JP2007333404A - Impulse radar - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an impulse radar which performs signal processing by the MUSIC (MUltiple SIgnal Classification) method with sufficient anti-noise measures. <P>SOLUTION: The impulse radar comprises a transmitting antenna for radiating an impulse wave, a receiving antenna for receiving a reflected wave of the impulse wave radiated from the transmitting antenna, and an arithmetic processing means for determining a reflected wave reflected by a target from received data obtained by A/D converting the received signal. The arithmetic processing means comprises a first arithmetic section for performing signal processing using the MUSIC method, a second arithmetic section which provides the data string of the received data with sections in the frequency domain and performs subarraying processing to create a plurality of new data strings by displacing the position of the sectioned data string by a few orders, and a third arithmetic section which determines a spatial smoothing covariance matrix on the basis of each of the plurality of the data strings created by the subarraying processing by the second arithmetic section and outputs it to the first arithmetic section. The second arithmetic section can adjust a data number, or the number of the data strings. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an impulse radar having an excellent real-time data collection function and widely used, and more particularly to an impulse radar having an improved S / N ratio (SNR) of a detection signal for an object. Specifically, the received data reflected from the object is sub-arrayed (hereinafter referred to as “snapshot”) in the frequency domain, and a spatial smoothing covariance matrix is obtained to obtain MUSIC (MUltiple) called super-resolution method. The present invention relates to an impulse radar that improves the SNR of the detection signal by adjusting the number of snapshot data points in an impulse radar that recognizes an object by performing signal processing using a SIgnal Classification method.

  Radars are classified into non-modulation (CW radar), frequency modulation (FMCW radar), and pulse modulation (impulse radar) depending on the modulation method of the transmission wave.

  In the FMCW radar, it has been proposed to use the MUSIC method for processing received signals (Patent Document 1).

The FMCW radar of Patent Document 1 transmits a frequency-modulated continuous wave, detects the received signal (reflected wave) returned to the target with the transmitted wave, obtains the autocorrelation matrix of the detection output, and uses the MUSIC method to calculate the time. An evaluation function is obtained, and the distance is measured from this evaluation function.
Japanese Patent Laid-Open No. 10-031065

  By using the MUSIC method in the above FMCW radar, it is possible to measure a target with high accuracy from reflected waves coming from multiple directions. However, since the FMCW radar has time in frequency sweep, it is disadvantageous in real-time data collection. It is.

  On the other hand, since impulse radar is excellent in real-time data collection, it is expected that the radar function will be greatly improved by introducing the MUSIC method to impulse radar.

  By the way, the present inventor performed a simulation for the purpose of detecting a target buried underground by using an impulse radar in which the MUSIC method was introduced. However, there was a lot of noise in the obtained result. The cause of this noise is considered to be the band and noise of the receiver itself that actually receives the reflected wave from the object, but currently no appropriate measures have been taken against these noises. is there.

  An object of the present invention is to provide an impulse radar that performs signal processing by the MUSIC method with sufficient noise countermeasures.

  According to a first configuration of an impulse radar that realizes the object of the present invention, as described in claim 1, a transmission antenna that radiates an impulse wave and a reception that receives a reflected wave of the impulse wave radiated from the transmission antenna. An arithmetic processing means for obtaining a reflected wave reflected by a target from reception data obtained by performing A / D conversion on a received signal received by the receiving antenna, and the arithmetic processing means comprises a MUSIC method A first arithmetic unit for performing signal processing using a signal, and a subarray configuration in which the received data is divided into data strings in the frequency domain, and a plurality of new data strings are created by shifting the position of the divided data strings several times A spatial computing covariance matrix is obtained on the basis of a second computing unit that performs processing and a plurality of data sequences created by subarray processing by the second computing unit, and is output to the first computing unit 3 and an operation portion, the second arithmetic unit is characterized in that the adjustable number of data is the number of the data sequence.

