A distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing

Abstract

This invention discloses a range ambiguity suppression method based on a joint reversible mismatch filter and nonlinear processing. The invention employs a medium-high PRF operating mode to achieve a larger unambiguous velocity measurement range; utilizes inter-pulse waveform agility technology to obtain range selectivity; designs a joint reversible mismatch filter, combined with an optimized two-dimensional nonlinear ambiguity suppression processing method, to suppress echo energy outside the desired range segment, reducing distortion and residual ambiguity of the desired signal; it updates the clutter target region by iterating the residual components of different range segments, reducing the impact of energy dispersion outside the desired range segment on clutter target region determination; the processing method has low computational complexity, good robustness, and does not require consideration of model mismatch and signal distortion issues, effectively improving radar detection performance and ambiguity suppression capability.

Description

Technical Field

[0001] This invention relates to the field of radar signal processing technology, and specifically to a range ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing. Background Technology

[0002] Pulse Doppler processing is widely used for target detection and tracking in ground-based, airborne, and shipborne radars. A typical pulse Doppler radar waveform uses a coherent pulse sequence with a constant pulse width and pulse repetition frequency (PRF), leading to range and velocity ambiguity. Typically, radars operate in a medium-to-high PRF mode to achieve a larger, unambiguous velocity measurement range, and the average power of the transmitted signal is increased by increasing the number of accumulated pulses. However, this also introduces severe range ambiguity, causing near-range ambiguous echoes to overlap with echoes from weak far-range targets, easily masking small, distant targets or generating false targets.

[0003] Existing literature on range ambiguity suppression mainly focuses on three aspects: First, using a multi-PRF scheme, the true range can be solved from the measured range using the Chinese remainder theorem. However, this increases the total search or monitoring time and may introduce ghosting or false targets in the case of multiple targets. Second, waveform diversity involves modulating different pulses intra-pulse, such as phase coding, frequency coding, frequency agility, and hybrid modulation, utilizing the isolation between different pulse signals to suppress ambiguity echo energy. However, in practical applications, due to distortion factors such as sampling points or Doppler shift, it is difficult to achieve good signal isolation. Furthermore, the energy from the blurred target is not eliminated but redistributed as a cross-correlation that increases the background noise of the signal. Third, nonlinear ambiguity suppression uses a combination of matched filtering and nonlinear processing techniques to eliminate the echo energy of blurred targets in the pulse compression domain, thereby reducing range ambiguity. However, this method leads to distortion of the desired signal imaging and some residual ambiguity energy.

[0004] Therefore, based on medium- and high PRF waveform agile radar, researching range ambiguity suppression methods to address the problems of distortion of the desired signal and residual ambiguity caused by nonlinear processing has important practical significance and application value. Summary of the Invention

[0005] In view of this, the present invention provides a range ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing, which effectively improves radar detection performance and ambiguity suppression capability.

[0006] The distance ambiguity suppression method based on a joint reversible mismatch filter and nonlinear processing of the present invention includes:

[0007] Step 1: The radar transmits phase-coded signals;

[0008] Step two: Identify clutter target regions in each range segment using a range-Doppler two-dimensional nonlinear fuzziness suppression processing method; specifically:

[0009] Two-dimensional coherent accumulation is performed for each distance segment. The two-dimensional coherent accumulation is as follows: the fast time dimension is processed by a joint reversible mismatch filter bank, and the slow time dimension is processed by windowed FFT. The joint reversible mismatch filter has low autocorrelation sidelobes and signal-to-noise ratio loss, as well as reversible filter spectrum and pulse compression sidelobe coherence.

[0010] The clutter target region I for this range is obtained sequentially based on the two-dimensional coherent accumulation results. p and I p After the region is nonlinearly zeroed and inversely transformed, the clutter target region I at each distance is iteratively updated by the residual components;

[0011] Step 3: For each range segment, using the clutter target region I determined in Step 2, perform two-dimensional coherent accumulation on signals outside the current range segment sequentially to obtain the clutter target region for the suppressed range segment. Will Nonlinear zeroing and inverse transformation are used to obtain the echo signals after ambiguity suppression for each range segment;

[0012] Step 4: Perform joint pulse-Doppler processing on the signals of each range segment, and then stitch them together sequentially to obtain a multi-range range-Doppler plane.

