Radar non-blind area detection method based on RLS-DCD

Abstract

The invention discloses a radar non-blind area detection method based on RLS-DCD, and relates to the field of radar signal processing, and the method comprises the steps: direct wave time delay measurement: obtaining the coefficient of a transmitting signal baseband waveform matching filter, and carrying out the peak search after pulse compression to obtain the position of a direct wave in a sampling signal; construction of a cancellation signal: constructing a transmitted waveform time delay signal at the position of the same direct wave by using a known transmitted wave, constructing a Wiener filter, solving by using a DCD algorithm, inputting the transmitted waveform time delay signal, ADC sampling and DDC processing to obtain a Wiener filtering result, namely the cancellation signal; and direct wave signal cancellation: performing ADC sampling, and subtracting the cancellation signal from the signal processed by the DDC to obtain a signal after cancellation. According to the method, the direct wave component and the echo direct wave are aligned in the time dimension, so that the correlation between the direct wave component and the echo direct wave is improved, the order required by Wiener filtering is effectively reduced, and the convergence process of the filter is more stable and reliable.

Description

Technical Field

[0001] This invention relates to the field of radar signal processing, and more specifically to a radar blind-spot-free detection method based on RLS-DCD. Background Technology

[0002] Radars with a transmit / receive switching system typically share an antenna. To ensure a sufficiently long detection range, they usually employ pulse compression technology and accumulate multiple echo cycles to improve the signal-to-noise ratio of the target. Within a single pulse repetition cycle, the power of the transmitted signal must be sufficiently high; the duty cycle of the transmitted signal is usually small (e.g., commonly 5% or 10%) to ensure a sufficiently long echo signal acquisition time.

[0003] With the continuous advancement of digital array radar, radar systems with separate transmit and receive terminals have become easier to implement, such as... Figure 1 As shown. Unlike radars with a transmit-receive switching system, radars with a separate transmit-receive system do not share antennas for transmitting and receiving. The receiving channel can continuously collect and process electromagnetic signals in space (including strong direct wave leakage signals from the transmitting antenna and weak target echo signals). Due to insufficient isolation of the RF front-end or transmit-receive antennas, the receiver's ADC may sample direct waves from the transmitter (such as...). Figure 1 As shown), compared to echoes with a signal-to-noise ratio (SNR) typically less than zero, this direct wave has a SNR much greater than 1. If the ADC is sampled directly (e.g....), Figure 2 As shown), the signal obtained by DDC (digital down-conversion) undergoes pulse compression processing. The pulse compression peak of the weak echo from the target will be submerged in the pulse compression result of the direct wave (as shown). Figure 7 As shown in the image, targets at closer distances will not be detected. Summary of the Invention

[0004] This invention provides a radar blind-spot-free detection method based on RLS-DCD, which solves the problems of existing technologies.

[0005] In a first aspect, the present invention provides a radar blind-spot-free detection method based on RLS-DCD, comprising: direct wave delay measurement: obtaining the coefficients of the baseband waveform matched filter of the transmitted signal, and obtaining the position of the direct wave in the sampled signal by peak search after pulse compression; cancellation signal construction: constructing a transmitted waveform delay signal with the same position as the direct wave using a known transmitted wave, constructing a Wiener filter and solving it using the DCD algorithm, and obtaining the Wiener filter result, i.e., the cancellation signal, by inputting the transmitted waveform delay signal, the ADC sampled signal, and the signal processed by the DCD; and direct wave signal cancellation: subtracting the cancellation signal from the signal processed by the ADC sampled signal and the signal processed by the DCD to obtain the cancelled signal.

