JP6223229B2 - Arrival wave number estimation device - Google Patents

Description

  The present invention relates to an arrival wave number estimation apparatus that estimates the number of target signals (hereinafter referred to as “arrival wave number”) included in a received signal of a sensor array system (for example, a radar or a sonar) that transmits and receives a plurality of coherent pulses. Is.

In the process of estimating the number of incoming waves, the determination probability and the erroneous determination probability of the number of incoming waves are usually important performance specifications.
The erroneous determination probability is a probability that the number of incoming waves of 1 or more is erroneously estimated although the target signal is not included in the received signal and only the receiver noise is included. In an actual sensor system, it is desired that excellent determination characteristics of the number of incoming waves can be obtained while controlling the erroneous determination probability to a desired constant value.
An excellent determination characteristic of the number of incoming waves is such that the determination probability of the number of incoming waves asymptotically approaches 100% at a high SNR (Signal to Noise Ratio), while the determination probability of a high number of incoming waves is exhibited even at a low SNR.

In the arrival wave number estimation device disclosed in Non-Patent Document 1 below, each eigenvalue constituting the correlation matrix of the received signal vector is obtained, and after sorting a plurality of eigenvalues in descending order, the eigenvalues sorted in descending order are used. An AIC (Akaike Information Criterion) normative value is obtained, and the number of incoming waves is estimated based on the minimum AIC normative value.
Alternatively, an MDL (Minimum Description Length) standard value is obtained from each eigenvalue constituting the correlation matrix, and the number of incoming waves is estimated based on the minimum MDL standard value.
However, regardless of which method is used to estimate the number of incoming waves, there is a problem that the determination probability of the number of incoming waves deteriorates under a low SNR condition where the difference between the target eigenvalue and the noise eigenvalue is small.
In the arrival wave number estimation devices disclosed in the following Patent Document 1 and Non-Patent Document 2, a SIGNED (Signal Number Detection by Eigenbeamforming and Doppler Filter) method is used to solve the problem that the determination probability of the arrival wave number deteriorates. Thus, the number of incoming waves is estimated.

In the SIGNED method, paying attention to the fact that the radar target signal is observed as a complex sine wave in the hit direction, an eigen beam weighting the eigen vectors corresponding to the eigen values arranged in descending order is obtained, and the eigen beam is calculated by DFT (Discrete Fourier). (Transform) and the like.
Then, a coherent integration value (hereinafter referred to as “MCI (Maximum Coherent Integration) value”) corresponding to the target Doppler bin is extracted from the coherent integration result, and an MDL norm value or an AIC norm value is obtained from the MCI value. The number of incoming waves is estimated based on the reference value or the minimum AIC reference value.
In the SIGNED method, since the eigen beam is coherently integrated before obtaining the MDL norm value or the AIC norm value, the difference between the target MCI value and the noise MCI value becomes larger than that in the case of the eigen value. For this reason, the determination probability of the number of incoming waves at a low SNR is improved.

JP 2010-25576 A (paragraph number [0006], FIG. 1)

M. Wax, T. Kailath, “Detection of signals by information theoretic criteria,” IEEE Trans. Acoustics, Speech, and Signal Processing, vol.33, no.2, pp.387-392, Apr. 1985 Takahashi, Hirata, Maniwa, “Radar target number estimation method using eigenbeam and pulse Doppler filter,” IEICE Transactions B vol.J94-B No.4, April 2011

  Since the conventional arrival wave number estimation apparatus is configured as described above, the use of the SIGNED method can increase the determination probability of the arrival wave number even under a low SNR condition. However, under a relatively high SNR condition in which the determination probability of the number of incoming waves by the MDL alone is asymptotic to 100%, the noise MCI value distribution does not become flat, so the determination probability of the number of incoming waves by the SIGNED method is 100% There is a problem that the estimation accuracy of the number of incoming waves is deteriorated because the asymmetry is not obtained.

  In the internal processing of the SIGNED method, since the MDL normative value or the AIC normative value is used, there is a problem that the misjudgment probability is uniquely determined and the misjudgment probability cannot be controlled. This is because the number of incoming waves is estimated based on the minimum MDL standard value or the minimum AIC standard value. The erroneous determination probability of the MDL normative value or the AIC normative value can be obtained numerically, for example, by inputting only the receiver noise, but it is difficult to match the same value with the specifications required for the sensor array system. It is.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to provide an arrival wave number estimation device that can accurately estimate the arrival wave number regardless of the SNR condition.

  An arrival wave number estimation apparatus according to the present invention includes a plurality of receiving means for receiving a reflected wave reflected by a target and outputting a received signal of the reflected wave, and a sum beam of received signals output from the plurality of receiving means And a target detection means for coherently integrating the sum beam and detecting a target range bin that is a range bin in which the target signal exists and a target Doppler bin in which the target signal exists from the coherent integration result. Eigenvector calculation means for calculating a correlation matrix of a plurality of reception signals in the target range bin among reception signals output from the plurality of reception means, and calculating an eigenvector corresponding to each eigenvalue constituting the correlation matrix; The eigenvectors calculated by the eigenvector calculation means for each of the received signals in the target range bin are And a coherent integration unit that coherently integrates each multiplication result, and the arrival wave number estimation unit obtains the coherent integration value of the target Doppler bin from each of the plurality of coherent integration results in the coherent integration unit. The number of coherent integral values larger than a preset threshold is counted as the wave number of.

  According to the present invention, the coherent integration unit that multiplies each of the received signals in the target range bin by the eigenvector calculated by the eigenvector calculation unit and coherently integrates each multiplication result is provided, and the arrival wave number estimation unit includes: Since the coherent integration value of the target Doppler bin is acquired from the plurality of coherent integration results in the coherent integration means, and the number of coherent integration values greater than a preset threshold is counted as the wave number of the incoming wave. There is an effect that the number of incoming waves can be estimated with high accuracy regardless of the SNR condition.

It is a block diagram which shows the arrival wave number estimation apparatus by Embodiment 1 of this invention. It is a block diagram which shows the arrival wave number estimation part 11 of the arrival wave number estimation apparatus by Embodiment 1 of this invention. It is a block diagram which shows the arrival wave number estimation part 11 of the arrival wave number estimation apparatus by Embodiment 2 of this invention. It is a block diagram which shows the arrival wave number estimation apparatus by Embodiment 3 of this invention. It is a block diagram which shows the arrival wave number estimation part 20 of the arrival wave number estimation apparatus by Embodiment 3 of this invention. It is a block diagram which shows the arrival wave number estimation part 20 of the arrival wave number estimation apparatus by Embodiment 4 of this invention. It is a block diagram which shows the arrival wave number estimation apparatus by Embodiment 5 of this invention. It is a block diagram which shows the arrival wave number estimation part 33 of the arrival wave number estimation apparatus by Embodiment 5 of this invention. It is a block diagram which shows the arrival wave number estimation part 33 of the arrival wave number estimation apparatus by Embodiment 6 of this invention.

Embodiment 1 FIG.
1 is a block diagram showing an arrival wave number estimation apparatus according to Embodiment 1 of the present invention.
In FIG. 1, element antennas 1-1 to 1-M constitute a planar array antenna, and a reflected wave reflected by a target enters. Here, it is assumed that M element antennas 1-1 to 1-M constitute a planar array antenna. However, even if M subarrays constitute a planar array antenna. Good.
The receivers 2-1 to 2-M receive reflected waves (carrier frequency band signals) incident from the element antennas 1-1 to 1-M, and convert the received signals into baseband signals.
The AD converters 3-1 to 3-M perform A / D conversion processing for converting the baseband received signals converted by the receivers 2-1 to 2-M into digital signals, and convert the digital received signals into digital signals. Output.
The element antenna 1-m, the receiver 2-m, and the AD converter 3-m (m = 1, 2,..., M) constitute receiving means.

The sum beam forming unit 4 forms a sum beam of M received signals by giving a sum beam weight to the M received signals output from the AD converters 3-1 to 3 -M. Processing for outputting the beam to the coherent integration processing unit 5 is performed.
The coherent integration processing unit 5 performs a process of coherently integrating the sum beam output from the sum beam forming unit 4 using a DFT (Discrete Fourier Transform) or the like, and outputting a pulse Doppler filter signal, which is a result of the coherent integration, to the signal detection unit 6. carry out.
The signal detection unit 6 performs a process of detecting a target range bin that is a range bin in which the target signal exists and a target Doppler bin in which the target signal exists from the pulse Doppler filter signal output from the coherent integration processing unit 5. To do.
The sum beam forming unit 4, the coherent integration processing unit 5, and the signal detection unit 6 constitute target detection means.

The correlation matrix calculation unit 7 calculates a correlation matrix of M reception signals in the target range bin detected by the signal detection unit 6 among the M reception signals output from the AD converters 3-1 to 3 -M. Perform the process.
The eigenvalue / eigenvector calculation unit 8 performs a process of calculating eigenvectors corresponding to the eigenvalues constituting the correlation matrix calculated by the correlation matrix calculation unit 7.
The correlation matrix calculation unit 7 and the eigenvalue / eigenvector calculation unit 8 constitute eigenvector calculation means.

