JPH0882662A - Direction detection apparatus - Google Patents

Abstract

PURPOSE: To obtain a direction detection apparatus by which the angle of a wave source can be estimated with high accuracy even when an error exists in an array manifold by a method wherein complex random numbers generated by a complex- random-numbers generator and signals from respective antennas are multipled by a multiplier and angle estimated values of the wave source are averaged by an average-arrival-time computation device. CONSTITUTION: Signals which are received respectively by M-pieces of antennas 1 are amplified and frequency-converted by respective receivers 2, and the signals and complex random numbers generated by a complex-random-numbers generator 9 are multiplied by a multiplier 3. On the basis of its output, a covariance matrix is found by a covariance-matrix computation device 4, and an eigenvalue and an eigenvector are found by an eigenvalue and eigenvector computation device 5. A mode-vector generator 10 finds an antenna output with reference to a virtual wave source, and a minimum-residual-matrix computation device 6 finds the Frobenius norm minimum on the basis of the eigenvector and a mode vector. An arrival-angle computation device 7 scans arrival angles of the signals, it finds the arrival angle which makes the minimum value of the Frobenius norm minimum, an average-arrival-angle computation device 8 computes an average on the basis of the value of the arrival angle at which a complex variable has been updated and repeated a plurality of times.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a direction detecting device for detecting the direction of a radio wave source or a sound source.

[0002]

2. Description of the Related Art Conventionally, an apparatus of this kind has been shown in FIG. This figure shows Ralph O. Schmid
T., “Multiple Emitter Loca”
tion and Signal Parameter
"Estimation" IEEE Trans. o
n Antena and Propagatio
n, vol. AP-34, No. 3, Marc
It is a block diagram of the direction detection apparatus based on the content described in h1986.

In FIG. 4, reference numeral 1 denotes an antenna for receiving signals from wave sources such as radio waves and sound waves, and a plurality of antennas are prepared. 2 is a receiver provided corresponding to each antenna, which amplifies the signal received by the antenna 1 and frequency-converts, 4 simultaneously samples the outputs of the M receivers, and snapshots multiple times. A covariance matrix calculator that calculates a covariance matrix, 5 is an eigenvalue from the covariance matrix, and an eigenvalue / eigenvector calculator that calculates an eigenvector. Vector generator, 20 is the eigenvalue
The mode vector generator 10 is added to the noise subspace based on the eigenvector obtained by the eigenvector calculator 5.
The projection of the mode vector generated in
A noise subspace projection length calculator for calculating the power, 21 scans the virtual wave source in space, and calculates the position of the virtual wave source when the projection length output from the noise subspace projection length calculator 20 becomes a minimum. It is a minimum value calculator.

Next, the operation will be described. Number of wave sources D
Is smaller than the number M of antennas 1, and the signals from the respective wave sources are uncorrelated with each other. Now, the outputs of the M antennas 1 are input to the receiver 2, respectively, and are amplified and frequency-converted.

The outputs of the M receivers 2 are input to the covariance matrix calculator 4, respectively. Here, when the outputs of the M receivers 2 are S _A , S _B , ..., S _M , the signal vector is expressed by equation (1).

[0006]

[Equation 1]

If P times of snapshots are performed in obtaining the covariance matrix, the covariance matrix calculator 4 calculates the value shown in the equation (2).

[0008]

[Equation 2]

Equation (2) is the output of the covariance matrix calculator 4. Here, * represents conjugate or conjugate transposition.

The eigenvalue / eigenvector calculator 5 calculates M eigenvalues on the basis of the obtained covariance matrix, calculates eigenvectors corresponding to the respective eigenvalues, and calculates the noise subspace projection length calculation results. Input to the container 20. When the above eigenvalues are λ ₁ , λ ₂ , ..., λ _M and the corresponding eigenvectors are X ₁ , X ₂ , ..., X _M , the following equation is given by equation (2).

[0011]

(Equation 3)

Generally, the covariance matrix is a positive semidefinite matrix, so that all eigenvalues are zero or more. The receiver noises of the receiver 2 are assumed to be all equal, and if this is σ ² , the following equation holds from the above assumption that the D wave sources are uncorrelated.

[0013]

[Equation 4]

The eigenvectors corresponding to the eigenvalues λ ₁ , λ ₂ , ..., λ _D are X ₁ , X ₂ , ..., X _D, and λ _{D + 1} , λ
_The eigenvectors corresponding to _{D + 2} , ..., λ _M are X _{D + 1} , X
_{D + 2,} ..., if _{X M X 1, X 2,} ..., signal subspace spanned by the X _D and _{X D + 1, X D +} 2, ..., noise subspace spanned by X _M are each It becomes an orthogonal complementary space.

