Sensors

Abstract

The present invention comprises: a complex transfer function calculation unit (30) which calculates a complex transfer function, regarding signals transmitted from N transmitting elements, from reception signals received by M receiving elements; a biological component extraction unit (40) which extracts biological information that is a variable component in a predetermined space; a correlation matrix calculation unit (50) which calculates a target correlation matrix from the biological information; a noise information storage unit (60) which records a noise correlation matrix; a first person count information calculation unit (61) which calculates first person count information that is an assumed value of the number of people existing in the predetermined space on the basis of a threshold value calculated from the noise correlation matrix and the target correlation matrix; a MUSIC spectrum calculation unit (70) which estimates candidates for a position in a living body by using the target correlation matrix, and outputs a likelihood spectrum in which the living body is present at the position; and a second person count information calculation unit (80) which estimates second person count information or position, which is the number of living bodies, with higher accuracy from position information based on the likelihood spectrum including a plurality of position candidates.

Description

[Technical field]

[0001] The present disclosure relates to sensors. [Background technology]

[0002] Techniques have been developed for detecting a detection target by utilizing a signal transmitted wirelessly (see, for example, Patent Document 1).

[0003] Patent Document 1 discloses that the number and positions of people to be detected can be determined by analyzing the eigenvalues ​​of components including Doppler shifts by applying a Fourier transform to signals received wirelessly. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2015-117972 A [Patent Document 2] JP 2014-228291 A [Patent Document 3] Patent No. 5047002 [Patent Document 4] Patent No. 5025170 [Non-patent literature]

[0005] [Non-Patent Document 1] "Number of people estimation method based on MUSIC spectrum in bistatic MIMO radar", IEICE Technical Report, AP2019-105, pp.137-142 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the technique disclosed in Patent Document 1 has a problem in that when the number of people to be detected changes frequently, the accuracy of estimating the number of people and their positions deteriorates.

[0007] The present disclosure has been made in consideration of the above-mentioned circumstances, and aims to provide a living body number and living body position sensor that can estimate the position and number of living bodies over a wider range by using wireless signals. [Means for solving the problem]

[0008] In order to achieve the above object, a sensor according to one embodiment of the present invention is a device for estimating the number of living organisms, and includes a complex transfer function calculation unit that calculates a complex transfer function from a signal transmitted from a transmitter having N (N is a natural number of 2 or more) transmitting antenna elements to a predetermined space and received by M (M is a natural number of 2 or more) receiving antenna elements for a predetermined period of time, a fluctuation component extraction unit that extracts biological information that is a fluctuation component of the predetermined space, a correlation matrix calculation unit that calculates an object correlation matrix from the biological information extracted by the fluctuation component extraction unit, and a noise information storage unit that records a noise correlation matrix measured in the predetermined space. the number of people calculation unit calculating first number of people information which is an assumed value of the number of people present in the specified space based on a threshold calculated from the noise correlation matrix and the target correlation matrix; a likelihood spectrum calculation unit using the number of people information and the target correlation matrix to estimate position candidates for the living body by a specified position estimation method and output a likelihood spectrum that the living body is present at the position; and a second number of people information calculation unit estimating second number of people information or a position which is a number of living bodies with higher accuracy by a specified method from position information based on the likelihood spectrum which may include a plurality of position candidates. Effect of the Invention

[0009] According to the sensor of the present disclosure, the number of living people can be estimated more accurately by using wireless signals. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram showing a configuration of a sensor according to the first embodiment. [Diagram 2] FIG. 2 is a conceptual diagram showing radio wave propagation in the first embodiment. [Diagram 3]FIG. 3 is a conceptual diagram showing a measurement environment for a noise correlation matrix in the first embodiment. [Figure 4] FIG. 4 is a graph showing an example of an eigenvalue distribution of a target correlation matrix in the first embodiment. [Diagram 5] FIG. 5 is a block diagram showing a configuration of a second number-of-persons information calculation unit according to the first embodiment. As shown in FIG. [Figure 6] FIG. 6 is a conceptual diagram illustrating the operation of the peak search unit in the first embodiment. [Figure 7] FIG. 7 is a conceptual diagram illustrating the operation of the ratio test unit in the first embodiment. [Figure 8] FIG. 8 is a flowchart showing the process of the sensor according to the first embodiment. [Figure 9] FIG. 9 is a flowchart showing a process of storing a noise correlation matrix of a sensor according to the first embodiment. [Figure 10] FIG. 10 is a flowchart illustrating a first number-of-persons information calculation process of the sensor according to the first embodiment. [Figure 11] FIG. 11 is a flowchart illustrating the second number of people information calculation process of the sensor according to the first embodiment. [Figure 12] FIG. 12 is a graph showing an example of eigenvalue distribution of a target correlation matrix and a noise correlation matrix in the second embodiment. [Figure 13] FIG. 13 is a graph of the success rate of number of people estimation showing the experimental results in the second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] (Findings on which the present invention is based) Techniques have been developed for detecting a detection target by utilizing a signal transmitted wirelessly (see, for example, Patent Documents 1 to 4 and Non-Patent Document 1).

