An external radiation source radar reference signal purification method based on segmented sliding window reconstruction

Abstract

The application discloses a kind of based on segmented sliding window reconstruction external radiation source radar reference signal purification method, belong to radar target detection signal processing field.The application implementation method is: using the frame structure characteristics of digital television signal, determine reference signal frame head mode and data type, and segmented according to frame number;Each segment signal carries out frequency offset estimation compensation and sampling rate deviation estimation, determine frame head starting position in combination with local PN sequence cross correlation, obtain initial direct wave signal sequence;Through adaptive equalization, compensate signal loss;According to digital television signal standard, data code stream is carried out constellation mapping, shaping filter and obtains original baseband transmitting signal as purified reference signal;Each segment purified reference signal is first adapted to the echo, then with the echo after cancellation is pulse compression, the result of each segment signal pulse compression is accumulated, to obtain the accumulated spliced echo signal, the whole accumulated spliced echo signal is Doppler filtered to detect target.

Description

TECHNICAL FIELD

[0001] The application relates to a reference signal purification method based on segmented sliding window reconstruction of an external radiation source radar, and belongs to the field of radar target detection signal processing. BACKGROUND

[0002] The external radiation source radar is a double (multi) -base radar system using a non-cooperative signal as an opportunity illuminator to realize motion target detection and positioning through coherent accumulation. The external radiation source radar uses the characteristic that the external radiation source radar does not actively emit radio waves, eliminates the influence of radio interference, and can effectively avoid detection and damage by the enemy. Compared with the traditional radar, the external radiation source radar has the advantages of excellent low-altitude coverage, strong detection capability for slow small targets, and more flexible installation and deployment. However, the transmission power, waveform and other parameters of the cooperative illuminator are uncontrollable, and in order to ensure the detection performance of the system, a suitable signal needs to be selected as the illuminator. The digital television signal has the characteristics of wide signal bandwidth, large transmission power and wide coverage, and is an ideal illuminator.

[0003] When the external radiation source radar performs target detection, the target echo signal of a monitoring channel and the direct wave signal of the illuminator of a reference channel can be received at the same time, and after the two channel signals are correlated, the distance, speed and coordinates of the target are extracted. The reference signal received by the reference channel has two important functions, one is to suppress the direct wave and clutter in the echo signal in the time domain, and the other is to detect the target by coherent accumulation with the echo signal. The quality of the reference signal affects the effectiveness of interference suppression and the reliability of target detection. In addition to the direct wave, the actual reference channel signal also has multipath interference and noise caused by the transmission process of the signal by houses, ground and the like, which poses new challenges to the reference signal purification method.

[0004] The time-domain reference signal purification method based on the reconstruction of the transmission signal learns from the signal demodulation method in communication, realizes the suppression of the interference of the reference channel, and can realize the suppression of non-coherent interference. The operation resources consumed in the sampling rate conversion process of this method are relatively large, and errors exist in the frequency offset estimation and sampling rate deviation estimation. The error accumulation causes problems such as constellation diagram divergence and phase rotation, which leads to the increase of the bit error rate of the reconstructed signal and affects the purity of the reconstructed signal. SUMMARY

[0005] The purpose of the present application is to provide an external radiation source radar reference signal purification method based on segmented sliding window reconstruction, which segments the reference signal by frame number, uses the segmented sliding window stepping method to compensate for frequency offset estimation and sampling rate deviation estimation of each segment of signal, reduces the requirement for deviation estimation accuracy, and improves the quality of the reconstructed signal; each segment of the reconstructed reference signal is first adaptively cancelled with the echo to suppress the direct wave and clutter in the echo, and then pulse compression is performed on the cancelled echo, the results of pulse compression of each segment of signal are accumulated to obtain the accumulated spliced echo signal, and the overall Doppler filtering is performed on the accumulated spliced echo signal to detect the target, thereby improving the adaptive cancellation gain and target signal-to-noise ratio, and improving the detection accuracy and efficiency of small targets by the external radiation source radar.

[0006] The purpose of the present application is achieved by the following technical solutions:

[0007] The external radiation source radar reference signal purification method based on segmented sliding window reconstruction disclosed by the present application determines the reference signal frame header mode and data type by utilizing the frame structure characteristics of digital television signals, and segments the signal by frame number; the frequency offset estimation compensation and sampling rate deviation estimation are performed on each segment of signal, the frame header starting position is determined by cross-correlation with the local PN sequence to obtain the initial direct wave signal sequence; the channel estimation value is determined according to the initial direct wave signal sequence and the local PN sequence, adaptive equalization is performed to compensate for the loss of the signal caused by channel transmission; each segment of signal is up-sampled to the least common multiple of the baseband data sampling rate and the preset baud rate for low-pass filtering, and then down-sampled to the preset baud rate for constellation demapping, the constellation mapping and shaping filtering are performed on the data stream according to the digital television signal standard to obtain the original baseband transmission signal as the purified reference signal; each segment of the purified reference signal is first adaptively cancelled with the echo, and then pulse compression is performed on the cancelled echo, the results of pulse compression of each segment of signal are accumulated to obtain the accumulated spliced echo signal, and the overall Doppler filtering is performed on the accumulated spliced echo signal to detect the target, thereby improving the adaptive cancellation gain and target signal-to-noise ratio, and improving the detection accuracy and efficiency of small targets by the external radiation source radar.

[0008] The external radiation source radar reference signal purification method based on segmented sliding window reconstruction of the present application comprises the following steps:

[0009] Step one: utilize the frame structure characteristics of digital television signals to perform autocorrelation on the reference signal, determine the reference signal frame header mode and data type in combination with the multiple relationship between the baseband data sampling rate and the preset baud rate, and segment the signal by frame number.