Now, the mode vector generator 10 stores the data of the output of the M antennas, that is, the mode vector when it is assumed that the wave source exists at an angle θ when the M antennas 1 are arranged. However, this angle typically spans a range.

Therefore, a mode vector at a certain angle θ is generated by the mode vector generator 10 and input to the noise subspace projection length calculator 20. On the other hand, in the noise subspace projection length calculator 20, the noise subspace is set based on the input eigenvalue and eigenvector, and the mode vector is projected onto the noise subspace. If this mode vector is a (θ), the square of the projection length given by the following equation is output from the noise subspace projection length calculator 20.

[0017]

(Equation 5)

In the minimum value calculator 21, the mode vector a
The square of the projection length of (θ) is obtained as a function of θ, and θ values θ ₁ , θ ₂ , ..., θ _D that give the minimum value are obtained.

Θ ₁ , θ ₂ , ..., θ _D are estimated values of angles at which D wave sources are present.

[0020]

Since the conventional direction finding device is constructed as described above, the antenna output stored in the mode vector generator 10 is changed by the change of the electromagnetic environment caused when the antenna is mounted. The output is different from the actual output, and as a result, the estimated value of the angle of the wave source deviates more than the true value.

The present invention has been made to solve the above problems, and an object thereof is to improve the accuracy of the estimated angle value of the wave source.

[0022]

A first embodiment of the present invention includes M antennas, M receivers connected to the respective antennas, a complex random number generator for generating a complex random number, and M multipliers capable of multiplying the output of the receiver and complex random numbers, a covariance matrix calculator that calculates the covariance matrix from the outputs of each multiplier, and an eigenvalue that calculates the eigenvalue and eigenvector from the covariance matrix.・ Eigenvector calculator, a mode vector generator that generates the expected antenna output when the wave source is assumed, that is, a mode vector, and the residual when minimizing the Frobenius norm of the residual matrix obtained from the eigenvector and the mode vector spanning the signal space A minimum residual matrix calculator that calculates the matrix and scans the mode vector within a predetermined angle range to minimize the minimum Frobenius norm of the residual matrix. Angle of arrival, that is, the angle of arrival calculator that calculates the angle of arrival, and the complex random number generated by the complex random number generator is changed to another complex random number, and the same procedure is repeated L times to obtain L arrival angles and averaged. It is provided with an average arrival angle calculator for calculating a value.

The second embodiment of the present invention is the first embodiment of the present invention.
A cumulative frequency distribution calculator that calculates the cumulative frequency distribution of the arrival angles of the wave sources instead of the average arrival angle calculator in the embodiment of
It is provided with a median calculator for calculating the median based on the obtained cumulative frequency distribution.

The third embodiment of the present invention is the first embodiment of the present invention.
In this embodiment, the order of the receiver and the multiplier is exchanged.

The fourth embodiment of the present invention is the third embodiment of the present invention.
A cumulative frequency distribution calculator that calculates the cumulative frequency distribution of the arrival angles of the wave sources instead of the average arrival angle calculator in the embodiment of
It is provided with a median calculator for calculating the median based on the obtained cumulative frequency distribution.

[0026]

In the first embodiment of the present invention, the complex random number generated by the complex random number generator and the signal from each antenna are multiplied by the multiplier, and the estimated angle of arrival of the wave source is calculated by the average arrival angle calculator. Is averaged, it is possible to estimate the angle of the wave source with high accuracy even when the array manifold has an error.

In the second embodiment of the present invention, since the median of the estimated angle value of the wave source is obtained by using the cumulative frequency distribution calculator and the median calculator, the average arrival angle calculation is performed depending on the characteristics of the estimated angle value. The angle of the wave source can be estimated with higher accuracy than in the case of using a detector.

In the third embodiment of the present invention, a multiplier is placed in front of the receiver as compared with the first embodiment of the present invention, and the complex random number generated by the complex random number generator is transmitted from the antenna. Since it is multiplied by the signal of 1), it can be multiplied independently of the receiver noise. Therefore, the angle estimation of the wave source can be obtained with higher accuracy than in the case of the first embodiment of the present invention.

In the fourth embodiment of the present invention, as compared with the third embodiment of the present invention, the median of the estimated angle value of the wave source is obtained by using the cumulative frequency distribution calculator and the median calculator. Depending on the characteristics of the estimated angle value, the angle of the wave source can be estimated with higher accuracy than in the case of using the average arrival angle calculator.