[0006] Furthermore, specifically:

[0007] Direct wave time delay measurement:

[0008] Step one: conjugate and flip the baseband waveform of the transmitted signal to obtain the coefficient of the matched filter;

[0009] Step two: send the baseband signal after sampling and digital down conversion into the matched filter for pulse compression to obtain the pulse compression result;

[0010] Step three: perform peak value search on the result obtained in step two to obtain the position of the direct wave in the sampled signal;

[0011] Construction of cancellation signal:

[0012] Step four: according to the position of the direct wave, use the known transmitted wave to construct a transmitted waveform time delay signal with the same length, delay τ, and the rest being 0;

[0013] Step five: use the least square criterion to construct the Wiener filter, use the DCD algorithm to solve the normal equation in the least square criterion, take the transmitted waveform time delay signal constructed in step one as the input signal of the filter, and take the ADC sampling and DCD processed signal as the reference input to obtain the Wiener filtering result, which is the cancellation signal;

[0014] Cancellation of direct wave signal:

[0015] Step six: subtract the ADC sampling and DCD processed signal from the cancellation signal to cancel the direct wave interference, obtain the output after cancellation, and perform pulse compression processing to detect the target submerged in the direct wave.

[0016] Steps one to six are steps 1-6 as follows:

[0017] Step 1: take the known transmitted wave s tx [n] to flip and conjugate to obtain the matched filter parameter:

[0018] h match [n]=conj(s tx [length(s tx )-n])

[0019] Where length(s tx ) is the sampling point number of the transmitted signal, and N s ;

[0020] Step 2: convolve the received signal with the matched filter to obtain the pulse compression result, as shown in Figure 7 ;

[0021] Step 3: Peak search is performed on the result in Step 2, and the peak position is the echo position of the direct wave, and the sampling point is recorded as delay time T delay , which is the delay of the analog front end and the transmission link τ;

[0022] Step 4: Delay the transmit signal by T delay , and zero the rest of the positions until the length is equal to the length N of the single-period received signal, to obtain the transmit waveform delay signal d[n];

[0023] Step 5: Use the least square criterion to construct a multi-tap Wiener filter, and use the transmit waveform delay signal d[n] obtained in Step 4 as the input, and the signal s rx after ADC sampling and DDC processing as the reference signal to perform Wiener filtering, and the output is the cancellation signal s c .

[0024] In Step 5, the accelerated flow of the DCD algorithm is as follows:

[0025] Step 5.1: Initialize the weight coefficient Residual error r(n), autocorrelation matrix R(n);

[0026] Step 5.2: Update the autocorrelation matrix R(n) at the current time according to the autocorrelation matrix R(n-1) at the previous time; the specific way is to multiply the autocorrelation matrix R(n-1) at the previous time by the forgetting factor λ and add the autocorrelation matrix x(n)x(n) of the current input signal x(n) H ;

[0027] Step 5.3: The filter output is the conjugate transpose of the filter increment at the previous time , multiplied by the input x(n) of the filter at the current time;

[0028] Step 5.4: The error at the current time is the reference signal d(n) at the current time minus the output signal y(n) at the current time;

[0029] Step 5.5: The output β0(n) of the auxiliary normal equation at the current time is the product of the forgetting factor λ and the residual error r(n-1) at the previous time, plus the product of the conjugate transpose of the error signal and the current input signal e(n) H x(n);

[0030] Step 5.6: Use the DCD method to solve the auxiliary normal equation R(n)Δh(n)=β0(n) to obtain the residual error r(n) and the weight coefficient increment

[0031] Step 5.6.1: Initialize the weight coefficient increment Residual error r, iteration step size α, and iteration count m;

[0032] Step 5.6.2: Calculate the maximum value of the absolute value of each position in the input β real and imaginary parts together, and record his position p; compare the absolute value of the maximum value with the set threshold, the threshold is related to the position of the maximum value mapped to the position on the autocorrelation matrix, if the absolute value of the maximum value is less than the set threshold, the iteration count m is added 1 and the threshold is adjusted according to the iteration step size, until the absolute value of the maximum value is greater than the threshold, exit the small loop, record the step size at the current time and the sign of the value; if the iteration count m reaches the preset number M b Still not exit the small loop, exit the DCD large loop, directly output the initialized weight coefficient and residual error;

[0033] Step 5.6.3: Calculate the weight coefficient increment at the current time as the weight coefficient increment at the previous time h n plus the product of the sign of the absolute maximum residual error position and the original input value r tmp of the real and imaginary part position, if it is positive, the result is multiplied by 1, if it is negative, the result is multiplied by -1, if it is real, the result is multiplied by 1, if it is imaginary, the result is multiplied by j;