[0013] Preferably, in step two, after all range segments have completed two-dimensional coherent accumulation, the residual components are subjected to two-dimensional coherent accumulation again. If the clutter target region is not updated, step three is executed; otherwise, two-dimensional coherent accumulation is performed again until the clutter target region is not updated, and then step three is executed.

[0014] Preferably, the phase-encoded signal is:

[0015] Within a CPI, there are N distinct pulse sequences, each pulse signal u n (n = 0N-1) is a phase-coded modulation of code length M [c n,0 ,c n,1 ,,c n,M-1 The time domain of the nth transmitted signal is:

[0016]

[0017]

[0018] in For the m-th phase encoding within the n-th pulse, φ n(m) is the phase modulation function, taking any value in [0, 2π), and t represents the fast time, T c T is the symbol width. p =MT c T is the pulse width. r The pulse repetition interval;

[0019] The frequency domain of the nth transmitted pulse signal is:

[0020]

[0021] Where f represents the fast time frequency.

[0022] Preferably, in step two, the method for obtaining the joint reversible mismatch filter bank is as follows:

[0023] S21, Construct the initialization mismatch filter:

[0024] The complex vector representation of the input pulse signal sampling is as follows: N s =T p f s f s The sampling frequency is given, and the mismatched filter length is P = KN. s K is the oversampling coefficient, typically on the order of 2 to 4. The mismatch filter coefficient vector is represented as h = [h1 h2 h... P ] T The result of pulse signal mismatch filtering is:

[0025]

[0026] Among them, when q-p+1≤0 and q-p+1>N s At that time, s q-p+1 =0, to represent equation (20) in matrix form, first define the dimension as ((K+1)N s -1)×KN s Convolution matrix:

[0027]

[0028] The output of pulse signal mismatch filtering can be expressed as:

[0029] y MMF =Ah (22)

[0030] First, the filter energy needs to be normalized, and the energy of the matched filter is defined as S = s H s, where H represents the Hermitian operator; next, a mismatched filter with low sidelobes and signal-to-noise ratio loss is designed based on the criterion of minimizing the integral sidelobe level. The optimization problem can be expressed as:

[0031]

[0032] Where ||·||2 represents the L2 norm, and e is the expected mismatch filter output;

[0033] By solving equation (23), the time-domain coefficients h of the mismatched filter are obtained. id and frequency domain response H id ;

[0034] S22, using the mismatched filter coefficients determined in S21 as initialization h0, calculate its amplitude-frequency response:

[0035] H0=Fh0 (25)

[0036] Where F is the P-point DFT matrix:

[0037]

[0038] Let H0 be the modulus of H0. Determine if it satisfies the requirement that an acceptable reversible mismatched filter has no zeros or minima. If it does, the required reversible mismatched filter is obtained. If not, solve the constrained QP problem using a nested iterative algorithm to gradually approximate the required reversible mismatched filter. The optimization problem is then updated to:

[0039]

[0040] Among them, threshold H id The element in the array is less than δ, and its index is denoted as k. id , will H id k in id The element at position δ is set as denoted as W kid Its function is to extract its Hadema product vector k. id The element at position k id The element at position 1 is 0, and all other elements are 0.

[0041] S23, based on the reversible mismatch filter obtained in S22, uses the dynamic alternating projection method to construct a joint reversible mismatch filter bank.

[0042] Preferably, in S21, the expected mismatch filter output e is set as follows: the main lobe value of e is set to the main lobe value of the energy-normalized matched filter result, and the rest are zero values.

[0043] Preferably, in S21, the least squares method is used or equation (23) is solved using the CVX toolbox.

[0044] A better approach is to use the CVX toolbox to solve equation (27); in each loop, check whether the cutoff condition is met:

[0045]

[0046] If not satisfied, update the constraint ε. id If the condition is met, the loop ends, and the desired joint reversible mismatch filter is obtained.