  According to a second configuration for realizing the object of the present invention, as set forth in claim 2, in the first configuration described above, the arithmetic processing means sets the number of data created by sub-array processing to M, and the data string. Assuming that the number of received wave samples is L, M is equal to or less than L / 2.

  According to the present invention, the S / N ratio can be improved by adjusting the number M of data created by the subarray processing, and the target can be searched with high accuracy. Further, the present invention can be applied to the field of environmental measurement for measuring the amount of snow and the thickness of ice in addition to those existing underground such as landmines / water pipes, underground cavities or detection of ruins as targets. .

  Hereinafter, the present invention will be described based on embodiments shown in the drawings.

  FIG. 1 is a system block diagram of an impulse radar, FIG. 2 is a diagram showing a frequency domain subarray array that is a means for obtaining a spatial averaged covariance matrix from detection data, and FIG. 3 is a diagram illustrating signal processing by the arithmetic unit 10 of FIG. It is a flowchart to show.

  In FIG. 1, the controller 1 repeatedly transmits (for example, at 1 μs intervals) a pulse serving as a transmission trigger for causing the impulse generator 2 to generate an impulse wave. The impulse generator 2 is repeatedly triggered by the controller 1 and repeatedly transmits an impulse wave having a maximum power of about 30 dBm and a pulse width of about 1 ns to the transmission antenna 3.

  The transmitting antenna 3 is a broadband antenna that efficiently radiates radio waves in the frequency range from 0.3 GHz to 2 GHz, for example, a bow-tie dipole antenna of a size that can be carried by humans (A5 size). It receives waves, converts this electrical signal into radio waves, and radiates electromagnetic waves toward the ground. At that time, the transmitting antenna 3 and a receiving antenna described later move along the ground surface (GL) in the operation direction indicated by the arrow.

  The receiving antenna 4 is a radio wave directly coupled from the radio wave radiated from the transmitting antenna 3, and the ground surface and the ground, or a target (13, 14, 15) buried in the ground, such as a buried pipe. A composite wave reflected from a non-metallic object such as a metal object or land mine is received, converted into an electric signal, and then transmitted to the sample and hold circuit 6 via the amplifier 5.

  The sample hold circuit 6 has a function of capturing a part of one wave of input signals instantaneously, for example, within 128 ps, and is a signal repeatedly input by a phase shifter 7 controlled by the controller 1. Is detected after being delayed by an interval of 128 ps, for example, and transmitted to the A / D converter 9 via a low-pass filter (LPF) 8.

  The A / D converter 9 repeatedly A / D-converts the electrical signal received from the sample hold circuit 6 and transmits it to the controller 1.

  The controller 1 stores the digital data received from the A / D converter 9 in an internal memory (not shown) provided inside. The internal memory stores a data amount of a power of 2 (in this embodiment, 128 detection parts or 256 detection parts) × a plurality of data.

  Here, “the data amount of power of 2” means that the electric signal output from the receiving antenna 4 can be digitally restored by combining them. “Multiple” means the amount of data accumulated for improving the S / N ratio (SNR), and is 100 in this embodiment.

  The arithmetic unit 10 is composed of a DSP (Digital Signal Processor) or the like, and sucks data stored in the internal memory of the controller 1 at regular intervals (for example, every 0.1 sec), and signals according to the flowchart shown in FIG. Processing is performed, level conversion is performed via the interface (I / F) 11, and then transmitted to the general-purpose personal computer (PC) 12.

  The general-purpose PC 12 loads the data received from the calculation unit 10 via the I / F 11 into the memory, and is one-dimensional (A mode: mode in which the horizontal axis is a time axis and the vertical axis is amplitude) or two-dimensional (B mode: for example, Data is displayed in a mode in which the horizontal axis is distance and the vertical axis is depth), and is stored in a built-in recording medium.

  The general-purpose PC 12 has software for controlling the arithmetic unit 10 and the controller 1 and has a function of performing measurement operations such as start and stop of measurement.