[0012] For example, Patent Document 1 discloses a technique for estimating the number and position of people to be detected by analyzing the eigenvalues ​​of components including Doppler shift using Fourier transform. Specifically, a Fourier transform is performed on the received signal, an autocorrelation matrix is ​​obtained for a waveform in which a specific frequency component is extracted, and the autocorrelation matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues. In general, each eigenvalue and eigenvector represents a propagation path of radio waves from a transmitting antenna to a receiving antenna, that is, one path. However, in the technique of Patent Document 1, components that do not include biological information are removed, so only paths corresponding to signals reflected by living bodies, their secondary reflections, and paths corresponding to noise appear in the eigenvalues ​​and eigenvectors. Here, the value of the eigenvalue corresponding to noise is smaller than the value of the eigenvalue corresponding to living bodies, so the number of living bodies can be estimated by counting the number of eigenvalues ​​that are larger than a predetermined threshold among the eigenvalues.

[0013] However, the technology disclosed in Patent Document 1 has a problem that when the target living body is far away or there are many living bodies, the difference between the eigenvalue corresponding to the living body and the eigenvalue corresponding to the noise is narrowed, and the accuracy of the number of people estimation decreases. This is because, in a situation where the Doppler effect is very weak, it becomes difficult to detect a weak signal that has a Doppler shift due to the influence of the internal noise of the receiver, interference waves coming from other than the detection target, and the presence of objects that generate a Doppler shift other than the detection target. In addition, since the living body to be measured has a certain size and the components of the living body are distributed across multiple eigenvalues, when the number of living bodies is large, it becomes difficult to completely separate the eigenvalues ​​of the living body, making it difficult to estimate the number of people.

[0014] Patent Document 2 and Non-Patent Document 1 disclose a technique for estimating the position of an object using a direction estimation algorithm such as the MUSIC (MUltiple SIgnal Classification) method. Specifically, the receiving station that receives a signal emitted by a transmitting station performs a Fourier transform on the received signal, finds an autocorrelation matrix for a waveform that extracts specific frequency components, and applies a direction estimation algorithm such as the MUSIC method. This enables highly accurate estimation of the direction of a living body. However, since the MUSIC method used in Patent Document 2 requires the number of living bodies to be detected in advance, it was found that the technique of Patent Document 2 requires a number of people to be estimated in advance. In addition, the technique of Non-Patent Document 1 requires the first number of people information, which is an assumed value of the number of people present in a space, to be appropriately set. However, the optimal value of the first number of people information changes depending on the environment and the number of people present in the space, and the method disclosed in Non-Patent Document 1 cannot track the optimal value.

[0015] Furthermore, for example, Patent Document 3 discloses a technology for estimating the number of incoming waves, i.e., the number of transmitters such as mobile phones, from the correlation between the eigenvectors of signals received by multiple antennas and the steering vectors of the range in which the radio waves may arrive.

[0016] Furthermore, for example, Patent Document 4 discloses a technology in which various numbers of incoming waves are assumed for signals received by multiple antennas, an evaluation function is calculated for each of them using a steering vector, and the number of incoming waves that maximizes the evaluation function is estimated as the true number of incoming waves.

[0017] However, it has been found that the techniques disclosed in Patent Documents 3 and 4 are techniques for estimating the number of transmitters that emit radio waves, and are not capable of estimating the number of living bodies.

[0018] Taking these factors into consideration, the inventors have come up with the invention of a sensor that can estimate a larger number of living organisms more accurately by using wireless signals, without requiring the target organisms to carry special equipment such as transmitters.

[0019] That is, a sensor according to one aspect of the present disclosure is a device for estimating the number of living organisms, and includes a complex transfer function calculation unit that calculates a complex transfer function from a signal transmitted from a transmitter having N (N is a natural number of 2 or more) transmitting antenna elements into a predetermined space and received by M (M is a natural number of 2 or more) receiving antenna elements for a predetermined period of time, a fluctuation component extraction unit that extracts biological information that is a fluctuation component of the predetermined space, a correlation matrix calculation unit that calculates an object correlation matrix from the biological information extracted by the biological component extraction unit, and a noise information calculation unit that records a noise correlation matrix measured in the predetermined space. the first number of people information calculation unit calculating first number of people information which is an assumed value of the number of people present in the specified space based on a threshold calculated from the noise correlation matrix and the target correlation matrix; a likelihood spectrum calculation unit using the person number information and the target correlation matrix to estimate position candidates for the living body by a specified position estimation method and output a likelihood spectrum that the living body is present at the position; and a second number of people information calculation unit estimating second number of people information or a position which is a number of living bodies with higher certainty by a specified method from the likelihood spectrum which may include a plurality of position candidates.

[0020] With this configuration, even if biological components are superimposed on eigenvalues ​​from which only noise should have been separated, the accuracy of number of people estimation can be improved by determining whether the spectral peak is a true peak from the shape of the likelihood spectrum.