The target range bin eigen beam forming unit 9 generates eigen values / eigen vectors for the target range bin received signal detected by the signal detection unit 6 among the M received signals output from the AD converters 3-1 to 3-M. A process for forming M eigenbeams (multiplication results of the received signal and the eigenvector) is performed by multiplying the eigenvectors calculated by the calculation unit 8 respectively.
The coherent integration processing units 10-1 to 10-M coherently integrate the eigen beam output from the target range bin eigen beam forming unit 9 by DFT or the like, and a pulse Doppler filter signal as a result of the coherent integration is supplied to the arrival wave number estimation unit 11. Perform the output process.
The target range bin eigen beam forming unit 9 and the coherent integration processing units 10-1 to 10-M constitute coherent integration means.

The arrival wave number estimation unit 11 acquires the coherent integration values of the target Doppler bins detected by the signal detection unit 6 from the M pulse Doppler filter signals output from the coherent integration processing units 10-1 to 10-M. Then, a process of counting the number of coherent integral values larger than a preset fixed threshold (threshold value) as the wave number of the incoming wave is performed.
The arrival wave number estimation unit 11 constitutes arrival wave number estimation means.

In the example of FIG. 1, a sum beam forming unit 4, a coherent integration processing unit 5, a signal detection unit 6, a correlation matrix calculation unit 7, an eigenvalue / eigenvector calculation unit 8, and target range bin eigenbeam formation, which are components of the arrival wave number estimation apparatus. Each of the unit 9, the coherent integration processing units 10-1 to 10-M and the arrival wave number estimation unit 11 is configured by dedicated hardware (for example, a semiconductor integrated circuit on which a CPU is mounted, or a one-chip microcomputer). However, these components may be configured by a computer.
When these components are configured by a computer, the sum beam forming unit 4, the coherent integration processing unit 5, the signal detecting unit 6, the correlation matrix calculating unit 7, the eigenvalue / eigenvector calculating unit 8, the target range bin eigenbeam forming unit 9, the coherent A program describing the processing contents of the integration processing units 10-1 to 10-M and the arrival wave number estimation unit 11 is stored in a memory of a computer so that the CPU of the computer executes the program stored in the memory. You can do it.

FIG. 2 is a block diagram showing the arrival wave number estimation unit 11 of the arrival wave number estimation apparatus according to Embodiment 1 of the present invention.
In FIG. 2, the fixed threshold determination units 11 a-1 to 11 a -M are the target Doppler bins detected by the signal detection unit 6 from the pulse Doppler filter signals output from the coherent integration processing units 10-1 to 10 -M. A process of obtaining a coherent integral value and determining whether the coherent integral value is larger than a preset fixed threshold is performed.
The arrival wave number recognition unit 11b counts the number of determination results indicating that the coherent integration value is larger than the fixed threshold among the determination results of the fixed threshold determination units 11a-1 to 11a-M, and determines the number of arrival signals. Implement a process that certifies the wave number.

式（１）の左辺のｕ_kは第ｋ番目の目標信号の到来方向を示すベクトルである。なお、明細書の文章中では、電子出願の関係上、ベクトルを表す文字を太文字で表記することができないため、細字のｕ_kで表記している。
以下、ｕ_k以外のベクトルを表す文字も、明細書の文章中では、ｕ_kと同様に細字で表記する。 Next, the operation will be described.
Before explaining the processing contents of the arrival wave number estimation apparatus in FIG. 1, the theory of the arrival wave number estimation method in this arrival wave number estimation apparatus will be described.
In the arrival wave number estimation apparatus of FIG. 1, M element antennas 1-1 to 1-M constitute a planar array antenna, and K target signals (K <M) arrive from different directions. To do.

Left side of u _k of formula (1) is a vector indicating the direction of arrival of the k-th target signal. In the sentence of the specification, the relationship between the electronic application, can not be represented characters representing the vector in bold, it is denoted by u _k of fine print.
Hereinafter, characters representing a vector other than u _k also in the text of the specification, written in fine print as with u _k.

式（２）において、ｘ（ｎ）は目標レンジビンにおける受信信号ベクトル、ｎはスナップショット（パルスヒット）番号、ｎ（ｎ）はガウス性受信機雑音ベクトルである。
また、ｓ（ｎ）は第ｋ番目の目標信号の複素振幅ｓ_k（ｎ）を要素とする下記の式（３）のような複素振幅ベクトルである。ただし、上添字Ｔは転置を表す記号である。

また、Ａは目標信号の到来方向ｕ_kに対応するステアリングベクトルａ（ｕ_k）を列ベクトルに有する下記の式（４）のようなアレーマニフォルドである。

In this case, the received signal vector in the target range bin (the range bin where the target signal exists) is given by the following equation (2).

In Equation (2), x (n) is a received signal vector in the target range bin, n is a snapshot (pulse hit) number, and n (n) is a Gaussian receiver noise vector.
Further, s (n) is a complex amplitude vector represented by the following formula (3) having the complex amplitude s _k (n) of the _kth target signal as an element. The superscript T is a symbol representing transposition.

Also, A is the array manifold, such as the following with the column vector of the steering vector a (u _k) corresponding to the arrival direction u _k of the target signal Equation (4).

ただし、ａ（ｕ_k）はＭ次元ベクトルであり、ｍは素子アンテナの番号であり、ｍ＝１，２，・・・，Ｍである。
また、（ｘ_m，ｙ_m）は第ｍ番目の素子アンテナの配置位置を表すアンテナ開口面上の座標であり、ｇ_m（ｕ_k，ｖ_k）は第ｍ番目の素子アンテナの到来方向ｕ_kに対するゲインである。λは波長である。 The steering vector a (u _k ) is expressed as the following equation (5).

Here, a (u _k ) is an M-dimensional vector, m is an element antenna number, and m = 1, 2,...
Further, (x _m , y _m ) is a coordinate on the antenna aperture plane that represents the arrangement position of the m-th element antenna, and g _m (u _k , v _k ) is the arrival direction u of the m-th element antenna. Gain for _k . λ is a wavelength.

Correlation matrix R _xx hat of received signal vector x (n) of snapshot number N (In the text of the description, the symbol “^” cannot be attached on the letter because of the electronic application. (shown as _xx hat) is given by the following equation (6). However, the superscript H represents the complex conjugate transpose.

If the correlation of the complex amplitude s _k (n) of the target signal is high, correlation suppression can be performed by applying the spatial averaging method disclosed in Non-Patent Document 3 below. In this case, the correlation matrix after applying the spatial averaging method corresponds to the equation.
[Non-Patent Document 3] Nobuyoshi Kikuma, Adaptive Signal Processing with Array Antenna, Science and Technology Publishing, 1999

ここで、Λハットは、下記の式（８）に示すように、固有値λ₁ハット，・・・，λ_Mハットを対角項に有する対角行列であり、固有値λ₁ハット，・・・，λ_Mハットを下記の式（９）のように降順に並べたものである。なお、相関行列Ｒ_xxハットはエルミート行列であるため、固有値λ₁ハット，・・・，λ_Mハットは実数である。

The correlation matrix R _xx hat can be expressed using eigenvalues and eigenvectors as shown in the following equation (7).

Here, the Λ hat is a diagonal matrix having eigenvalues λ ₁ hat,..., Λ _M hat in the diagonal terms, as shown in the following equation (8), and the eigenvalues λ ₁ hat,. , Λ _M hats are arranged in descending order as in the following equation (9). Since the correlation matrix R _xx hat is a Hermitian matrix, the eigenvalues λ ₁ hat,..., Λ _M hat are real numbers.

Moreover, E-hat, as shown in the following equation (10), M eigenvalues lambda ₁ hat,., Eigenvector e ₁ hat corresponding to lambda _M hat, ..., and e _M hat column vector Is a matrix. Since the correlation matrix R _xx hat is a Hermitian matrix, the M eigenvectors e ₁ hat,..., E _M hat are orthogonal to each other.

Below, the eigenvalues λ ₁ hat corresponding to the K-number of the target signal, ···, λ _K hat and is referred to as a signal eigenvalues, the eigenvalues λ ₁ hat, ···, λ _K eigenvectors e ₁ hat corresponding to the hat, ... • The e _K hat is called the signal eigenvector.
Further, (M-K) corresponding to the number of receiver noise eigenvalue lambda _{K + 1} hat, ..., referred to as noise eigenvalues lambda _M hat, eigenvalue lambda _{K + 1} hat, ..., the lambda _M hat The corresponding eigenvector e _{K + 1} hat,..., E _{K + M} hat is referred to as a noise eigenvector.
A matrix having a noise eigenvector as a column vector is defined as the following equation (11).

