KR102416604B1 - Mehtod and device for precise positioning for wireless communication system - Google Patents

Abstract

The present invention discloses a positioning method and apparatus. More particularly, the present invention relates to a method and apparatus for precise positioning of a wireless communication system capable of calculating the coordinates of a target node, which is a moving body, in a short-range wireless data communication system such as WiFi.
According to an embodiment of the present invention, CSI (Channel State Information) data is extracted from a signal transmitted and received between a target node and an anchor, a plurality of algorithms are applied to derive two or more estimated distances, and a learning model advanced through deep learning By determining the estimated value using , and generating the coordinates of the target node based on this, there is an effect that more precise positioning of the moving object is possible.

Description

Precise positioning method and device of wireless communication system

The present invention relates to a positioning method and apparatus, and to a precise positioning method and apparatus in a wireless communication system capable of calculating the coordinates of a target node, which is a moving body, in a short-range wireless data communication system such as WiFi.

With the rapid development of electronic and communication technologies, various wireless communication services using a wireless network are being provided. Accordingly, the service provided by the wireless communication system using the wireless communication network is developing into a multimedia communication service that transmits multimedia contents such as packet data as well as a limited voice service such as a telephone call.

Recently, methods for measuring the location of a mobile terminal indoors using APs (Access Points) located in buildings are being studied. This enables related services according to the location information of a user carrying a mobile terminal connected to communicate with the AP using the AP location information. In general, the AP location information is calculated in advance and signal strength is used to positioning will be performed.

As a related art, Korean Patent Registration No. 10-1174542 discloses a system and method for measuring a position of a mobile terminal, which measures the reception intensity values of a plurality of radio signals transmitted from a plurality of APs to the mobile terminal, and After setting a path loss model for a plurality of APs based on the wireless signal transmission environment between each AP and the mobile terminal, the distance between each of the plurality of APs and the mobile terminal is calculated from the reception intensity values of the plurality of wireless signals using the set path loss. Disclosed is a technique for calculating and measuring the position of a mobile terminal.

In addition, the Proceedings of the Information and Communication Equipment Conference "Implementation of a system that provides indoor location information service to smartphone users using a wireless sensor network and indoor positioning algorithm (2011.08.25. ~ 27.)" utilizes a wireless sensor network. It relates to a system that provides a location information service to a smartphone user in an indoor environment. Sensors built indoors communicate with the user's smartphone to receive the user's signal strength, and each sensor node transmits the user's signal strength to the gateway. After transmitting to the server through the, a technique for calculating the user's location based on the signal strength received from the server is described.

The location information calculation based on the AP may use a measuring device or a mobile terminal, but in any case, the AP location information depends on a simple calculation based on a single measurement result using a single device. There is a limitation in that an accurate calculation of .

Inaccurate AP location calculation leads to inaccurately providing the user's location, which causes user dissatisfaction with related services.

Korean Patent Publication No. 10-1174542 (2012.08.09.)

The present invention uses a mobile object such as a smart terminal as a target node in a WiFi environment, which is one of the commercialized wireless communication systems, and an AP (Access Point) of fixed coordinates as an anchor, so that the estimated distance between the target node and the anchor, the reception angle, etc. There is a problem to provide a precise positioning method and apparatus of a wireless communication system for estimating the coordinates of a target node using information.

In order to solve the above problems, the precise positioning method of a wireless communication system according to an embodiment of the present invention is a target node using a reception signal according to wireless communication between a mobile target node and one or more AP (Access Point) anchors. As a positioning method, CSI (Channel State Information) extracted from the received signal of the target node is collected, a first algorithm is applied to the CSI to identify a propagation path, and FTM (Fine Time Measurement) is calculated, and the propagation path and calculating a first estimated distance value between the target node and the anchor by applying a second algorithm to the FTM estimate, and applying the third algorithm to the angle of arrival (AoA) estimated using the CSI and a plurality of FTM samples to correct the multipath error and calculate a second estimated distance value between the AP and the target node; inputting a value and selectively extracting one with high accuracy to determine a distance estimate; and applying a fourth algorithm to the distance estimate and the phase-relaxed AoA estimate for the AoA to generate the coordinates of the target node may include steps.

In addition, in order to solve the above problems, the precision positioning device of the wireless communication system according to an embodiment of the present invention is a target using a reception signal according to wireless communication between a mobile target node and one or more AP (Access Point) anchors. As a positioning device for a node, collecting CSI (Channel State Information) extracted from the received signal of the target node, applying a first algorithm to the CSI, identifying a propagation path, calculating FTM (Fine Time Measurement), and A first distance estimating means for calculating a first estimated distance value between the target node and the anchor by applying a second algorithm to the propagation path and FTM estimate, AoA (Angle of Arrival) estimated using the CSI, and a plurality of FTM samples Learning learned through a deep learning procedure using a second distance estimation means for correcting multipath errors by applying a third algorithm to the AP and calculating a second estimated distance value between the AP and the target node, a plurality of learning data and environmental parameters A deep learning prediction unit that inputs the first and second estimated distance values to a model and selectively extracts one with high accuracy to determine a distance estimate; 4 It may include a target node output unit for generating the coordinates of the target node by applying the algorithm.

According to an embodiment of the present invention, CSI (Channel State Information) data is extracted from a signal transmitted and received between a target node and an anchor, a plurality of algorithms are applied to derive two or more estimated distances, and a learning model advanced through deep learning By determining the estimated value using , and generating the coordinates of the target node based on this, there is an effect that more precise positioning of the moving object is possible.

