JP2007003325A - Method of calculating covariance matrix, weight calculation circuit, adaptive array antennas and radar installation - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of calculating a covariance matrix which enables the shortening of the throughput of adaptation weight calculation. <P>SOLUTION: In the method, target reflection signals received by way of antennas (#1 to #M) are stored in a data storage part 3 in accordance with a reception timing. The data storage part 3 has memories corresponding to a specific length (N times) processing range cells, which are split into a plurality (Q) on time axis, and covariance matrix in the range cells in the section is calculated. Therefore, in the complex number operation of covariance matrix calculation in N total processing range cells, the calculation times counts (N/Q) in comparison with the calculation times (N) of a conventional covariance matrix in each range cell by the sliding window method, and operation quantity can be drastically reduced. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention provides a covariance matrix calculation method suitable for detecting a reflected signal from a target by suppressing unnecessary waves by weight control, a weight calculation circuit using the covariance matrix calculation method, and a weight calculation circuit thereof. The present invention relates to an adopted adaptive array antenna, a radar apparatus incorporating the adaptive array antenna, and the like.

  In a pulse radar that uses a recent adaptive array antenna, by forming a reception composite beam that makes the directivity in the direction in which an unwanted wave such as an interference wave arrives zero by weight control on the received signal in the adaptive array antenna, that is, It is configured to detect the target by performing adapter null steering by phase and amplitude control.

  In an adaptive array antenna, weight control is required so that a received combined beam is properly formed even in an environment where a large number of delayed signals arrive or an environment where unnecessary waves such as clutter and interference waves exist. It is done.

  Therefore, in adaptive array antennas, a weight control method that employs a space-time adaptive signal processing (STAP) method improves SINR (Signal to Interference plus Noise Ratio) and improves the arrival direction of unwanted waves. Attention has been focused on being able to perform good beam formation with directivity close to zero.

  The spatio-temporal adaptive signal processing (STAP) method is to receive a target reflected signal by a plurality (M) of antennas (element antennas, that is, channels) arranged in an array, and the received signal corresponds to the received pulse width. It is assumed that a range cell with a specified width is stored in the corresponding cell positions of all processing range cells formed with a predetermined length on the time axis, and the target signal is assumed to be included from the stored data. A covariance matrix is calculated from data of a range cell excluding a range cell (referred to as a processing application range cell), that is, a cell that is assumed to be formed only from unwanted waves, and the beam synthesis circuit calculates an adaptation calculated based on the covariance matrix. Weight control is performed by weight.

  In weight control in this spatio-temporal adaptive signal processing method, covariance matrix calculation in the range cell is performed in the weight calculation circuit in order to calculate the adaptive weight, but so-called sliding window method is adopted for conventional covariance matrix calculation. Is done. (For example, refer nonpatent literature 1.).

  As described above, the covariance matrix operation is to perform a covariance matrix operation on cell data that is assumed to be formed only from unnecessary waves. In the covariance matrix operation by the sliding window method, FIG. As shown in FIG. 5, the calculation based on the training data in the front and rear portions is performed for all N processing range cells.

Therefore, the antenna is composed of M element antennas arranged in an array at equal intervals, the number of training of the covariance matrix (or also called the number of snapshots) is K, and the number of pulses of the pulse delay circuit in the beam synthesis circuit Is L, and the delay time in the pulse delay circuit is the pulse repetition time (period) T of the pulse reception signal, the covariance matrix in each range cell can be obtained by K × (M × L) ^two complex operations. It has been known.

Therefore, in the conventional covariance matrix operation employing the sliding window method, N × K × (M × L) is used to calculate adaptive (optimum) weights for all N processing range cells as described below. ^Two complex operations of the covariance matrix are required.

That is, when the direction matrix of the arrival direction of the received signal X is A, the complex amplitude vector is S, and the thermal noise given by the mean 0 and the variance σ ² is n, the received signal X is expressed by the following equation (1). Is done.

When a target signal is received by M element antennas #m (m: 1 to M) arranged in an array with an interval h, the wavelength of the received frequency signal is λ (Λ) and D arrivals. The steering vector a (θ _d ) that determines the arrival direction of the target signal d (d: 1 to _D ) is expressed by the following equation (2), and the direction matrix A _θ for the spatial sequence is expressed by the following equation (3).