上述したＭＤＬやＡＩＣでは、上記の式（９）に示すように、雑音固有値より信号固有値の大きさが十分に大きいことを利用して目標数（到来波の波数）を推定している。
しかし、上記の非特許文献２で指摘されているように、ＳＮＲが低くなるほど、信号固有値と雑音固有値の差が小さくなり、目標数の推定精度が劣化していく問題がある。 At this time, since the noise subspace spanned by (M−K) noise eigenvectors and the subspace spanned by the K signal eigenvectors are orthogonal, the steering vector corresponding to the noise subspace and the arrival direction θ _k is They are orthogonal as in equation (12). However, 0 _{M × 1} is an M × 1 zero vector.

In the above-mentioned MDL and AIC, as shown in the above equation (9), the target number (the wave number of the incoming wave) is estimated using the fact that the magnitude of the signal eigenvalue is sufficiently larger than the noise eigenvalue.
However, as pointed out in Non-Patent Document 2 above, there is a problem that the lower the SNR, the smaller the difference between the signal eigenvalue and the noise eigenvalue, and the target number estimation accuracy deteriorates.

ここで、Ｍ本の固有ビームのうち、信号固有ベクトルｅ₁ハット，・・・，ｅ_KハットによるＫ本の信号固有ビームｙ_m（ｎ）ハットは、下記の式（１４）のようになる。ただし、１≦ｍ≦Ｋであり、上添字＊は複素共役を表している。

And beam weight eigenvectors e _m hat to the received signal vector x (n) beam y _m (n) hat (hereinafter referred to as "intrinsic beam") is expressed as the following equation (13).

Here, among the M eigenbeams, _K signal eigenbeams y _m (n) by the signal eigenvectors e ₁ hat,..., E _K hat are expressed by the following equation (14). However, 1 ≦ m ≦ K, and the superscript * represents a complex conjugate.

Next, consider (M−K) number of noise eigenbeams by noise eigenvectors e _{K + 1} hat,..., E _{K + M} hat.
Considering the property of the above expression (12), the target signal included in the received signal vector x (n) is canceled, and the eigen beam y _m (n) hat represented by the expression (13) has only a noise component. It becomes.
Therefore, the noise eigen beam is expressed as the following equation (15). However, K + 1 ≦ m ≦ M.

式（１６）において、Ａ_kは複素振幅値、ｆ_kは規格化ドップラ周波数である。 Here, a target signal in the radar is considered.
The s _k (n), which are K radar target signals sampled at a constant PRI (Pulse Repetition Interval), are expressed as the following equation (1) as a complex sine wave signal having a Doppler frequency corresponding to the radial velocity: 16).

In Expression (16), A _k is a complex amplitude value, and f _k is a normalized Doppler frequency.

式（１７）より、信号固有ビームｙ_m（ｎ）ハットは、Ｋ個の目標信号を合成したものであることがわかる。 Here, the K normalized Doppler frequencies are different from each other. Further, it is assumed that s _k (n) that are K radar target signals belong to the same Doppler bin by a pulse Doppler filter based on a discrete Fourier transform DFT described later.
For formula (14) representing the signal-specific beam y _m (n) hat, substituting s _k (n) of equation (16), the signal eigenbeams y _m (n) hat in radar, the following formula ( 17).

From equation (17), it can be seen that the signal eigen beam y _m (n) hat is a combination of K target signals.

ただし、ｌは０から（Ｎ−１）のレンジビン番号である。 Therefore, when the signal eigenbeam and noise eigenbeam are input to the pulse Doppler filter by the discrete Fourier transform DFT based on the characteristics of the signal eigenbeam represented by Equation (17) and the noise eigenbeam represented by Equation (15). The Z _m (l) hat, which is the output of the pulse Doppler filter, is expressed by the following equation (18).

Here, l is a range bin number from 0 to (N−1).

式（１９）において、ｗは和ビームウェイトである。 At this time, as already described, the K target signals are coherently integrated into one range bin, and will be referred to as target Doppler bin l _s .
The target Doppler bin l _s is known in the target detection process prior to the estimation process of the number of incoming waves. That is, if a pulse Doppler filter based on the discrete Fourier transform DFT is applied to the sum beam y (n) of the received signal vector x (n) and signal detection is performed on the output, the Doppler bin from which the signal is detected becomes clear. It is to become.

In equation (19), w is the sum beam weight.

よって、下記の式（２１）に示すように、スレッショルド判定による目標信号の検出回数を求めれば、到来波の波数を推定することができる。

Subsequently, threshold determination is performed on | Z _m (l _s ) hat | ² (or | Z _m (l _s ) hat |), which is the detection output of the pulse Doppler filter of the target Doppler bin l _s .
As a result, M determination results b (m) are obtained.

Therefore, as shown in the following equation (21), the number of incoming waves can be estimated by obtaining the number of times of detection of the target signal by threshold determination.

Hereinafter, the processing content of the arrival wave number estimation apparatus of FIG. 1 is demonstrated concretely.
After the pulse signal is radiated from the radar, the element antennas 1-1 to 1-M are incident on the reflected wave (carrier frequency band signal) of the pulse signal reflected on the target.
When the reflected waves are incident from the element antennas 1-1 to 1-M, the receivers 2-1 to 2-M receive the reflected waves and convert the received signals into baseband signals.
When the receivers 2-1 to 2-M convert the received signals into baseband signals, the AD converters 3-1 to 3-M convert the baseband received signals into digital signals. Output the received signal.
The received signal vector x (n) shown in the above equation (2) is a vector representation of the M received signals output from the AD converters 3-1 to 3-M.

When the sum beam forming unit 4 receives M received signal vectors x (n) which are M received signals from the AD converters 3-1 to 3 -M, as shown in the above equation (19), the received signal By giving a sum beam weight w to the vector x (n), a sum beam y (n) of M received signals is formed, and the sum beam y (n) is output to the coherent integration processing unit 5.
When the coherent integration processing unit 5 receives the sum beam y (n) from the sum beam forming unit 4, the coherent integration unit 5 coherently integrates the sum beam y (n) by DFT or the like, and outputs a pulse Doppler filter signal as a result of the coherent integration. Output to the detector 6.

When the signal detection unit 6 receives the pulse Doppler filter signal from the coherent integration processing unit 5, the signal detection unit 6 compares the coherent integration value of each range bin included in the pulse Doppler filter signal with a preset threshold, A target range bin that is a range bin in which the target signal exists is detected as a range bin whose threshold is higher than the threshold, and the detection result of the target range bin is used as the correlation matrix calculation unit 7 and the target range bin eigen beam forming unit 9.
The signal detection unit 6 detects the target Doppler bin l _s is a Doppler bin where target signal is present, and outputs a detection result of the target Doppler bin l _s arrival wave number estimating unit 11.

When the correlation matrix calculation unit 7 receives the detection result of the target range bin from the signal detection unit 6, as shown in the above equation (6), the M receptions output from the AD converters 3-1 to 3 -M. Of the received signal vector x (n), which is a signal, a correlation matrix R _xx hat of the received signal vector x (n) corresponding to the target range bin is calculated.
When the correlation matrix calculation unit 7 calculates the correlation matrix R _xx hat of the received signal vector x (n) corresponding to the target range bin, the eigenvalue / eigenvector calculation unit 8 calculates the correlation matrix R based on the above equation (7). Each eigenvalue λ ₁ hat,..., λ _M hat constituting the _xx hat is calculated.
In addition, the eigenvalues and eigenvectors calculation unit 8, the eigenvalues λ ₁ hat,..., Eigenvectors e ₁ hat corresponding to the λ _M hat, ..., to calculate the e _M hat, the eigenvector e ₁ hat, ... .., E _M hat is output to the target range bin eigen beam forming unit 9.

When the target range bin eigenbeam forming unit 9 receives the eigenvector e ₁ hat,..., E _M hat from the eigenvalue / eigenvector calculation unit 8, as shown in the above equation (13), the AD converters 3-1 to 3-1 Among the received signal vectors x (n) that are M received signals output from 3-M, the eigenvector e ₁ hat,..., E _M for the received signal vector x (n) of the target range bin. by multiplying the hat, to form the M-specific beam y _m (n) hat, outputs the M eigenbeams y _m (n) is hat coherent integration processor 10-1 to 10-M.
When the coherent integration processing units 10-1 to 10-M receive the eigen beam y _m (n) hat from the target range bin eigen beam forming unit 9, the eigen beam y _m ( n) Coherently integrate the hat with DFT or the like, and output a pulse Doppler filter signal Z _m (l) hat, which is a result of the coherent integration, to the arrival wave number estimating unit 11.

When the fixed threshold determination units 11a-1 to 11a-M of the arrival wave number estimation unit 11 receive the pulse Doppler filter signal Z _m (l) hat from the coherent integration processing units 10-1 to 10-M, the pulse Doppler filter signal is received. from the Z _m (l) hat, the detected coherent integration value of the target Doppler bin l _s by the signal detection unit 6 | Z _m (l _s) hat | ² (or, | Z _m (l _s) hat |) of Detect.
Fixed thresholds determining unit 11a-1 to 11 a-M is the coherent integration value of the target Doppler bin l _s | when detects the ^2, the coherent integration values | | Z _m (l _s) hat Z _m (l _s) hat | ² is It is determined whether or not it is larger than a preset fixed threshold T _m ^{(1), and} a determination result b (m) as shown in the above equation (20) is output to the incoming wave number authorization unit 11b.