[0047] Beneficial effects:

[0048] (1) This invention is based on the characteristics of medium-high PRF waveform agile radar, which can obtain a large unambiguous velocity measurement range. It improves the original nonlinear ambiguity suppression method. In the range-Doppler domain, it reduces the influence of ambiguity energy dispersion on the selection of clutter target regions by iteratively updating the clutter target region at different range segments. Furthermore, it uses a joint reversible mismatch filter (JRVMMF) to reduce the peak sidelobe level, restore sidelobe coherence, and solve the range sidelobe modulation (RSM) effect. The processing method of this invention has low computational complexity, good robustness, and does not need to consider model mismatch and signal distortion, effectively improving radar detection performance and ambiguity suppression capability.

[0049] (2) In view of the problem that nonlinear processing can easily lead to distortion of the desired signal and residual ambiguity, the present invention designs a reversible mismatch filter and uses this reversible mismatch filter as initialization. The joint reversible mismatch filter is constructed by using the alternating projection algorithm, which can effectively reduce the peak sidelobe level, restore the sidelobe coherence, and solve the range sidelobe modulation effect while maintaining reversibility and stability. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of range gating based on inter-pulse agile waveforms and receiving filter banks for different range segments;

[0051] Figure 2 Here is a flowchart of distance-Doppler two-dimensional nonlinear fuzziness suppression.

[0052] Figure 3 Here is a flowchart of a distance ambiguity suppression method based on a joint reversible mismatch filter and nonlinear processing;

[0053] Figure 4 A comparison of the inverse spectra of the joint reversible mismatched filter bank and the matched filter bank;

[0054] Figure 5 a represents the pulse compression result based on the inter-pulse agile waveform matched filter bank;

[0055] Figure 5 b represents the pulse Doppler processing result based on the inter-pulse agile waveform matched filter bank;

[0056] Figure 6 a represents the pulse compression result based on the pulse agile waveform combined with the reversible mismatch filter bank;

[0057] Figure 6 b represents the pulse Doppler processing result based on the pulse agile waveform combined with the reversible mismatch filter bank;

[0058] Figure 7 a represents the real-time echo after inverse processing of conjugate matched filtering following fast-time-dimensional nonlinear fuzziness suppression;

[0059] Figure 7 b is the real part time-domain echo after inverse processing of joint reversible mismatch filtering following fast time-dimensional nonlinear fuzziness suppression;

[0060] Figure 8 Comparison of pulse compression results for matched filtering, nonlinear fuzzy suppression based on conjugate matched filter and joint invertible mismatch filter;

[0061] Figure 9 a represents the result of multi-range segment joint pulse Doppler processing after two-dimensional nonlinear fuzzy suppression processing based on inter-pulse agile waveform and reversible mismatch filter group;

[0062] Figure 9 b represents multi-range pulse Doppler processing based on inter-pulse agile waveform matched filter banks;

[0063] Figure 10 a represents the measured data of the inter-pulse agility waveform and the results of multi-range pulse Doppler processing before fuzziness suppression in this invention (three-view drawing);

[0064] Figure 10 b represents the measured data of the inter-pulse agile waveform and the result of multi-range pulse Doppler processing after fuzz suppression in this invention (three-view drawing);

[0065] Figure 10 c represents the measured data of the inter-pulse agile waveform and the multi-range pulse Doppler processing results after fuzz suppression in this invention (top view);

[0066] Figure 10 d represents the measured data of the inter-pulse agility waveform and the result of multi-range pulse Doppler processing after fuzz suppression in this invention (enlarged view of the target area). Detailed Implementation

[0067] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0068] This invention provides a range ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing. It employs a medium-high PRF operating mode to achieve a larger unambiguous velocity measurement range; utilizes inter-pulse waveform agility technology to obtain range selectivity; designs a joint reversible mismatch filter, combined with an optimized two-dimensional nonlinear ambiguity suppression processing method, to suppress echo energy outside the desired range segment, reducing distortion and residual ambiguity of the desired signal; it updates the clutter target region by iterating the residual components of different range segments, reducing the impact of energy dispersion outside the desired range segment on clutter target region determination; the processing method has low computational complexity, good robustness, and does not require consideration of model mismatch and signal distortion issues, effectively improving radar detection performance and ambiguity suppression capability.