[0021] First, when the signal-to-noise ratio is small, such as when the living body is far away, it becomes difficult to distinguish the eigenvalue corresponding to the living body from the eigenvalue corresponding to the noise by using a threshold value in the method using eigenvalues. Also, since the living body to be measured has a certain size and the components of the living body are distributed across multiple eigenvalues, when the number of living bodies increases, it becomes impossible to completely separate the eigenvalues ​​of the living body, making it difficult to estimate the number of people.

[0022] Therefore, in the present invention, the likelihood spectrum is calculated by assuming first number of people information, which is the number of people that can be located, obtained from the environment in which the sensor is installed and the number of antenna elements of the sensor. Therefore, the number of people actually present within the detection range may differ from the first number of people information used to calculate the likelihood spectrum.

[0023] A typical example of the likelihood spectrum is the MUSIC spectrum by the MUSIC method, and hereafter, the MUSIC spectrum will be described as a representative example. In the MUSIC method, when the first number of people information is fewer than the number of people actually present, the number of peaks in the MUSIC spectrum will be fewer than the number of people actually present, resulting in oversight. In addition, when the first number of people information is greater than the number of people actually present, the MUSIC spectrum will have peaks called virtual images that appear at positions where there are no living bodies, but the inventors have found that the virtual images have characteristics such as low peak heights and gentle slopes. Therefore, the inventors have found that it is possible to estimate the number of people with high accuracy by determining the virtual images of the MUSIC spectrum by a threshold value for the peak value, a ratio of each peak value, or a machine learning determination, and counting the number of true peaks.

[0024] The present disclosure may be realized not only as an apparatus, but also as an integrated circuit including a processing means included in such an apparatus, as a method having steps corresponding to the processing means constituting the apparatus, as a program for causing a computer to execute the steps, or as information, data, or signals indicating the program. These programs, information, data, and signals may be distributed via a recording medium such as a CD-ROM or a communication medium such as the Internet.

[0025] Hereinafter, the embodiments of the present disclosure will be described in detail with reference to the drawings. Note that the embodiments described below all show preferred specific examples of the present disclosure. The numerical values, shapes, materials, components, the arrangement and connection forms of the components, steps, and the order of steps shown in the following embodiments are merely examples and are not intended to limit the present disclosure. In addition, among the components in the following embodiments, components that are not described in the independent claims showing the highest concept of the present disclosure will be described as any components constituting a more preferred embodiment. In addition, in this specification and the drawings, components having substantially the same functional configuration will be given the same reference numerals to omit repeated explanations.

[0026] (Embodiment 1) Hereinafter, a method for estimating the number of people by the sensor 1 in the first embodiment will be described with reference to the drawings.

[0027] [Sensor 1 Configuration] Fig. 1 is a block diagram showing a configuration of a sensor 1 in embodiment 1. The sensor 1 shown in the figure includes a transmitter 10, a receiver 20, a complex transfer function calculation unit 30, a biological component extraction unit 40, a correlation matrix calculation unit 50, a noise information storage unit 60, a first number of people information calculation unit 61, a MUSIC spectrum calculation unit 70, and a second number of people information calculation unit 80.

[0028] [Transmitter 10] The transmitter 10 has a transmission antenna. Specifically, as shown in FIG. 1, the transmitter 10 is composed of a transmission section 11 and a transmission antenna section 12. The transmission antenna section 12 is a T It is composed of an array antenna of elements, for example a four-element patch array antenna with the array elements spaced half a wavelength apart.

[0029] The transmitting unit 11 generates a high-frequency signal used to estimate the presence / absence, location, and number of people of the living body 200. For example, the transmitting unit 11 generates a 2.4 GHz CW (Continuous Wave) and transmits the CW as a transmission wave from the transmitting antenna unit 12. Note that the signal to be transmitted is not limited to the CW, and may be a modulated signal such as OFDM (Orthogonal Frequency Division Multiplexing).

[0030] [Receiver 20] The receiver 20 includes a receiving antenna section 21 and a receiving section 22 .

[0031] The receiving antenna section 21 is M R The receiving antenna unit 21 is an array antenna of elements, for example a four-element patch array antenna with an array element spacing of half a wavelength. The receiving antenna unit 21 receives high-frequency signals with the array antenna.

[0032] The receiver 22 converts the high-frequency signal received by the receiving antenna unit 21 into a low-frequency signal that can be processed using, for example, a down converter. If the transmitter 10 transmits a modulated signal, the receiver 22 also demodulates the modulated signal. The receiver 22 transmits the converted low-frequency signal to the complex transfer function calculation unit 30.

[0033] In the present embodiment, the frequency used is 2.4 GHz, but any frequency such as 5 GHz or a millimeter wave band may be used.