) 성분에 대한 모식도(a) 및, 그 신호 파형에 대한 그래프(b)를 나타낸 도면이다.
도 7은 MUSIC 알고리즘을 이용한 CSI 분석에 따라 추출된 스펙트럼 데이터를 나타내는 그래프(a) 및, 그 스펙트럼 데이터에 따른 거리값에 대한 다이어그램(b)을 나타낸 도면이다.
도 8은 FTM 프로토콜의 데이터 플로우를 예시한 도면이다.
도 9는 4개의 앵커로부터 표적노드의 좌표를 산출하기 위한 좌표 그래프를 예시한 도면이다.
도 10은 두 개의 직선이동을 기반으로 하는 단일 앵커 환경에서 측위를 위한 좌표 그래프를 나타낸 도면이다.
도 11은 표적노드가 임의의 삼각형을 따라 이동하는 단일 앵커 환경에서 측위를 위한 좌표 그래프를 나타낸 도면이다.1 is a diagram schematically showing the overall structure of a precision positioning device of a wireless communication system according to an embodiment of the present invention.
2 is a diagram illustrating a precise positioning method of a wireless communication system in a pipeline according to an embodiment of the present invention.
3 to 5 are diagrams illustrating a precise positioning method of a wireless communication system by the precision positioning device of FIG. 2 as a flowchart.
6 is an angle of an incoming signal for an AP having two antennas (

) is a diagram showing a schematic diagram (a) of the component and a graph (b) of the signal waveform.
7 is a diagram showing a graph (a) showing spectrum data extracted according to CSI analysis using the MUSIC algorithm (a) and a diagram (b) of a distance value according to the spectrum data.
8 is a diagram illustrating a data flow of an FTM protocol.
9 is a diagram illustrating a coordinate graph for calculating the coordinates of a target node from four anchors.
10 is a diagram illustrating a coordinate graph for positioning in a single anchor environment based on two linear movements.
11 is a diagram illustrating a coordinate graph for positioning in a single anchor environment in which a target node moves along an arbitrary triangle.

Prior to the description, when a part in the entire specification "includes" or "includes" a certain component, it does not exclude other components, but may further include other components unless otherwise stated. means there is In addition, terms such as "...unit", "...server", "...system", etc. described in the specification mean a unit that processes at least one function or operation and may be implemented by hardware, software, or a combination of hardware and software.

In addition, the term "embodiment" in this specification means serving as an illustration, example, or illustration, but the subject matter of the invention is not limited by such examples. Also, "comprising", "comprising", "having" and other similar terms are used, but when used in the claims, " as an open-ended transition word that does not exclude any additional or other element. It is used generically in a manner analogous to the term "comprising".

The various techniques described herein may be implemented in conjunction with hardware or software, or a combination of both, where appropriate. As used herein, terms such as "...Unit", "...Server", "...System" and the like likewise refer to computer-related entities, i.e. hardware , a combination of hardware and software, software or software in execution may be treated as equivalent. In addition, each function implemented in the system of the present invention may be configured as a module unit program, and may be recorded in one physical memory or may be recorded while being distributed between two or more memories and recording media.

In addition, in the following description, the term "precision positioning device of a wireless communication system" of the present invention may be abbreviated as "precision positioning device" for convenience of description.

Hereinafter, a precise positioning method and apparatus of a wireless communication system according to an embodiment of the present invention will be described in detail with reference to the drawings.

1 is a diagram schematically showing the overall structure of a precision positioning device of a wireless communication system according to an embodiment of the present invention.

1, the precision positioning device 100 of the wireless communication system according to the embodiment of the present invention, a mobile terminal such as a smartphone or a tablet PC as a target node in a WiFi environment, a location supporting WiFi connection It is characterized in that the target node is precisely positioned in a wireless communication system using a fixed access point (AP) as an anchor.

To this end, the precision positioning device 100 according to an embodiment of the present invention estimates the coordinates of the target node using a plurality of algorithms from the estimated distance from the target node to the anchor, the reception angle and the absolute coordinates of the anchor, where the anchor It is assumed that the absolute coordinate of is a value known in advance as a fixed node.

In detail, the precision positioning device 100 of the present invention is a positioning device of the target node 10 using a received signal according to wireless communication between the target node 10, which is a mobile body, and one or more APs (50) anchors. Collects CSI (Channel State Information) extracted from the received signal of the node 10, identifies the propagation path by applying the first algorithm to the CSI, calculates FTM (Fine Time Measurement), and adds the information to the propagation path and FTM estimate. The first distance estimating means 110 for calculating a first estimated distance value between the target node 10 and the AP 50 by applying the 2 algorithm, Fine Time Measurement (FTM) and Angle of Arrival (AoA) using CSI The second distance estimation means 120 for calculating the second estimated distance value between the AP 50 and the target node 10 by applying a third algorithm for correcting the multipath error for A deep learning prediction unit 130 that determines a distance estimate by selectively extracting a high-accuracy one by inputting first and second estimated distance values to a learning model learned through a deep learning procedure using, and a distance estimate, AoA The target node output unit 140 may include a target node output unit 140 that generates coordinates of the target node by applying the fourth algorithm to the phase-relaxed AoA estimate.

The first distance estimating means 110 collects CSI in the WiFi environment of the present invention, and derives the result value from the MUSIC (Multiple Emitter Location and Signal Parameter Estimation) algorithm, which is a first algorithm based thereon, and the error therefor The first estimated distance value may be calculated by correcting the value through a second algorithm, When FTM Discovered MUSIC (FUSIC).

In detail, the first distance estimating unit 110 extracts CSI from the PHY of the received signal through the product of the number of transmitter antennas, the number of receiver antennas, and the number of subcarriers of the target node 10 according to its role. (111), a spectrum extractor 112 that extracts a pseudo spectrum by classifying the CSI multiple signal by applying the first algorithm, identifies a propagation path based on the peak of the pseudo spectrum, and based on the propagation path ToF (Time of Flight) and the predominance path calculation unit 113 that extracts the estimate of radio wave intensity and the second algorithm to detect the inaccurate return value of the FTM, including the NLOS (Non Line of Sight) environment, and correct the error. It may be divided into a first distance calculator 114 that calculates one estimated distance value.