[0034] Step 5.6.4: Calculate the residual error at the current time as the residual error at the previous time r minus the product of the sign of the absolute maximum value position p mapped to the column R (n) of the autocorrelation matrix at the current time and the original input value r tmp of the absolute maximum residual error position;

[0035] Step 5.6.5: Repeat the loop until the iteration number N u is reached, exit the loop to get the weight coefficient increment and residual error calculated by DCD;

[0036] Step 5.7: Get the weight coefficient at the current time by using the weight coefficient increment and the weight coefficient at the previous time solved by DCD algorithm;

[0037] Step 5.8: Repeat the loop steps 5.2-5.8 until the loop number is reached, exit the loop to get the weight coefficient.

[0038] Step 6: Subtract the cancellation signal s rx from the received signal s c to get the cancellation signal s after , pulse compression is performed on s after to get two previously buried peak values (such as Figure 8The two peaks are the target real position. It can be seen that the pulse compression peak value generated by the direct wave has been greatly attenuated after cancellation.

[0039] After measuring the time delay of the direct wave, the application constructs a time delay transmission signal by a known transmission waveform, takes the signal after ADC sampling and DDC processing as a reference, and uses the least square criterion to perform adaptive filtering on the signal;

[0040] The radar blind area detection method based on RLS-DCD provided by the application can ensure that the cancellation signal constructed is consistent in energy with the direct wave component in the echo: by performing time delay processing on the transmission signal, the direct wave component therein is aligned with the echo direct wave in the time dimension, thereby improving the correlation between the two while effectively reducing the order required by the Wiener filter, ensuring that the convergence process of the filter is more stable and reliable. At the same time, the signal obtained after ADC sampling and DDC processing is taken as a reference signal, and the transmission waveform time delay signal is approximated through the filter, which can ensure that the cancellation signal generated contains the amplitude and phase imbalance introduced by the wideband radar wave in the radio frequency front end and the transmission link, thereby further enhancing the cancellation performance. BRIEF DESCRIPTION OF DRAWINGS

[0041] The accompanying drawings described herein are used to provide a further understanding of the embodiments of the application, constitute a part of the application, and do not constitute a limitation on the embodiments of the application. In the drawings:

[0042] Figure 1 A schematic diagram of the cause of the radar direct wave.

[0043] Figure 2 A signal time domain waveform diagram after ADC sampling and DDC processing without cancellation.

[0044] Figure 3 A schematic diagram of the near distance blind area (interference range) caused by direct wave leakage.

[0045] Figure 4 A digital domain adaptive direct wave cancellation method processing flowchart.

[0046] Figure 5 An adaptive direct wave cancellation algorithm implementation structure diagram in FPGA.

[0047] Figure 6 A transmission time domain waveform diagram.

[0048] Figure 7 A result diagram of directly pulse compression on the echo signal.

[0049] Figure 8 A pulse compression result diagram of the cancellation signal. DETAILED DESCRIPTION

[0050] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention.

[0051] Definitions:

[0052] RLS-DCD is a recursive least squares method for binary coordinate descent;

[0053] The center frequency of a single-transmitter, single-receiver radar is f. c The interval between the two transmitted pulses is T, and the transmitted wave is an FMCW signal with a duty cycle of D. Let the received signal be s. rx [n], with a length of N = 2T / T s If the simulated front-end transmit / receive delay is τ, then the signal-to-noise ratio of the direct wave will be much greater than that of the target echo signal within the time range of τ to τ+D*T. Figure 2 As shown, if pulse compression processing is directly applied to the signal obtained from ADC sampling and DDC (digital down-conversion), the pulse compression peak of the target's weak echo will be submerged in the pulse compression result of the direct wave (e.g., Figure 7 (As shown), this will cause arrive Targets within range cannot be captured by pulse compression, such as Figure 3 As shown.

[0054] To address the aforementioned issues, this invention proposes a high duty cycle radar blind-zone-free detection method based on RLS-DCD adaptive cancellation, and uses the DCD (binary classification coordinate descent) algorithm for acceleration, facilitating its efficient implementation in FPGA.