[0010] Step 1.1: perform autocorrelation on the reference signal r(n), as shown in formula (1):

[0011] R(n)=r(n)*r *(-n) (1)

[0012] Based on the ratio between the baseband data sampling rate and the preset baud rate f, the frame length L of the three signals is... i The frame length L, converted to baseband data sampling rate, is shown in equation (2):

[0013]

[0014] In the formula f s L represents the actual baseband data sampling rate. i To determine the frame length of different types of digital television signal frames at a preset baud rate, the relationship between the peak position of the signal autocorrelation result R(n) and the frame length L is compared to determine the frame header mode and data type of the reference signal.

[0015] Step 1.2: Based on the frame header mode and data type corresponding to the reference signal r(n), segment the reference signal r(n) and the echo signal e(n). Each segment of the reference signal is x(n), and each segment of the echo signal is y(n). The number of segments is m, and the length of each segment is N. m As shown in equation (3), the step N of each signal segment L As shown in equation (4):

[0016] N m =N i / 8+1 i=420,595,945 (3)

[0017] N L =N i / 8 i=420,595,945 (4)

[0018] In the formula N i This refers to the number of signal frames for different types of digital television signals within the acquisition time at a preset baud rate.

[0019] Step 2: Perform carrier frequency offset estimation and compensation, and sampling rate deviation estimation for each segment of the reference signal. A segmented sliding window method is used to estimate and compensate for the carrier frequency offset segment by segment, suppressing the constellation diagram rotation problem after demapping caused by deviation accumulation. A segmented sliding window method is also used to estimate the sampling rate deviation segment by segment, which is used for segmented sampling rate deviation compensation in subsequent Step 4. Frequency offset compensation is performed for each segment of the echo signal. Based on the reference signal data type obtained in Step 1, the local PN sequence of this data type is cross-correlated with each segment of the frequency offset-compensated reference signal to obtain the starting position of the frame header of each compensated reference signal segment. Starting from the frame header, the initial reference signal sequence s(n) for each segment is obtained.

[0020] Step 2.1: Use the FFT method to estimate the carrier frequency offset of each reference signal x(n). The N-point FFT transform of a signal segment is shown in Equation (5):

[0021]

[0022] where X(k) is the result of N-point FFT of the segment reference signal, and the frequency offset value k is calculated according to the position of the peak spectrum line f As shown in equation (6):

[0023]

[0024] The carrier frequency offset CFO of the segment signal is:

[0025] CFO=k f / (f s N) (7)

[0026] The CFO of each segment signal is obtained, and the frequency offset compensation is performed on each segment reference signal x(n) and each segment echo signal y(n), and the compensated reference signal x p (n) is shown in equation (8), and the echo signal y p (n) is shown in equation (9):

[0027] x p (n) = x(n) * exp(-j2π*CFO*nf s ) (8)

[0028] y p (n) = y(n) * exp(-j2π*CFO*nf s ) (9)

[0029] According to equations (8) and (9), the carrier frequency offset is estimated and compensated by using the segment sliding window method, and the rotation problem of the demapped constellation diagram caused by the deviation accumulation is suppressed.

[0030] Step 2.2: Sampling rate deviation estimation is performed on each segment frequency offset compensated reference signal x p (n).

[0031] Taking the clock frequency B s of the transmitting station as the reference, the nominal value of the sampling frequency is f s , and the actual value is f s ', and the theoretical position of the kth correlation peak corresponding to the kth frame is shown in equation (10):

[0032] pn n = round(kN s f s / B s ) (10)

[0033] where N s is the number of symbols of one signal frame. The clock frequency refers to the symbol rate.

[0034] The actual position of the correlation peak corresponding to the kth frame is shown in equation (11):

[0035] pn' n = round(kN s f s ' / B s ) (11)

[0036] The sampling rate offset SFO is shown in equation (12):

[0037]

[0038] The sampling rate offset is estimated by using the piecewise sliding window method according to equation (12), which is used for the sampling rate offset compensation in the subsequent step four.

[0039] Step 2.3: According to the reference signal data type obtained in step one, the local PN sequence of the data type is correlated with each piece of frequency offset compensated reference signal x p (n) to obtain the starting position of the frame header of each piece of compensated signal x p (n), and the initial reference signal sequence s(n) of each piece is obtained by taking the frame header as the starting point.

[0040] Step three: According to the initial reference signal sequence s(n) of each piece obtained in step two, the channel estimation value is determined by using the mean square distortion criterion MMSE in combination with the local PN sequence of the data type, and s(n) is adaptively equalized to obtain the MMSE equalized signal s eq (n).

[0041] Step 3.1: The system function H'(Z) of the equalization filter and the channel transfer function H(Z) satisfy:

[0042] H'(Z)H(Z) = 1 (13)

[0043] For a stationary signal, the equalization filter coefficient W opt is shown in equation (14):

[0044]

[0045] In the equation, R xx is the autocorrelation matrix of the signal, and R xs is the cross-correlation matrix of the signal. The equalizer coefficient calculated by the MMSE criterion is:

[0046]

[0047] Step 3.2: Each piece of reference signal s(n) is filtered by the equalization filter coefficient to obtain the MMSE equalized signal s eq(n), the equalization process in the frequency domain is represented as:

[0048] S eq (n) = W opt S(n) (16)

[0049] Step four: convert the equalized reference signal s eq (n) obtained in step three to a preset baud rate for constellation demapping, segment the sampling rate deviation according to the sampling rate deviation estimation in step two, and compensate the sampling rate deviation of each segment of the sampling rate converted signal based on the segment sampling rate deviation estimation result to suppress the constellation divergence problem caused by deviation accumulation after demapping. Perform constellation mapping and shaping filtering on the data stream according to the digital television signal standard to obtain the original baseband transmission signal as the purified reference signal c(n).