[0030]

【Example】

First Embodiment Next, a first embodiment of the present invention will be described with reference to FIG.
In FIG. 1, 1 is M antennas, 2 is a receiver that amplifies and frequency converts signals received by M antennas,
9 is a complex random number generator for generating complex random numbers, 3 is a receiver 2
M multipliers for multiplying the signal output from M and the complex random number generated by the complex random number generator 9, and 4 is a covariance matrix calculation for calculating a covariance matrix based on the outputs of the M multipliers. And 5 are eigenvalue / eigenvector calculators that calculate eigenvalues and eigenvectors based on the covariance matrix output from the covariance matrix calculator 4, and 10 are virtual based on the arrayed M antennas 1. A mode vector generator for calculating the antenna output for the wave source, and 6 is a residue obtained from the eigenvector selected based on the eigenvalue output from the eigenvalue / eigenvector calculator 5 and the mode vector output from the mode vector generator 10. A minimum residual matrix calculator for calculating a residual matrix when the Frobenius norm of the matrix is minimized, and 7 scans the arrival angle of the signal from the virtual wave source to calculate the residual matrix. AoA calculator for calculating an arrival angle of a minimum of Niusunorumu minimized, 8 complex random number generator 9
It is an average arrival angle calculator that calculates the average from a plurality of arrival angles obtained by repeating the process from the multiplier 3 to the arrival angle calculator 7 a plurality of times by updating the complex random number output from

Next, the operation will be described. Number of wave sources D
Is smaller than the number M of antennas 1, and the signals from the respective wave sources are uncorrelated with each other. The outputs from the M antennas 1 are input to the receiver 2 for amplification and frequency conversion.

The outputs of the M receivers 2 are input to the M multipliers. On the other hand, the complex random number generator 9 outputs M complex random numbers represented by the following equation and inputs them to the multiplier 3.

[0033]

(Equation 6)

However, the average value and variance of the random numbers α _i and β _i are represented by the following equations, respectively.

[0035]

(Equation 7)

The multiplier 3 multiplies the output of the receiver by a complex random number, and the result Q _i (i = A, B, ... M) is input to the covariance matrix calculator 4. These inputs are shown as a matrix in the following equation.

[0037]

[Equation 8]

The covariance matrix calculator 4 calculates the covariance matrix shown in the following formula for P snapshots based on the vector Q of formula (11) and inputs it to the eigenvalue / eigenvector calculator 5. To be done.

[0039]

[Equation 9]

In the eigenvalue / eigenvector calculator 5, the equation (1
Based on the covariance matrix shown in 2), M eigenvalues are obtained, eigenvectors are calculated corresponding to the respective eigenvalues, and these eigenvalues and eigenvectors are input to the minimum residual matrix calculator 6.

If the above eigenvalues are μ ₁ , μ ₂ , ..., μ _M and the corresponding eigenvectors are Y ₁ , Y ₂ , ..., Y _M , then equation (12) is expressed by the following equation.

[0042]

[Equation 10]

If the receiver noises of the receiver 2 are all equal, and if this is σ ² , these eigenvalues satisfy the following equation from the above-mentioned assumption that the D wave sources are uncorrelated.

[0044]

[Equation 11]

The eigenvectors corresponding to the eigenvalues μ ₁ , μ ₂ , ..., μ _D are Y ₁ , Y ₂ , ..., Y _D, and μ _{D + 1} , μ
_The eigenvectors corresponding to _{D + 2} , ..., μ _M are Y _{D + 1} , Y
_{If D + 2} , ..., Y _M , the signal subspace spanned by Y ₁ , Y ₂ , ..., Y _D and the noise subspace spanned by Y _{D + 1} , Y _{D + 2} , ..., Y _M are mutually It becomes an orthogonal complementary space.

Now, the eigenvectors Y ₁ , Y ₂ , ..., Y _D
Is given by the following equation.

[0047]

[Equation 12]

At this time, the minimum residual matrix calculator 6 satisfies the relational expression (16) between the vector Y and the mode vector a (θ) output from the mode vector generator 10, and the expression (17) is minimized. Δ vectors Y and Δa are calculated. These are given by equation (18). Here, the vector X is the weight of the column vector of (vector Y + Δ vector Y), and ‖ · ‖ _F represents the Frobenius norm of the matrix. Formula (16), formula (17), and formula (18) are shown below.

[0049]

[Equation 13]

In the equation (18), Z ₁₂ and Z ₂₂ are defined by the following equation.

[0051]

[Equation 14]

However, it is assumed that the following expression is satisfied.