[0034] [Complex transfer function calculation unit 30] The complex transfer function calculation unit 30 calculates a complex transfer function representing the propagation characteristics between the array antenna of the receiving antenna unit 21 and the transmitting antenna unit 12 of the transmitter 10 from a signal observed by the array antenna of the receiving antenna unit 21. More specifically, the complex transfer function calculation unit 30 calculates the M T The transmitting antenna elements and the receiving antenna section M RA complex transfer function expressing the propagation characteristics between the receiving antenna elements is calculated. The complex transfer function calculated by the complex transfer function calculation unit 30 may include reflected waves and scattered waves, which are signals that are part of the transmission waves transmitted from the transmitting antenna unit 12 reflected and scattered by the living body 200. The complex transfer function calculated by the complex transfer function calculation unit 30 also includes reflected waves that do not pass through the living body 200, such as direct waves from the transmitting antenna unit 12 and reflected waves from fixed objects. The amplitude and phase of the signals reflected or scattered by the living body 200, i.e., the reflected waves and scattered waves that pass through the living body 200, always fluctuate due to biological activities such as breathing and heartbeat of the living body 200, and noise contained in the space, the transmitter 10, and the receiver 20. Hereinafter, the noise contained in the space, the transmitter 10, and the receiver 20 will simply be called noise.

[0035] 2 shows a radio wave propagation model from the transmitting antenna unit 12 to the receiving antenna unit 21 assumed in this embodiment. There are K human bodies (K is an integer equal to or greater than 0) in the space, and L waves are reflected or scattered by the bodies and received by the receiving antenna. In other words, there are L propagation paths that pass through the bodies. Here, K is the number of people to be estimated, and L is an unknown quantity since it changes depending on the propagation environment.

[0036] In the configuration diagram of Figure 1, the transmitter and receiver are shown adjacent to each other, but in reality, they may be placed apart as shown in Figure 2. The transmitting antenna and the receiving antenna may be combined. The antenna may also be shared with the hardware of wireless devices such as Wi-Fi routers and child devices.

[0037] [Biomedicine Extraction Unit 40] The biological component extraction unit 40 extracts a fluctuation component, which is a time-varying component transmitted from the transmitting antenna unit 12, from a signal observed by the receiving array antenna of the receiving antenna unit 21. This fluctuation component may include a biological component, which is a signal component reflected or scattered by one or more living organisms 200, in addition to fluctuation due to noise. The biological component extraction unit 40 is also referred to as a fluctuation component extraction unit.

[0038] More specifically, the biocomponent extraction section 40 records the complex transfer functions calculated by the complex transfer function calculation section 30 in a time series in the order in which the signals were observed. Then, the biocomponent extraction section 40 extracts signals having specific fluctuation components from the changes in the complex transfer functions recorded in time series.

[0039] Here, the method of extracting the fluctuation components includes, for example, a method of extracting only a predetermined frequency component after conversion to the frequency domain by Fourier transform or the like, or a method of extracting by calculating the difference between complex transfer functions at two different times. By these methods, the components of the direct wave and the reflected wave that passes through a fixed object are removed, and only the biological component that passes through the living body 200 and noise remain. For example, a complex transfer function for 5 seconds is used to extract the components from 0.3 Hz to 3 Hz, and the fluctuation components including the respiratory component that exists even when the living body is stationary are extracted.

[0040] In this embodiment, the components between 0.3 Hz and 3 Hz are extracted as an example, but if it is desired to extract slower or faster movements, it goes without saying that the extracted frequency components can be changed accordingly.

[0041] In this embodiment, the transmitting antenna array is made up of M T The receiving array antenna is made up of M R Since there are multiple fluctuating components in the complex transfer functions corresponding to the transmitting and receiving array antennas, there are also multiple fluctuating components. Hereinafter, these are collectively referred to as a fluctuating component channel matrix F(f) with M rows and N columns, and expressed as (Equation 1).

[0042]

number

[0043] Each element F of the biological component complex transfer function matrix F(f) ij is each element of the above complex transfer function h ijThe biological component complex transfer function matrix F(f) is a function of frequency or a similar differential period, and includes information corresponding to a plurality of frequencies.

[0044] [Correlation matrix calculation unit 50] The correlation matrix calculation unit 50 rearranges the elements of the fluctuation component channel matrix, which is composed of M rows and N columns and calculated by the biological component extraction unit 40, to generate a fluctuation component channel vector F vec The method of rearrangement is, for example, as shown in (Formula 1), but the order does not matter as long as it is an operation to rearrange a matrix.

[0045]

number

[0046] Thereafter, the correlation matrix calculation unit 50 calculates a correlation matrix from the fluctuation component channel vector. More specifically, the correlation matrix calculation unit 50 calculates a correlation matrix R of the fluctuation component channel vector composed of a plurality of fluctuation components due to the living body 200 and noise according to (Equation 3).

[0047]

number

[0048] Here, E[ ] in (Equation 3) represents the averaging operation, and the operator H represents the complex conjugate transpose. Here, by averaging the above fluctuation component channel vector, which contains multiple frequency components in the correlation matrix calculation, in the frequency direction, it becomes possible to perform sensing that simultaneously uses the information contained in each frequency.