The CSI data extraction unit 111 may extract CSI (Channel State Information) through the received signal of the target node 10 . The precision positioning device 100 according to an embodiment of the present invention uses various information included in CSI, as an example, attenuation for carrier n and θn value of a phase shift, etc. for the coordinate calculation of the target node 10 FTM and AoA estimates can be obtained, and the CSI data extraction unit 111 extracts CSI from the received signal of the target node 10 .

The spectrum extractor 112 may extract a pseudo spectrum from the extracted CSI using the MUSIC algorithm. A pseudo spectrum is a multi-signal classification for data, and the spectrum extractor 112 extracts a spectrum of CSI and identifies a propagation path through a peak value. In addition, the spectrum extraction unit 112 may extract the ToF and the radio wave intensity estimate from the identified propagation path.

The dominant path calculation unit 113 may identify the dominant path using the propagation path extracted by the spectrum extraction unit 112 . A propagation path is identified based on the peak of the pseudo spectrum, and when the ToF (Time of Flight) and radio wave intensity estimates are extracted based on the propagation path, the signal strength and ToF for each path identified by the first algorithm, the MUSIC algorithm, are calculated. can be used to calculate the relative intensity for each path.

The first distance calculator 114 may detect an inaccurate return value of the FTM including a non-line of sight (NLOS) environment through the second algorithm, correct the error, and calculate the first distance estimate value. To this end, the first distance calculator 114 can quantify the contribution of the direct path to the spectrum according to the MUSIC algorithm through the new parameter 'R', and the propagation path described above through the FUSIC algorithm as the second algorithm. The FTM measurement error can be calculated and the FTM measurement error can be calculated with the estimated ToF value.

In particular, the FUSIC algorithm can provide a distance estimate by finally removing the effect of an error. A detailed description of the MUSIC algorithm, including the FUSIC algorithm, and a detailed description of a calculation method of each algorithm will be described later.

In addition, the precision positioning device 100 of the wireless communication system according to an embodiment of the present invention is characterized in that it determines a more accurate distance estimation value through deep learning prediction learned based on situational awareness from two distance estimation results, Accordingly, the second distance estimating means 120 may be further included.

Specifically, the second distance estimating unit 120 includes the FTM collection unit 121 that collects M (M is a natural number greater than or equal to 20) FTM sample data from a plurality of links, the dominant path calculation unit, and the first algorithm. Receive the estimated Angle of Arrival (AoA) based on the extracted pseudo-spectrum peak, calculate the estimated AoA phase shift and FTM error using a third algorithm, and calculate the average and Akaike Information (AIC) results of the calculation. Criterion), a second distance calculator 123 that determines a path having a minimum positive average as a range estimate, and additionally collects FTMs from two additional APs, and collects them from three or more APs including an existing anchor A positioning correction unit 125 for generating a second distance estimation value by correcting the positioning calculated by the second distance calculation unit 123 by applying a multi-position calculation algorithm based on the least squares method to the FTM estimate may be included. .

The FTM collection unit 121 transmits the FTM message to the transmitter (FTM-I) through the receiver (FTM-R) as separate sample data and waits for an ACK, M (M is a natural number greater than or equal to 20) for each link The distance error can be calculated by collecting sample data.

The second distance calculation unit 123 refers to the AoA estimated by applying the first algorithm to the aforementioned CSI, and compares the distance error calculation result by applying the third algorithm with the average and AIC, respectively, to obtain a minimum amount The path with the mean can be determined as the range estimate.

Here, the AIC allows the performance of several statistical models to be compared with each other, the lower the error, the less the error, and a model that is not overfit or underfit may be selected. In particular, the aforementioned AIC is used to identify the optimal number of clusters in the Gaussian Mixture Model (GMM), and the second distance calculator 123 infers the optimal number of paths through the lowest AIC and then The path with the mean will be used as the range estimate.

The above-described third algorithm is SPRING (Smartphone Positioning with Radio Measurements from a Single WiFi) that corrects multipath errors through CSI for distance and direction targeting FTM and AoA in commercial chipsets to which IEEE 802.11-2016 and 802.11ac amendments are applied. Access Point) algorithm.

The positioning correction unit 125 may further collect FTMs from two additional APs, and using the estimated distances from the receivers (FTM-R) of three or more APs including the AP 50 above, the least squares method ( A second distance estimation value may be calculated by correcting the positioning calculated by the second distance calculator 123 by applying a multilateration algorithm based on Liner Least Squares (LLS).

Accordingly, the precise positioning apparatus 100 according to an embodiment of the present invention acquires first and second distance estimate values, and selects one more accurate value through a situation-aware deep learning technique, which will be described later.

As a configuration for this, the deep learning prediction unit 130 collects various situations at the time of communication positioning as learning data, and selects a more accurate distance estimation value through reinforcement learning among known deep learning techniques. characterized in that

That is, the deep learning prediction unit 130 can select any one of the first and second distance estimation values calculated by the above-described first and second distance estimation means 110 and 120 through a learning model according to reinforcement learning. have.

The deep learning prediction unit 130 performs reinforcement learning by inputting a plurality of learning data including first and second distance estimation values and environmental parameters for learning to the learning model through a learning procedure, thereby performing reinforcement learning later in the real environment. From among the first and second distance estimates, a more accurate distance estimate is selectively extracted under an actual environmental parameter value.