Upon receiving the determination result b (m) from the fixed threshold determination units 11a-1 to 11a-M, the arrival wave number determination unit 11b of the arrival wave number estimation unit 11 receives the determination result b (m) from among the determination results b (m). Count the number of determination results indicating that Z _m (l _s ) hat | ² is larger than the fixed threshold T _m ⁽¹⁾ .
That is, as shown in the above equation (21), the arrival wave number authorization unit 11b has b (m) = 1 among the determination results b (m) output from the fixed threshold determination units 11a-1 to 11a-M. The number K of the determination results is counted.
When the number of arriving waves certifying section 11b counts the number K hats of the determination result of b (m) = 1, the number K hats is certified as the wave number (target number) of arriving waves.

As is apparent from the above, according to the first embodiment, the received signal vector x (n), which is M received signals output from the AD converters 3-1 to 3-M, A target range bin eigen beam forming unit 9 that forms M eigen beams y _m (n) hats by multiplying the eigen vectors e ₁ hat,..., E _M hat calculated by the eigen vector calculation unit 8; Coherent integration processing units 10-1 to 10-M that perform coherent integration of the eigenbeams y _m (n) hats and output pulse Doppler filter signals Z _m (l) hats that are the coherent integration results are provided. wave number estimating unit 11, from the pulse Doppler filtered signal Z _m (l) hat that have been outputted from the coherent integration processor 10-1 to 10-M, coherent target Doppler bin l _s Integral value | Z _m (l _s) hat | ² detects the as wave number K hat incoming wave, preset fixed threshold T _m ⁽¹⁾ greater than coherent integration value | Z _m (l _s) hat | ² Therefore, the number of incoming waves can be accurately estimated regardless of the SNR condition.
In other words, the first embodiment solves the problem of the SIGNED method, and the target number determination probability can be made asymptotic to 100% under a relatively high SNR condition. Further, by appropriately changing the fixed threshold T _m ⁽¹⁾ used by the arrival wave number estimation unit 11, the erroneous determination probability can be controlled to be constant.

Embodiment 2. FIG.
FIG. 3 is a block diagram showing the arrival wave number estimation unit 11 of the arrival wave number estimation apparatus according to the second embodiment of the present invention. In the figure, the same reference numerals as those in FIG.
The fixed threshold determination units 11a-1 to 11a-M receive the pulse Doppler filter signal Z _m (l) hat output from the coherent integration processing units 10-1 to 10-M, as in the first embodiment. Is done.
That is, the pulse Doppler filter signal Z ₁ (l) hat input to the fixed threshold determination unit 11 a-1 is an eigen beam obtained by multiplying the received signal output from the AD converter 3-1 by the eigen vector e ₁ hat. y ₁ (n) is a result of the coherent integration of the hat, and the pulse Doppler filter signal Z ₂ (l) hat that is input to the fixed threshold determination unit 11a-2 corresponds to the received signal output from the AD converter 3-2. This is a coherent integration result of the eigen beam y ₂ (n) hat multiplied by the eigen vector e ₂ hat.
The pulse Doppler filter signal Z _M (l) hat input to the fixed threshold determination unit 11a-M is an eigen beam obtained by multiplying the reception signal output from the AD converter 3-M by the eigen vector e _M hat. y _M (n) Coherent integration result of the hat.
In this case, the eigenvector e ₁ hat, ..., eigenvalues corresponding to e _M hat lambda ₁ hat, ..., the magnitude relationship of lambda _M hat are as described above for formula (9).

The arrival wave number authorization unit 11c estimates the wave number K hat of the arrival wave in the same manner as the arrival wave number authorization unit 11b of FIG. 2, but at this time, the determination result b (m) of a certain fixed threshold determination unit 11a-m ) Is 1 ⁽ indicating that the coherent integral value | Z _m (l _s ) hat | ² is greater than the fixed threshold T _m ⁽¹⁾ ), one before the fixed threshold determination unit 11a-m. If the determination result b (m−1) of the fixed threshold determination units 11a- (m−1) arranged side by side is 0 (coherent integration value | Z _m (l _s ) hat | ² is the fixed threshold T _m ^(1). The determination result b (m) of the fixed threshold determination unit 11a-m is excluded from the number of objects to be counted.

Specifically, it is as follows.
Here, for convenience of explanation, of the M eigen beams y _m (n) hats formed by the target range bin eigen beam forming unit 9, the first to Kth eigen beams y ₁ (n) hat to y _The target signal is included in the _K (n) hat, but the target signal is not included in the (K + 1) th to Mth eigenbeams y _{K + 1} (n) hat to y _M (n) hat. It is assumed that only receiver noise is included.
Based on this, we will proceed with explanations using specific situations. For example, consider the case where M = 8 and K = 2.

しかし、例えば、受信機雑音だけを含む固有ビームｙ₄（ｎ）ハットのコヒーレント積分結果から検波された目標ドップラビンｌ_sのコヒーレント積分値｜Ｚ_m（ｌ_s）ハット｜²が偶発的に固定スレッショルドＴ_m ⁽¹⁾より大きくなることで、固定スレッショルド判定部１１ａ−４の判定結果ｂ（４）に誤りが発生すると、固定スレッショルド判定部１１ａ−１〜１１ａ−８の判定結果ｂ（１）〜ｂ（８）は、下記の式（２３）のようになる。

Under these circumstances, the determination results b (1) to b (8) of the fixed threshold determination units 11a-1 to 11a-8 are essentially as shown in the following equation (22).

However, for example, the coherent integration value | Z _m (l _s ) hat | ^{2 of} the target Doppler bin l _s detected from the coherent integration result of the eigen beam y ₄ (n) hat including only receiver noise is accidentally fixed threshold. If an error occurs in the determination result b (4) of the fixed threshold determination unit 11a-4 by becoming larger than T _m ⁽¹⁾ , the determination results b (1) to 11a-8 of the fixed threshold determination units 11a-1 to 11a-8 are generated. b (8) is represented by the following equation (23).

In the first embodiment, the number-of-arrivals recognition unit 11b in FIG. 2 also includes the determination result b (4) of the fixed threshold determination unit 11a-4 in which erroneous detection has occurred in the count target. The estimation result is 3 (K hat = 3), and the arrival wave determination probability deteriorates.
In the second embodiment, in order to prevent the deterioration of the arrival wave determination probability even if the above-described erroneous detection occurs, the arrival wave number authorization unit 11c uses the determination result of “1” as the wave number K hat of the arrival wave. Only the determination results that are consecutive are counted, and even if the previous determination result of “0” is “1”, it is excluded from the counting targets.
When the determination results b (1) to b (8) of the fixed threshold determination units 11a-1 to 11a-8 are the expression (23), the determination result b (1) of the fixed threshold determination unit 11a-1 and the fixed threshold determination unit Since the determination result b (2) of 11a-2 is “1” continuously, it is counted, but the determination result b (4) of the fixed threshold determination unit 11a-4 is the previous fixed threshold. Since the determination result b (3) of the determination unit 11a-3 is “0”, it is excluded from the counting target.
As a result, the estimation result of the wave number of the incoming wave becomes 2 (K hat = 2), and the deterioration of the incoming wave determination probability is suppressed.

Embodiment 3 FIG.
FIG. 4 is a block diagram showing an arrival wave number estimating apparatus according to Embodiment 3 of the present invention. In the figure, the same reference numerals as those in FIG.
The arrival wave number estimation unit 20 calculates an adaptive threshold T _m ⁽²⁾ , which is a threshold necessary for achieving a certain erroneous determination probability P _fa ^(Proposed) required by a preset sensor system, and a coherent integration processing unit The coherent integration value | Z _m (l _s ) of the target Doppler bin l _s detected by the signal detector 6 from among the M pulse Doppler filter signals Z _m (l) hats output from 10-1 to 10-M. ) Hat | ² is acquired, and the process of counting the number of coherent integral values | Z _m (l _s ) hat | ² larger than the adaptive threshold T _m ^{(2) is performed} as the wave number K hat of the incoming wave. The arrival wave number estimation unit 20 constitutes arrival wave number estimation means.

In the example of FIG. 4, the sum beam forming unit 4, the coherent integration processing unit 5, the signal detection unit 6, the correlation matrix calculation unit 7, the eigenvalue / eigenvector calculation unit 8, and the target range bin eigenbeam formation, which are components of the arrival wave number estimation apparatus. Each of the unit 9, the coherent integration processing units 10-1 to 10-M and the arrival wave number estimation unit 20 is configured by dedicated hardware (for example, a semiconductor integrated circuit on which a CPU is mounted or a one-chip microcomputer). However, these components may be configured by a computer.
When these components are configured by a computer, the sum beam forming unit 4, the coherent integration processing unit 5, the signal detecting unit 6, the correlation matrix calculating unit 7, the eigenvalue / eigenvector calculating unit 8, the target range bin eigenbeam forming unit 9, the coherent A program describing the processing contents of the integration processing units 10-1 to 10-M and the arrival wave number estimation unit 20 is stored in the memory of a computer, and the CPU of the computer executes the program stored in the memory. You can do it.