[0069] Specifically, the steps include the following:

[0070] Step S1, suppressing distance ambiguity based on nonlinear processing of phase-coded signals, specifically includes the following sub-steps:

[0071] Step S11: Construct a phase-coded signal model with code length M and pulse number N, and analyze the time-frequency domain characteristics of the signal.

[0072] Assume there are N different pulse sequences within a CPI, each pulse signal u n (n = 0N-1) is a phase-coded modulation of code length M [c n,0 ,c n,1 ,,c n,M-1 If the nth transmitted signal is given by the given information, then the time domain representation can be expressed as:

[0073]

[0074]

[0075] in For the m-th phase encoding within the n-th pulse, φ n (m) is the phase modulation function, taking any value in [0, 2π), and t represents the fast time, T c T is the symbol width. p =MT c T is the pulse width. r This is the pulse repetition interval.

[0076] The frequency domain representation of the nth transmitted pulse signal can be expressed as:

[0077]

[0078] Where f represents the fast time frequency.

[0079] Equation (3) can be rewritten in two parts:

[0080] Un (f)=A n (f)B n (f) (4)

[0081] in:

[0082]

[0083]

[0084] A n (f) is the ripple term generated by the phase transition, B n (f) is the envelope term generated by the signal symbol envelope, which determines the total bandwidth and spectral envelope of the signal.

[0085] Step S12: Construct an echo model for distance-ambiguous scenarios and introduce a traditional one-dimensional nonlinear ambiguity suppression method based on conjugate matched filters.

[0086] The blurred scene of the pulse agility waveform and the range gating diagram of the receiving filter bank at different range segments are shown below. Figure 1 As shown, the current PRT receiving range may receive ambiguous echo signals from outside the current range segment. Different receiving filter banks can be set for different range segments to achieve coherent processing of the echoes from the region of interest. Assuming the transmitted signal u... n There is a distance ambiguity, meaning that the echoes of the first n pulses are folded into the (n+1)th receiving interval. Assuming the current processing u... n Let be the desired echo signal, and the others be ambiguous echo signals. Without loss of generality, assume that the complex amplitude of all echo signals is 1 and the ambiguity delay is 0. Then the expression for the total echo s(t) is as follows:

[0087]

[0088] The matched filter for the k-th pulse signal is defined as:

[0089]

[0090] Step 1: Fast time-dimensional matched filtering of the k-th fuzzy signal to be suppressed: the k-th fuzzy signal is then the autocorrelation response r. kk :

[0091]

[0092] in, Represents convolution;

[0093] And the desired signal u n (t) represents the cross-correlation response r nk :

[0094]

[0095] Step 2: Nonlinear zeroing of the autocorrelation response of the fuzzy signal. This is achieved by setting a threshold α(t) to suppress the autocorrelation response of the fuzzy signal while preserving the cross-correlation response of the desired signal. The nonlinear zeroing operator is defined as Γ. α :

[0096]

[0097] Step 3: Fast Time Dimension Conjugate Matched Filtering u k (t) Inverse processing: the cross-correlation response of the desired signal is deduced into a time-domain echo signal, and the fuzziness suppression operator Θ is defined. k Used to suppress the k-th ambiguous signal:

[0098]

[0099] For u n To obtain the desired echo signal channel, ambiguity suppression needs to be performed sequentially according to adjacent pulses from near to far, ultimately yielding the ambiguity-suppressed echo signal, which is then subjected to fast time-dimensional matched filtering. Then u n Channel output y n (t) is:

[0100]

[0101] in, Indicates repeated convolution, q kk (t) is r kk (t) The result after nonlinear zeroing, let t m If the interval is the zero point of the main lobe, then the ideal q kk (t) can be expressed as:

[0102]

[0103] The first term of equation (13) reveals the distortion caused by nonlinear fuzzy suppression, and the second term is the residual fuzzy quantity.