[0050] Step 4.1: upsample s eq (n) to the actual baseband data sampling rate f s and the public multiple of the preset baud rate, and filter it using a square root raised cosine (SSRC) filter with a roll-off coefficient a of 0.05. The filter frequency response H s (f) is:

[0051]

[0052] where f N is the Nyquist frequency, which is 1 / 2 of the input symbol period.

[0053] Step 4.2: downsample each upsampled data to the preset baud rate, and also perform low-pass filtering during the downsample process. At the same time, segmentally compensate the sampling rate deviation of each sampling rate converted signal according to formula (12) during the downsample process to suppress the constellation divergence problem caused by deviation accumulation after demapping.

[0054] For each sampling rate converted signal, increase SFO points as the sampling rate deviation compensation value based on the SFO value obtained according to formula (12) on the basis of the original number of signal points.

[0055] Use the middle frame of the segment signal to perform phase estimation, and the phase estimation value φ i is as shown in formula (18):

[0056] φ i = angle(s eqm (n)*PN') (18)

[0057] where s eqm (n) is the reference signal s eq(n) is the intermediate frame data, PN is the local PN sequence of this data type, s' eq (n) is shown in formula (19):

[0058] s' eq (n) = s eq (n) * exp(-jφ i ) (19)

[0059] Step 4.3: Separate the frame header, digital television signal system information and frame body data of each segment of reference signal s' eq (n), according to the modulation type x of the frame body data, for the data condensed in the constellation diagram, s' eq (n) is deconstellation mapped using the qamdemod function:

[0060] b(n) = qamdemod(s' eq (n), x) (20)

[0061] In the formula, b(n) is the data bit stream after deconstellation mapping, and b(n) is directly mapped to obtain the directly mapped signal s cg (n) using the qammod function according to the modulation type x:

[0062] s cg (n) = qammod(b(n), x) (21)

[0063] Step 4.4: Up-sample the directly mapped reconstructed signal s cg (n) to a common multiple of the actual baseband data sampling rate f s and the preset baud rate, and then down-sample it to the actual baseband data sampling rate f s . Similarly, low-pass filtering, sampling rate deviation compensation and phase compensation are needed in the up-sampling and down-sampling process, that is, the operations in steps 4.1 and 4.2 are repeated to obtain the reconstructed signal c(n) of each segment of reference signal at the baseband sampling rate f s .

[0064] Step five: According to each purified reference signal c(n) obtained in step four, first perform adaptive cancellation with the corresponding segment of echo y p (n) after frequency offset compensation; then pulse compress the echo after cancellation, accumulate the pulse compression results of each segment of signal, obtain the accumulated and spliced echo signal, and perform Doppler filtering on the whole accumulated and spliced echo signal to detect the target, improve the adaptive cancellation gain and target signal-to-noise ratio, and improve the detection accuracy and efficiency of the external radiation source radar on small targets.

[0065] Step 5.1: c(n) and yp (n) using variable step size frequency domain block NLMS algorithm to get the cancelled echo d(n).

[0066] Step 5.2: determine the pulse length Lseg, segment c(n) and d(n), the number of segments is Nseg, the number of FFT points in each segment is FFTNumber, and the pulse compression result p(n) is shown in equation (22):

[0067]

[0068] According to the purified reference signal c(n) in each segment in step one and the corresponding segment echo y p (n) with frequency offset compensation, the number of segments is m segments, and the m segment pulse compression results are spliced and accumulated to get P(n) as shown in equation (23):

[0069] P(n) = [p1(n), p2(n), … p m (n)] (23)

[0070] Step 5.3: window FFT operation is performed on the m segment pulse compression accumulation result P(n) along the pulse dimension, the window type is hamming window, the FFT point number is Nseg*8, and finally the Doppler filtering result D(n) is obtained as shown in equation (24), and the distance, speed and amplitude of the target are extracted.

[0071] D(n) = fftshift{fft(P(n).*window)} (24)

[0072] The accumulated and spliced echo signal as shown in equation (23) is used to detect the target according to equation (24) for overall Doppler filtering, so as to improve the adaptive cancellation gain and the target signal-to-noise ratio, and improve the detection accuracy and efficiency of small targets by the external radiated source radar.

[0073] According to the digital television signal standard, as a preferred, the baud rate is selected as 7.56MHz.

[0074] Advantages:

[0075] 1. The external radiation source radar reference signal purification method based on segmented sliding window reconstruction disclosed in the application, which compensates for carrier frequency offset estimation and sampling rate deviation estimation by performing carrier frequency offset estimation and compensation on each segment of the reference signal, suppresses the constellation rotation problem after demapping caused by deviation accumulation by using the segmented sliding window method to segmentally estimate and compensate for the carrier frequency offset, segmentally estimates the sampling rate deviation by using the segmented sliding window method, and performs segmented compensation for the sampling rate deviation in the subsequent steps; and compensates for the frequency offset of each segment of the echo signal. According to the data type of the reference signal, the starting position of the frame header of each compensated reference signal is obtained by correlating the local PN sequence of the data type with each segment of the frequency offset compensated reference signal, and the initial reference signal sequence of each segment is obtained starting from the frame header.

[0076] 2. The external radiation source radar reference signal purification method based on segmented sliding window reconstruction disclosed in the application, which converts each segment of the equalized and filtered reference signal to a preset baud rate for demapping, compensates for the sampling rate deviation of each segment of the sampling rate converted signal based on the segmented sampling rate deviation estimation result, and suppresses the constellation divergence problem after demapping caused by deviation accumulation. The original baseband transmission signal is obtained as the purified reference signal by performing constellation mapping and shaping filtering on the data stream according to the digital television signal standard.

