JP2014059225A - Receiver, noise suppression method, and noise suppression program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a receiver that can acquire an accurate result even if noise included in received signals is irregular among a plurality of antennas.SOLUTION: The receiver according to an embodiment includes: an antenna; an RF conversion part; an analog-to-digital conversion part; first and second calculation parts; a noise suppression part; and an eigenvalue decomposition part. The first calculation part calculates a first correlation matrix about a plurality of digital signals having received electric waves converted. The second calculation part calculates the first correlation matrix about a plurality of digital signals, and performs a similarity conversion of the calculated first correlation matrix to a second correlation matrix, by using a conversion matrix. The noise suppression part eliminates a difference between the first correlation matrix and the second correlation matrix to thereby prepare a third correlation matrix having a noise component suppressed. The eigenvalue decomposition part performs eigenvalue decomposition to the third correlation matrix, and calculates an eigenvalue and an eigenvector.

Description

  Embodiments described herein relate generally to a receiving device that receives radio waves using a plurality of antennas, and a noise suppression method and a noise suppression program used in the receiving device.

  A conventional receiving apparatus calculates a correlation matrix from reception signals received by a plurality of antenna elements, decomposes the calculated correlation matrix into eigenvalues, and calculates eigenvalues and eigenvectors of the correlation matrix. The receiving apparatus performs processing such as wave number estimation using the eigenvalue and eigenvector thus calculated.

  However, the conventional receiving apparatus assumes that the noise included in the received signal is uniform among the plurality of antenna elements. However, the noise included in the received signal may be non-uniform among the plurality of antenna elements. In such a case, an accurate result cannot be obtained by the above-described processing such as wave number estimation.

JP 2007-309847 A

  As described above, in the conventional receiving apparatus, when the noise included in the received signal is uneven among the plurality of antenna elements, the result obtained using the calculated eigenvalue and eigenvector includes an error. Therefore, there is a problem that an accurate result cannot be obtained.

  Therefore, the purpose is to provide a receiving device capable of obtaining an accurate result even when noise included in the received signal is non-uniform among a plurality of antennas, and a noise suppression method and a noise suppression program used in the receiving device. It is to provide.

  According to the embodiment, the reception device includes an antenna, an RF conversion unit, an analog-digital conversion unit, first and second calculation units, a noise suppression unit, and an eigenvalue decomposition unit. The antenna receives radio waves including noise components. The RF conversion unit converts a plurality of radio waves received by the plurality of antennas into a preset frequency band. The analog-digital conversion unit converts the plurality of frequency-converted signals into digital signals. The first calculation unit calculates a first correlation matrix for the plurality of digital signals. The second calculation unit calculates the first correlation matrix for the plurality of digital signals, and performs similarity conversion of the calculated first correlation matrix into a second correlation matrix using a conversion matrix. The noise suppression unit creates a third correlation matrix in which the noise component is suppressed by taking a difference between the first correlation matrix and the second correlation matrix. The eigenvalue decomposition unit performs eigenvalue decomposition on the third correlation matrix to calculate eigenvalues and eigenvectors.

It is a block diagram which shows the function structure of the receiver which concerns on this embodiment. It is a figure which shows the flowchart at the time of the signal processing part shown in FIG. 1 calculating an eigenvalue and an eigenvector. It is a figure which shows the receiving spectrogram of the signal given to the receiver shown in FIG. It is a figure which shows the simulation result of the eigenvalue calculated by the eigenvalue decomposition | disassembly part shown in FIG. It is a figure which shows the simulation result of the azimuth | direction spectrum calculated by the azimuth | direction calculation part shown in FIG. It is a figure which shows the simulation result of the eigenvalue calculated based on the 1st correlation matrix. It is a figure which shows the simulation result of the azimuth | direction spectrum calculated from the eigenvector calculated based on a 1st correlation matrix. It is a block diagram which shows the other example of a function structure of the receiver which concerns on this embodiment.

  Hereinafter, embodiments will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a functional configuration of the receiving apparatus according to the present embodiment. 1 includes antennas 10-1 to 10-n (n is a natural number of 1 or more), a frequency conversion unit 20, an analog-digital conversion unit 30, and a signal processing unit 40.

  The antennas 10-1 to 10-n receive radio waves coming from the outside. Radio waves contain noise. In the present embodiment, it is assumed that the level of noise included in the radio wave is non-uniform between the antennas 10-1 to 10-n.

