiment of the present invention. [Figure 9] 2 is a flowchart of a wireless positioning method according to an embodiment of the present invention. [Figure 10] 4 is a detailed flowchart of a step of performing positioning of a mobile terminal according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0033] The terms and words used in this specification and the claims should not be interpreted in a limited manner based on their ordinary or dictionary meaning, but should be interpreted in a meaning and concept that corresponds to the technical idea of ​​the present invention, based on the principle that the inventor can appropriately define the concept of the term in order to best describe his / her invention.

[0034] Therefore, it should be understood that the embodiments described in this specification and the configurations illustrated in the drawings are merely the most preferred embodiments of the present invention and do not represent the entire technical idea of ​​the present invention, and that there may be various equivalents and modifications that can replace them at the time of this application.

[0035] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art to which the present invention pertains can easily carry out the present invention.

[0036] FIG. 3 is a diagram showing the overall configuration of a wireless positioning system according to an embodiment of the present invention, and FIG. 4 is a block diagram of the wireless positioning system according to an embodiment of the present invention.

[0037] Hereinafter, a radio positioning system according to an embodiment of the present invention will be described with reference to FIG. 3 and FIG.

[0038] The wireless positioning system according to the embodiment of the present invention may include a server 100, a mobile terminal 200 and an anchor terminal 300.

[0039] A plurality of anchor terminals 300 may be fixedly installed in an indoor space where positioning of the mobile terminal 200 is to be performed. In addition, position (coordinate) information of the anchor terminal 300 may be preset according to the location where the anchor terminal 300 is installed, and may be transmitted to the mobile terminal 200 through wireless communication.

[0040] The mobile terminal 200 is a device that a user can carry around with him or her, and may include a mobile phone, a smart phone, a laptop computer, a digital broadcasting terminal, a PDA (Personal Digital Assistant), a PMP (Portable Multimedia Player), a navigation system, and the like.

[0041] The mobile terminal 200 may include a communication unit for performing wireless communication with the server 100 and the anchor terminal 300. As such wireless communication technologies, Wireless LAN (WLAN) (Wi-Fi), Wireless broadband (Wibro), World Interoperability for Microwave Access (Wimax), High Speed ​​Downlink Packet Access (HSDPA), General Packet Radio Service (GPRS), CDMA, WCDMA (registered trademark), Long Term Evolution (LTE), 5G, and 6G (not limited thereto) may be used.

[0042] In addition, the mobile terminal 200 can perform short-range communication with the anchor terminal 300 through a defined wireless communication protocol. Such short-range communication technology may include, but is not limited to, UWB (Ultra Wideband), ZigBee, Bluetooth (registered trademark), RFID (Radio Frequency Identification), and IrDA (infrared Data Association).

[0043] The mobile terminal 200 can measure the communication distances with each of the anchor terminals 300. Specifically, the mobile terminal 200 can measure the communication distance between the mobile terminal 200 and the anchor terminal 300 by performing communication with the anchor terminal 300 and using channel state information (CSI) of a radio signal received from the anchor terminal 300.

[0044] Here, channel state information (CSI) refers to data related to a wireless channel such as CIR (Channel Impulse Response) and CFR (Channel Frequency Response), and the mobile terminal 200 can select and collect channel state information (CSI) that can be easily collected for each wireless communication technology.

[0045] The position (coordinate) information of the mobile terminal 200 can be calculated using the position information of a plurality of anchor terminals 300 and the communication distance between the mobile terminal 200 and the anchor terminal 300 .

[0046] For example, but not limited to, the mobile terminal 200 can calculate the location of the mobile terminal 200 using a triangulation method.

[0047] Here, the position of the mobile terminal 200 does not mean an absolute position but a relative position based on a plurality of anchor terminals 300 .

[0048] In the following, a method for calculating the two-dimensional position coordinates (x, y) of the mobile terminal 200 using a trilateral measurement method will be described, but the three-dimensional position coordinates can also be calculated using the same method.

[0049] The position of the mobile terminal 200 can be expressed by two-dimensional position coordinates (x, y). In the case of two-dimensional position coordinates, the positions of at least three anchor terminals 300 and the communication distances between the mobile terminal 200 and at least three anchor terminals 300 must be measured.