FIG. 5 is a block diagram showing the arrival wave number estimation unit 20 of the arrival wave number estimation apparatus according to Embodiment 3 of the present invention.
In FIG. 5, the adaptive threshold coefficient calculation unit 20a uses a certain erroneous determination probability P _fa ^(Proposed) and the number of snapshots N required by the sensor system, and the erroneous determination probability P _fa ^(Proposed) becomes constant. A process for calculating the adaptive threshold coefficient α _m ⁽²⁾ is performed.
The adaptive threshold determination units 20b-1 to 20b-M are average values (σ _m ^{(1)) of} noise detection outputs from the pulse Doppler filter signals Z _m (l) hats output from the coherent integration processing units 10-1 to 10-M. Hat) ² , and a fixed erroneous determination probability P _fa from the average value (σ _m ⁽¹⁾ hat) ^{2 of} the noise detection output and the adaptive threshold coefficient α _m ⁽²⁾ calculated by the adaptive threshold coefficient calculation unit 20a. A process of calculating an adaptive threshold T _m ⁽²⁾ necessary for achieving ^(Proposed) is performed.
In addition, the threshold value calculation part is comprised from the adaptive threshold coefficient calculation part 20a and the adaptive threshold determination parts 20b-1 to 20b-M.

The adaptive threshold determination units 20b-1 to 20b-M are signal detection units 6 from among the M pulse Doppler filter signals Z _m (l) hats output from the coherent integration processing units 10-1 to 10-M. To obtain a coherent integral value | Z _m (l _s ) hat | ² of the target Doppler bin l _s detected by, and the coherent integral value | Z _m (l _s ) hat | ² is larger than the adaptive threshold T _m ^(2). A process of determining whether or not is performed.
Of the determination results b (1) to b (M) of the adaptive threshold determination units 20b-1 to 20b-M, the arrival wave number recognition unit 20c has a coherent integral value | Z _m (l _s ) hat | ^{2 that} is an adaptive threshold T. _m ⁽²⁾ Counts the number of determination results indicating that the value is larger than the number, and performs processing to recognize the number as the wave number of the incoming wave.

Ｎ’＝Ｎ−Ｎ_g−１

ただし、Ｎ_gは目標ドップラビンｌ_sの両側にある複数のドップラビン（以降、「ガードドップラビン」と称する）である。 Next, the operation will be described.
Other than the arrival wave number estimation unit 20 is the same as that of the first embodiment, so only the processing content of the arrival wave number estimation unit 20 will be described.
The adaptive threshold coefficient calculation unit 20a of the arrival wave number estimation unit 20 uses a certain erroneous determination probability P _fa ^(Proposed) required by the sensor system and the number of snapshots N as shown in the following equation (24). Then, an adaptive threshold coefficient α _m ^{(2) at} which the erroneous determination probability P _fa ^(Proposed) becomes constant is calculated.

N ′ = N−N _g −1

Here, N _g is a plurality of Doppler bins (hereinafter referred to as “guard Doppler bins”) on both sides of the target Doppler bin l _s .

Adaptive threshold determination unit 20b-1~20b-M receives the pulse Doppler filtered signal Z _m (l) hat from the coherent integration section 10-1 to 10-M, the pulse Doppler filter signal Z _m (l) hat From this, the average value of the noise detection output (σ _m ⁽¹⁾ hat) ² is calculated.
That is, the adaptive threshold determination units 20b-1 to 20b-M obtain the target Doppler bin l _s and the guard from the detection outputs (= pulse Doppler filter signal Z _m (l) hat) corresponding to the number of snapshots N. By excluding the detection output of Doppler bin N _g , the average value (σ _m ⁽¹⁾ hat) ² of noise detection outputs in N ′ Doppler bins consisting only of receiver noise is calculated.

When the adaptive threshold determination units 20b-1 to 20b-M calculate the average value (σ _m ⁽¹⁾ hat) ² of the noise detection output, as shown in the following equation (25), the average value of the noise detection output (Σ _m ⁽¹⁾ hat) ² and the adaptive threshold coefficient T _m ⁽² required to achieve a certain erroneous determination probability P _fa ^(Proposed) from the adaptive threshold coefficient α _m ⁽²⁾ calculated by the adaptive threshold coefficient calculator 20a. ⁾ Is calculated.

After calculating the adaptive threshold T _m ⁽²⁾ , the adaptive threshold determination units 20b-1 to 20b-M calculate the pulse Doppler filter signal Z _m (l) hat output from the coherent integration processing units 10-1 to 10-M. The coherent integral value | Z _m (l _s ) hat | ² (or | Z _m (l _s ) hat |) of the target Doppler bin l _s detected by the signal detection unit 6 is detected.
Adaptive threshold determination unit 20b-1~20b-M is the coherent integration value of the target Doppler bin l _s | when detects the ^2, the coherent integration values | | Z _m (l _s) hat Z _m (l _s) hat | ² is It is determined whether or not it is larger than the adaptive threshold T _m ^{(2), and} a determination result b (m) as shown in the above equation (20) is output to the arrival wave number recognition unit 20c.

Upon receiving the determination result b (m) from the adaptive threshold determination units 20b-1 to 20b-M, the arrival wave number recognition unit 20c includes the coherent integration value | Z _m (l _s ) among the determination results b (m). The number of determination results indicating that hat | ² is larger than the adaptive threshold T _m ⁽²⁾ is counted.
That is, as shown in the above equation (21), the arrival wave number certification unit 20c has b (m) = 1 among the determination results b (m) output from the adaptive threshold determination units 20b-1 to 20b-M. The number K of the determination results is counted.
The incoming wave number recognition unit 20c, after counting the number K hat of the determination result of b (m) = 1, recognizes the number K hat as the wave number (target number) of the incoming wave.

As is apparent from the above, according to the third embodiment, the adaptive threshold determination units 20b-1 to 20b-M have the adaptive threshold T _m ⁽² required for achieving a certain erroneous determination probability P _fa ^{(Proposed). ) And} the determination is made using the adaptive threshold T _m ^{(2). In} addition to the effects similar to those of the first embodiment, the average value of the noise detection output (σ _m ⁽¹⁾ (Hat) Regardless of the fluctuation of ² , there is an effect that the misjudgment probability of the sensor system can be controlled to a desired constant value.

Embodiment 4 FIG.
FIG. 6 is a block diagram showing an arrival wave number estimation unit 20 of the arrival wave number estimation apparatus according to Embodiment 4 of the present invention. In the figure, the same reference numerals as those in FIG.
Similarly to the third embodiment, the pulse threshold Doppler filter signal Z _m (l) hat output from the coherent integration processing units 10-1 to 10-M is input to the adaptive threshold determination units 20b-1 to 20b-M. Is done.
That is, the pulse Doppler filter signal Z ₁ (l) hat input to the adaptive threshold determination unit 20 b-1 is an eigen beam obtained by multiplying the received signal output from the AD converter 3-1 by the eigen vector e ₁ hat. y ₁ (n) is a result of the coherent integration of the hat, and the pulse Doppler filter signal Z ₂ (l) hat input to the adaptive threshold determination unit 20b-2 corresponds to the received signal output from the AD converter 3-2. This is a coherent integration result of the eigen beam y ₂ (n) hat multiplied by the eigen vector e ₂ hat.
The pulse Doppler filter signal Z _M (l) hat input to the adaptive threshold determination unit 20b-M is an eigen beam obtained by multiplying the received signal output from the AD converter 3-M by the eigen vector e _M hat. y _M (n) Coherent integration result of the hat.
In this case, the eigenvector e ₁ hat, ..., eigenvalues corresponding to e _M hat lambda ₁ hat, ..., the magnitude relationship of lambda _M hat are as described above for formula (9).

The arrival wave number recognition unit 20d estimates the wave number K hat of the arrival wave by the same method as the arrival wave number recognition unit 20c of FIG. 5, but at this time, the determination result b (m) of a certain adaptive threshold determination unit 20b-m ) Is 1 (which indicates that the coherent integral value | Z _m (l _s ) hat | ² is larger than the adaptive threshold T _m ⁽²⁾ ), one before the adaptive threshold determination unit 20b-m. If the determination result b (m-1) of the adaptive threshold determination units 20b- (m-1) arranged side by side is 0 (coherent integral value | Z _m (l _s ) hat | ² is the adaptive threshold T _m ^(2). The determination result b (m) of the adaptive threshold determination unit 20b-m is excluded from the number of objects to be counted.

Specifically, it is as follows.
Here, for convenience of explanation, of the M eigen beams y _m (n) hats formed by the target range bin eigen beam forming unit 9, the first to Kth eigen beams y ₁ (n) hat to y _The target signal is included in the _K (n) hat, but the target signal is not included in the (K + 1) th to Mth eigenbeams y _{K + 1} (n) hat to y _M (n) hat. It is assumed that only receiver noise is included.
Based on this, we will proceed with explanations using specific situations. For example, consider the case where M = 8 and K = 2.