  The frequency conversion unit 20 converts the received signals received by the antennas 10-1 to 10-n into IF (Intermediate Frequency) signals in the intermediate frequency band. Note that the frequency conversion unit 20 may convert radio waves received by the antennas 10-1 to 10-n into baseband BB signals.

  The analog-digital conversion unit 30 converts the IF signal supplied from the frequency conversion unit 20 into a digital signal in digital format. The analog-digital conversion unit 30 outputs a digital signal to the signal processing unit 40.

  The signal processing unit 40 includes, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory) and a RAM (Random Access Memory) that store application programs and data for the CPU to execute processing. The signal processing unit 40 has the following functions by causing the CPU to execute an application program. That is, the signal processing unit 40 includes a first correlation matrix calculation unit 41, a second correlation matrix calculation unit 42, a noise suppression unit 43, an eigenvalue decomposition unit 44, a wave number estimation unit 45, an azimuth calculation unit 46, and a measurement unit 47. Prepare.

  The first correlation matrix calculation unit 41 calculates a first correlation matrix that represents the correlation between all digital signals based on the n digital signals supplied from the analog-digital conversion unit 30.

  Below, the calculation process of the 1st correlation matrix by the 1st correlation matrix calculating part 41 is demonstrated in detail.

First, the digital signal X supplied from the analog-digital conversion unit 30 is expressed as X = [x ₁ (mT) x ₂ (mT)... X _N (mT)] ^T (1)
It is defined as Here, m represents a sample number, T represents a time interval of samples, and [•] ^T represents a transposition operation. Then, the first correlation matrix R _xx ^A is
R _xx ^A = <X · X ^H > (2)
It can be expressed. Here, <•> indicates an ensemble average, and [•] ^H indicates a conjugate complex transpose. The first correlation matrix R _xx ^A is
R _xx ^A = ASS ^H A ^H + NN ^H (3)
And disassembled. Here, A indicates a steering matrix including information on the direction of a signal, S indicates a matrix including a signal component, and N indicates a matrix including an average power of a noise component.

  The second correlation matrix calculation unit 42 calculates a second correlation matrix that represents the correlation between all digital signals based on the n digital signals supplied from the analog-digital conversion unit 30. Note that the second correlation matrix calculation unit 42 calculates the second correlation matrix by a method different from that of the first correlation matrix calculation unit 41.

  Below, the calculation process of the 2nd correlation matrix by the 2nd correlation matrix calculating part 42 is demonstrated in detail.

In general, when noise included in radio waves received between the antennas 10-1 to 10-n is uniform, a matrix representing the noise is
<NN ^H > = σ ² I (4)
It is expressed. Here, σ ² indicates an average noise power when received by a single antenna, and I indicates a unit matrix. In this case, when the correlation matrix is subjected to eigenvalue decomposition, the acquired eigenvalue is obtained by adding a fixed noise component to the eigenvalue of the signal component. That is, the noise component essentially does not affect the result of the eigenvalue decomposition.

On the other hand, when the noise included in the radio waves received between the antennas 10-1 to 10-n is uneven, the matrix representing the noise is
<NN ^H > = diag [σ ₁ ² σ ₂ ² ... Σ _N ² ] (5)
It is expressed. In this case, when the eigenvalue decomposition is performed on the correlation matrix, the size of the noise component added for each eigenvalue differs.

When calculating the second correlation matrix, first, the second correlation matrix calculator 42 sets a transformation matrix. Although various structures can be considered as the transformation matrix, in the present embodiment, the second correlation matrix calculator 42 uses the transformation matrix Z as an example.
Z = [1 kk ² ... K ^N−1 ] (6)
And set. k is a real number, and 0 <k <1. The second correlation matrix calculation unit 42 changes the value of k in the transformation matrix Z according to C / N (Carrier-Noise Ratio) measured by the measurement unit 47. That is, the second correlation matrix calculation unit 42 sets k to a value close to 1 or 1 when C / N is higher than a predetermined value, and sets k to a value close to 0 or 0 when C / N is low. Set. Note that the value of k may be manually changed by the operator of the receiving apparatus according to C / N.

The second correlation matrix calculator 42 calculates the first correlation matrix R _xx ^A based on the digital signal X supplied from the analog-digital converter 30. The second correlation matrix calculation unit 42 converts the calculated first correlation matrix R _xx ^A into a second correlation matrix R _xx ^B using a conversion matrix. That is, the second correlation matrix calculation unit 42 performs the following on the first correlation matrix R _xx ^A.
R _xx ^B = ZR _xx ^A Z ⁻¹ (7)
And the second correlation matrix R _xx ^B is calculated. At this time, the similarity transformation of <N · N ^H > is
Z <N · N ^H > Z ⁻¹ = <N · N ^H > (8)
It becomes.