[0050] That is, to calculate the two-dimensional position coordinates, it is necessary to find the equations of three circles whose centers are the positions of the three anchor terminals 300 and whose radii are the communication distances between the mobile terminal 200 and the three anchor terminals 300. At this time, the point where the three circles intersect is calculated as the position of the mobile terminal 200.

[0051] The mobile terminal 200 calculates equations for three circles whose centers are the current positions of the three anchor terminals 300 and whose radii are the communication distances between the mobile terminal 200 and the anchor terminal 300, as shown in Equation 1 below.

number

[0052] Here, xi and yi (where i is 1, 2, 3) are the two-dimensional position coordinates of the anchor terminal, and d1, d2, d3 are the communication distances between the mobile terminal 200 and the anchor terminal 300.

[0053] In this way, by simultaneously solving the three equations defined in Equation 1, the two-dimensional position coordinates (x, y) of the mobile terminal 200 can be calculated.

[0054] Of course, it goes without saying that the anchor terminal 300 also includes a communication unit for performing communication functions and communication distance measurement functions.

[0055] FIG. 5 is a block diagram of a server of a wireless positioning system according to an embodiment of the present invention.

[0056] Referring to FIG. 5, the server 100 may include a collection unit 110, a preprocessing unit 120, a first learning unit 130, an analysis unit 140 and a second learning unit 150.

[0057] The collection unit 110 can collect a plurality of CIR data 10 and distance data between the anchor terminal 300 and the mobile terminal 200 by the mobile terminal 200 communicating with a plurality of anchor terminals 300 installed in the space in which the positioning of the mobile terminal 200 is to be performed.

[0058] Here, the collection unit 110 can label the distance data to the CIR data 10 .

[0059] Meanwhile, the CIR data 10 is received energy over time, and is data in the form of a time series. In order for the first learning unit 130 to learn the CIR data 10, the first learning unit 130 must grasp the relationship between hundreds of peaks, but there is a limit in that it is difficult to grasp the relationship between peaks far away from the CIR data 10.

[0060] Accordingly, the pre-processing unit 120 can receive a plurality of CIR data 10 labeled with distance data in order to grasp the relationship between peaks that are far apart, and pre-process the CIR data 10 to generate a plurality of pre-processed data 11.

[0061] Here, the plurality of pre-processed data 11 may comprise a plurality of pixels in the form of a two-dimensional image showing the relationship between the plurality of peaks of the CIR data 10 .

[0062] The pre-processing unit 120 can generate pre-processed data 11 by pre-processing the CIR data 10 using the following Equations 2 to 5.

number

number

number

number

[0063] Here, S represents the CIR data 10, X represents the pre-processed data 11, and n represents the number of peaks in the CIR data 10.

[0064] And μ s means the average value of the 10 peaks of the CIR data, and f i,jis the normalized value of the difference between each peak.

[0065] The first learning unit 130 receives a plurality of preprocessing data 11 and distance data from the preprocessing unit 120, and can train the first artificial neural network 410 based on these data.

[0066] Here, artificial neural network (ANN) is a statistical learning algorithm in machine learning and cognitive science inspired by biological neural networks (particularly the brain of an animal's central nervous system). Artificial neural network refers to a general model in which artificial neurons that form a network through synaptic connections have the ability to solve problems by changing the strength of synaptic connections through learning.

[0067] The first learning unit 130 can learn all kinds of artificial neural networks, but it is preferable to select an artificial neural network that is suitable for an environment in which positioning of the mobile terminal 200 is performed. For example, the first learning unit 130 can learn based on a Convolutional Neural Network (CNN).

[0068] Thereafter, when the pre-processed data is input to the trained first ANN 410, a plurality of distances between the anchor terminal 300 and the mobile terminal 200 can be output.

[0069] FIG. 6 is a diagram illustrating the structure of a first artificial neural network according to an embodiment of the present invention.

[0070] Referring to FIG. 6, the first artificial neural network 410 may be composed of an input layer to which a plurality of pieces of channel state information (CSI1 to CSIn) are input, an output layer to which a plurality of pieces of distance data between the anchor terminal 300 and the mobile terminal 200 are output, and a hidden layer located between the input layer and the output layer.

[0071] The first learning unit 130 can train the first artificial neural network 410 by inputting a large number of pieces of channel state information (CSI1 to CSIn) and a plurality of pieces of distance data between the anchor terminal 300 and the mobile terminal 200 to an input layer. Here, by training the first artificial neural network 410 using sufficient data until the accuracy reaches a target value, a high-performance and high-computation first artificial neural network 410 can be constructed.