  In the impulse radar according to the present embodiment having the above-described configuration, the arithmetic unit 10 performs fast Fourier transform (FFT) processing as preprocessing on data (raw data) acquired by a radar that detects a substance in the time domain. Thus, frequency domain data is calculated, and the calculated frequency domain data is combined with signal processing by the MUSIC method.

  In the present embodiment, a signal processing algorithm based on the MUSIC method will be described below in accordance with the flowchart shown in FIG.

  In the flowchart shown in FIG. 3, steps (abbreviated as S) 1 to S5 indicate processing of raw data before processing by the MUSIC method.

  As described above, an impulse radar signal, which is raw data acquired by a radar, is generally represented by the equation (1). In Equation (1), L indicates the number of received wave samples to be described later.

Equation (1) is a signal in the time domain, and is obtained by performing fast Fourier transform processing (hereinafter referred to as “FFT processing”) as preprocessing (because the data to be processed is frequency in the MUSIC method). Data in the frequency domain is expressed by equation (2).

Here, f is a frequency, τ _i is a delay time, and y _i is an input signal represented by a complex quantity at an arbitrary reference position.

  On the other hand, since the received signal includes internal noise and signal noise independently of each other, Equation (2) can be rewritten as Equation (3).

Here, w (ω) represents a noise vector.
The details of the expression (2) are shown by the expression (3a), but for the sake of convenience of explanation, the expression (3b) is simplified.

  The general formula for applying the MUSIC method is given by equation (4).

x = Ay + w (4)
Here, when compared with the equation (3b), the following equation can be used.

  Here, the vector A is a delay parameter matrix of L rows and k columns, and L described therein is the number of received wave samples. y is a reflection coefficient for the Lth reflection point (sample) at the frequency fk. w is a noise vector. The autocorrelation matrix S of x is expressed as in equation (5).

Here, * is a transposed matrix.

The input waveform and noise are uncorrelated. The input wave represented by element y is assumed to be uncorrelated. That is, (yy ^* = P) is a diagonal matrix with L rows and L columns. However, if the input waveform is partially or completely related by the element y, (yy ^* = P) is not a diagonal matrix. When the reaching wave and the internal noise are not related, the correlation matrix S of x is expressed by Expression (8).

However, the average of the noise vector w is zero, σ ² is a variance value, and I is a unit matrix.

The delay time at each reflection position is obtained from the peak position of the MUSIC function (P _music ).

Here, a (τ) is a delay time vector, E _N is an L × (L−k) noise matrix, and (L−k) rows are noise eigenvalues.

  Perform decorrelation processing to remove radar signal problems. In the present embodiment, this decorrelation processing is performed by spatial smoothing processing (hereinafter referred to as “SSP (Spatial Smoothing Process) processing”).

  The SSP process will be described below.

In FIG. 2, the received data is arranged in a frequency domain array having L reflection coefficients such as (1, 2, 3,... N, N + 1,... L-1, L). That is, when the number of samples of the received wave is L, a region of length N (frequency f ₁ to f _N ) is extracted from this frequency region (this extracted region is called a snapshot) It creates so that M snapshots shifted by 1 are overlapped (hereinafter referred to as sub-array division). The relationship is expressed as in equation (10).

In the snapshot obtained by the sub-array division, x ₁ (array number) is the first snapshot, x ₂ is the second snapshot, and the snapshots exist up to x _M so as to overlap. The number M of snapshots is set arbitrarily. In equation (4), the kth element is written as in equation (11).

Then, the autocorrelation matrix S _k of x _k is given as shown in Equation (12)

  The spatial smoothing covariance matrix is defined as an average value of snapshots and is expressed as in equation (13).

  In order to perform signal processing by the MUSIC method in the arithmetic unit 10 shown in FIG. 1, the data x (ω) in the frequency domain of the impulse radar is divided into subarrays as shown in FIG.

A complex conjugate matrix x _k ^* of each snapshot x _k is created, and an autocorrelation matrix S _k is calculated as shown in equation (12).

Furthermore, as shown in equation (13), the SSP process, determined by eigenvalue decomposition eigenvectors E _N of the formula (14).