The noise suppression unit 43 subtracts the second correlation matrix R _xx ^B calculated by the second correlation matrix calculation unit 42 from the first correlation matrix R _xx ^A calculated by the first correlation matrix calculation unit 41. By doing so, the third correlation matrix R _xx is calculated. That is, the noise suppression unit 43
^{_{R xx = <R xx A>}} - <R xx B> (9)
To calculate a third correlation matrix R _xx . Accordingly, the noise component contained in the first correlation matrix R _xx ^A is suppressed. Since the third correlation matrix R _xx has a noise component eliminated,

It is expressed.

The eigenvalue decomposition unit 44 performs eigenvalue decomposition on the third correlation matrix R _xx represented by Expression (10). When the third correlation matrix R _xx represented by Equation (10) is subjected to eigenvalue decomposition,
R _xx = EΛE ^H (11)
It is expressed. Here, Λ represents an eigenvalue, and E represents an eigenvector.

  The wave number estimation unit 45 estimates the number of signals using the eigenvalue calculated by the eigenvalue decomposition unit 44. Since the third correlation matrix represented by Expression (10) is a Vandermonde matrix, if the number of signals is r, the number of signal subspaces of the eigenvalue represented by Expression (11) is 2r. The wave number estimation unit 45 estimates the number of signals by executing an algorithm such as a known AIC, for example, using eigenvalues while considering this law. The wave number estimation unit 45 outputs information on the estimated number of signals to the subsequent stage and the direction calculation unit 46.

The azimuth calculation unit 46 uses the eigenvector calculated by the eigenvalue decomposition unit 44 and the number of signals estimated by the wave number estimation unit 45 to create information on the arrival direction of the radio wave. The bearing calculation unit 46 outputs the calculated information to the subsequent stage. For example, the azimuth calculation unit 46 calculates the MUSIC spectrum shown in Expression (12) based on the calculated eigenvector.

Here, a indicates a steering vector. The azimuth calculation unit 46 creates information related to the arrival azimuth based on the MUSIC spectrum shown in Expression (12).

  The measurement unit 47 measures C / N based on the digital signal supplied from the analog-digital conversion unit 30.

  Next, the noise suppression operation by the receiving apparatus configured as described above will be described according to the processing procedure of the signal processing unit 40. FIG. 2 is a diagram illustrating a flowchart when the signal processing unit 40 calculates eigenvalues and eigenvectors.

  When n digital signals are received from the analog-digital conversion unit 30, the signal processing unit 40 calculates the first and second correlation matrices by the first and second correlation matrix calculation units 41 and 42, respectively. (Step S21).

  The signal processing unit 40 calculates a third correlation matrix in which the noise component is suppressed by subtracting the second correlation matrix from the first correlation matrix by the noise suppression unit 43 (step S22).

  The signal processing unit 40 performs eigenvalue decomposition on the third correlation matrix by the eigenvalue decomposition unit 44, calculates eigenvalues and eigenvectors (step S23), and ends the processing.

  Next, a processing result by the receiving apparatus according to the present embodiment will be described. FIG. 3 is a diagram showing a reception spectrogram of a signal given to the receiving apparatus.

  When the signal shown in FIG. 3 is given to the receiving device, the receiving device calculates a third correlation matrix by the noise suppression unit 43, and calculates an eigenvalue based on the third correlation matrix by the eigenvalue decomposition unit 44. FIG. 4 is a diagram illustrating a simulation result of the eigenvalue calculated by the eigenvalue decomposition unit 44. According to FIG. 4, as the eigenvalue number increases, the relative level of the eigenvalue decreases.

  When the signal shown in FIG. 3 is given to the receiving device, the receiving device calculates a third correlation matrix by the noise suppression unit 43, and calculates an eigenvector based on the third correlation matrix by the eigenvalue decomposition unit 44. The azimuth spectrum is calculated by the azimuth calculation unit 109. FIG. 5 is a diagram illustrating a simulation result of the orientation spectrum calculated by the orientation calculation unit 109. According to FIG. 5, a peak with a clearly high level appears in the vicinity of 300 degrees.