しかし、例えば、受信機雑音だけを含む固有ビームｙ₄（ｎ）ハットのコヒーレント積分結果から検波された目標ドップラビンｌ_sのコヒーレント積分値｜Ｚ_m（ｌ_s）ハット｜²が偶発的に適応スレッショルドＴ_m ⁽²⁾より大きくなることで、適応スレッショルド判定部２０ｂ−４の判定結果ｂ（４）に誤りが発生すると、適応スレッショルド判定部２０ｂ−１〜２０ｂ−８の判定結果ｂ（１）〜ｂ（８）は、下記の式（２７）のようになる。

Under these circumstances, the determination results b (1) to b (8) of the adaptive threshold determination units 20b-1 to 20b-8 are essentially as shown in the following equation (26).

However, for example, the coherent integration value | Z _m (l _s ) hat | ^{2 of} the target Doppler bin l _s detected from the coherent integration result of the eigen beam y ₄ (n) hat including only the receiver noise is an adaptive threshold. When an error occurs in the determination result b (4) of the adaptive threshold determination unit 20b-4 by becoming larger than T _m ⁽²⁾ , the determination results b (1) to 20b-8 of the adaptive threshold determination units 20b-1 to 20b-8 are generated. b (8) is represented by the following equation (27).

In the third embodiment, the number-of-arrivals recognition unit 20c in FIG. 5 also includes the determination result b (4) of the adaptive threshold determination unit 20b-4 in which erroneous detection has occurred in the counting target. The estimation result is 3 (K hat = 3), and the arrival wave determination probability deteriorates.
In the fourth embodiment, in order to prevent the deterioration of the arrival wave determination probability even if the above-described erroneous detection occurs, the arrival wave number authorization unit 20d determines the determination result of “1” as the wave number K hat of the arrival wave. Only the determination results that are consecutive are counted, and even if the previous determination result of “0” is “1”, it is excluded from the counting targets.
When the determination results b (1) to b (8) of the adaptive threshold determination units 20b-1 to 20b-8 are the expression (27), the determination result b (1) of the adaptive threshold determination unit 20b-1 and the adaptive threshold determination unit Since the determination result b (2) of 20b-2 is “1” continuously, it is counted, but the determination result b (4) of the adaptive threshold determination unit 20b-4 is the previous adaptive threshold. Since the determination result b (3) of the determination unit 20b-3 is “0”, it is excluded from the counting target.
As a result, the estimation result of the wave number of the incoming wave becomes 2 (K hat = 2), and the deterioration of the incoming wave determination probability is suppressed.

Embodiment 5. FIG.
FIG. 7 is a block diagram showing an arrival wave number estimation apparatus according to Embodiment 5 of the present invention. In the figure, the same reference numerals as those in FIG.
The noise range bin eigen beam forming unit 31 is other than the target range bin detected by the signal detection unit 6 among the received signal vectors x (n) that are M received signals output from the AD converters 3-1 to 3-M. to the received signal vector x of the noise range bin is a range bin (n), the eigenvector e _m hat calculated by multiplying each by eigenvalue-eigenvector computing section 8, M pieces of eigenbeams y _m '(n) The process which forms a hat is implemented.
The coherent integration processing units 32-1 to 32 -M coherently integrate the eigen beam y _m ′ (n) hat output from the noise range bin eigen beam forming unit 31 by DFT or the like, and a pulse Doppler filter signal which is a result of the coherent integration. A process of outputting the Z _m ′ (l) hat to the arrival wave number estimation unit 33 is performed.
The noise range bin eigen beam forming unit 31 and the coherent integration processing units 32-1 to 32-M constitute a second coherent integration unit.
In the fifth embodiment, the target range bin eigen beam forming unit 9 and the coherent integration processing units 10-1 to 10-M constitute the first coherent integration means.

The incoming wave number estimation unit 33 uses the pulse Doppler filter signal Z _m ′ (l) hat output from the coherent integration processing units 32-1 to 32 -M to perform a certain error required by a preset sensor system. An adaptive threshold T _m ⁽³⁾ , which is a threshold necessary for achieving the determination probability P _fa ^(Proposed) , is calculated, and M pulse Doppler filter signals Z _m output from the coherent integration processing units 10-1 to 10-M are calculated. (L) The coherent integral value | Z _m (l _s ) hat | ² of the target Doppler bin l _s detected by the signal detection unit 6 is acquired from the hat, and the adaptive threshold is set as the wave number K hat of the incoming wave. A process of counting the number of coherent integral values | Z _m (l _s ) hat | ² greater than T _m ⁽³⁾ is performed. The arrival wave number estimation unit 33 constitutes arrival wave number estimation means.

In the example of FIG. 7, a sum beam forming unit 4, a coherent integration processing unit 5, a signal detection unit 6, a correlation matrix calculation unit 7, an eigenvalue / eigenvector calculation unit 8, and target range bin eigenbeam formation, which are components of the arrival wave number estimation apparatus. Unit 9, coherent integration processing units 10-1 to 10 -M, noise range bin eigen beam forming unit 31, coherent integration processing units 32-1 to 32 -M, and arrival wave number estimation unit 33, respectively. A semiconductor integrated circuit on which a CPU is mounted or a one-chip microcomputer or the like is assumed. However, these components may be configured by a computer.
When these components are configured by a computer, the sum beam forming unit 4, the coherent integration processing unit 5, the signal detecting unit 6, the correlation matrix calculating unit 7, the eigenvalue / eigenvector calculating unit 8, the target range bin eigenbeam forming unit 9, the coherent Programs describing the processing contents of the integration processing units 10-1 to 10-M, the noise range bin eigen beam forming unit 31, the coherent integration processing units 32-1 to 32-M, and the arrival wave number estimation unit 33 are stored in the memory of the computer. The program may be stored and the CPU of the computer may execute the program stored in the memory.

FIG. 8 is a block diagram showing the arrival wave number estimation unit 33 of the arrival wave number estimation apparatus according to Embodiment 5 of the present invention.
In FIG. 8, the adaptive threshold coefficient calculation unit 33a uses a certain erroneous determination probability P _fa ^(Proposed) , the number of snapshots N and the number of noise range bins N ″ required in the sensor system, to determine the erroneous determination probability P _fa ^{(Proposed )} Is performed to calculate an adaptive threshold coefficient α _m ⁽³⁾ that makes constant.
The adaptive threshold determination units 33b-1 to 33b-M use the pulse Doppler filter signal Z _m ′ (l) hat output from the coherent integration processing units 32-1 to 32-M and use the average value (σ _m ⁽²⁾ hat) ² is calculated, and a certain error is calculated from the average value (σ _m ⁽²⁾ hat) ^{2 of} the noise detection output and the adaptive threshold coefficient α _m ⁽³⁾ calculated by the adaptive threshold coefficient calculation unit 33a. A process of calculating an adaptive threshold T _m ⁽³⁾ necessary for achieving the determination probability P _fa ^(Proposed) is performed.
In addition, the threshold value calculation part is comprised from the adaptive threshold coefficient calculation part 33a and the adaptive threshold determination parts 33b-1 to 33b-M.

Also, the adaptive threshold determination units 33b-1 to 33b-M are signal detection units 6 from among the M pulse Doppler filter signals Z _m (l) hats output from the coherent integration processing units 10-1 to 10-M. The coherent integral value | Z _m (l _s ) hat | ² of the target Doppler bin l _s detected by the above is acquired, and the coherent integral value | Z _m (l _s ) hat | ² is larger than the adaptive threshold T _m ^(3). A process of determining whether or not is performed.
The incoming wave identification portion 33c of the determination result of the adaptive threshold determination unit 33b-1~33b-M b (1 ) ~b (M), the coherent integration value | Z _m (l _s) hat | ² adaptation threshold T _m ⁽³⁾ Counts the number of determination results indicating that the value is larger than the number, and performs processing to recognize the number as the wave number of the incoming wave.

Next, the operation will be described.
Since the noise range bin eigen beam forming unit 31, the coherent integration processing units 32-1 to 32-M and the arrival wave number estimating unit 33 are the same as those in the first embodiment, the noise range bin eigen beam forming unit 31, coherent integration processing is performed. Only the processing contents of the units 32-1 to 32-M and the arrival wave number estimation unit 33 will be described.
It is assumed that only the receiver noise is included in the received signal vector x (n) of the range bin other than the target range bin detected by the signal detection unit 6.

When the noise range bin eigen beam forming unit 31 receives the eigen vector e ₁ hat,..., E _M hat from the eigen value / eigen vector calculating unit 8, as shown in the above equation (13), the AD converter 3-1 to 3-1. Among the received signal vectors x (n) that are M received signals output from 3-M, the received signal vector x (n) of a noise range bin that is a range bin other than the target range bin is its eigenvector e ₁ hat. ,..., E _M hats are multiplied to form M eigen beams y _m ′ (n) hats, and the M eigen beams y _m ′ (n) hats are coherent integration processing unit 32-1. Output to ~ 32-M.
When the coherent integration processing units 32-1 to 32 -M receive the eigen beam y _m ′ (n) hat from the noise range bin eigen beam forming unit 31, as shown in the above equation (18), the eigen beam y _m The '(n) hat is coherently integrated by DFT or the like, and the pulse Doppler filter signal Z _m ′ (l) hat which is the result of the coherent integration is output to the arrival wave number estimating unit 33.