[0053]

(Equation 15)

In the arrival angle calculator 7, since the mode vector a (θ) output from the mode vector generator 10 is a function of θ, the equation (18) for the mode vector when scanning θ over a predetermined angle range is given. The minimum value of is obtained, and the angle ^ θ corresponding to this minimum value is calculated.

Next, the complex random number output from the complex random number generator 9 is updated to another complex random number, and the process from the multiplier 3 to the arrival angle calculator 7 is repeated L times, and the arrival angle calculator 7 ^ Θ ₁ , ^ θ ₂ , ..., ^ θ _L are input to the average arrival angle calculator 8.

The average arrival angle calculator 8 calculates the average arrival angle ^ θ _AVE according to the following equation.

[0057]

[Equation 16]

Second Embodiment Next, a second embodiment of the present invention will be described with reference to FIG.
A cumulative frequency distribution calculator 11 and a median calculator 12 are provided instead of the average arrival angle calculator 8 described in the first embodiment of the present invention.
The cumulative frequency distribution is calculated from the plurality of estimated arrival angles input to the cumulative frequency distribution calculator 11.

The median calculator 1 calculates the obtained cumulative frequency distribution.
2 is input to obtain an arrival angle corresponding to ½ of the total number of estimated arrival angles.

Third Embodiment Next, a third embodiment of the present invention will be described with reference to FIG.
3 has almost the same configuration as that of FIG. 1 except that the multiplier 3 is placed in front of the receiver 2.

The outputs of the M antennas 1 are input to the multiplier 3, respectively. On the other hand, the M complex random numbers output from the complex random number generator 9 are input to the multiplier 3 and are multiplied by the output from the antenna 1. The multiplication result is output from the multiplier 3 and input to the receiver 2. The other operation is the same as that of the first embodiment of the present invention.

Fourth Embodiment Next, a fourth embodiment of the present invention will be described. A cumulative frequency distribution calculator 11 and a median calculator 12 are used instead of the average arrival angle calculator 8 described in the third embodiment of the present invention, and a plurality of estimated arrivals input to the cumulative frequency distribution calculator 11 are used. The cumulative frequency distribution is obtained from the angles, and the obtained cumulative frequency distribution is input to the median calculator 12 to find the arrival angle corresponding to 1/2 of the total number of estimated arrival angles.

[0063]

As described above, according to the first embodiment of the present invention, the estimation angle accuracy of the wave source is deteriorated due to the change of the mode vector due to the change of the electromagnetic environment caused when the antenna is mounted. Can be made smaller.

Further, according to the second embodiment of the present invention, since the average arrival angle calculator in the first embodiment of the present invention is replaced with the cumulative frequency distribution calculator and the median calculator, the cumulative frequency distribution calculation is performed. The estimated angle of the wave source can be obtained with higher accuracy than the first embodiment of the present invention, depending on the statistical characteristics of the arrival angle input to the vessel.

Further, according to the third embodiment of the present invention, by placing the multiplier in front of the receiver, the signal output from the antenna and the complex random number output from the complex random number generator can be multiplied. Therefore, as compared with the case where the multiplier is placed after the receiver, only the signal and the complex random number can be multiplied without the influence of the receiver noise. Therefore, the angle of the wave source can be estimated with higher accuracy than that of the first embodiment of the present invention. be able to.

According to the fourth embodiment of the present invention,
Since the average arrival angle calculator in the third embodiment of the present invention is replaced with the cumulative frequency distribution calculator and the median calculator, the present invention may be changed depending on the statistical characteristics of the arrival angle input to the cumulative frequency distribution calculator. The estimated angle of the wave source can be obtained with higher accuracy than in the third embodiment.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a direction finding device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing configurations of a cumulative frequency distribution calculator and a median calculator in the direction finding device according to the second embodiment of the present invention.

FIG. 3 is a diagram showing a configuration of a direction finding device according to a third embodiment of the present invention.

FIG. 4 is a diagram showing a configuration of a conventional direction finding device.

[Explanation of symbols]

 1 antenna 2 receiver 3 multiplier 4 covariance matrix calculator 5 eigenvalue / eigenvector calculator 6 minimum residual matrix calculator 7 arrival angle calculator 8 average arrival angle calculator 9 complex random number generator 10 mode vector generator 11 cumulative frequency Distribution calculator 12 Median calculator 20 Noise subspace projection length calculator 21 Minimum value calculator

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Moto Yanagisawa 325 Kamimachiya, Kamakura City Mitsubishi Electric Corporation Kamakura Factory

Claims (4)