[0072] Although the first artificial neural network 410 exhibits high accuracy, it is difficult to operate in real time within the mobile terminal 200 because it requires a high level of computation.

[0073] The analysis unit 140 can analyze the first artificial neural network 410 using explainable artificial intelligence (XAI) technology to select and extract important positioning data used for positioning the mobile terminal 200 from the multiple pre-processed data 11.

[0074] That is, the analysis unit 140 can obtain a data area that is mainly used when measuring the location of the mobile terminal 200 in the first artificial neural network 410 using an explainable artificial intelligence (XAI) (e.g., LIME), and can define the corresponding data area as optimized positioning important data.

[0075] Such positioning-important data refers to data that has a significant impact on the positioning of the mobile terminal 200 in the environment in which the first artificial neural network 410 was trained. By optimizing the first artificial neural network 410 to utilize only the positioning-important data, it is possible to significantly reduce the weight of the first artificial neural network 410 while still achieving a similar level of accuracy to the first artificial neural network 410.

[0076] Meanwhile, a data selection process is important in order to apply explainable artificial intelligence technology to the first artificial neural network 410. If data that causes a neural network to malfunction, such as an outlier, is applied to explainable artificial intelligence, the interpretation accuracy of the first artificial neural network 410 will decrease due to incorrect explanations.

[0077] Accordingly, the analysis unit 140 can use a data selection algorithm to remove connections between any neural network nodes from the first artificial neural network 410, and construct a plurality of third artificial neural networks 405 whose comparison accuracy with the first artificial neural network 410 is equal to or greater than a reference value.

[0078] The analysis unit 140 can input a plurality of pre-processed data to a plurality of third artificial neural networks 405, and select pre-processed data that outputs the same distance data from a reference number or more of the third artificial neural networks 405 as selected data.

[0079] The analysis unit 140 can evaluate the importance of each of the plurality of pixels of the preprocessed data 11, particularly the selected data, and select a plurality of important positioning data consisting of a plurality of pixels whose importance is equal to or greater than a reference value.

[0080] For example, the analysis unit 140 may evaluate the importance of a pixel as low as possible so that there is no difference between the distance data output when the entire selected data is input to the first artificial neural network 410 and when the selected data is input to the first artificial neural network 410 after removing a specific pixel.

[0081] Although not shown in the drawings, the wireless positioning system according to the embodiment of the present invention may further include a third learning unit.

[0082] The third learning unit can train an AutoEncoder 420 based on a plurality of pre-processed data or selected data and positioning-important data.

[0083] Here, the trained autoencoder 420 can output positioning important data when the preprocessed data 11 is input. Specifically, the autoencoder 420 can output positioning important data of a size (m, m) when the preprocessed data 11 of a size (n, n) is input. For example, n can be 2036 and m can be 20.

[0084] The second learning unit 150 can train the second artificial neural network 430 based on the plurality of positioning important data and distance data output from the auto-encoder 420. Here, the positioning important data can be labeled with distance data.

[0085] The trained second ANN 430 is trained by inputting only a minimum amount of important positioning data, so it is lighter than the first ANN 410, and noise data is removed, so it can show high accuracy.

[0086] FIG. 7 is a diagram illustrating the structure of a second artificial neural network according to an embodiment of the present invention.

[0087] 7, the second artificial neural network 430 may be composed of an input layer to which important positioning data (PCI1 to PCIm, where m is an integer smaller than n) is input, an output layer to which distance data between a plurality of anchor terminals 300 and the mobile terminal 200 is output, and a hidden layer present between the input layer and the output layer, where the number of hidden layers is smaller than that of the first artificial neural network 410.

[0088] FIG. 8 is a block diagram of a mobile terminal in a wireless positioning system according to an embodiment of the present invention.

[0089] The mobile terminal 200 may include a pre-processing unit 120, a trained auto-encoder 420 and a second artificial neural network 430.

[0090] The mobile terminal 200 can perform positioning of the mobile terminal 200 using the on-board pre-processing unit 120, the auto-encoder 420 and the second artificial neural network 430.

[0091] Specifically, the mobile terminal 200 wirelessly communicates with a plurality of anchor terminals 300 to measure and collect CIR data 10 used in the communications.