[0055] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0056] Example 1: In this example, the received signal and the transmitted waveform are input into the FPGA as follows: Figure 5 As shown, the signal sampled by the ADC and processed by the DDC, along with the transmitted waveform, is input into the FPGA. This method can estimate the time delay parameters of the direct wave in the propagation link and model and compensate for the amplitude and phase distortions introduced by the RF front-end and the transmission process. Based on this, a cancellation signal is generated to effectively cancel the direct wave component in the receiver, achieving blind-zone-free detection for a high-bandwidth, high-duty-cycle radar.

[0057] The processing flow of the algorithm is shown as follows: Figure 4

[0058] The direct wave time delay estimation, the cancellation signal construction and the direct wave cancellation are not described here again according to the content of the description;

[0059] The simulation of the embodiment adopts the following:

[0060] The radar transmits the FMCW signal in single-receiving and single-transmitting mode, the bandwidth is 100M, the time width is 50μs, the waveform duty cycle is 80%, the distance between the receiving antenna and the transmitting antenna is 15m, the two targets are respectively 75m and 150m away from the radar, and the time-domain waveform of the radar transmission is shown as follows: Figure 6

[0061] The steps 1-6 of the algorithm in the description are implemented, and the algorithm includes the following table:

[0062]

[0063]

[0064] It should be understood that the present application is not limited to the precise construction which has been described above and illustrated in the accompanying drawings, and that various modifications and changes can be made without departing from the scope thereof. The only scope of the present application is limited by the appended claims.​​

Claims

1. A radar non-blind zone detection method based on RLS-DCD, characterized in that, The method comprises the following steps: Direct wave time delay measurement: obtaining the coefficients of the matched filter of the baseband waveform of the transmitted signal, and searching for the peak value after pulse compression to obtain the position of the direct wave in the sampled signal; Anti-jamming signal construction: constructing a time-delay signal of the transmitted waveform at the position of the direct wave using the known transmitted wave, constructing a Wiener filter using a DCD algorithm, inputting the time-delay signal of the transmitted waveform, the ADC-sampled and DDC-processed signal to obtain the Wiener filtering result, i.e., the anti-jamming signal; Direct wave signal cancellation: subtracting the anti-jamming signal from the ADC-sampled and DDC-processed signal to obtain the post-cancellation signal.

2. The RLS-DCD based radar non-blind zone detection method of claim 1, wherein, The direct wave time delay measurement specifically comprises: Step 1: obtaining the coefficients of the matched filter by conjugate flipping the baseband waveform of the transmitted signal; Step 2: inputting the baseband signal sampled and digitally down-converted into the matched filter to perform pulse compression, and obtaining the pulse compression result; Step 3: searching for the peak value of the result obtained in step 2 to obtain the position of the direct wave in the sampled signal.

3. The RLS-DCD based radar non-blind-zone detection method of claim 2, wherein, The anti-jamming signal construction specifically comprises: Step 4: constructing a time-delay signal of the transmitted waveform with the same length, delay τ and 0 elsewhere according to the position of the direct wave; Step 5: constructing a Wiener filter using the least mean square criterion, solving the normal equation in the least mean square criterion using a DCD algorithm, inputting the time-delay signal of the transmitted waveform constructed in step 1 as the input signal of the filter, inputting the ADC-sampled and DDC-processed signal as the reference input, and obtaining the Wiener filtering result, which is the anti-jamming signal.

4. The RLS-DCD based radar non-blind-zone detection method of claim 3, wherein, The specific algorithmic process of steps 1 to 3 in the direct wave time delay measurement is as follows: Step 1: Take the known transmitted wave s tx [n] Perform a flip and conjugation to get matched filter parameters: h match [n] = conj(s tx [length(s tx )-n]) where length(s tx ) is the number of sample points of the transmit signal, and N s ; Step 2: convolving the received signal with the matched filter to obtain the pulse compression result; Step 3: Peak search is performed on the result of Step 2, and the peak position is the direct wave echo position. The sampling point is recorded as delay time T delay is the delay τ of the analog front end and the transmission link.