Therefore, since the steering vector a (f _d ) for determining the arrival direction of the target signal d is expressed by the following equation (4), the direction matrix A _f for the time series is expressed by the following equation (5).

Therefore, the direction matrix A _{θ, f} is given by the following equation (7) using the spatio-temporal steering vector a (θ _d , f _d ) expressed by the following equation (6).

Here, since the training number is K, the covariance matrix R is expressed by the following equation (8).

  In the above equation (8), E |. | Represents an operation for obtaining an expected value (ensemble average).

Therefore, when the covariance matrix R is considered as an addition average of covariance matrices from time k to time (k + K−1), the covariance matrix R is expressed by the following equation (9).

In the above equation (7) _, the order of the direction matrix A _{θ, f} is the number of antennas (M) × the number of pulses (L), that is, (M × L), so the number of elements of the covariance matrix R is (M × L) ² , and the covariance matrix R is calculated by averaging the number of elements (M × L) ² matrix by the number of training (K) times.

In this way, since the covariance matrix R in each range cell is obtained by K × (M × L) ^two complex operations, in the conventional covariance matrix operation employing the sliding window method, all N blocks are obtained. To calculate the adaptive weight for the processing range cell, N × K × (M × L) ^two times of complex operation of the covariance matrix is required. In all the processing range cells having N processing range cells, N × K × (M × L) ^two times of the complex operation of the covariance matrix R is performed, and the adaptation calculated based on the calculated covariance matrix R is performed. Weights are supplied to the beam synthesis circuit, and beam formation is performed by suppressing the unnecessary waves in time and space by weight control in the beam synthesis circuit.

  However, as a method of reducing the amount of calculation of the covariance matrix in the weight calculation circuit, the total reception area in the azimuth direction, that is, the total number N of processing range cells is divided into a plurality of pieces, and the divided reception data of one section is obtained. There is also known a method (so-called Fixed, Slide & Freeze method) in which only the covariance matrix is calculated, and the calculated covariance matrix is applied to all the other remaining sections to calculate adaptive weights. (For example, see also Non-Patent Document 1.)

In any case, in an adaptive array antenna, a covariance matrix is obtained by complex operation, and based on the covariance matrix, for example, an MSN (Maximum Signal to Noise ratio) that operates on the assumption that the arrival direction of a desired signal is known is known. The adaptive weight is calculated by an algorithm, an MMSE (Minimum Mean Square Error) algorithm that is estimated from a correlation value between a received signal and a reference signal highly correlated with a desired signal.
"IEEE-Radar Conference 2004" 2004/3/29, published by MTI Lincoln Laboratory

  As described above, in the conventional covariance matrix calculation method based on the sliding window method, complex processing of the covariance matrix by training (averaging processing) is performed on all the processing range cells from cell data that is assumed to be formed only from unnecessary waves. Since the processing is performed for a number (N), the processing calculation amount in the weight calculation circuit increases as the number of data increases, and a long processing time is required.

  In the covariance matrix calculation method by the Fixed, Slide & Freeze method, the covariance matrix calculated for the received data of one division among the total number N of divided processing range cells is adapted to the remaining other divisions. Although the amount of calculation of the covariance matrix by the complex operation can be reduced, in this method, the adaptive weight for the N total processing range cells is determined as one, so that there is no unnecessary wave in the calculation target cell or the target signal When the covariance matrix of cell data including is calculated, there arises a problem that the SINR performance is greatly deteriorated.

  Accordingly, the present invention has been made to solve the above-described conventional problems, and provides a covariance matrix calculation method capable of reducing the processing load and the calculation time, a weight calculation circuit using this method, and this weight. It is an object of the present invention to provide an adaptive array antenna using a calculation circuit and a radar apparatus incorporating the adaptive array antenna.

  In the covariance matrix calculation method according to the first aspect of the present invention, the target reflected signal received via the antenna is stored in correspondence with the cell position along the reception timing of the processing range cell having a predetermined length. The processing range cell in which the reflection signal is stored is divided into a plurality on the time axis, and a covariance matrix in the range cell in the division is calculated in each divided division.