[0092] The mobile terminal 200 then pre-processes the CIR data 10 through the pre-processing unit 120 to generate pre-processed data 11, inputs the pre-processed data 11 to the auto-encoder 420 to output positioning important data, and inputs the positioning important data to the second artificial neural network 430 to estimate distances between the multiple anchor terminals 300 and the mobile terminal 200.

[0093] The mobile terminal 200 can perform positioning of the mobile terminal 200 using the position (coordinate) information of the plurality of anchor terminals 300 and the estimated distances between the plurality of anchor terminals 300 and the mobile terminal 200 .

[0094] In this way, the wireless positioning system of the present invention can analyze complex neural networks using explainable artificial intelligence (XAI) technology to extract several tens of key positioning important data, and can easily reduce the weight of the neural network by using the extracted positioning characteristic data.

[0095] Through this, the wireless positioning system of the present invention exhibits higher accuracy than conventional complex machine learning-based positioning technologies, while enabling much lighter positioning than conventional methods, and has the effect of enabling quick and convenient search for key positioning characteristic data in new environments.

[0096] Furthermore, the wireless positioning system of the present invention can be expected to achieve a dramatic cost reduction effect by minimizing the minimum required computing power of the mobile terminal for positioning.

[0097] FIG. 9 is a flowchart of a wireless positioning method according to an embodiment of the present invention, and FIG. 10 is a detailed flowchart of a step of performing positioning of a mobile terminal according to an embodiment of the present invention.

[0098] Hereinafter, a radio positioning method according to an embodiment of the present invention will be described, but the same parts as those described above will be omitted.

[0099] First, referring to FIG. 9, a mobile terminal 200 wirelessly communicates with a plurality of anchor terminals 300 installed in a space in which positioning is to be performed, and measures a plurality of CIR data 10 and distance data between the plurality of anchor terminals 300 and the mobile terminal 200.

[0100] Next, the server 100 receives and collects the CIR data 10 measured by the mobile terminal 200 and distance data between a plurality of anchor terminals 300 and the mobile terminal 200 (S110). At this time, the CIR data 10 may be labeled with distance data.

[0101] Next, the server 100 pre-processes the plurality of CIR data 10 to generate a plurality of pre-processed data 11 (S120).

[0102] Next, the server 100 trains the first artificial neural network 410 based on the plurality of pre-processed data 11 and the distance data (S130).

[0103] Next, the server 100 analyzes the first artificial neural network 410 using explainable artificial intelligence (XAI) technology to extract and select a plurality of important positioning data used for positioning the mobile terminal 200 from the plurality of pre-processed data 11 (S140).

[0104] Next, the server 100 trains the autoencoder 420 based on the plurality of preprocessed data 11 or the selected data and the important positioning data (S150). After the training is completed, when the plurality of preprocessed data 11 is input to the autoencoder 420, the important positioning data is output.

[0105] Next, the server 100 trains a second artificial neural network based on the plurality of positioning important data and the distance data outputted to the autoencoder 420 (S160).

[0106] Next, referring to FIG. 10, the pre-processing unit 120, the auto-encoder 420 and the second artificial neural network 430 are installed in the mobile terminal 200 (S210).

[0107] Next, the mobile terminal 200 communicates with a plurality of anchor terminals 300 and measures the CIR data 10 of the radio signals used in the communication.

[0108] Next, the preprocessing unit 120 preprocesses the CIR data 10 to generate preprocessed data 11 (S220).

[0109] Next, the mobile terminal 200 inputs the pre-processed data 11 to the auto-encoder 420 and outputs positioning important data (S230).

[0110] Next, the mobile terminal 200 inputs the positioning important data to the second artificial neural network 430 to estimate the distances between the plurality of anchor terminals 300 and the mobile terminal 200, thereby performing positioning of the mobile terminal 200 (S240).

[0111] In this way, the wireless positioning method of the present invention can analyze a complex neural network using explainable artificial intelligence technology (XAI) to extract up to several tens of key positioning characteristic data, and can easily reduce the weight of the neural network by using the extracted positioning characteristic data.

[0112] Through this, the wireless positioning method of the present invention exhibits higher accuracy than conventional complex machine learning-based positioning technologies, while enabling much lighter positioning than conventional methods, and has the effect of enabling quick and convenient search for key positioning characteristic data in new environments.