The target node output unit 140 may supplement the positioning by the distance estimation value extracted by the deep learning prediction unit 130 . Specifically, the target node output unit 140 may generate the coordinates of the target node by applying the fourth algorithm to the distance estimation value and the phase-relaxed AoA estimation value for AoA.

Here, the fourth algorithm is Accurate and Distributed Localization (ADL) for obtaining accurate positioning using a predetermined number of APs 50 based on the above-described distance estimation value and the AoA estimation value in which the phase error due to phase calibration is alleviated. An extended ADL algorithm in consideration of the algorithm or the movement of the target may be applied. In this case, the extended ADL algorithm has a feature that enables positioning in a poor environment in which only one AP 50 can be used.

According to the general ADL algorithm, the target node output unit 140 provides angle and distance information between the target node 10 and the AP 50 for each AP 50 when a plurality of APs 50 exist. The coordinates of the target node 10 may be calculated by calculating the estimated position of the target node 10 using the approximation, and calculating the average of the position estimates from all APs 50 .

In addition, the target node output unit 140 says that when one AP 50 exists according to the extended ADL algorithm, the target node 10 moves along an arbitrary triangle in a virtual coordinate system having the starting coordinate as the origin. To calculate the coordinates of the target node 10 by measuring the moving direction of the target node 10, calculating the distance between the rotation point and the starting point of the target node 10, and measuring the rotation angle or movement distance can

A description of the specific coordinate calculation method of the target node 10 using the above-described ADL algorithm and the extended ADL algorithm will be described later.

Hereinafter, with reference to the drawings, a precise positioning method of a wireless communication system by the above-described precision positioning device will be described.

2 is a diagram showing a precise positioning method of a wireless communication system according to an embodiment of the present invention in a pipeline, and FIGS. 3 to 5 are flowcharts showing a precise positioning method of a wireless communication system by the precision positioning device of FIG. It is a drawing.

In the following description, even if there is no separate description, the execution entity of each step is the precise positioning method of the wireless communication system of the present invention.

3 to 5 , the precise positioning method according to an embodiment of the present invention is a positioning method of a target node using a reception signal according to wireless communication between a mobile target node and one or more AP (Access Point) anchors. , collects CSI (Channel State Information) extracted from the received signal of the target node, identifies the propagation path by applying the first algorithm to the CSI, calculates FTM (Fine Time Measurement), and calculates the second Calculating the first estimated distance value between the target node and the anchor by applying the algorithm (S100), applying the third algorithm to the AoA (Angle of Arrival) estimated using CSI and a plurality of FTM samples to eliminate the multipath error Compensating and calculating a second estimated distance value between the AP and the target node (S200), input the first and second estimated distance values to the learning model learned through a deep learning procedure using a plurality of learning data and environmental parameters Selecting one with high accuracy to determine the distance estimate (S300) and applying the fourth algorithm to the distance estimate and the phase-relaxed AoA estimate for AoA to generate the coordinates of the target node (S400) may include

First, the CSI extracted from the received signal of the target node is collected, the first algorithm is applied to the CSI to identify the propagation path, and the FTM is calculated, and the second algorithm is applied to the propagation path and FTM estimate between the target node and the anchor. In the step of calculating the first estimated distance value (S100), the CSI of the PHY is extracted from the radio reception signal between the target node and the anchor. This step S100 may be subdivided into steps S110 to S140 of FIG. 4 .

Specifically, step S110 extracts CSI from the PHY of the received signal through the product of the number of transmitter antennas, the number of receiver antennas, and the number of subcarriers of the target node. In an embodiment of the present invention, analysis is performed based on CSI. In the case of wireless communication of a WiFi terminal having both transceivers and multiple array antennas, CSI can be extracted from the PHY of all multipath reception signals. Here, the number of all multi-paths may be calculated as 'the number of transmitter antennas × the number of receiver antennas × the number of subcarriers'.

Next, in step S120, a pseudo spectrum is extracted by classifying the CSI multi-signal by applying the first algorithm.

Next, in step S130, the propagation path is identified based on the peak of the pseudo spectrum, and time of flight (ToF) and an estimate of the radio wave intensity are extracted based on the propagation path.

6 illustrates a schematic diagram (a) of an angle (θ) component of an incoming signal arriving with respect to an AP having two antennas, and a graph (b) of the signal waveform.

It is known that the above-described data analysis of CSI is capable of capturing minor changes in the surrounding environment with a higher probability than considering a Receiver Signal Strength Indicator (RSSI). Therefore, it is easy to estimate the direct path (EDP) of the received signal, and thus EDP-based distance estimation has the advantage of being more robust than RSSI because it is much less sensitive to multi-path reflection.

The first algorithm is a method of processing a signal received from a sensor array to know the location of a multiple emitter, and is based on asymptotic analysis of an unbiased estimator, Multiple Emitter Location and Signal Parameter (MUSIC) Estimation) algorithm may be used, the data elements provided to it include the number of projected wavefront (or incident wavefront) signals present, the direction of arrival (DOA) or emitter positions of the signals, the intensity and cross-correlation between the incident waveforms; There are noise and noise blocking strength, polarization, and the like.

The above factors can be derived by distinguishing signals based on predictable phase changes when radio waves are received at specific locations. For example, in the case of WiFi, it relies on measurements obtained from each subcarrier of an OFDM (Orthogonal Frequency Division Multiplexing) system, and for a given ToF, as a difference in signal frequency creates a difference in phase at the receiving side, extraction is possible.

And, by analyzing the CSI through the MUSIC algorithm, a result in the form of a pseudo-spectrum can be obtained, and a propagation path can be identified by taking the peak from this spectrum. The propagation path above can provide an estimate of the ToF and propagation intensity.