  A weight calculation circuit according to a second aspect of the present invention provides a covariance matrix calculation for calculating a covariance matrix in a range cell in each divided section of the processing range cell by the covariance matrix calculation method according to the first aspect. Means, a correlation vector calculation means for calculating a correlation vector from the covariance matrix in the range cell in each section calculated by the covariance matrix calculation means, and a correlation from the correlation vector calculated by the correlation vector calculation means A correlation matrix calculating unit that calculates a matrix; and an adaptive weight calculating unit that calculates an adaptive weight by performing an inverse matrix operation on the correlation matrix calculated by the correlation matrix calculating unit.

  The adaptive array antenna according to a fourth aspect of the present invention is configured such that the antenna includes a plurality of element antennas, and the adaptive weight calculated by the weight calculation circuit according to the second or third aspect provides the antenna A beam combining circuit for performing weight control on a target reflected signal received via an antenna and forming a reception combined beam at the antenna is provided.

  A radar apparatus according to a fifth aspect of the present invention includes the adaptive array antenna according to the fourth aspect, and is configured to detect a target based on an output signal from the adaptive array antenna. Features.

  A synthetic aperture radar apparatus according to a sixth aspect of the present invention incorporates the adaptive array antenna according to the fourth aspect and detects a target shape based on an output signal from the adaptive array antenna. It is structured.

  As described above, the covariance matrix calculation method of the present invention divides the processing range cell in which the target reflection signal is stored into a plurality of pieces on the time axis, and in each divided division, the range cell in the division Since the covariance matrix is calculated, the number of operations of the covariance matrix for all processing range cells in the range direction is the number of divisions. When this covariance matrix calculation method is applied to the weight calculation circuit, the adaptive weight calculation processing The amount can be greatly reduced.

  As described above, the weight calculation circuit of the present invention uses the covariance matrix calculation method according to the present invention, which can significantly reduce the processing amount, in calculating the covariance matrix necessary for weight calculation. Calculation time can be shortened.

  As described above, the adaptive array antenna according to the present invention employs the weight calculation circuit according to the present invention capable of shortening the time for weight calculation, so that a good combined beam can be formed in a short time.

  Since the radar apparatus of the present invention incorporates the adaptive array antenna of the present invention capable of forming a combined beam in a short time as described above, the target can be quickly captured.

  As described above, the synthetic aperture radar apparatus of the present invention incorporates an adaptive array antenna to form a combined beam in a short time, and thus can quickly grasp the target shape.

  Hereinafter, a weight calculation circuit employing the covariance matrix calculation method according to the present invention and an adaptive array antenna using the weight calculation circuit will be described in detail with reference to FIGS.

  FIG. 1 is a schematic configuration diagram of an adaptive array antenna employing one embodiment of a weight calculation circuit according to the present invention.

  In FIG. 1, an adaptive array antenna includes element antennas #m (m: 1 to M) arranged in an array with an element interval h and M element antennas #m. Commonly connected receiving unit 1, data storing unit 2 connected to the receiving unit 1 and configured by a memory such as a RAM, and connected to the data storing unit 2, a weight calculation circuit 31 and a beam synthesis circuit And a spatio-temporal adaptive signal processing unit 3 configured by 32 and a reference (reference) signal estimation that is connected to the data storage unit 2 and selectively outputs a reference (reference) estimation signal Ref to the weight calculation circuit 31 And a reference that can be switched to the reference estimation signal from the reference signal estimation unit 4 and the reference signal estimation unit 4 as necessary, and to supply a reference (reference) signal Ref to the weight calculation circuit 31. In order to obtain a reference (or reference estimation) signal having a high correlation with the target (desired wave) for the source (reference) signal generation unit 5, and the reference signal generation unit 5 and the reference signal estimation unit 4, the reception range cell The excitation unit 6 supplies a corresponding timing signal.

  The excitation unit 6 that outputs the timing signal synchronized with the reception range cell is generally a coherent signal synchronized with the transmission repetition signal T in the pulse radar transmission unit, and the transmission repetition signal transmitted from the pulse radar transmission unit (not shown). Generated based on

  Therefore, a plurality of (M) element antennas #m arranged linearly with an element interval h 1 receive the reflected pulse reception signal from the target and supply it to the receiving unit 1.