[0113] Furthermore, the wireless positioning method of the present invention can be expected to achieve a dramatic cost reduction effect by minimizing the minimum required computing power of the mobile terminal for positioning.

[0114] The wireless positioning method according to the embodiment of the present invention may be implemented as a processor-readable code on a medium having a program recorded thereon. Examples of such a processor-readable medium include ROM, RAM, CD-ROM, magnetic tape, flexible disk, optical data storage device, etc., and may be implemented in the form of a carrier wave (e.g., transmission through the Internet).

[0115] The above detailed description is illustrative of the present invention. Moreover, the above description merely illustrates and describes preferred embodiments of the present invention, and the present invention can be used in various other combinations, modifications, and environments. That is, changes or modifications are possible within the scope of the inventive concept disclosed herein, the scope of the disclosure as written, and / or the scope of the skill or knowledge of the art. The above examples are intended to illustrate the best conditions for carrying out the present invention, and may be implemented in other conditions known in the art to utilize other inventions such as the present invention, and various modifications may be required for the specific application field and use of the invention. Therefore, the above detailed description of the invention is not intended to limit the present invention to the disclosed embodiments. Moreover, the appended claims should be construed to include other embodiments. [Explanation of symbols]

[0116] 100 servers 110 Collection Department 120 Pre-processing section 130 First Learning Section 140 Analysis Department 150 Second Learning Section 200 Mobile terminal 300 Anchor Terminal

Claims

1. a collection unit that communicates with a plurality of anchor terminals installed in a space where the positioning of the mobile terminal is to be performed and collects a plurality of CIR data and distance data between the anchor terminals and the mobile terminal; a pre-processing unit that pre-processes the plurality of CIR data to generate a plurality of pre-processed data consisting of a plurality of pixels in a two-dimensional image form that indicates a relationship between a plurality of peaks of the CIR data; a first learning unit configured to train a first artificial neural network based on the plurality of pre-processed data and the distance data; an analysis unit that evaluates the importance of each of a plurality of pixels of the preprocessed data and selects a plurality of data consisting of a plurality of pixels whose importance is equal to or greater than a reference value as a plurality of important data that have a significant effect on the positioning; a second learning unit configured to learn a second artificial neural network based on the plurality of important data and the distance data; inputting the plurality of important data into the second artificial neural network to estimate the distance data; A wireless positioning system using explainable artificial intelligence technology (XAI) that performs positioning of the mobile terminal using location information of the plurality of anchor terminals and distance data between the plurality of anchor terminals estimated and the mobile terminal.

2. The analysis unit 2. The wireless positioning system of claim 1, further comprising removing any connections between neural network nodes from said first artificial neural network to construct a plurality of third artificial neural networks having an accuracy equal to or greater than a reference value.

3. The analysis unit 3. The wireless positioning system according to claim 2, further comprising: inputting the plurality of pre-processed data to a plurality of third artificial neural networks; and selecting, as selected data, pre-processed data that output the same distance data from a reference number or more of the plurality of third artificial neural networks.

4. The location information of the mobile terminal The wireless positioning system according to claim 1 , wherein the calculation is performed using location information of the anchor terminal and a communication distance between the mobile terminal and the anchor terminal.

5. The mobile terminal The wireless positioning system of claim 1 , further comprising: the preprocessing unit and a second artificial neural network for performing positioning of the mobile terminal.

6. a step of communicating with a plurality of anchor terminals installed in a space where the positioning of the mobile terminal is to be performed, and collecting a plurality of CIR data and distance data between the anchor terminals and the mobile terminal; pre-processing the plurality of CIR data to generate a plurality of pre-processed data consisting of a plurality of pixels in the form of a two-dimensional image indicating a relationship between a plurality of peaks of the CIR data; training a first artificial neural network based on the plurality of pre-processed data and the distance data; evaluating the importance of each of the plurality of pixels of the preprocessed data, and selecting a plurality of data consisting of a plurality of pixels whose importance is equal to or greater than a reference value as a plurality of important data that significantly affect the positioning; training a second artificial neural network based on the plurality of key data and the distance data; inputting the plurality of important data into the second artificial neural network to estimate the distance data; A wireless positioning method using explainable artificial intelligence technology (XAI) that performs positioning of the mobile terminal using location information of the plurality of anchor terminals and distance data between the plurality of anchor terminals estimated and the mobile terminal.

Images