  On the other hand, FIG. 6 is a diagram illustrating a simulation result of eigenvalues calculated when the noise suppression unit 43 does not calculate the third correlation matrix, that is, based on the first correlation matrix. According to FIG. 6, the relative level of the eigenvalues hardly changes when the eigenvalue number is 3 or more. Further, depending on the reception system, the relative level of the eigenvalue hardly changes from the eigenvalue number 1.

  FIG. 7 is a diagram illustrating a simulation result of the azimuth spectrum calculated from the eigenvector calculated based on the first correlation matrix when the noise suppression unit 43 does not calculate the third correlation matrix. According to FIG. 7, there are high level peaks in multiple directions.

  Referring to FIG. 3, it can be seen that a high-intensity signal given to the receiving apparatus is one wave. Therefore, the number of eigenvalue signal subspaces is two. Thus, according to FIGS. 4 and 6, it can be seen that eigenvalue numbers equal to or higher than eigenvalue number 3 correspond to noise. In FIG. 4, if the threshold value is set to about −35 dB in the relative level, it is possible to treat eigenvalue number 3 or more as noise. On the other hand, in FIG. 6, eigenvalue number 3 or more cannot be treated as noise regardless of the threshold value. Accordingly, the eigenvalue of eigenvalue number 3 shown in FIG. 4 is improved by about 10 dB compared to the eigenvalue of FIG. 6, and the eigenvalue of eigenvalue number 4 shown in FIG. 4 is improved by about 20 dB compared to the eigenvalue of FIG. The eigenvalue of eigenvalue number 5 shown in FIG. 4 is improved by about 30 dB compared to the eigenvalue of FIG. 6, and the eigenvalue of eigenvalue number 6 shown in FIG. 4 is improved by about 40 dB compared to the eigenvalue of FIG.

  Further, as shown in FIG. 3, although the signal given to the receiving apparatus is one wave, the azimuth spectrum shown in FIG. 7 seems to have a plurality of arrival directions. On the other hand, in FIG. 5, it is possible to specify one arrival direction.

  As described above, in the above embodiment, the first and second correlation matrix calculators 41 and 42 calculate the first and second correlation matrices. The noise suppression unit 43 suppresses the noise component included in the first correlation matrix using the second correlation matrix. The eigenvalue decomposition unit 44 performs eigenvalue decomposition on the third correlation matrix in which the noise component is suppressed, and calculates eigenvalues and eigenvectors. This makes it possible to calculate eigenvalues and eigenvectors while suppressing the influence of non-uniform noise components between antennas.

  Patent Document 1 proposes a technique for suppressing the correlation of noise components by performing a cyclic shift process on a received signal and dealing with non-uniform noise included in the received signal. The shift width of the cyclic calculation of the cyclic correlation matrix has an appropriate value depending on the number of antennas and the nature of the signal. For this reason, in the case of digital communication, there is a great limitation on the estimation of the number of signals having different communication speeds. Further, the shift width is greatly limited by the autocorrelation relationship of the analog signal. In the receiving apparatus according to the present embodiment, noise components are suppressed using two types of correlation matrices, and thus there is no limitation such as the shift width described in Patent Document 1.

In the above embodiment, the second correlation matrix calculation unit 42 changes the value of the conversion matrix according to the C / N measured by the measurement unit 47. The value of the transformation matrix affects the degree to which the relative level of the eigenvalue shown in FIG. 4 decreases according to the eigenvalue number. As a result, it is possible to accurately calculate eigenvalues and eigenvectors while taking C / N into account. Therefore, according to the receiving apparatus according to the present embodiment, noise included in the received signal is not generated between a plurality of antennas. Even if it is uniform, accurate results can be obtained.

  In the above embodiment, the wave number estimation unit 45 estimates the number of received signals using the eigenvalue calculated by the eigenvalue decomposition unit 44 using the third correlation matrix. Thereby, according to the receiving apparatus which concerns on this embodiment, it becomes possible to estimate the number of signals more correctly.

  In the above embodiment, the azimuth calculation unit 46 uses the eigenvector calculated by the eigenvalue decomposition unit 44 using the third correlation matrix to calculate information related to the arrival azimuth of the received signal. Thereby, according to the receiving apparatus which concerns on this embodiment, it becomes possible to calculate the information regarding an arrival direction more correctly.