[0049] [Noise information storage unit 60] The noise information storage unit 60 stores the correlation matrix calculated by the correlation matrix calculation unit 50 under conditions corresponding to an unattended condition where no living body is present in the predetermined area to be measured. Hereinafter, the stored correlation matrix is ​​referred to as the noise correlation matrix R. Nis called the "power of the fluctuation component". FIG. 3 shows a schematic diagram of the propagation environment when no one is present. As shown in FIG. 3, only noise consisting of external noise and internal noise is received by the receiving antenna unit 21. As a method for determining whether or not the user is present, the user may input the result of judging whether or not the user is present, or information on a time period when it is known in advance that the user is present may be used. In addition, the fact that the power of the fluctuation component is larger when a living body is present than when the user is present may be used. Specifically, the power of the fluctuation component or the eigenvalues ​​of the correlation matrix may be recorded in chronological order, and the correlation matrix at the time when the power falls below a predetermined threshold or when it is the minimum within a predetermined period, such as one day, may be regarded as the correlation matrix when the user is present.

[0050] If it is difficult to measure an unattended state, measurement results in a state where there is relatively little body movement, such as during sleep, may be used instead. The power of the fluctuation component or the eigenvalue of the correlation matrix may be used as an indicator of whether or not it can be used instead.

[0051] Alternatively, even if it is not possible to measure the noise correlation matrix in the space where the sensor 1 is to be installed, the sensor 1 may be installed in an uninhabited space that simulates the space where the sensor 1 is to be installed, and the noise correlation matrix may be measured. In this case, it is desirable to reproduce as accurately as possible the distance between the transmitter 10 and receiver 20 of the sensor 1 and the positional relationship with metal walls or the like that reflect radio waves well.

[0052] The noise information storage unit 60 only needs to store noise information derived from received signals under conditions equivalent to an unmanned vehicle, and instead of storing the correlation matrix, it may store eigenvalues ​​that are the result of eigenvalue decomposition of the correlation matrix.

[0053] [First number of people information calculation unit 61] The first number of people information calculation unit 61 calculates the first number of people information to be used in the MUSIC spectrum calculation unit 70 by comparing the correlation matrix calculated by the correlation matrix calculation unit with the noise correlation matrix stored in the noise information storage unit. Here, the first number of people information is 1 or more, M R ×M THereinafter, the correlation matrix calculated by the correlation matrix calculation unit is called a target correlation matrix R in order to distinguish it from the noise correlation matrix. The first number of people information calculation unit 61 performs eigenvalue decomposition on the target correlation matrix and the noise correlation matrix to obtain their respective eigenvalues. The target correlation matrix is

number

number

[0054]

number

[0055] Note that the noise power may be calculated from the average or sum of the diagonal terms of the noise correlation matrix. Next, the first number of people information calculation unit 61 counts the number of eigenvalues ​​of the target correlation matrix that are equal to or greater than the threshold, which is the noise power or a value obtained by dividing the noise power by a predetermined numerical value. Figure 4 shows an example of the eigenvalues ​​and noise power of the target correlation matrix. The vertical axis of Figure 4 is the magnitude of the eigenvalue, and the horizontal axis is the index number when the eigenvalues ​​are sorted in descending order. The graph indicated by the circles in Figure 4 shows the eigenvalues ​​of the target correlation matrix sorted in descending order, and the threshold noise power P W and P W The horizontal line indicates the division by 4 because the receiving antenna unit 21 is composed of four elements, and the value can be converted to a value per element. W When using the above, the first number of people information is 4, and the threshold is P WWhen 4 / 4 is used, the first number of people information becomes 5. Note that a small number such as 1 or 2 may be added to the first number of people information obtained by the above procedure to obtain the final first number of people information.

[0056] Note that instead of using (Equation 4) as the sum, the sum of some eigenvalues ​​based on (Equation 4) may be used as the sum.

[0057] Normally, to calculate the MUSIC spectrum, the number of arriving waves, or in this embodiment the number of people, needs to be known, but the first number of people information calculation unit 61 calculates the first number of people information as a temporary number of people, making it possible to calculate the MUSIC spectrum.

[0058] [MUSIC spectrum calculation unit 70] The MUSIC spectrum calculation unit 70 corresponds to a likelihood spectrum calculation unit.

[0059] When the correlation matrix calculated by the correlation matrix calculation unit 50 is decomposed into eigenvalues,

number

[0060] Where:

number

number

number

number

[0061] In addition, the steering vector (direction vector) of the transmitting array antenna is

number

number

number

[0062] That is, the MUSIC spectrum calculation unit 70 calculates an evaluation function P music Calculate the spectrum of (θ).

[0063]

number

[0064] It should be noted that the MUSIC spectrum can also be substituted by spectra produced by the Beamformer method or the Capon method, but these have lower accuracy than the MUSIC method and cannot provide highly accurate estimations by themselves.

[0065] [Second number of people information calculation unit 80] The second number of people information calculation unit 80 calculates second number of people information, which is the number of living beings present in the target area, based on the MUSIC spectrum calculated by the MUSIC spectrum calculation unit 70. In principle, a peak for the correct number of people appears in the MUSIC spectrum calculated by inputting the correct number of people, but in this embodiment, the MUSIC spectrum is calculated using first number of people information that is greater than the correct number of people, so that a virtual image is generated that has a peak even though there is no living being. Therefore, the second number of people information calculation unit 80 distinguishes peaks that appear in the MUSIC spectrum that are not virtual images. The calculation of the second number of people information can be performed, for example, by using a ratio method for the peak value of the spectrum, by counting the number of blocks in which the area above a predetermined threshold value is continuous in the MUSIC spectrum, or by treating the MUSIC spectrum as an image and using machine learning such as a convolutional neural network. In this embodiment, a calculation method of the second number of people information using the ratio method will be described as an example.