Here, D is at canonical form of E _N, is a matrix with eigenvalues E _N on the main diagonal. The matrix E _N is a modal matrix, and each column is an eigenvector of S _SSP . To suppress interference between signals reflected from the plurality of objects (target), substitutes E _N in equation (9) is an evaluation function of the MUSIC method. The flow of this process is shown in S6 and S7 in the flowchart of FIG.

  By the way, the relationship between the SNR of the input signal for performing the FFT processing and the SNR of the output signal subjected to the FFT processing is substantially linear.

  On the other hand, since signal processing by a general MUSIC method may be greatly affected by noise, the behavior when the SNR of the input signal is low is unclear.

  Therefore, in the present embodiment, when noise is mixed in the input signal, the performance (SNR of the output signal) of the MUSIC method is selected by selecting a numerical value of the number M of snapshots for executing the spatial smoothing process (SSP). It has improved.

  The performance improvement simulation of this MUSIC method is described below.

The SNR (S / N) _i of the input signal is given by equation (15).

Here, S _i is the total energy of the input signal, and N _i is the total energy of the input noise. These are expressed by equations (16) and (17).

  First, in the first stage of simulation, signal processing by the MUSIC method is executed only with a signal (that is, without noise).

The signal shown in the upper part of FIG. 4A is raw data (time domain data), and the signal shown in the lower part is a frequency domain spectrum of the target signal after FFT processing. In the time domain data after the MUSIC process indicating the response to noise in FIG. 4B, it is calculated from the amplitude amount measured from 0 to the peak value (V _O-P ).

  Next, in the second stage of the simulation, the MUSIC process is executed only with noise. In this case, noise as shown in FIG. 4 (c) is set (upper row shows raw data, lower row shows after FFT processing). In the case of noise, the calculation is performed in the same manner as in the case of only the target signal (FIG. 4 (d)).

In the case of noise (noise output: N _O ), it is evaluated by rms (root mean square) of all data, and is calculated by equation (18).

  In the third stage of the simulation, as illustrated in FIG. 4E, the target signal and noise are mixed in the received signal (the upper stage shows the raw data, and the lower stage shows after the FFT processing). Then, as described above, signal processing by the MUSIC method is performed. The result is the waveform shown in FIG. In this case, the SNR is calculated as shown in Equation (19).

  The inventor has found that the SNR on the output side and the SNR on the input side are independent, but this tendency depends on the number M of snapshots.

  Fig. 5 shows the simulation results when the number of snapshots M is a parameter (10, 20, 30, 50, 75, 100, 125, 150, 175, 200, 230, 240) and the target is present in 50 ns. It was confirmed that increasing the number M of snapshots improved the decorrelation and SNR, and decreasing M decreased the decorrelation and SNR performance. Taking the detection of underground objects as an example, Time (ns) on the horizontal axis corresponds to the depth of the target.

  The simulation result shown in FIG. 6 explains that the decorrelation and SNR are improved when the number M of snapshots described above is increased, and the decorrelation and SNR performance are reduced when the number M of snapshots is decreased.

  In FIG. 6, when the number M of snapshots is set to 10, the SNR on the input side is −2.52 dB and the SNR on the output side is 11.10 dB.

  On the other hand, when the number M of snapshots is increased to 240, the SNR on the input side is -2.39 dB and the SNR on the output side is increased up to 23.97 dB. Therefore, it is possible to increase the SNR on the output side by increasing the number M of snapshots.

  By the way, in this simulation, noise is discrete and random, so that noise in the output signal is suppressed. In signal processing using the MUSIC method, it has been confirmed that the number of snapshots M is greater than 2 and the input signal SNR is -20 dB or more. However, the number of snapshots M is less than 2. In this case, even if the SNR of the input signal is 10 dB, it cannot be processed.

Therefore, in the present embodiment, the value of the number of snapshots M is set to 10 or a value greater than 10. The calculation conditions at this time are as follows:
Input signal SNR -5.8dB
Number of received wave samples (L) 150
It is.