  The reflected pulse signal from the target supplied to the receiving unit 1 is amplified and subjected to frequency conversion, and then supplied to the data storage unit 2 and sequentially stored in a memory such as a RAM.

  The data accumulating unit 2 includes at least a range cell having a length corresponding to the pulse width of the received data in the range (distance) direction, that is, a plurality of (on the time axis, as shown in FIG. 2A). N) It has a storage capacity for all the processing range cells, and the data storage unit 2 stores the reception data supplied via the reception unit 1 in a memory corresponding to the reception cell position.

  The received data stored in the data storage unit 2 is appropriately read and supplied to the weight calculation circuit 31 and the beam synthesis circuit 32 of the spatiotemporal adaptive signal processing unit 3, respectively.

  As shown in FIG. 2B, the weight calculation circuit 31 first receives data read from the data storage unit 2, that is, data of a plurality (N) of target reflected pulse signals as shown in FIG. Then, a plurality of (P) pieces are divided into a plurality of Q (Q = (N / P)) on the inter-axis, and within each divided section, a covariance matrix in the range cell in the section is calculated (ie, In this embodiment, N = P × Q).

As described above, the order of the direction matrix A _{θ, f} is the number of antennas (M) × the number of pulses (L) (= ML), and the number of elements of the covariance matrix R is (ML) ^2. When the number is K, the covariance matrix R is calculated by averaging the matrix of (ML) ² elements K times.

Therefore, in the weight calculation circuit 31 of this embodiment, the received data stored in the data storage unit 2 is sequentially read in the direction of all the processing range cells including N range cells, and is arranged as shown in FIG. The N total processed data cells are divided into a plurality (Q) of a plurality (P) of range cells, and, as shown in FIG. A covariance matrix R is calculated by performing a complex operation of × (M × L) ² .

Therefore, in the calculation of the covariance matrix R in the weight calculation circuit 31 of the present embodiment, the complex operation of K × (M × L) ² is performed (N / P) times (that is, a plurality (Q) times as the number of divisions). Complex calculation = (N / P) × K × (M × L) ² , so that the number of times of obtaining the covariance matrix by calculation for all the processing range cells consisting of N in the conventional sliding window method shown in FIG. = N × K × (M × L) ² ), the number of operations can be greatly reduced.

  In the above description, the relationship between the total processing range cell number N and the processing data number P is described as P <N. However, the number corresponding to all processing range cells in the distance direction (N) and the processing data number P are described. May be equal (that is, P = N).

  Next, as shown in FIG. 3, the weight calculation circuit 31 of this embodiment uses the estimated reference (estimation reference) signal Ref supplied from the reference signal estimation unit 4 from the covariance matrix calculated for each partitioned cell. Alternatively, a correlation vector is calculated based on the reference (reference) signal Ref supplied from the reference signal generation unit 5, and an input signal between channels (element antennas) for each processing data number P from the calculated correlation vector. A correlation matrix (1, 2,... (N / P)) representing a correlation value (coherence) is calculated, an adaptive weight is calculated by inverse matrix operation for the calculated correlation matrix, and a beam synthesis circuit 31.

  It should be noted that the adaptive weight can be calculated more simply and quickly by using the Cholesky decomposition method in order to obtain the adaptive weight by inverse matrix calculation for the correlation matrix.

  As described above, the weight calculation circuit 31 of the adaptive array antenna includes the covariance matrix means for calculating the covariance matrix, the correlation vector calculation means for calculating the correlation vector from the calculated covariance matrix, and the correlation matrix from the calculated correlation vector. A correlation matrix calculating means for calculating the adaptive weight, and an adaptive weight calculating means for calculating an adaptive weight from the calculated correlation matrix by inverse matrix calculation. 32.

  The beam combining circuit 32 that has received supply of adaptive weights in each range cell from the weight calculation circuit 31 reads the received input signal of each element antenna (channel) from the data storage unit 2 as in the prior art, and performs a pulse repetition period (PRI). : Pulse Repetition Interval) Beam combining is performed by weight control for data between T, and a reception combined beam is formed so that directivity in the direction in which an unnecessary wave arrives becomes zero (null).