7 illustrates a graph (a) showing spectrum data extracted according to CSI analysis using the MUSIC algorithm, and a diagram (b) for a distance value according to the spectrum data. In the graph (a), each channel is a boundary , and channel 1 is indicated by a bold line.

In addition, after spectral extraction, the dominant path can be identified using this. The dominant path can be identified using a CSI matrix, which has an attenuation for carrier n, θn value of the phase shift, so it is used to obtain FTM and angle of arrival (AoA) estimates using the first algorithm. will be used

Next, in step S140, an inaccurate return value of the FTM including the NLOS (Non Line of Sight) environment is detected through a second algorithm, and the error is corrected to calculate the first estimated distance value.

On the other hand, Wi-Fi Location™ of Wi-Fi Alliance® defines the FTM (Fine Timing Measurement) protocol according to IEEE 802.11-2016, and standardizes the provision of distance information with meter accuracy to provide the latest function was added. Accordingly, recently commercialized AP devices and mobile communication terminals support the corresponding function, and tests for interoperability authentication are continuously being performed. 8 illustrates the data flow of the FTM protocol, and in the embodiment of the present invention, the characteristics of the FTM are utilized for estimating the location of the target road.

In particular, according to an embodiment of the present invention, the FUSIC algorithm is used as the second algorithm, and a distance estimate thereof is used.

Specifically, the FUSIC algorithm, which is the second algorithm, may consist of logic that detects when the FTM returns an incorrect value by utilizing the results of the incorrect FTM and MUSIC in the NLOS environment, and corrects the error as necessary.

In other words, an approach that converges FTM and MUSIC is used for the purpose of extending the line of sight (LOS) accuracy of FTM to the NLOS environment. , an error correction mechanism that fuses data from FTM and MUSIC.

Here, it is required to detect when correction of FTM is required, and for this, a new parameter 'R' is introduced to quantify the contribution of the direct path to the entire MUSIC spectrum.

At this time, it is necessary to set a threshold value (R_threshold) of the parameter 'R', and it is required to select a correct value for the R threshold value. This can lead to unnecessary errors as the FUSIC algorithm changes the correct FTM measurement if the threshold is too low. Also, vice versa, it may not be possible to correct erroneous FTM measurements.

When the analysis procedure by the FUSIC algorithm is described in detail, the dominant path is first identified using the MUSIC algorithm. If this is expressed as an equation, it is as shown in Equation 1 below.

Here, P _k and t _k are signal intensity and ToF of path 'k', respectively.

Thereafter, the relative intensity with respect to other paths may be calculated by Equation 2 below.

), 현재 FTM 거리 추정치를 유지하고, 'R'이 'R 임계값'보다 작을 경우, 평균초과지연을 이하의 수학식 3에 의해 계산한다.When 'R' calculated according to Equation 2 above is equal to or greater than 'R threshold (R_threshold)' (

), the current FTM distance estimate is maintained, and when 'R' is less than the 'R threshold', the average excess delay is calculated by Equation 3 below.

Next, the corrected distance estimate is calculated by the following Equation (4).

Here, d _fusic is the distance estimate by the second algorithm, and d _ftm is the FTM distance estimate.

Summarizing the procedures for Equations 1 to 4 above, as a procedure for correcting the FTM output, it is assumed that there are only two propagation paths between transmission and reception, that is, if the LOS path is blocked by an obstacle and it is a reflected path, the FTM is a reflector. It will print the length of the row. Accordingly, the FTM error is expressed as the difference between the length of the output and the LOS path.

And, since the difference between the estimated ToFs for the two paths generated by the MUSIC algorithm is accurate, the FUSIC algorithm can calculate the FTM measurement error using the estimated ToF values of the propagation paths analyzed by the MUSIC algorithm.

In addition, the above-mentioned average excess delay is an FTM measurement error estimated in relation to the ToF, and may be calculated as a metric average of the ToF difference with respect to the ToF of the direct path.

Accordingly, the FUSIC algorithm finally removes the effect of the error and outputs the corrected distance estimate.

According to the above-described procedure, in the precise positioning method of the wireless communication system of the present invention, the first distance estimation value by the first and second algorithms can be calculated based on the CSI, hereinafter, the second distance by the third algorithm A method for calculating the estimated value will be described.

Specifically, applying the third algorithm to the AoA and a plurality of FTM samples estimated using CSI to correct the multipath error and to calculate a second estimated distance value between the AP and the target node (S200), 802.11- In commercial chipsets to which the 2016 and 802.11ac amendments are applied, the second estimated distance value is calculated through a third algorithm that corrects multipath errors through CSI for distance and direction targeting FTM and angle of arrival (AoA), This step S200 may be subdivided into steps S210 to S240 of FIG. 5 .

First, in step S2100, the first algorithm is applied to classify the CSI multi-signal to extract a pseudo spectrum, which corresponds to the step of extracting the CSI spectrum through the first algorithm used in step S100 described above.

Also, step S220 is a step of estimating an angle of arrival (AoA) based on the peak of the pseudo spectrum, and corresponds to the step of estimating the FTM and AoA based on the CSI in step S100 described above. In this case, the highest peak in the spectrum according to the MUSIC algorithm may be used to estimate the strongest path.

Next, in step S230, M (M is a natural number greater than or equal to 20) FTM sample data is collected from a plurality of links, the AoA phase relaxation and FTM error estimated using the third algorithm are calculated, and the calculation results are averaged. and Akaike Information Criterion (AIC) to determine the path with the least positive mean as the range estimate.

In detail, in order to collect the FTM measurement result, the receiver (FTM-R) transmits an FTM message to the transmitter (FTM-I) and waits for an ACK, and uses M=20 sample data for each link to detect the distance error. will be calculated Then, the calculation result is compared with the average and AIC. Here, the latter is commonly used to identify the optimal number of clusters in Gaussian Mixture Model (GMM), inferring the optimal number of paths using the lowest AIC and then taking the path with the least positive mean as the range estimate. use.