[0066] 5 is a block diagram of a second number of people information calculation unit 80 that calculates second number of people information using the ratio method. The second number of people information calculation unit 80 includes a peak search unit 81, a false peak determination unit 82, a peak sorting unit 83, and a ratio test unit 84.

[0067] <Peak Search Section 81> The peak search unit 81 searches for a peak having a maximum value in the MUSIC spectrum. The set of peaks found here is defined as a first peak set. In order to exclude small peaks due to noise, it is desirable to limit the peak value to only those having a maximum value in a predetermined range x. FIG. 6 is a conceptual diagram showing the processing of the peak search unit 81 using a one-dimensional MUSIC spectrum 1000. In FIG. 6, there are four peaks 1001-A, 1001-B, 1001-C, and 1001-D, but there are three peaks 1001-A, 1001-B, and 1001-D that have a maximum value within a distance of 0.5 m from the peaks represented by 1002-A, 1002-B, 1002-C, and 1002-D. Therefore, the peak search unit 81 extracts the three peaks 1001-A, 1001-B, and 1001-D from the MUSIC spectrum 1000 and defines them as a first peak set.

[0068] <False peak determination unit 82> The false peak determination unit 82 calculates the y% value of the values ​​included in a predetermined x range surrounding the peak values ​​included in the first peak set, extracts peaks whose difference or ratio between the peak values ​​and the y% value is equal to or greater than a predetermined threshold z, and sets them as a second peak set. This makes it possible to exclude peaks that have gentle slopes from the peaks included in the first peak set. For example, when the predetermined range has a radius of 0.5 m, the y value is 70%, and the z value is 0.4 dB, only peaks that are 0.4 dB or more greater than the 70% value of the values ​​included in a surrounding radius of 0.5 m are extracted for each peak value included in the first peak set.

[0069] <Peak sorting section 83> The peak sorting unit 83 sorts the peak values ​​of the second peak set in descending order. The peak sorting unit may add a virtual peak to the second peak set that is smaller than the minimum peak value included in the second peak set by w. For example, when w is set to 3.4 dB and the minimum peak is -3 dB from the maximum peak, the virtual peak to be added is -6.4 dB from the maximum peak.

[0070] <Ratio Testing Department 84> The ratio test unit 84 estimates the second number of people information by calculating the ratio between adjacent peak values ​​for the second peak set sorted by the peak sort unit 83. More specifically, the ratio or difference between the i-th peak and the i+1-th peak of the second peak set sorted in descending order is calculated, and the i with the maximum ratio or difference is output as the second number of people information. Here, i is an integer of 1 or more and the number of elements of the second peak set or less. FIG. 7 is a conceptual diagram showing the processing of the ratio test unit 84. In FIG. 7, the elements 1101-A, 1101-B, 1101-C, and 1102 of the second peak set are sorted in descending order by peak value and entered. The virtual peak 1102 is a virtual peak added by the peak sort unit 83. The peak search unit 81 calculates the differences 1103-A, 1103-B, and 1103-C between adjacent peaks of the second peak set, and finds the combination of peaks with the maximum difference. In the example of FIG. 7, 1103-B is the maximum, and the difference between the second peak 1101-B and the third peak 1101-C is the maximum, so the calculated second number of people information is 2.

[0071] In this embodiment, since the purpose is to estimate the number of people, the second number of people information is the sensor output, but the location of the living body may be estimated using the MUSIC spectrum, and the location information of the living body may be the sensor output.

[0072] In this embodiment, an example of a MIMO (Multiple-Input Multiple-Output) configuration in which both the transmitting antenna and the receiving antenna are multiple has been described, but a configuration in which either the transmitting or receiving antenna is a single antenna may be used. In this case, the MUSIC spectrum output by the MUSIC spectrum calculation unit becomes one-dimensional, but even in this case, it is possible to estimate the second number of people information by peak search as in the two-dimensional case.

[0073] It is also possible to perform a judgment based on the magnitude of the maximum eigenvalue, the power of the fluctuating component of the complex transfer function, and the correlation value with the unoccupied state only when the room is occupied, and perform the MUSIC spectrum calculation by the MUSIC spectrum calculation unit 70. In this way, it is possible to omit the calculations required for the MUSIC spectrum calculation when the room is unoccupied.

[0074] [Sensor 1 operation] A process of estimating the number of living organisms by the sensor 1 configured as above will be described below. Fig. 8 is a flowchart showing the process of estimating the number of living organisms by the sensor 1 in the first embodiment.