[0104] Step S2 involves analyzing the problems of desired signal imaging distortion and residual blur that occurred in Step S1, as well as the sidelobe modulation effect, and designing a joint reversible mismatch filter, including the following sub-steps:

[0105] Step S21: Analyze the causes of the expected signal imaging distortion and residual blur that occurred in step S1, and transform the problem into the design of a reversible mismatch filter.

[0106] Without loss of generality, consider the two-pulse case, i.e., s(t) = u1(t) + u0(tT). rTaking channel u1 as an example, the coherent accumulation processing filter for the k-th pulse signal is defined as h. k (t), the inverse processing filter is defined as h invk (t), its output is:

[0107]

[0108] For ease of analysis, the output is converted to the frequency domain, and the result is as follows:

[0109] The time delay corresponds to the frequency shift, where The frequency shift factor is used. The first term in the above equation is the distortion term of the pulse compression result of the desired signal u1, and the second term is the residual ambiguity term from the nonlinear processing. When the matched filter from step 1 is used... Conjugate matched filter, h invk (t)=u k (t), H invk (f)=U k (f), then equations (15) and (16) are transformed into:

[0110]

[0111]

[0112] Where ⊙ represents the Hadmar product, and the related function r mn (t) and R mn (f) are Fourier transform pairs. From equations (17) and (18), it can be seen that the pulse compression result of the desired signal u1 is affected by the pulse compression result of the ambiguous signal u0, thus causing distortion. If the time domain result of the first two filters is a Dirac function, that is, the frequency domain result is a vector of all 1s, then the desired signal u1 can be recovered without distortion, thereby reducing the distortion of the desired signal. Obviously, R in equation (18) 00 (f)=H0(f)⊙H inv0 (f)=|U0(f)| 2 Since H is not a vector of all 1s, the problem is transformed into: making H k (f)⊙H invk (f) is an all-1 vector, meaning the second-stage filter is the inverse of the first-stage filter:

[0113] H invk (f)=[H k (f)] -1 (19)

[0114] [H k (f)] -1 Indicates H k(f) Taking the reciprocal of each element, as shown in equation (4), it is known that directly constructing an inverse filter from a matched filter is unstable because the spectrum of a matched filter has zeros and minima. Therefore, a mismatch filter needs to be designed so that its spectrum has no zeros or minima. Furthermore, as shown in equation (14), the residual ambiguity is the sidelobe portion remaining after the main lobe is zeroed. Therefore, the residual ambiguity can be reduced by concentrating energy on the main lobe and reducing the sidelobes. Finally, the problem is transformed into designing a filter with low sidelobes and signal-to-noise ratio loss, as well as a reversible spectrum, which is called a reversible mismatch filter.

[0115] Step S22: Design a reversible mismatch filter that meets the requirements of step S21.

[0116] The complex vector representation of the input pulse signal sampling is as follows: N s =T p f s f s The sampling frequency is given, and the mismatched filter length is P = KN. s (K is typically on the order of 2 to 4), the mismatch filter coefficient vector is represented as h = [h1 h2 …h P ] T The result of pulse signal mismatch filtering is:

[0117]

[0118] Among them, when q-p+1≤0 and q-p+1>N s At that time, s q-p+1 =0, to represent equation (20) in matrix form, first define the dimension as ((K+1)N s -1)×KN s Convolution matrix:

[0119]

[0120] The output of pulse signal mismatch filtering can be expressed as:

[0121] y MMF =Ah (22)

[0122] First, the filter energy needs to be normalized, and the energy of the matched filter is defined as S = s H s, where H represents the Hermitian operator. Next, a mismatched filter with low sidelobes and signal-to-noise ratio loss is designed based on the criterion of minimizing the integrated sidelobe level (ISL). The optimization problem can be expressed as:

[0123]

[0124] Where ||·||2 represents the L2 norm, and e is the expected mismatch filter output. Because oversampling may occur, resulting in multiple data points within the main lobe, the influence of the main lobe width cannot be ignored when setting the expected mismatch filter output e. Otherwise, ignoring the out-of-band spectrum limitation will cause a significant signal-to-noise ratio loss. Therefore, considering the main lobe width and signal-to-noise ratio loss, the expected mismatch filter output e should be set reasonably. Here, we refer to the known energy-normalized matched filtering results and give the corresponding setting criteria: the main lobe value of e is set to the main lobe value of the energy-normalized matched filtering result, and the rest are zero values. Therefore, the objective function (23) reduces ISL while constraining the main lobe energy, effectively improving the signal-to-noise ratio loss.