Here, the AIC allows the performance of several statistical models to be compared with each other, and the lower the value, the better the result can be obtained, and the AIC has a feature of selecting a model having a small over fit or under fit.

In addition, as the third algorithm, a Smartphone Positioning with Radio Measurements from a Single WiFi Access Point (SPRING) algorithm may be used.

The above AoA estimate is extracted from CSI extracted from commercial chipsets using the standard 802.11-2016 and 802.11ac amendments. will do

In addition, as FTM is a new feature of the 802.11 standard, no model to improve noise has been presented so far, so the FTM protocol is used to analyze the main factors and parameters affecting the range performance and measure the FTM. Make it possible to handle the city noise.

Next, in step S240, FTMs are further collected from two additional APs, and a multi-location calculation algorithm based on the least squares method is applied to the FTM estimates collected from three or more APs including the existing anchor to apply the second estimated distance value. to calculate

This is according to the characteristic that the third algorithm described above requires only information about one anchor for SPRING and that three FTM APs are required for multiple evaluation, and in the case of multiple evaluation, using the estimated distance from each FTM-R The initial position estimation is performed using a multilateration algorithm based on the least squares method (LLS).

According to the above-described procedure, in the precise positioning method of the wireless communication system of the present invention, a corrected distance estimate is output through reception of a corrected input from CSI by alleviating the influence of multipath in FTM measurement. This can be seen as a procedure corresponding to the second distance estimate in the entire pipeline.

Thereafter, a procedure of selecting any one of the first and second distance estimation values calculated through the steps S100 and S200 described above according to an embodiment of the present invention as the coordinates of the target node is performed.

In detail, the first and second estimated distance values are input to the learning model learned through a deep learning procedure using a plurality of learning data and environmental parameters according to the calculation of the above-described estimated distance calculation value to select one with high accuracy As a system (S300) for extracting and determining a distance estimate, a well-known deep learning technique, as an example, a reinforcement learning technique may be used.

According to the precision positioning method according to the embodiment of the present invention, situation-aware deep learning-based prediction collects various situations at the time of communication positioning as learning data so that a more correct distance estimation value can be selected by the method of reinforcement learning. do. This is to enable a more correct value to be selected from among the first and second distance estimation values derived by the transferred procedure.

Accordingly, in the present invention, by performing reinforcement learning by inputting a large amount of learning data including the first distance estimate value, the second distance estimate value, and the environment parameter to the learning model by performing a test in advance, the two distances input in the real environment later Among the estimated values, it is possible to select and extract a more accurate distance estimation value under the actual environmental parameter value.

Reinforcement learning, a field of machine learning, originates from behavioral psychology, and is known as a method of selecting an action or action sequence that maximizes reward among selectable actions by recognizing the current state of an agent defined in an environment. Because these problems are very comprehensive, they are used in research in the fields of game theory, control theory, operation science, information theory, simulation-based optimization, multi-agent systems, statistics, and genetic algorithms, and are also suitable for positioning estimation of the present invention.

In addition, a number of environmental parameters may be considered for distance estimation value prediction. These environmental parameters include the anchors used for measurement and the number of antennas of each anchor, the number of subcarriers transmitted by each anchor and capacity such as the frequency range, the number of anchors and NLOS anchors in the LOS environment, the number of antennas of the target node and environment variables, and geometrical location information of anchors, etc. may be included.

Next, with respect to any one distance estimate selected through machine learning, positioning may be supplemented through the fourth algorithm, Accurate and Distributed Localization (ADL) algorithm or Extended ADL (Extended ADL) algorithm.

As a procedure for this, the step (S400) of generating coordinates of the target node by applying the fourth algorithm to the distance estimate and the phase-relaxed AoA estimate for AoA (S400) includes: It may be a step of calculating the positioning using a predetermined anchor through the AoA estimation value that alleviates the phase error due to calibration, or calculating the positioning according to the movement of the target node with respect to one anchor.

When there are a plurality of anchors, the fourth algorithm calculates the position estimate of the target node by using angle and distance information of each anchor, receives the position estimate from all anchors, and calculates the position estimate of the target node. It may consist of calculating the coordinates of the target node using the average.

In detail, when there are a plurality of anchors, an estimate of the position of the target node is calculated using the angle and distance information of each anchor, and FIG. 9 illustrates a coordinate graph for calculating the coordinates of the target node from the four anchors. .

In the example of FIG. 9 , the location of the target node may be calculated by Equation 5 below.

,

는 각각 표적노드와 앵커간의 신호 각도를 가리킨다.Here, t _i,1 , t _i,2 are the estimated coordinates of the target coordinates, x _i,1 , x _i,2 are the actual coordinates of each anchor, d _i is the distance between the target node and each anchor,

,

denotes the signal angle between the target node and the anchor, respectively.

Next, the step of receiving the position estimates from all anchors and calculating the coordinates of the target node using the average of the position estimates of the target node satisfies Equation 6 below.

In addition, when the position of the target node is calculated using the average of the position estimates of all anchors, efficiency can be improved and accuracy can be guaranteed.

는 표적노드의 좌표 추정치, x₀는 표적노드의 좌표, m는 앵커의 개수, T_i는 각 앵커별 표적노드의 위치 추정치를 가리킨다.In Equations 6 and 7 above,

is the estimated coordinates of the target node, x ₀ is the coordinates of the target node, m is the number of anchors, and T _i is the estimated position of the target node for each anchor.

On the other hand, even when the number of anchors is less than three, positioning of the target node is possible. However, as the ADL algorithm performs positioning calculations using both angle and distance information, more accurate results can be expected in the same number of anchor situations.