[0077] 3. The external radiation source radar reference signal purification method based on segmented sliding window reconstruction disclosed in the application, which first performs adaptive cancellation between each segment of the purified reference signal and the corresponding segment of the frequency offset compensated echo, then performs pulse compression on the echo after cancellation, accumulates the results of pulse compression of each segment of the signal, obtains the accumulated and spliced echo signal, and performs Doppler filtering on the whole accumulated and spliced echo signal to detect the target, thereby improving the adaptive cancellation gain and the signal-to-noise ratio of the target, and improving the detection accuracy and efficiency of the external radiation source radar on small targets. BRIEF DESCRIPTION OF DRAWINGS

[0078] Figure 1 is an external radiation source radar reference antenna signal receiving diagram;

[0079] Figure 2 is a flow chart of the external radiation source radar reference signal purification method based on segmented sliding window reconstruction;

[0080] Figure 3 is a reference signal autocorrelation result diagram;

[0081] Figure 4 is a constellation diagram of the reference signal frame body data after demapping, wherein Figure 4 (a) is the 864th frame constellation diagram resolved by the segmented sliding window reconstruction, Figure 4 (b) is the 864th frame constellation diagram resolved by the whole segment data reconstruction.

[0082] Figure 5is a time-domain and frequency-domain result chart of the echo self-adaptive cancellation of the purified reference signal and the frequency offset compensation;

[0083] Figure 6 is a time-delay-Doppler chart, wherein Figure 6 (a) is a result chart of pulse compression Doppler filtering of the reference signal and the echo signal after the segmented sliding window reconstruction, Figure 6 (b) is a result chart of pulse compression Doppler filtering of the original reference signal and the echo signal. DETAILED DESCRIPTION

[0084] In order to better illustrate the purposes and advantages of the present application, the summary will be further described in combination with the drawings and examples.

[0085] In the example, the reference signal used for analysis is a Beijing Pinggu TV tower digital television signal collected by a reference antenna of an external radiation source radar in a certain experimental site in Beijing Pinggu, and the frequency is 674 MHz; the echo signal is a measured signal of the echo antenna pointing to the target detection space, and the target is a DJI unmanned aerial vehicle.

[0086] As shown in Figure 1 , the reference signal received by the reference antenna of the embodiment contains not only the direct wave signal, but also the multipath interference and noise influence caused by buildings, ground and the like, as shown in formula (25):

[0087]

[0088] In the formula, r r (n) is the signal of the transmitting station, a k , and τ k are the amplitude attenuation coefficient, time delay and phase shift of the corresponding direct wave interference relative to the direct wave of the main transmitting station, K is the number of multipaths, g r (n) represents Gaussian white noise.

[0089] Based on MATLAB, the reference signal r(n) and the echo signal e(n) of one second of time are analyzed.

[0090] As shown in Figure 2 , the embodiment applies a reference signal purification method of an external radiation source radar based on segmented sliding window reconstruction of the present application to purify the collected 674 MHz reference signal, and analyzes the target signal-to-noise ratio, and the specific implementation steps are as follows:

[0091] Step one: using the frame structure characteristics of the digital television signal, self-correlating the reference signal, combining the baseband data sampling rate and the multiple relationship of the preset baud rate 7.56 MHz, determining the reference signal frame header mode and data type, and segmenting the signal according to the frame number.

[0092] Step 1.1: Perform autocorrelation on the reference signal r(n), as shown in equation (26):

[0093] R(n)=r(n)*r * (-n) (26)

[0094] Based on the relationship between the baseband data sampling rate and the preset baud rate of 7.56MHz, the frame length L of the three signals is determined. i The frame length L, converted to baseband data sampling rate, is shown in equation (27):

[0095]

[0096] In the formula f s L represents the actual baseband data sampling rate. i To determine the frame length of different types of digital television signal frames at a preset baud rate of 7.56MHz, the relationship between the peak position of the signal autocorrelation result R(n) and the frame length L is compared to determine the frame header mode and data type of the reference signal.

[0097] In the embodiment, the autocorrelation result of the reference signal r(n) is as follows: Figure 3 As shown, the actual baseband data sampling rate f s With a sampling rate of 10MHz, the peak position difference of the autocorrelation result R(n) of r(n) is 5788, which is consistent with the frame length of a single-carrier signal with a frame header of PN595 at a sampling rate of 10MHz. Therefore, the 674MHz reference signal is a single-carrier digital television signal of PN595.

[0098] Step 1.2: Based on the frame header mode and data type corresponding to the reference signal r(n), segment the reference signal r(n) and the echo signal e(n). Each segment of the reference signal is x(n), and each segment of the echo signal is y(n). The number of segments is m, and the length of each segment is N. m As shown in equation (28), the step N of each signal segment L As shown in equation (29):

[0099] N m =N i / 8+1 i=420,595,945 (28)

[0100] N L =N i / 8 i=420,595,945 (29)

[0101] In the formula N i This represents the number of signal frames for different types of digital television signals within the acquisition time at a preset baud rate of 7.56MHz.

[0102] In the embodiment, the frame header is a single carrier signal with PN595, the number of signal frames in one second is 1728 frames, and the segmentation length N of the signal is 217 frames, m is 8 segments, and N m is 216 frames. L is 216 frames, and m is 8 segments.