  Therefore, a radar receiver (not shown) connected to the adaptive array antenna having the above configuration can accurately detect and display a target position by introducing a reception signal in which unnecessary waves are suppressed.

  As described above, according to the weight calculation method according to this embodiment, the covariance matrix is divided into P cells on the time axis for all processing range cells, and one covariance matrix is calculated for each division. In order to calculate the application weight for each sequential processing range cell by calculating the correlation vector using the calculated covariance matrix, calculating the correlation matrix using the calculated correlation vector, and inverse matrix calculation for the correlation row, conventional sliding Compared with the window method, the amount of calculation is greatly reduced, and a good SINR can be obtained to capture the target.

  Therefore, in the adaptive array antenna so that beam forming is performed based on the adaptive weight calculated by the weight calculation device according to the present embodiment, unnecessary waves are appropriately and satisfactorily suppressed by weight control that is continuous at high speed for each range cell. Is output.

  Therefore, the radar apparatus incorporating the adaptive array antenna can quickly capture the target position, that is, the three-dimensional information of the distance-azimuth-elevation angle, from the combined output signal from the adaptive array antenna.

  Furthermore, by incorporating the adaptive array antenna into the synthetic aperture radar device, a synthetic output signal from the adaptive array antenna can be introduced, and the target shape can be quickly captured and output.

  In the description of the above embodiment, the operations in the weight calculation circuit 31 are described as if they were processed by hardware means, but actually, the calculation processing is performed by software on a computer.

1 is a schematic configuration diagram of an adaptive array antenna that constitutes a weight calculation device according to an embodiment of a covariance matrix calculation method according to the present invention and adopts the weight calculation device. FIG. 2A is processing data of the weight calculation apparatus in the adaptive array antenna shown in FIG. 1, and FIG. 2B is an explanatory diagram of covariance matrix calculation in the processing data shown in FIG. It is explanatory drawing of the correlation matrix calculation in the weight calculation apparatus of the adaptive array antenna shown in FIG. It is explanatory drawing of covariance matrix calculation in the conventional weight calculation apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Reception part # 1- # M Element antenna 2 Data storage part 3 Space-time adaptive signal processing part 31 Weight calculation circuit 32 Beam combining circuit 4 Reference (reference) signal estimation part 5 Reference (reference) signal generation part 6 Excitation part

Claims (6)

The target reflected signal received via the antenna is stored in correspondence with the cell position along the reception timing of the processing range cell having a predetermined length,
The processing range cell in which the target reflection signal is stored is divided into a plurality on the time axis,
A covariance matrix calculation method characterized by calculating a covariance matrix in a range cell in each of the divided sections. Covariance matrix calculation means for calculating a covariance matrix in a range cell in each of the divided sections of the processing range cell by the covariance matrix calculation method according to claim 1;
Correlation vector calculation means for calculating a correlation vector from the covariance matrix in the range cell in each section calculated by the covariance matrix calculation means;
Correlation matrix calculating means for calculating a correlation matrix from the correlation vector calculated by the correlation vector calculating means;
A weight calculation circuit comprising: an adaptive weight calculation unit that calculates an adaptive weight by performing an inverse matrix operation on the correlation matrix calculated by the correlation matrix calculation unit.   The weight calculation circuit according to claim 2, wherein the correlation vector calculation unit calculates a correlation vector using a reference (reference) signal or an estimated reference (estimation reference) signal corresponding to a target signal. The antenna is composed of a plurality of element antennas,
4. A beam combining circuit that performs weight control on a target reflected signal received via the antenna by the adaptive weight calculated by the weight calculating circuit according to claim 2 or 3 and forms a reception combined beam at the antenna. An adaptive array antenna comprising:   A radar apparatus comprising the adaptive array antenna according to claim 4 and configured to detect a target based on an output signal from the adaptive array antenna.   5. A synthetic aperture radar apparatus comprising the adaptive array antenna according to claim 4 and configured to detect a target shape based on an output signal from the adaptive array antenna.

Images