In addition, in a poor environment where only one anchor can be used in consideration of the movement of the target node, the extended ADL algorithm can be used.

According to the extended ADL algorithm, it is assumed that the target node moves in the virtual coordinate system with the starting coordinate as the origin, and the anchor measures the direction in which the target node moves and transmits the measurement result to the target node.

10 is a coordinate graph for positioning in a single anchor environment based on two linear movements.

)하고 다시 특정거리에 도달시, 이하의 수학식 8에 기초하여 표적노드의 실제 좌표를 산출하는 단계를 포함할 수 있다.Referring to FIG. 10 , if there is only one anchor, the fourth algorithm may include calculating the coordinates of the target node based on two linear movements or based on triangular movement through the target node movement distance and rotation angle. In particular, in the step of calculating the coordinates of the target node based on two linear movements through the movement distance and rotation angle of the target node, it is assumed that the target node moves in a virtual coordinate system with the starting coordinate as the origin, and the anchor is Steps to measure the moving direction of the target node and transmit the measurement result to the target node, and while the target node is moving (ex. d ₄ meters), rotate to the right or left (

) and when the specific distance is reached again, calculating the actual coordinates of the target node based on Equation 8 below.

Here, x and y are the moving positions of the target node, θ is the rotation angle of the moving target node, and d is the moving distance of the target node.

11 is a coordinate graph for positioning in a single anchor environment in which a target node moves along an arbitrary triangle.

Referring to FIG. 11 , in the step of calculating the coordinates of the target node based on the movement of the triangle, it is assumed that the target node moves along an arbitrary triangle in a virtual coordinate system having the starting coordinate as the origin, and the anchor is the target. It may include measuring the moving direction of the node and transmitting the measurement result to the target node, and calculating by Equation 9 below when the origin is reached during the movement of the target node.

Here, x and y indicate the movement position of the target node.

Although many matters are specifically described in the above description, these should be construed as examples of preferred embodiments rather than limiting the scope of the invention. Accordingly, the invention should not be defined by the described embodiments, but should be defined by the claims and equivalents to the claims.

10: target node 50: AP
100: precision positioning device 110: first distance estimating means
111: CSI data extraction unit 112: spectrum extraction unit
113: dominant path calculation unit 114: first distance calculation unit
120: second distance estimation means 121: FTM collection unit
123: second distance calculation unit 125: positioning correction unit
130: deep learning prediction unit 140: target node output unit

Claims (16)

As a positioning method of a target node using a received signal according to wireless communication between a mobile target node and one or more AP (Access Point) anchors,
Collects CSI (Channel State Information) extracted from the received signal of the target node, identifies a propagation path by applying a first algorithm to the CSI, calculates FTM (Fine Time Measurement), and calculates the propagation path and FTM estimate. calculating a first estimated distance value between the target node and the anchor by applying a second algorithm;
correcting a multipath error by applying a third algorithm to the angle of arrival (AoA) and a plurality of FTM samples estimated using the CSI and calculating a second estimated distance value between the AP and a target node;
inputting the first and second estimated distance values to a learning model learned through a deep learning procedure using a plurality of learning data and environmental parameters, and selectively extracting one with high accuracy to determine a distance estimate; and
generating coordinates of the target node by applying a fourth algorithm to the distance estimate value and the phase-relaxed AoA estimate value for the AoA;
Precise positioning method of a wireless communication system comprising a. The method of claim 1,
Calculating the first estimated distance value comprises:
extracting the CSI from the PHY of the received signal by multiplying the number of transmitter antennas, the number of receiver antennas, and the number of subcarriers of the target node;
extracting a pseudo spectrum by classifying the CSI multi-signal by applying the first algorithm;
identifying the propagation path based on the peak of the pseudo spectrum, and extracting an estimate of Time of Flight (ToF) and radio wave intensity based on the propagation path; and
Detecting an incorrect return value of the FTM including a Non Line of Sight (NLOS) environment through the second algorithm, correcting an error, and calculating the first estimated distance value
Precise positioning method of a wireless communication system comprising a. 3. The method of claim 2,
Detecting the incorrect return value of the FTM including the NLOS (Non Line of Sight) environment through the second algorithm and correcting the error to calculate the first estimated distance value,
identifying a dominant path by applying a first algorithm to the CSI;
calculating a relative intensity with respect to another path by the following equation;
formula

(Where R is the parameter quantifying the contribution of the direct path to the CSI spectrum, K/k is the path, P _k is the signal intensity of the path k, t _k is the ToF)
maintaining the distance estimate according to the FTM when R is equal to or greater than a threshold value, and calculating an average excess delay by the following equation when R is less than a threshold value; and
formula

calculating the corrected distance estimate by the following equation
formula

(where d _fusic is the distance estimate by the second algorithm, and d _ftm is the FTM distance estimate)
Precise positioning method of a wireless communication system comprising a. The method of claim 1,
Calculating the second estimated distance value comprises:
extracting a pseudo spectrum by classifying the CSI multi-signal by applying the first algorithm;
estimating an angle of arrival (AoA) based on the peak of the pseudo spectrum;
M (M is a natural number greater than or equal to 20) FTM sample data is collected from a plurality of links, phase relaxation of the AoA and FTM error estimated using the third algorithm are calculated, and the calculation results are averaged and AIC (Akaike) Information Criterion) and determining the path with the least positive mean as the range estimate; and
Calculating the second estimated distance value by further collecting FTMs from two additional APs, and applying a multi-location calculation algorithm based on the least squares method to the FTM estimates collected from three or more APs including the existing anchor
Precise positioning method of a wireless communication system comprising a. The method of claim 1,
The environment parameter is
The number of antennas of the anchor and the additional AP, the number of subcarriers transmitted by the anchor and the capability of the frequency range, the number of anchors and NLOS anchors in the LOS environment, the number of antennas and environment variables of the target node, and geometric location information of the anchor of, one or more
A precise positioning method of a wireless communication system comprising a. The method of claim 1,
The fourth algorithm is
calculating a position estimate of the target node by using angle and distance information of each anchor when the anchors are plural; and
receiving the position estimates from all anchors, and calculating the coordinates of the target node using the average of the position estimates of the target node
Precise positioning method of a wireless communication system comprising a. 7. The method of claim 6,
If there are a plurality of anchors, calculating the position estimate of the target node using angle and distance information of each anchor comprises:
the following formula,