As shown in the following equation (28), the adaptive threshold coefficient calculation unit 20a of the arrival wave number estimation unit 20 has a certain erroneous determination probability P _fa ^(Proposed) required by the sensor system, the number of snapshots N, and the number of noise range bins. N ″ is used to calculate an adaptive threshold coefficient α _m ⁽³⁾ that makes the erroneous determination probability P _fa ^(Proposed) constant.

Upon receiving the pulse Doppler filter signal Z _m ′ (l) hat output from the coherent integration processing units 32-1 to 32 -M, the adaptive threshold determination units 33 b-1 to 33 b -M receive the pulse Doppler filter signal Z _m. '(L) The average value (σ _m ⁽²⁾ hat) ² of the noise detection output of each Doppler bin in the hat is calculated.
When the adaptive threshold determination units 33b-1 to 33b-M calculate the average value (σ _m ⁽²⁾ hat) ² of the noise detection output, as shown in the following equation (29), the average value of the noise detection output (Σ _m ⁽²⁾ hat) ² and the adaptive threshold coefficient T _m ⁽³ required to achieve a certain erroneous determination probability P _fa ^(Proposed) from the adaptive threshold coefficient α _m ⁽³⁾ calculated by the adaptive threshold coefficient calculator 33a. ⁾ Is calculated.

After calculating the adaptive threshold T _m ⁽³⁾ , the adaptive threshold determination units 33b-1 to 33b-M calculate the pulse Doppler filter signal Z _m (l) hat output from the coherent integration processing units 10-1 to 10-M. The coherent integral value | Z _m (l _s ) hat | ² (or | Z _m (l _s ) hat |) of the target Doppler bin l _s detected by the signal detection unit 6 is detected.
Adaptive threshold determination unit 33b-1~33b-M is the coherent integration value of the target Doppler bin l _s | when detects the ^2, the coherent integration values | | Z _m (l _s) hat Z _m (l _s) hat | ² is It is determined whether or not it is larger than the adaptive threshold T _m ^{(3), and} a determination result b (m) as shown in the above equation (20) is output to the arrival wave number recognition unit 33c.

Upon receiving the determination result b (m) from the adaptive threshold determination units 33b-1 to 33b-M, the arrival wave number recognition unit 33c includes the coherent integration value | Z _m (l _s ) of the determination results b (m). The number of determination results indicating that hat | ² is larger than the adaptive threshold T _m ⁽³⁾ is counted.
That is, as shown in the above equation (21), the arrival wave number authorization unit 33c has b (m) = 1 among the determination results b (m) output from the adaptive threshold determination units 33b-1 to 33b-M. The number K of the determination results is counted.
The incoming wave number authorization unit 33c, when counting the number K hat of the determination result of b (m) = 1, authorizes the number K hat as the wave number (target number) of the incoming wave.

As is apparent from the above, according to the fifth embodiment, the adaptive threshold determination units 33b-1 to 33b-M output the pulse Doppler filter signal Z _m output from the coherent integration processing units 32-1 to 32-M. '(L) A hat is used to calculate an adaptive threshold T _m ⁽³⁾ necessary to achieve a certain erroneous determination probability P _fa ^(Proposed) , and the adaptive threshold T _m ⁽³⁾ is used for determination. In addition to the same effects as in the first embodiment, the misdetection probability of the sensor system is set to a desired constant value regardless of the fluctuation of the average value (σ _m ⁽¹⁾ hat) ² of the noise detection output. It is possible to control to a higher level and to achieve a higher target number determination probability.

Embodiment 6 FIG.
FIG. 9 is a block diagram showing the arrival wave number estimation unit 33 of the arrival wave number estimation apparatus according to Embodiment 6 of the present invention. In the figure, the same reference numerals as those in FIG.
Similarly to the fifth embodiment, the adaptive threshold determination units 33b-1 to 33b-M output from the coherent integration processing units 10-1 to 10-M and the coherent integration processing units 32-1 to 32-M. Pulse Doppler filter signals Z _m (l) hat and Z _m ′ (l) hat are input.
That is, the pulse Doppler filter signals Z ₁ (l) hat and Z ₁ ′ (l) hat input to the adaptive threshold determination unit 33b-1 are the eigenvector e for the received signal output from the AD converter 3-1. _The pulse Doppler filter signal Z ₂ (l) hat, Z ₂ ′ (l) hat, which is a coherent integration result of the eigen beam y ₁ (n) hat multiplied by ₁ hat and is input to the adaptive threshold determination unit 33b-2 Is a coherent integration result of the eigen beam y ₂ (n) hat obtained by multiplying the reception signal output from the AD converter 3-2 by the eigen vector e ₂ hat.
The pulse Doppler filter signals Z _M (l) hat and Z _M ′ (l) hat input to the adaptive threshold determination unit 33b-M are eigenvectors e with respect to the received signal output from the AD converter 3-M. _This is a coherent integration result of the eigen beam y _M (n) hat multiplied by the _M hat.
In this case, the eigenvector e ₁ hat, ..., eigenvalues corresponding to e _M hat lambda ₁ hat, ..., the magnitude relationship of lambda _M hat are as described above for formula (9).

The arrival wave number recognition unit 33d estimates the wave number K hat of the arrival wave by a method similar to that of the arrival wave number recognition unit 33c of FIG. 8, but at this time, the determination result b (m) of a certain adaptive threshold determination unit 33b-m ) Is 1 (which indicates that the coherent integral value | Z _m (l _s ) hat | ² is larger than the adaptive threshold T _m ⁽³⁾ ), one before the adaptive threshold determination unit 33b-m. If the determination result b (m-1) of the adaptive threshold determination units 33b- (m-1) arranged side by side is 0 (coherent integral value | Z _m (l _s ) hat | ² is the adaptive threshold T _m ^(3). In other words, the determination result b (m) of the adaptive threshold determination unit 33b-m is excluded from the number of objects to be counted.

Specifically, it is as follows.
Here, for convenience of explanation, of the M eigen beams y _m (n) hats formed by the target range bin eigen beam forming unit 9, the first to Kth eigen beams y ₁ (n) hat to y _The target signal is included in the _K (n) hat, but the target signal is not included in the (K + 1) th to Mth eigenbeams y _{K + 1} (n) hat to y _M (n) hat. It is assumed that only receiver noise is included.
Based on this, we will proceed with explanations using specific situations. For example, consider the case where M = 8 and K = 2.

しかし、例えば、受信機雑音だけを含む固有ビームｙ₄（ｎ）ハットのコヒーレント積分結果から検波された目標ドップラビンｌ_sのコヒーレント積分値｜Ｚ_m（ｌ_s）ハット｜²が偶発的に適応スレッショルドＴ_m ⁽³⁾より大きくなることで、適応スレッショルド判定部３３ｂ−４の判定結果ｂ（４）に誤りが発生すると、適応スレッショルド判定部３３ｂ−１〜３３ｂ−８の判定結果ｂ（１）〜ｂ（８）は、下記の式（３１）のようになる。

Under these circumstances, the determination results b (1) to b (8) of the adaptive threshold determination units 33b-1 to 33b-8 are essentially as shown in the following equation (30).

However, for example, the coherent integration value | Z _m (l _s ) hat | ^{2 of} the target Doppler bin l _s detected from the coherent integration result of the eigen beam y ₄ (n) hat including only the receiver noise is an adaptive threshold. If an error occurs in the determination result b (4) of the adaptive threshold determination unit 33b-4 by becoming larger than T _m ⁽³⁾ , the determination results b (1) to 33b-8 of the adaptive threshold determination unit 33b-1 to 33b-8 b (8) becomes like the following formula | equation (31).

In the above-described fifth embodiment, the arrival wave number authorization unit 33c of FIG. 8 also includes the determination result b (4) of the adaptive threshold determination unit 33b-4 in which erroneous detection has occurred in the count target. The estimation result is 3 (K hat = 3), and the arrival wave determination probability deteriorates.
In the sixth embodiment, in order to prevent deterioration of the arrival wave determination probability even if the above-described erroneous detection occurs, the arrival wave number authorization unit 33d uses the determination result of “1” as the wave number K hat of the arrival wave. Only the determination results that are consecutive are counted, and even if the previous determination result of “0” is “1”, it is excluded from the counting targets.
When the determination results b (1) to b (8) of the adaptive threshold determination units 33b-1 to 33b-8 are the expression (31), the determination result b (1) of the adaptive threshold determination unit 33b-1 and the adaptive threshold determination unit Since the determination result b (2) of 33b-2 is “1” continuously, it is counted, but the determination result b (4) of the adaptive threshold determination unit 33b-4 is the previous adaptive threshold. Since the determination result b (3) of the determination unit 33b-3 is “0”, it is excluded from the counting target.
As a result, the estimation result of the wave number of the incoming wave becomes 2 (K hat = 2), and the deterioration of the incoming wave determination probability is suppressed.