  In this embodiment, it has been confirmed that even if noise is mixed in the input signal, it has high resolution and is robust against noise if a value with a large number M of snapshots is selected.

  From these facts, if the number of received wave samples is L and the number of snapshots for the sample number L is M,

By adjusting the value of the number of snapshots M so that SNR becomes, an SNR improvement effect can be obtained with respect to the SNR (signal-to-noise ratio) of the received wave.

  As described above, the present embodiment has an effect of improving the SNR by adjusting the number of snapshots. The present embodiment is applicable to the field of environmental measurement in which the detection of a cavity, the amount of snow and the ice thickness are measured in addition to the underground buried object detection device whose target is a landmine / water pipe as described above. Can do.

  The flow of the above arithmetic processing will be briefly described below with reference to FIG.

  For example, the arithmetic unit 10 constituted by a DSP reads the raw data x (t) shown in the expression (1) from the internal memory, and first, in S1, the raw data is subjected to FFT processing and replaced with time domain data x (ω) A vector x (ω) shown in Expression (2) is obtained (S2).

Next, in S3, it performs SSP process in Formula (13) sets the number M of the snapshot, in S4, determining the eigenvectors _{E N} from spatial smoothing covariance matrix _{S SSP.}

  In S5, in order to restore the data to the time domain, the delay amount a (τ) is set from the equation (4c) using the radar parameters. At that time, for example, δτ = 128 ps, L = 128 or 256.

Then, in S6, P _MUSIC is calculated by the MUSIC method evaluation function shown in Expression (9) to create one-dimensional data (A mode). The A mode is a mode in which the horizontal axis is time and the vertical axis is amplitude, as shown in FIGS.

In S7, the number of measurement data items y (when the transmitting antenna 3 and the receiving antenna 4 are scanned in parallel with the ground, data is acquired at a certain interval, in this embodiment, several centimeters, and the total number of data obtained as a result is obtained. Data) is generated, two-dimensional data (B mode) P _MUSIC (τ, y) is generated, and an image is displayed on the display unit of the PC.

  The image displayed on the image display unit is, for example, a two-dimensional display of a target embedded underground.

The figure which shows schematic structure of embodiment of the impulse radar by this invention. The figure explaining subarray-ization of frequency domain data. The flowchart which shows the data processing of the impulse radar of FIG. The wave form diagram which shows the confirmation by simulation. The wave form diagram which shows the confirmation by simulation. The wave form diagram which shows the confirmation by simulation. The wave form diagram which shows the confirmation by simulation. The wave form diagram which shows the confirmation by simulation. The wave form diagram which shows the confirmation by simulation. The wave form diagram which shows the number M of snapshots, and a response state. The figure which shows the SNR characteristic with respect to the number M of snapshots.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Controller 2 Impulse generator 3 Transmitting antenna 4 Receiving antenna 5 Amplifier 6 Sample hold circuit 7 Phase shifter 8 Low pass filter (LPF)
9 A / D converter 10 Calculation unit 11 Interface (I / F)
12 General-purpose personal computer (PC)
13, 14, 15 targets

Claims (2)

A transmission antenna that radiates an impulse wave, a reception antenna that receives a reflected wave of the impulse wave radiated from the transmission antenna, and a reception data obtained by A / D converting a reception signal received by the reception antenna Arithmetic processing means for obtaining a reflected wave reflected by the mark,
The arithmetic processing means includes a first arithmetic unit for performing signal processing using the MUSIC method, a section for the received data in the data domain in the frequency domain, and a new position by shifting the position of the partitioned data string several times. A second arithmetic unit that performs sub-array processing for creating a plurality of data strings; and a first smoothing covariance matrix based on each of the plurality of data strings created by the sub-array processing by the second arithmetic unit. An impulse radar, characterized in that the second arithmetic unit can adjust the number of data, which is the number of the data strings. The arithmetic processing means is characterized in that M is equal to or less than L / 2, where M is the number of data created by the subarray processing, and L is the number of received wave samples as the data string. The impulse radar according to claim 1.

Images