[0103] Step two: carrier frequency offset estimation compensation and sampling rate deviation estimation are performed on each segment of the reference signal, the segmented estimation and compensation of the carrier frequency offset are performed by using the segmented sliding window method, the rotation problem of the constellation diagram after demapping caused by deviation accumulation is suppressed, the segmented estimation of the sampling rate deviation is performed by using the segmented sliding window method, and is used for subsequent sampling rate deviation segmented compensation in step four; and the frequency offset compensation is performed on each segment of the echo signal. According to the reference signal data type obtained in step one, the local PN sequence of the data type is correlated with each segment of the frequency offset compensated reference signal to obtain the starting position of each segment of the compensated reference signal frame header, and the initial reference signal sequence s(n) of each segment is obtained starting from the frame header.

[0104] Step 2.1: the carrier frequency offset estimation is performed on each segment of the reference signal x(n) by using the FFT method, and the N-point FFT transformation of a segment of the signal is shown in formula (30):

[0105]

[0106] In the formula, X(k) is the result of N-point FFT of the segment of the reference signal, the frequency offset value k is calculated according to the position corresponding to the peak spectrum line f , as shown in formula (31):

[0107]

[0108] The carrier frequency offset CFO of the segment of the signal is:

[0109] CFO=k f / (f s N) (32)

[0110] The CFO of each segment of the signal is obtained, the frequency offset compensation is performed on each segment of the reference signal x(n) and each segment of the echo signal y(n), the compensated reference signal x p (n) is shown in formula (33), and the echo signal y p (n) is shown in formula (34):

[0111] x p (n)=x(n)*exp(-j2π*CFO*nf s ) (33)

[0112] y p (n)=y(n)*exp(-j2π*CFO*nf s ) (34)

[0113] According to formula (33) (34), the segmented sliding window method is adopted to segmentally estimate and compensate the carrier frequency offset, and the demapping constellation rotation problem caused by deviation accumulation is inhibited.

[0114] In the embodiment, the calculated carrier frequency offset CFO is 168, and each segment of the reference signal and the echo signal is compensated.

[0115] Step 2.2: Correlate the reference signal x p (n) after each segment of frequency offset compensation with the local PN sequence of the data type of the reference signal data obtained in step one to obtain each segment of compensated signal x p (n).

[0116] Taking the clock frequency B s of the transmitting station as the reference, the nominal value of the sampling frequency is f s , and the actual value is f s '. The theoretical position of the kth correlation peak corresponding to the kth frame is shown in formula (35):

[0117] pn n = round(kN s f s / B s ) (35)

[0118] In the formula, N s is the number of symbols of one signal frame. The clock frequency refers to the symbol rate.

[0119] The actual position of the correlation peak corresponding to the kth frame is shown in formula (36):

[0120] pn' n = round(kN s f s ' / B s ) (36)

[0121] The sampling rate deviation SFO is shown in formula (37):

[0122]

[0123] According to formula (37), the segmented sliding window method is adopted to segmentally estimate the sampling rate deviation, which is used for segmentally compensating the sampling rate deviation in step four.

[0124] In the embodiment, the calculated sampling rate deviation SFO is 1.037.

[0125] Step 2.3: Correlate the reference signal data type obtained in step one with the local PN sequence of the data type and each segment of the frequency offset compensated reference signal x p (n) to obtain each segment of the compensated signal x p(n) the start position of the frame header, each initial reference signal sequence s(n) is obtained starting from the frame header.

[0126] In the embodiment, the start positions of the frame header of the 8 reference signal segments are {927, 928, 929, 929, 930, 931, 931, 932}, and the initial reference signal sequence s(n) is obtained starting from the start position of the frame header.

[0127] Step three: according to each initial reference signal sequence s(n) obtained in step two, the channel estimation value is determined using the MMSE criterion of the local PN sequence of the data type, the adaptive equalization is performed on s(n), and the MMSE equalized signal s eq (n) is obtained.

[0128] Step 3.1: the system function H'(Z) of the equalization filter and the channel transfer function H(Z) satisfy:

[0129] H'(Z)H(Z) = 1 (38)

[0130] For a stationary signal, the equalization filter coefficient W opt is shown in equation (39):

[0131]

[0132] In the equation, R xx is the autocorrelation matrix of the signal, and R xs is the cross-correlation matrix of the signal. The equalizer coefficient calculated by the MMSE criterion is:

[0133]

[0134] In the embodiment, the equalization filter coefficient w obtained by the MMSE criterion is:

[0135]

[0136] Step 3.2: each reference signal s(n) is filtered by the equalization filter coefficient to obtain the MMSE equalized signal s eq (n), and the equalization process in the frequency domain is represented as:

[0137] S eq (n) = W opt S(n) (41)

[0138] Step four: according to each equalized reference signal s eq(n) is converted to preset baud rate 7.56MHz for de-constellation mapping, according to step two, the sampling rate deviation is segmented and estimated, and based on the segmented estimation result of the sampling rate deviation, the sampling rate converted signal of each segment is compensated for the segmented sampling rate deviation to suppress the divergence problem of the de-mapped constellation caused by deviation accumulation. The data stream is constellation mapped and shaped filtered according to the digital television signal standard to obtain the original baseband transmission signal as the purified reference signal c(n).

[0139] Step 4.1: s eq (n) is up-sampled to the actual baseband data sampling rate f s is a multiple of the preset baud rate 7.56MHz, and a square root raised cosine SSRC filter is used for filtering, the roll-off coefficient a of which is 0.05, and the filter frequency response H s (f) is:

[0140]

[0141] where f N is the Nyquist frequency, which is 1 / 2 of the input symbol period.

[0142] Step 4.2: Each up-sampled data is down-sampled to the preset baud rate 7.56MHz, and low-pass filtering is also needed in the down-sampling process. At the same time, according to formula (37), the sampling rate converted signal of each segment is compensated for the segmented sampling rate deviation to suppress the divergence problem of the de-mapped constellation caused by deviation accumulation.