  In the present invention, within the scope of the invention, any combination of the embodiments, or any modification of any component in each embodiment, or omission of any component in each embodiment is possible. .

  1-1 to 1-M element antenna (receiving unit), 2-1 to 2-M receiver (receiving unit), 3-1 to 3-M AD converter (receiving unit), 4 sum beam forming unit (target) Detection means), 5 coherent integration processing section (target detection means), 6 signal detection section (target detection means), 7 correlation matrix calculation section (eigenvector calculation means), 8 eigenvalue / eigenvector calculation section (eigenvector calculation means), 9 target Range bin eigen beam forming unit (coherent integration unit, first coherent integration unit), 10-1 to 10-M coherent integration processing unit (coherent integration unit, first coherent integration unit), 11 arrival wave number estimation unit (arrival wave number) Estimation means), 11a-1 to 11a-M fixed threshold determination unit, 11b, 11c arrival wave number recognition unit, 20 arrival wave number estimation unit (arrival wave number estimation unit) ), 20a Adaptive threshold coefficient calculation unit (threshold value calculation unit), 20b-1 to 20b-M Adaptive threshold value determination unit (threshold value calculation unit), 20c, 20d Arrival wave number certification unit, 31 Noise range bin eigen beam forming unit (second Coherent integration unit), 32-1 to 32-M coherent integration processing unit (second coherent integration unit), 33 arrival wave number estimation unit (arrival wave number estimation unit), 33a adaptive threshold coefficient calculation unit (threshold calculation unit), 33b −1 to 33b-M Adaptive threshold value determination unit (threshold value calculation unit), 33c, 33d Arrival wave number certification unit.

Claims (9)

A plurality of receiving means for receiving a reflected wave reflected by the target and outputting a reception signal of the reflected wave;
A sum beam of reception signals output from the plurality of receiving means is formed, the sum beam is coherently integrated, and a target range bin that is a range bin in which a target signal exists from the coherent integration result and the target signal exist Target detection means for detecting target Doppler bins, which are Doppler bins,
An eigenvector calculating unit that calculates a correlation matrix of a plurality of received signals in the target range bin among the received signals output from the plurality of receiving units, and calculates an eigenvector corresponding to each eigenvalue constituting the correlation matrix; ,
Coherent integration means for multiplying a plurality of received signals in the target range bin by eigenvectors calculated by the eigenvector calculation means, respectively, and coherently integrating each multiplication result;
The number of arrival waves for obtaining the coherent integration value of the target Doppler bin from a plurality of coherent integration results in the coherent integration means, and counting the number of the coherent integration values larger than a preset threshold as the number of arrival waves. An arrival wave number estimation device comprising: an estimation means. The arrival wave number estimating means includes:
Among the coherent integration results of the coherent integration means, obtain a coherent integration value of the target Doppler bin, a plurality of fixed threshold determination unit for determining whether the coherent integration value is larger than a preset threshold;
Among the determination results of the plurality of fixed threshold determination units, the number of determination results indicating that the coherent integral value is greater than the threshold value is counted, and a wave number certifying unit that certifies the number as the wave number of the incoming wave; The incoming wave number estimating apparatus according to claim 1, comprising: A plurality of eigenvectors calculated by the eigenvector calculating means are sorted in the order of the corresponding eigenvalues;
When the plurality of fixed threshold determination units are arranged according to the eigenvector sorting result related to the coherent integration result given from the coherent integration means,
The wave number recognition unit is arranged so that a fixed threshold is arranged one before a certain fixed threshold determination unit even if a determination result of a certain fixed threshold determination unit indicates that the coherent integral value is larger than the threshold. 3. The determination result of a certain fixed threshold determination unit is excluded from the number of counting targets if the determination result of the determination unit indicates that the coherent integral value is smaller than the threshold value. Arrival wave number estimation device. A plurality of receiving means for receiving a reflected wave reflected by the target and outputting a reception signal of the reflected wave;
A sum beam of reception signals output from the plurality of receiving means is formed, the sum beam is coherently integrated, and a target range bin that is a range bin in which a target signal exists from the coherent integration result and the target signal exist Target detection means for detecting target Doppler bins, which are Doppler bins,
An eigenvector calculating unit that calculates a correlation matrix of a plurality of received signals in the target range bin among the received signals output from the plurality of receiving units, and calculates an eigenvector corresponding to each eigenvalue constituting the correlation matrix; ,
Coherent integration means for multiplying a plurality of received signals in the target range bin by eigenvectors calculated by the eigenvector calculation means, respectively, and coherently integrating each multiplication result;
A threshold value corresponding to a preset error determination probability is calculated, and a coherent integration value of the target Doppler bin is obtained from a plurality of coherent integration results in the coherent integration unit, and the wave number of an incoming wave is calculated from the threshold value. An arrival wave number estimation device comprising: an arrival wave number estimation means for counting the number of large coherent integration values. The arrival wave number estimating means includes:
A threshold value calculation unit for calculating a threshold value corresponding to a preset erroneous determination probability;
A plurality of adaptive threshold determinations for acquiring a coherent integration value of the target Doppler bin from the coherent integration results of the coherent integration means and determining whether the coherent integration value is larger than a threshold value calculated by the threshold value calculation unit. And
Among the determination results of the plurality of adaptive threshold determination units, a number of determination results indicating that the coherent integral value is larger than the threshold value is counted, and a wave number certifying unit that certifies the number as the wave number of the incoming wave; 5. The arrival wave number estimation apparatus according to claim 4, wherein the arrival wave number estimation apparatus comprises: A plurality of eigenvectors calculated by the eigenvector calculating means are sorted in the order of the corresponding eigenvalues;
When the plurality of adaptive threshold determination units are arranged according to the eigenvector sorting result related to the coherent integration result given from the coherent integration means,
The wave number recognition unit is arranged so that an adaptive threshold lined up before a certain adaptive threshold determination unit even if a determination result of a certain adaptive threshold determination unit indicates that the coherent integral value is larger than the threshold. 6. The determination result of an adaptive threshold determination unit is excluded from the number of counting targets if the determination result of the determination unit indicates that the coherent integration value is smaller than the threshold value. Arrival wave number estimation device. A plurality of receiving means for receiving a reflected wave reflected by the target and outputting a reception signal of the reflected wave;
A sum beam of reception signals output from the plurality of receiving means is formed, the sum beam is coherently integrated, and a target range bin that is a range bin in which a target signal exists from the coherent integration result and the target signal exist Target detection means for detecting target Doppler bins, which are Doppler bins,
An eigenvector calculating unit that calculates a correlation matrix of a plurality of received signals in the target range bin among the received signals output from the plurality of receiving units, and calculates an eigenvector corresponding to each eigenvalue constituting the correlation matrix; ,
First coherent integration means for multiplying a plurality of received signals in the target range bin by eigenvectors calculated by the eigenvector calculation means, respectively, and coherently integrating each multiplication result;
Among the received signals output from the plurality of receiving means, a plurality of received signals in range bins other than the target range bin are respectively multiplied by the eigenvectors calculated by the eigenvector calculating means, and each multiplication result is coherently integrated. Second coherent integrating means to
Using each coherent integration result by the second coherent integration means, a threshold value corresponding to a preset misjudgment probability is calculated, and the target coherent integration result is selected from among a plurality of coherent integration results in the first coherent integration means. An arrival wave number estimation device comprising: an arrival wave number estimation unit that obtains a coherent integration value of Doppler bins and counts the number of coherent integration values larger than the threshold as the wave number of the arrival wave. The arrival wave number estimating means includes:
Using a plurality of coherent integration results in the second coherent integration means, a threshold value calculation unit for calculating a threshold value corresponding to a preset erroneous determination probability;
A plurality of coherent integration values of the target Doppler bin are obtained from the coherent integration results of the first coherent integration means, and a plurality of determinations are made as to whether or not the coherent integration values are greater than a threshold value calculated by the threshold value calculation unit. An adaptive threshold determination unit;
Among the determination results of the plurality of adaptive threshold determination units, a number of determination results indicating that the coherent integral value is larger than the threshold value is counted, and a wave number certifying unit that certifies the number as the wave number of the incoming wave; 8. The arrival wave number estimation apparatus according to claim 7, wherein A plurality of eigenvectors calculated by the eigenvector calculating means are sorted in the order of the corresponding eigenvalues;
When the plurality of adaptive threshold determination units are arranged according to the eigenvector sorting result related to the coherent integration result given from the first coherent integration means,
The wave number recognition unit is arranged so that an adaptive threshold lined up before a certain adaptive threshold determination unit even if a determination result of a certain adaptive threshold determination unit indicates that the coherent integral value is larger than the threshold. 9. The determination result of an adaptive threshold determination unit is excluded from the number of counting targets if a determination result of the determination unit indicates that the coherent integration value is smaller than the threshold value. Arrival wave number estimation device.

Images