[0125] Equation (23) is an unconstrained quadratic programming (QP) problem, and the closed-form solution of the mismatch filter coefficients can be obtained by the least squares method:

[0126] h LS =(A H A+δI) -1 A H e (24)

[0127] Where I is the identity matrix and δ is a small constant. To avoid ill-conditioned points (zeros in the sidelobes) caused by signal oversampling, it can also be solved using the CVX toolbox.

[0128] At this point, the design of the mismatched filter considering the main lobe width and signal-to-noise ratio loss to minimize the ISL criterion is completed. Then, further constraints are added to constrain the minimum value of the amplitude-frequency response of the mismatched filter, ensuring that the designed mismatched filter is reversible and stable, that is, the spectrum has no zeros or minima.

[0129] Using the mismatched filter coefficients generated above as initialization h0, calculate its amplitude-frequency response:

[0130] H0=Fh0 (25)

[0131] Where F is the P-point DFT matrix:

[0132]

[0133] Modulo H0 and denote it as H0. Determine whether it meets the requirements of an acceptable reversible mismatch filter having no zeros and minimum values ​​(H0 > δ, and each element of H0 is greater than δ). If the requirements are met, the required reversible mismatch filter is obtained.

[0134] If the requirements are not met, a nested iterative algorithm is used to solve the constrained QP problem, gradually approximating the reversible mismatched filter that meets the requirements. The optimization problem is then updated to:

[0135]

[0136] Among them, threshold H id The element in the array is less than δ, and its index is denoted as k. id , will H id k in id The element at position δ is set as denoted as Its function is to extract its Hadema product vector k. id The element at position k id The element at position 1 is 0, and all other elements are 0.

[0137] Equation (27) is a constrained QP problem, which can be solved using the CVX toolbox, such as the GUROBI and SDPT3 solvers based on the interior point method. Each loop checks whether the cutoff condition is met.

[0138]

[0139] If the conditions are not met, update the constraints; if they are met, the loop ends, yielding the desired reversible mismatch filter. The design process is summarized in Table 1.

[0140] Table 1

[0141]

[0142] Step S23: Considering the sidelobe modulation effect of the inter-pulse agile waveform, a joint reversible mismatch filter is constructed using the dynamic alternating projection method.

[0143] Due to waveform agility, the pulse compression sidelobes corresponding to each waveform are non-coherent. After slow-time dimension coherent accumulation, the sidelobe energy diffuses throughout the Doppler domain, which may obscure weak targets. In order to maintain the coherence of the diversity waveform pulse compression sidelobes, it is desirable that the pulse compression output of each waveform satisfies the following:

[0144]

[0145] The above equation, when converted to frequency domain, is:

[0146]

[0147] The spectrum H of the reversible mismatch filter constructed above n (f) is used as initialization. The result of noncoherent superposition of the power spectra of pulse compression of N waveform diversity signals:

[0148]

[0149] Then, calculate the power spectrum correction vector for each signal:

[0150]

[0151] To avoid maxima and significant signal-to-noise ratio loss, i.e., to ensure minimal changes in each iteration, the nth mismatch filter is updated as follows:

[0152] H(f)⊙β(f)

[0153] Substitute the updated result of the mismatched filter in equation (33) into equation (31) to make the pulse compression output of all mismatched filters approximately the same through continuous iterative adjustments, thereby restoring the sidelobe coherence.

[0154] Step S3: Utilize the joint reversible mismatch filter obtained in step S2 combined with the distance-Doppler two-dimensional nonlinear fuzziness suppression processing method. Figure 2 As shown in Table 2, the clutter target region identification process is as follows:

[0155] Table 2

[0156]

[0157]

[0158] P represents the maximum ambiguity suppression range. In order to reduce the impact of energy dispersion outside this range on clutter and target area identification, the clutter target area is updated by iterating the residual components.