[0075] First, the sensor 1 observes and records the noise correlation matrix as shown in FIG. 8 (S1). Details of S1 will be explained using the flowchart in FIG. 9. The receiver 20 observes the received signal for a predetermined period of time in the absence of a living body (S2). Next, the sensor 1 calculates a complex transfer function from the received signal (S3). Then, the sensor 1 records each of the calculated complex transfer functions in time series, and calculates a fluctuation component channel matrix by extracting fluctuation components from the recorded time series complex transfer functions (S4). Next, the sensor 1 calculates a correlation matrix of the extracted fluctuation component channel matrix (S5). Finally, the sensor 1 stores the correlation matrix calculated in S5 as a noise correlation matrix.

[0076] Returning to the flowchart of FIG. 8, the sensor 1 next observes the received signal in the receiver 20 for a predetermined period (S10). Next, the sensor 1 calculates a complex transfer function from the received signal (S20). Then, the sensor 1 records each of the calculated complex transfer functions in a time series, and calculates a fluctuation component channel matrix by extracting a fluctuation component from the recorded time series complex transfer function (S30). Next, the sensor 1 calculates a correlation matrix of the extracted fluctuation component channel matrix and sets it as a target correlation matrix (S40). Then, the sensor 1 calculates the first number of people information by a predetermined method (S50). The calculation method of the first number of people information will be described in detail below with reference to the flowchart of FIG. 10. First, the sensor 1 calculates the eigenvalues ​​of the noise correlation matrix and the target correlation matrix (S51). Next, the sensor 1 calculates the noise power from the eigenvalues ​​of the noise correlation matrix (S52). Finally, the sensor 1 counts the number of eigenvalues ​​of the target correlation matrix that are equal to or greater than the noise power, and sets it as the first number of people information (S53).

[0077] After that, the sensor 1 calculates the MUSIC spectrum based on the first number of people information calculated in S50 and the correlation matrix calculated in S40 (S60). Finally, the sensor 1 calculates the second number of people information from the MUSIC spectrum calculated in S60 and outputs it as the number of living organisms (S70). For example, S70 can be performed using a machine learning method such as a convolutional neural network that treats the MUSIC spectrum as an image, a method that counts the number of consecutive blocks with areas equal to or greater than a predetermined value in the MUSIC spectrum, or a method that uses a ratio method for the peak value of the spectrum.

[0078] Here, the method using the ratio method will be described with reference to the flowchart of FIG. 5. First, the sensor 1 extracts the peaks of the MUSIC spectrum that are the maximum values ​​in a predetermined range, and sets them as a first peak set (S71). Next, the sensor 1 calculates the y% value of the value included in a predetermined range around the peak for each peak included in the first peak set (S72). Then, the sensor 1 extracts peaks whose difference between the y% value and the peak value is equal to or greater than a predetermined threshold value, and sets them as a second peak set (S73). Next, the sensor 1 sorts the peak values ​​included in the second peak set in descending order (S74). Finally, the sensor 1 calculates the ratio or difference between the i-th peak and the i+1-th peak in the second peak set, and outputs the i with the maximum ratio or difference as the second number of people information. Here, i is an integer between 1 and the number of elements in the second peak set.

[0079] [Effects, etc.] According to the sensor 1 and the living organism number estimation method of the present embodiment, the number of living organisms present can be estimated with high accuracy using a wireless signal. Furthermore, according to the sensor 1 of the present embodiment, even when the number of living organisms is large and the noise and the living organism components cannot be sufficiently separated by eigenvalue decomposition, the location of the living organism can be estimated with high accuracy. More specifically, the existing number of people estimation method using eigenvalues ​​is based on the premise that the eigenvalues ​​corresponding to noise and the eigenvalues ​​corresponding to living organisms are clearly distinguished, but in reality, the living organism components are also slightly superimposed on the noise components, causing a decrease in accuracy of the existing living organism number estimation method. In the present invention, the spectrum is calculated by the MUSIC method using the above first number of people information, which may be greater than the actual number of people, and only the true peaks are extracted from the characteristics of the erroneous peaks of the spectrum, so that the number of people can be estimated even when the eigenvalues ​​of noise and the eigenvalues ​​of living organisms cannot be sufficiently separated.

[0080] (Embodiment 2) In the first embodiment, the first number of people information calculation unit 61 calculates the first number of people information using noise power calculated from the eigenvalue of the noise correlation matrix. In the second embodiment, a method of calculating the first number of people information using the maximum eigenvalue of the noise correlation matrix will be described. Note that the configuration other than the first number of people information calculation unit 61 is the same as in the first embodiment, so the description will be omitted here.

[0081] The first number of people information calculation unit 61 calculates the first number of people information to be used in the MUSIC spectrum calculation unit 70 by comparing the correlation matrix calculated by the correlation matrix calculation unit with the noise correlation matrix stored in the noise information storage unit. Here, the first number of people information is 1 or more,

[0082]

number

number

number

[0083] [Effects, etc.] By using the sensor of embodiment 2, the amount of calculations in the first number of people information calculation unit can be reduced compared to the sensor of embodiment 1, the processing device capacity required for real-time processing can be reduced, and number of people estimation can be achieved at low cost.