A precise positioning method of _a wireless communication system that calculates the position estimate _{of the} target node _by The actual coordinates of each anchor, d _i is the distance between the target node and each anchor,

,

is the signal angle between the target node and the anchor, respectively). 8. The method according to claim 6 or 7,
receiving the position estimates from all the anchors, and calculating the coordinates of the target node using the average of the position estimates of the target node,
the following formula,

A precise positioning method of a wireless communication system that calculates the position estimate of the target node by

is the estimated coordinates of the target node, x ₀ is the coordinates of the target node, m is the number of anchors, and T _i is the estimated position of the target node for each anchor). 6. The method of claim 5,
The fourth algorithm is
If there is only one anchor, calculating the coordinates of the target node based on two linear movements or based on triangular movement through the movement distance and rotation angle of the target node
Precise positioning method of a wireless communication system comprising a. 10. The method of claim 9,
Calculating the coordinates of the target node based on two linear movements through the target node movement distance and rotation angle comprises:
Assuming that the target node moves in a virtual coordinate system having a starting coordinate as an origin, measuring the moving direction of the anchor by the anchor and transmitting the measurement result to the target node; and
During the movement of the target node, when it rotates to the right or left and reaches a specific distance again, the following equation,

step calculated by
A precise positioning method of a wireless communication system comprising

is the rotation angle of the moving target node, and d is the moving distance of the target node). 10. The method of claim 9,
Calculating the coordinates of the target node based on the triangle movement comprises:
It is assumed that the target node moves along an arbitrary triangle in a virtual coordinate system having a starting coordinate as an origin, and the anchor measures the moving direction of the target node and transmits the measurement result to the target node; and
During the movement of the target node, when the origin is reached, the following equation,

step calculated by
A precise positioning method of a wireless communication system comprising a (where, x, y is the movement position of the target node). As a positioning device of a target node using a received signal according to wireless communication between a mobile target node and one or more AP (Access Point) anchors,
Collects CSI (Channel State Information) extracted from the received signal of the target node, identifies a propagation path by applying a first algorithm to the CSI, calculates FTM (Fine Time Measurement), and calculates the propagation path and FTM estimate. first distance estimation means for calculating a first estimated distance value between the target node and the anchor by applying a second algorithm;
Second distance estimation means for correcting multipath error by applying a third algorithm to the angle of arrival (AoA) and a plurality of FTM samples estimated using the CSI and calculating a second estimated distance value between the AP and the target node ;
a deep learning prediction unit that inputs the first and second estimated distance values to a learning model learned through a deep learning procedure using a plurality of learning data and environmental parameters, and selectively extracts one with high accuracy to determine a distance estimate; and
A target node output unit for generating coordinates of the target node by applying a fourth algorithm to the distance estimation value and the phase-relaxed AoA estimation value for the AoA
A precision positioning device of a wireless communication system comprising a. 13. The method of claim 12,
The first distance estimating means,
a CSI data extraction unit for extracting the CSI from the PHY of the received signal by multiplying the number of transmitter antennas, the number of receiver antennas, and the number of subcarriers of the target node;
a spectrum extractor for classifying the CSI multi-signal by applying the first algorithm to extract a pseudo spectrum;
a dominant path calculation unit for identifying the propagation path based on the peak of the pseudo spectrum and extracting estimated values of Time of Flight (ToF) and radio wave intensity based on the propagation path; and
A first distance calculator that detects an incorrect return value of the FTM including a non-line of sight (NLOS) environment through the second algorithm, corrects an error, and calculates the first estimated distance value
A precision positioning device of a wireless communication system comprising a. 14. The method of claim 13,
The second distance estimation means,
an FTM collection unit that collects M (M is a natural number greater than or equal to 20) FTM sample data from a plurality of links;
An Angle of Arrival (AoA) estimated based on the peak of the pseudo spectrum extracted by the first algorithm is received from the dominant path calculator, and the phase relaxation of the AoA estimated using the third algorithm and FTM error is received. a second distance calculation unit for calculating , and comparing the calculation result with an average and Akaike Information Criterion (AIC) to determine a path having a minimum positive average as a range estimate; and
FTMs are further collected from two additional APs, and the multi-position calculation algorithm based on the least squares method is applied to the FTM estimates collected from three or more APs, including the existing anchor, to calculate the positioning calculated by the second distance calculator. Positioning correction unit to generate a second distance estimate by correcting
A precision positioning device of a wireless communication system comprising a. 13. The method of claim 12,
The target node output unit,
A precision positioning device of a wireless communication system for calculating the coordinates of the target node by calculating the position estimate of the target node using the angle and distance information between the target node and a plurality of anchors, and calculating the average of the position estimates from all anchors. 13. The method of claim 12,
The target node output unit,
It is assumed that the target node moves along an arbitrary triangle in a virtual coordinate system having the starting coordinate as the origin, measures the moving direction of the target node, calculates the distance between the rotation point of the target node and the starting point, A precision positioning device for a wireless communication system for calculating the coordinates of the target node by measuring a rotation angle or a moving distance.

Images