5. The RLS-DCD based radar non-blind-zone detection method of claim 4, wherein, The specific algorithmic process of steps 4 and 5 in the anti-jamming signal construction comprises: Step 4: Delay the transmit signal by T delay Zero-pad the rest of the positions until the length is equal to the length of the single-cycle received signal N, resulting in the transmit waveform time-delayed signal d[n] Step 5: Construct a multi-tap Wiener filter using the minimum mean square criterion, taking the time-delayed signal d[n] obtained in Step 4 as input, and the signal s after ADC sampling and DDC processing rx Wiener filtering for reference signal, output is the cancellation signal s c .

6. The RLS-DCD based radar non-blind-zone detection method of claim 5, wherein, In step 5, the acceleration process of the DCD algorithm is as follows: Step 5.1: Initialization of the weight coefficients Residual r(n), autocorrelation matrix R(n); Step 5.2: update the autocorrelation matrix R(n) of the current time instant from the autocorrelation matrix R(n-1) of the previous time instant; the specific way is to multiply the autocorrelation matrix R(n-1) of the previous time instant by the forgetting factor λ and add the autocorrelation matrix x(n)x(n) of the current input signal x(n) H ; Step 5.3: Filter output is the conjugate transpose of the previous time step filter increment Multiply by the input x(n) of the current time step filter; Step 5.4: the error at the current time is the reference signal d(n) at the current time minus the output signal y(n) at the current time; Step 5.5: The output β0(n) of the current time instant auxiliary normal equation is the product of the forgetting factor λ and the residual r(n-1) of the previous time instant plus the product of the conjugate transpose of the error signal e(n) and the current time instant input signal x(n) H x(n) Step 5.6: Solve the auxiliary normal equation R(n) Ah(n) = β0(n) using the DCD method to obtain the residual r(n) and weight coefficient increment Step 5.6.1 : Initialization of weight coefficient increment residual r, iteration step size a, iteration count m; Step 5.6.2: calculating the maximum value of the absolute values of the real and imaginary parts of each position in the input β, recording the position p of the maximum value, comparing the absolute value of the maximum value with the set threshold value, and adjusting the threshold value according to the iteration step length if the absolute value of the maximum value is less than the set threshold value, until the absolute value of the maximum value is greater than the threshold value, exiting the small loop, and recording the step length at the current time and the sign of the value; If the iteration count m reaches the preset number M b If the small loop is still not exited, the DCD large loop is exited, and the initialized weight coefficients and the residual are directly output. Step 5.6.3: Calculate the current time weight coefficient increment as the previous time weight coefficient increment h n plus the product of the sign of the current time iteration step size a and the original input value r at the position of the largest absolute value of the residual tmp , if positive, the result is multiplied by 1, if negative, the result is multiplied by -1, if real, the result is multiplied by 1, and if imaginary, the result is multiplied by j; Step 5.6.4: Calculate the residual for the current time instance result as the residual r for the previous time instance minus the product of the sign of the maximum of the absolute values of the residual r and the real and imaginary part positions p of the maximum of the absolute values of the residual r mapped to the column R of the autocorrelation matrix for the current time instance and the iteration step size a for the current time instance (n) and the original input value r of the position of the maximum of the absolute values of the residual r tmp of the maximum of the absolute values of the residual r Step 5.6.5: Repeat the loop until the iteration number N is reached u , exit the loop to obtain the weight coefficient increment and the residual calculated by DCD; Step 5.7: weight coefficient increment solved by using DCD algorithm and the weight coefficient of the previous moment to obtain the weight coefficient of the current moment; Step 5.8: repeating steps 5.2-5.8 until the number of cycles is reached, exiting the loop, and obtaining the weight coefficient.

7. The RLS-DCD based radar non-blind-zone detection method of claim 6, wherein, The specific steps of the direct wave signal cancellation include: subtracting the received signal s rx from the cancellation signal s c to obtain the cancelled signal s after , and performing pulse compression on s after to obtain two previously buried peak values, i.e. the target real position.

Images