  Moreover, although the said embodiment demonstrated the case where the direction calculation part 46 outputs the information regarding the calculated arrival direction to a back | latter stage, it is not necessarily limited to this. For example, as shown in FIG. 8, the receiving device may include the accumulating unit 48 at the subsequent stage of the azimuth calculating unit 46. The information calculated by the azimuth calculation unit 46 is a calculation result of one snapshot. The accumulating unit 48 accumulates the calculation results of one snapshot calculated by the azimuth calculating unit 46. The accumulation unit 48 outputs the accumulation result to the subsequent stage. As a result, the accumulation unit 48 accumulates the calculation results of a plurality of snapshots, so that it is possible to suppress stationary noise, instantaneous noise, and the like. For this reason, it becomes easy to detect the signal component in the azimuth spectrum based on the accumulated result.

  Although the embodiment of the present invention has been described, this embodiment is presented as an example and is not intended to limit the scope of the invention. This embodiment can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. This embodiment and its modifications are included in the scope of the present invention and the gist thereof, and are also included in the invention described in the claims and the equivalent scope thereof.

  10-1 to 10-n antenna, 20 frequency conversion unit, 30 analog-digital conversion unit, 40 signal processing unit, 41 first correlation matrix calculation unit, 42 second correlation matrix calculation unit, 43 ... Noise suppression unit, 44 ... Eigen value decomposition unit, 45 ... Wave number estimation unit, 46 ... Direction calculation unit, 47 ... Measurement unit, 48 ... Accumulation unit

Claims (9)

A plurality of antennas for receiving radio waves including noise components;
An RF converter that converts a plurality of radio waves received by the plurality of antennas into a preset frequency band;
An analog-to-digital converter that converts the plurality of frequency-converted signals into digital signals;
A first calculation unit for calculating a first correlation matrix for the plurality of digital signals;
A second arithmetic unit that calculates the first correlation matrix for the plurality of digital signals, and performs similarity conversion of the calculated first correlation matrix into a second correlation matrix using a conversion matrix;
A noise suppression unit that creates a third correlation matrix in which the noise component is suppressed by taking a difference between the first correlation matrix and the second correlation matrix;
A receiving apparatus comprising: an eigenvalue decomposition unit that performs eigenvalue decomposition on the third correlation matrix and calculates eigenvalues and eigenvectors.   The receiving apparatus according to claim 1, further comprising a wave number estimating unit that estimates a wave number based on the eigenvalue.   The receiving apparatus according to claim 2, further comprising an azimuth calculating unit that calculates information related to an arrival direction of the radio wave with reference to the eigenvector and the estimated wave number. Using the eigenvector and the estimated wave number, an azimuth calculating unit that calculates an azimuth spectrum indicating the arrival azimuth of the radio wave;
The receiving apparatus according to claim 2, further comprising an accumulating unit that accumulates the orientation spectrum. A measuring unit for measuring a CN (Carrier-Noise Ratio) of the received radio wave;
The receiving apparatus according to claim 1, wherein the second calculation unit changes a value of the conversion matrix according to a value of the CN. Receives multiple digital signals in which radio waves containing noise components are converted to digital format,
Calculating a first correlation matrix for the plurality of digital signals;
Calculating the first correlation matrix for the plurality of digital signals, and transforming the calculated first correlation matrix into a second correlation matrix using a transformation matrix;
A difference between the first correlation matrix and the second correlation matrix is taken to create a third correlation matrix that suppresses the noise component;
A noise suppression method for performing eigenvalue decomposition on the third correlation matrix and calculating eigenvalues and eigenvectors. Measure the CN (Carrier-Noise Ratio) of the received radio wave,
The noise suppression method according to claim 6, wherein the value of the transformation matrix is changed according to the value of the CN. A noise suppression program used in a receiving device that receives radio waves including noise components by a plurality of antennas and converts the received radio waves into a plurality of digital signals in a digital format,
Processing to calculate a first correlation matrix for the plurality of digital signals;
A process of calculating the first correlation matrix for the plurality of digital signals, and converting the calculated first correlation matrix into a second correlation matrix using a conversion matrix;
A process of creating a third correlation matrix in which the noise component is suppressed by taking a difference between the first correlation matrix and the second correlation matrix;
The noise suppression program which makes the computer of the said receiver perform the process which performs eigenvalue decomposition | disassembly with respect to a said 3rd correlation matrix, and calculates an eigenvalue and an eigenvector. A process of measuring a CN (Carrier-Noise Ratio) of the received radio wave;
The noise suppression program according to claim 8, further causing the computer to execute a process of changing the value of the transformation matrix according to the value of the CN.

Images