[0143] For each sampling rate converted signal, the SFO value obtained according to formula (37) on the basis of the original number of signal points is increased by SFO points as the sampling rate deviation compensation value. In the embodiment, the sampling rate deviation SFO compensation point is 1.037.

[0144] The middle frame of the segment is used for phase estimation, and the phase estimation value φ i As shown in formula (43):

[0145] φ i = angle(s eqm (n)*PN') (43)

[0146] where s eqm (n) is the middle frame data of the segment reference signal s eq (n), PN is the local PN sequence of the data type, and each segment reference signal s' eq (n) after phase compensation at the preset baud rate 7.56MHz is as shown in formula (44):

[0147] s' eq (n) = s eq(n) * exp(-jφ i ) (44)

[0148] Step 4.3: Separate the frame header, digital television signal system information and frame body data of each segment of the reference signal s eq (n), draw the constellation diagram of the frame body data, and according to the modulation type x of the frame body data, condense the data of the constellation diagram, and use the qamdemod function to demodulate s eq (n) to obtain the demodulated signal s eq (n):

[0149] b(n) = qamdemod(s cg (n), x) (45)

[0150] In the formula, b(n) is the demodulated data bit stream, and according to the modulation type x, the qammod function is used to directly map b(n) to obtain the directly mapped signal s cg (n):

[0151] s cg (n) = qammod(b(n), x) (46)

[0152] In the embodiment, the 674MHz signal transmitted by the Beijing Pinggu TV tower adopts 4QAM modulation, and the system information analysis shows that it adopts LDPC code rate 3 and symbol interleaving mode 2. The constellation diagram of the 864th frame using the segmented sliding window reconstruction is shown in Figure 4 (a), and the constellation diagram of the 864th frame after frequency offset compensation and sampling rate deviation compensation for one second of data is shown in Figure 4 (b), and the method of segmented sliding window demapping effectively suppresses the rotation and divergence of the demapped constellation diagram caused by deviation accumulation.

[0153] Step 4.4: Up-sample the directly mapped reconstructed signal s s (n) to a public multiple of the preset baud rate 7.56MHz, and then down-sample it to the actual baseband data sampling rate f s . In the up-sampling and down-sampling process, low-pass filtering, sampling rate deviation compensation and phase compensation are needed, that is, the operations in steps 4.1 and 4.2 are repeated to obtain the reconstructed signal c(n) of each segment of the reference signal at the baseband sampling rate f s .

[0154] Step five: According to each purified reference signal c(n) obtained in step four, it is first compensated for frequency offset with the corresponding segment of the echo y p(n) adaptive cancellation is performed; the pulse compressed results of each segment of signals are accumulated to obtain the accumulated spliced echo signals, the whole accumulated spliced echo signals are subjected to Doppler filtering to detect targets, the adaptive cancellation gain and the target signal-to-noise ratio are improved, and the detection precision and efficiency of the external radiation source radar on small targets are improved.

[0155] Step 5.1: c(n) and y p (n) The cancellation is performed using a variable step size frequency domain block NLMS algorithm to obtain the cancelled echo d(n).

[0156] In the embodiment, the purified reference signal c(n) of each segment and the corresponding segment of echo y p (n) The FFT point number of adaptive cancellation is 4096, the initial step size is 0.1, the maximum step size is 0.02, the minimum step size is 0.006, the coefficient forgetting factor is 0.99, the step size adjustment coefficient is 0.007, and the cancellation filter order is 2048.

[0157] c(n) and y p (n) The adaptive cancellation gain is improved compared with the adaptive cancellation gain of the original reference signal r(n) and the original echo signal e(n), and the improvement results are shown in the following table:

[0158]

[0159] Taking the fourth segment of signals (frames 649-865) as an example, the purified reference signal c(n) of this segment and the corresponding segment of echo y p (n) The time domain and frequency domain results of adaptive cancellation are shown in Figure 5 .

[0160] Step 5.2: Determine the pulse length Lseg, segment c(n) and d(n), the number of segments is Nseg, the FFT point number of each segment of signals is FFTNumber, and the pulse compression result p(n) is shown in formula (47):

[0161]

[0162] According to step one, the purified reference signal c(n) of each segment and the corresponding segment of echo y p (n) The number of segments is m, and the spliced and accumulated results of m segments of pulse compression are obtained as P(n) shown in formula (48):

[0163] P(n) = [p1(n), p2(n), … p m (n)] (48)

[0164] In the embodiment, m is 8, the pulse length Lseg of each segment of the reference signal and echo signal pulse compression is 1248, the number of segments Nseg is 1001, the number of FFT points FFTNumber of each segment is 2048, pulse compression is performed on each segment, and the results of pulse compression of the 8 segments are spliced to obtain the results P(n) of pulse compression of the reconstructed reference signal and echo signal in one second.

[0165] Step 5.3: Windowed FFT operation is performed on the m-segment pulse compression accumulation result P(n) along the pulse dimension, the window type is a Hamming window, the number of FFT points is Nseg*8, and finally the Doppler filtering result D(n) is obtained as shown in formula (49), and the distance, velocity and amplitude of the target and other information are extracted.

[0166] D(n) = fftshift {fft(P(n).*window)} (49)

[0167] The accumulated and spliced echo signal as shown in formula (48) is subjected to overall Doppler filtering according to formula (49) to detect the target, improve the adaptive cancellation gain and target signal-to-noise ratio, and improve the detection accuracy and efficiency of the external radiation source radar on small targets.