[Claims] 1. A plurality of antennas, a plurality of receivers respectively connected to the antennas, a complex random number generator for outputting a complex random number by a number corresponding to the number of the receivers, and a receiver corresponding to the receivers. A multiplier that multiplies the output of the receiver with the output of the complex random number generator, a covariance matrix calculator that calculates the covariance matrix from the outputs of the above multipliers, and an eigenvalue and eigenvector from the covariance matrix An eigenvalue / eigenvector calculator that calculates, a mode vector generator that calculates the antenna output for the virtual wave source based on the array of the antenna, an eigenvector selected from the eigenvector based on the eigenvalue, and the mode A minimum residual matrix calculator that calculates the minimum value of the Frobenius norm of the residual matrix from the antenna output from the vector generator,
An arrival angle calculator that scans the arrival angle of the signal from the virtual wave source to calculate an arrival angle that minimizes the minimum value of the Frobenius norm, and outputs the complex random number generator to update the output from the multiplier. A direction finding device comprising: an average arrival angle calculator that repeats the process up to the arrival angle calculator a plurality of times and obtains the average of the arrival angles from the obtained plurality of arrival angles. 2. A plurality of antennas, a plurality of receivers respectively connected to the antennas, a complex random number generator for outputting a complex random number in a number corresponding to the number of the receivers, and a receiver corresponding to the receivers. A multiplier that multiplies the output of the receiver and the output of the complex random number generator, a covariance matrix calculator that calculates the covariance matrix from the outputs of the above multipliers, and an eigenvalue and eigenvector from the above covariance matrix An eigenvalue / eigenvector calculator, a mode vector generator that calculates the antenna output for the virtual wave source based on the array of the above antennas, an eigenvector selected from the above eigenvectors based on the above eigenvalues, and the above mode vector generation Minimum residual matrix calculator that calculates the minimum value of the Frobenius norm of the residual matrix from the antenna output from the detector, and the signal from the virtual source The arrival angle calculator that scans the arrival angle of to calculate the arrival angle that minimizes the minimum value of the Frobenius norm, and updates the output of the complex random number generator from the multiplier to the arrival angle calculator. The process is repeated multiple times, and the cumulative frequency distribution calculator that calculates the cumulative frequency distribution of the arrival angle from the obtained plural arrival angles and the median based on the obtained cumulative frequency distribution are calculated. A direction finding device comprising a median calculator for calculating. 3. A plurality of antennas, a complex random number generator that outputs a complex random number in a number corresponding to the number of the antennas, and a multiplier that multiplies the antenna output and the output of the complex random number generator corresponding to the antennas. , A plurality of receivers connected to the multiplier, a covariance matrix calculator that calculates a covariance matrix from the output of the receiver, and an eigenvalue / eigenvector calculator that calculates an eigenvalue / eigenvector from the covariance matrix And a mode vector generator for calculating the antenna output for the virtual wave source based on the array of the antennas, and an eigenvector selected based on the eigenvalue from the eigenvector and the antenna output from the mode vector generator. A minimum residual matrix calculator that calculates the minimum value of the Frobenius norm of the residual matrix, and scans the arrival angle of the signal from the virtual source to The arrival angle calculator that calculates the arrival angle that minimizes the minimum value of the Frobenius norm and the output of the complex random number generator are updated, and the process from the multiplier to the arrival angle calculator is repeated multiple times to obtain A direction finding device, comprising: an average arrival angle calculator that calculates an average of arrival angles from a plurality of the obtained arrival angles. 4. A plurality of antennas, a complex random number generator that outputs a complex random number in a number corresponding to the number of the antennas, and a multiplier that multiplies the antenna output and the output of the complex random number generator corresponding to the antennas. , A plurality of receivers connected to the multiplier, a covariance matrix calculator that calculates a covariance matrix from the output of the receiver, and an eigenvalue / eigenvector calculator that calculates an eigenvalue / eigenvector from the covariance matrix And a mode vector generator for calculating the antenna output for the virtual wave source based on the array of the antennas, and an eigenvector selected based on the eigenvalue from the eigenvector and the antenna output from the mode vector generator. A minimum residual matrix calculator that calculates the minimum value of the Frobenius norm of the residual matrix, and scans the arrival angle of the signal from the virtual source to The arrival angle calculator that calculates the arrival angle that minimizes the minimum value of the Frobenius norm and the output of the complex random number generator are updated, and the process from the multiplier to the arrival angle calculator is repeated multiple times to obtain A cumulative frequency distribution calculator that calculates a cumulative frequency distribution that calculates the cumulative frequency distribution of arrival angles from the obtained multiple arrival angles, and a median calculator that calculates the median based on the obtained cumulative frequency distribution. A direction finding device characterized by being configured.