[0084] Here, a method for setting an appropriate threshold value and the estimation accuracy of the sensor of this embodiment will be described based on experimental results of the success rate of number of people estimation when the threshold value used in the first number of people information calculation unit is changed.

[0085] First, the conditions of this experiment are shown. The transmitting and receiving antennas used were 4-element patch arrays with a center frequency of 2.47125 GHz, and CW of the same frequency was used for the transmitting wave. The transmitting and receiving antennas were placed 4 m apart in a 4 m square area, and the fluctuating component channel matrix was measured. In this experiment, measurements were performed in an unmanned state to measure the noise correlation matrix, and in a state with one, two, and three subjects to estimate the number of people. From these measurement results, the success rate of the number of people estimation was obtained when the value β added to the above threshold λN1 was changed in various ways. The results are shown in Figure 13. The horizontal axis of Figure 13 is the correction value to be added, and the vertical axis shows the success rate of the number of people estimation when that β was used. From these results, the best accuracy was obtained when β was set to about 5 dB, and the success rate of the number of people estimation in that case was about 70%. Note that the results of this experiment are only an example, and the optimal value may change depending on the number of antenna elements used, etc.

[0086] Although one or more aspects have been described based on the embodiment, the present invention is not limited to this embodiment. As long as it does not deviate from the spirit of the present invention, various modifications conceived by a person skilled in the art to this embodiment and forms constructed by combining components in different embodiments may also be included within the scope of one or more aspects. [Industrial Applicability]

[0087] The present invention can be applied to measuring devices that measure the number and positions of living bodies, home appliances that perform control according to the number and positions of living bodies, and monitoring devices that detect the intrusion of living bodies. [Explanation of symbols]

[0088] 1 Sensor 10 Transmitter 11 Transmitter 12 Transmitting antenna section 20 Receiver 21 Receiving antenna section 22 Receiving section 30 Complex transfer function calculation section 40 Biological component extraction section 50 Correlation matrix calculation section 60 Noise information storage unit 61 First People Information Calculation Department 70 MUSIC spectrum calculation section 80 Second people information calculation section 81 Peak Search Section 82 False peak judgement section 83 Peak sorting section 84 Proportion Test Section 200, 200-1, 200-K Living body 300-1, 300-L propagation path 1000, 2100 MUSIC Spectrum 1001-A, 1001-B, 1001-C, 1001-D MUSIC spectrum peaks 1002-A, 1002-B, 1002-C, 1002-D Predefined areas around each peak 1101-A, 1101-B, 1101-C Peaks included in the second peak set 1102 Virtual Peak 1103-A, 1103-B, 1103-C Difference between adjacent peaks

Claims

1. A sensor for estimating a live population, comprising: A signal transmitted from a transmitter having N (N is a natural number equal to or greater than 2) transmitting antenna elements to a predetermined space is expressed as: a complex transfer function calculation unit that calculates a complex transfer function from received signals received by M (M is a natural number equal to or greater than 2) receiving antenna elements for a predetermined period of time; a variation component extraction unit that extracts biometric information that is a variation component in the predetermined space; a correlation matrix calculation unit that calculates a target correlation matrix from the biological information extracted by the fluctuation component extraction unit; A noise information storage unit that records a noise correlation matrix measured by receiving a second signal transmitted by the transmitter to the specified space with the receiving antenna element under a condition corresponding to when the specified space is unoccupied; A first number of people information calculation unit that calculates first number of people information, which is an assumed value of the number of people present in the predetermined space, based on a threshold value calculated from the noise correlation matrix and the target correlation matrix; a likelihood spectrum calculation unit that estimates a position candidate of the living body by a predetermined position estimation method using the first number of people information and the object correlation matrix, and outputs a likelihood spectrum that the living body is present at the position; and a second number-of-people information calculation unit that estimates second number-of-people information, which is the number of living bodies, or the positions of the living bodies, by using a ratio or a difference between adjacent peak values ​​when peaks included in a peak set of the likelihood spectrum, which may include a plurality of position candidates, are sorted in descending order. Sensor.

2. The sensor according to claim 1 , wherein the first number of people information calculation unit calculates the first number of people information based on eigenvalues ​​of the target correlation matrix and eigenvalues ​​of the noise correlation matrix.

3. The sensor according to claim 2 , wherein the first number of people information calculation unit outputs, as first number of people information, a number of eigenvalues ​​obtained by performing eigenvalue decomposition on the target correlation matrix having a value greater than the threshold.

4. The sensor according to claim 1 , wherein the first number of persons information calculation unit uses a maximum value or a sum of eigenvalues ​​of the noise correlation matrix as a threshold value.

5. The sensor according to any one of claims 1 to 3, wherein the threshold value used by the first number of people information calculation unit is a value obtained by increasing or decreasing the maximum value or the sum of the eigenvalues ​​of the noise correlation matrix by a predetermined ratio.

6. The threshold value used by the first number of people information calculation unit is a value obtained by adding or subtracting a predetermined value from the maximum value or the sum of the eigenvalues ​​of the noise correlation matrix. The sensor according to any one of claims 1 to 3.

Images