[0168] In the embodiment, windowed FFT is performed on the pulse compression accumulation result P(n), the number of FFT points is 8008, the result D(n) of pulse compression Doppler filtering of the reconstructed reference signal and echo signal in one second is obtained, and the range-Doppler plot thereof is shown in Figure 6 (a). The result of pulse compression Doppler filtering of the original signal and echo signal in one second is shown in Figure 6 (b), the reference signal is purified by using the segmented sliding window reconstruction method, the target signal-to-noise ratio is improved from 32.1 dB to 35.9 dB, and is improved by 3.8 dB.

[0169] In summary, the segmented sliding window reconstruction-based external radiation source radar reference signal purification method provided by the application reduces the requirement for the estimation accuracy of frequency deviation and sampling rate deviation, effectively suppresses the problems of constellation plot deflection and divergence after demapping caused by deviation accumulation, improves the adaptive cancellation gain and target signal-to-noise ratio, and improves the detection accuracy and efficiency of the external radiation source radar on small targets.

[0170] The above specific description further details the purpose, technical scheme and beneficial effects of the application, and it should be understood that the above description is only a specific embodiment of the application and is not used to limit the protection scope of the application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the application should be included in the protection scope of the application.

Claims

1. A method for reference signal purification for an external source radar based on piecewise sliding window reconstruction, characterized in that: Comprising the following steps, Step one: using the frame structure characteristics of digital television signal, autocorrelation is performed on the reference signal, the frame header mode and data type of the reference signal are determined in combination with the multiple relationship between the baseband data sampling rate and the preset baud rate, and the signal is segmented according to the frame number; Step two: carrier frequency offset estimation compensation and sampling rate deviation estimation are performed on each segment of the reference signal, the carrier frequency offset is estimated and compensated by using the segmented sliding window method, the rotation problem of the constellation diagram after demapping caused by deviation accumulation is suppressed, the sampling rate deviation is estimated by using the segmented sliding window method, and is used for sampling rate deviation segmented compensation in step four; the echo signal is compensated for frequency offset; according to the reference signal data type obtained in step one, the local PN sequence of the data type is correlated with each segment of the frequency offset compensated reference signal to obtain the starting position of each segment of the compensated reference signal frame header, and the initial reference signal sequence s(n) is obtained from the starting frame header; Step three: according to each initial reference signal sequence s(n) obtained in step two, using mean square distortion criterion MMSE in combination with the local PN sequence of the data type to determine the channel estimation value, performing adaptive equalization on s(n) to obtain the MMSE equalized and filtered signal s eq (n); Step four: according to the equalized filtered reference signal s of each segment obtained in step three eq (n), convert it to a preset baud rate for de-constellation mapping, segment the sampling rate deviation according to the sampling rate deviation estimation in step two, and compensate the sampling rate deviation of each segment of the sampling rate converted signal based on the sampling rate deviation estimation result, to suppress the divergence of the de-mapped constellation caused by deviation accumulation; and perform constellation mapping and shaping filtering on the data stream according to the digital television signal standard to obtain the original baseband transmission signal as the purified reference signal c(n). Step five: according to the purified reference signal c(n) of each segment obtained in step four, it is firstly self-adaptively cancelled with the corresponding segment of echo y p (n) after frequency offset compensation; then pulse compression is performed on the echo after cancellation, the pulse compression results of each segment are accumulated to obtain the accumulated spliced echo signal, and the whole accumulated spliced echo signal is Doppler filtered to detect the target.

2. A method for reference signal purification for an external source radar based on segment sliding window reconstruction as claimed in claim 1, characterized in that: The implementation method of step one is, Step 1.1: autocorrelation is performed on the reference signal r(n), as shown in formula (1): R(n) = r(n) * r * (-n) (1) In combination with the relationship between the baseband data sampling rate and the multiple of the preset baud rate f, the frame length L of the three signals is converted into the frame length L under the baseband data sampling rate, as shown in formula (2): i In combination with the relationship between the baseband data sampling rate and the multiple of the preset baud rate f, the frame length L of the three signals is converted into the frame length L under the baseband data sampling rate, as shown in formula (2): i where f s is the actual baseband data sampling rate, L i is the frame length of different types of digital television signals at a preset baud rate; the peak position of the correlation result R(n) of the comparison signal is compared with the frame length L to determine the frame header mode and data type of the reference signal; Step 1.2: According to the frame header mode and data type corresponding to the reference signal r(n), the reference signal r(n) and the echo signal e(n) are segmented, each segment of the reference signal is x(n), each segment of the echo signal is y(n), the number of segments is m, and the length of each signal is N m As shown in formula (3), the step N of each signal L As shown in formula (4): N m = N i / 8 + 1 i = 420,595,945 (3) N L = N i / 8 i = 420,595,945 (4) where N i is the number of signal frames of different types of digital television signals in the acquisition time at the preset baud rate.

3. A method for reference signal purification for an external source radar based on segmented sliding window reconstruction as claimed in claim 2, characterized in that: The implementation method of step two is, Step 2.1: the carrier frequency offset of each segment of the reference signal x(n) is estimated by using the FFT method, and the N-point FFT transformation of a segment of signal is as shown in formula (5): In the formula, X(k) is the result of N-point FFT of the reference signal segment, and the frequency offset value k is calculated according to the position of the peak spectrum line f As shown in formula (6): The carrier frequency offset CFO of the segment of signal is: CFO = k f / (f s N) (7) The CFO of each segment signal is obtained, and the frequency offset compensation is performed on each segment reference signal x(n) and each segment echo signal y(n), and the compensated reference signal x p (n) is as shown in formula (8) p (n) as shown in formula (9): x p (n) = x(n) * exp(-j2π*CFO*nf s ) (8) y p (n) = y(n) * exp(-j2π*CFO*nf s ) (9) The carrier frequency offset is estimated and compensated by using the segmented sliding window method according to formula (8) (9), and the rotation problem of the constellation diagram after demapping caused by deviation accumulation is suppressed; Step 2.2: Compute the reference signal x for each segment after frequency offset compensation p (n) performing a sampling rate offset estimation; with the clock frequency B of the transmitting station s The nominal value of the actual baseband data sampling rate is f s , the actual value is f s The theoretical position of the kth correlation peak corresponding to the kth frame is shown in equation (10): pn n = round(kN s f s / B s ) (10) where N s is the number of symbols in a signal frame; the clock frequency refers to the symbol rate; The actual position of the correlation peak corresponding to the kth frame is as shown in formula (11): pn' n = round(kN s f s ' / B s ) (11) The sampling rate deviation SFO is as shown in formula (12): The sampling rate deviation is estimated by using the segmented sliding window method according to formula (12), and is used for sampling rate deviation segmented compensation in step four; Step 2.3: According to the reference signal data type obtained in step 1, use the local PN sequence of the data type to multiply each piece of the frequency offset compensated reference signal x p (n) to obtain the cross-correlation of each piece of the compensated signal x p (n) The starting position of the frame header, and the initial reference signal sequence s(n) is obtained starting from the frame header.

4. A method for reference signal purification for an external source radar based on segment sliding window reconstruction as claimed in claim 3, characterized in that: The implementation method of step three is, Step 3.1: the system function H'(Z) of the equalization filter and the channel transfer function H(Z) satisfy: H'(Z)H(Z)=1 (13) For a stationary signal, the equalization filter coefficients W opt As shown in equation (14): where R xx is the autocorrelation matrix of the signal, R xs is the cross-correlation matrix of the signal; the equalizer coefficients calculated from the MMSE criterion are: Step 3.2: Filter each segment of reference signal s(n) with the equalization filter coefficients to obtain the MMSE equalized signal s eq (n), the equalization process is represented in the frequency domain as: S eq (n) = W opt S(n)(16).

5. A method for reference signal purification for an external source radar based on segment sliding window reconstruction as claimed in claim 3, characterized in that: The implementation method of step four is, Step 4.1: s eq (n) up to the actual baseband data sampling rate f s and a preset baud rate multiple, using a square root raised cosine SSRC filter, with a roll-off coefficient a of 0.05, the filter frequency response H s (f) is: where f N is the Nyquist frequency, which is 1 / 2 of the input symbol period. Step 4.2: the up-sampled data is down-sampled to the preset baud rate, and low-pass filtering also needs to be performed in the down-sampling process; meanwhile, the sampling rate deviation segmented compensation is performed on the signal after sampling rate conversion according to formula (12) in the down-sampling process, and the divergence problem of the constellation diagram after demapping caused by deviation accumulation is suppressed; For each segment of the signal after sampling rate conversion, the SFO value obtained on the basis of the original signal point number according to formula (12) is increased by SFO points as the sampling rate deviation compensation value; The phase estimate φ is performed using the middle frame of the signal segment i As shown in equation (18): φ i = angle(s eqm (n)*PN') (18) where s eqm (n) is the intermediate frame data of the segment reference signal s eq (n), PN is the local PN sequence of this data type, and each segment reference signal s' eq (n) after phase compensation at the preset baud rate is shown in equation (19): s' eq (n) = s eq (n) * exp(-jφ i ) (19) Step 4.3: Separate the frame header, digital television signal system information and frame body data of each segment of reference signal s' eq (n), draw the constellation diagram of frame body data, and according to the modulation type x of frame body data, demap s' eq (n) using the qamdemod function: b(n) = qamdemod(s eq (n), x) (20) In the formula, b(n) is the data bit code stream after constellation demapping, and b(n) is directly mapped to obtain the signal s after direct mapping by using the qammod function according to the modulation type x cg (n): s cg (n) = qammod(b(n), x) (21) Step 4.4: Reconstruct the signal s cg (n) up to the actual baseband data sampling rate f s with a preset multiple of baud rate, and down to the actual baseband data sampling rate f s , and the same low-pass filtering, sampling rate deviation compensation and phase compensation are needed in the process of up-sampling and down-sampling, i.e. repeat the operations in steps 4.1 and 4.2 to obtain the reconstructed signal c(n) of each reference signal at the baseband sampling rate f s .

6. A method for reference signal purification for an external source radar based on segment sliding window reconstruction as claimed in claim 5, characterized in that: The implementation method of step five is, Step 5.1: c(n) with y p (n) using a variable step size frequency domain block NLMS algorithm to obtain the cancelled echo d(n); Step 5.2: the pulse length Lseg is determined, c(n) and d(n) are segmented, the segment number is Nseg, the FFT point number of each segment of signal is FFTNumber, and the pulse compression result p(n) is as shown in formula (22): According to the reference signal c(n) purified in each segment in step one and the corresponding segment echo y p (n) segment number is m segment, the m segment pulse compression result splicing accumulation P(n) is shown in formula (23): P(n) = [p1(n), p2(n),... p m (n)] (23) Step 5.3: the m segments of pulse compression cumulative results P(n) are subjected to windowed FFT operation along the pulse dimension, the window type is hamming window, the FFT point number is Nseg*8, and finally the Doppler filtering result D(n) is as shown in formula (24), and the distance, velocity and amplitude information of the target is extracted; D(n) = fftshift{fft(P(n) * window)} (24) The accumulated spliced echo signal as shown in formula (23) is subjected to overall Doppler filtering according to formula (24) to detect the target, improve adaptive cancellation gain and target signal-to-noise ratio, and improve the detection precision and efficiency of the external radiation source radar on small targets.

7. A method for reference signal purification for an external radiation source radar based on segmental sliding window reconstruction according to any one of claims 2 to 6, characterized in that: The baud rate is 7.56 MHz.

Images