[0159] Step S4: Using the clutter target regions I at different ranges obtained in step S3, two-dimensional nonlinear ambiguity suppression is performed sequentially on the different ranges. The signals at different ranges after ambiguity suppression are then subjected to joint pulse-Doppler processing. The overall nonlinear ambiguity suppression process is shown in Table 3.

[0160] Table 3

[0161]

[0162] N is the maximum imaging range segment. The pulse-Doppler planes of each range segment after blur suppression in step S4 are sequentially stitched together to obtain a multi-range range-Doppler plane.

[0163] Step S3, clutter target region identification, and step S4, nonlinear ambiguity suppression processing together constitute a range ambiguity suppression method based on a joint reversible mismatch filter and nonlinear processing. The flowchart is as follows: Figure 3 As shown.

[0164] The present invention provides the following embodiments to illustrate the method:

[0165] The radar waveform parameters are shown in Table 4:

[0166] Table 4

[0167]

[0168] Designing joint reversible mismatch filters according to the waveform parameters in Table 4 and the method in step S2, we obtain 32 joint reversible mismatch filters. The spectrum of these filters is compared with that of the matched filters as follows: Figure 4 As shown, the spectrum of the joint reversible mismatch filter is constrained above the threshold, with no zeros or minima.

[0169] The pulse compression and pulse Doppler processing results based on traditional matched filter banks are as follows: Figure 5 As shown, due to the non-coherence of the sidelobes at the output of the matched filter for agile signals, the sidelobe energy diffuses throughout the Doppler domain. The pulse compression and pulse Doppler processing results based on the joint reversible mismatched filter bank are as follows: Figure 6 As shown, after joint reversible mismatch filtering, the range sidelobe recovers coherence, which can effectively suppress the range sidelobe modulation effect and is beneficial to subsequent nonlinear fuzziness suppression processing, reducing the distortion of the desired signal and the amount of residual fuzziness.

[0170] Assume target 1 is located at 0.25R. u As the desired objective, objective 2 is located at 1.75R. u To blur the target, the target echo amplitude is equal. This represents the maximum unambiguous distance, where C represents the speed of light. Following the fast time-dimensional nonlinear fuzziness suppression processing based on the conjugate matched filter in step S1 and the joint reversible mismatch filter in step S2, the real-time echoes before and after fuzziness suppression are as follows: Figure 7 a and Figure 7 As shown in b. The Ambiguous Power Suppressed Ratio (APSR) is defined as the ratio of the fuzzy echo power after fuzzy suppression processing to that before processing:

[0171]

[0172] Where, N am For fuzzy echo complex data s A With [n] sampling points, the APSRs of the two methods are -29.3203dB and -80.7483dB, respectively. The fast time-dimensional nonlinear fuzziness suppression method based on the joint reversible mismatch filter is superior to the fast time-dimensional nonlinear fuzziness suppression method based on the conjugate matched filter, and the desired signal distortion is less. The fuzz-suppressed echo is then subjected to fast time-dimensional matched filtering and joint reversible mismatch filtering of the desired signal. The output is compared with the direct matched filter output. Figure 8 As shown, the joint reversible mismatch filter effectively reduces the distortion and residual ambiguity of the desired signal in nonlinear processing, while also improving the signal-to-noise ratio loss.

[0173] The fast time-dimensional nonlinear fuzziness suppression processing is extended to the range-Doppler dimension, and a clutter target region identification process for each range segment is added in step S3. Together with the nonlinear fuzziness suppression processing in step S4, this constitutes a range fuzziness suppression method based on a joint reversible mismatch filter and nonlinear processing. Based on this method, the range-Doppler plane of the fuzz-suppressed desired signal is obtained as follows: Figure 9 As shown in (a), the distance-Doppler plane to the desired signal obtained by the direct matched filter bank is as follows: Figure 9 As shown in (b), the comparison results show that this method effectively suppresses blur energy and improves imaging performance.

[0174] To further demonstrate the effectiveness of the method, this invention was also verified using experimental data. The experimental process was conducted on the ground, with the transmitted signal being an inter-pulse code agile signal, folded clutter distributed in the first 8 range segments, and the target located in the 11th range segment.

[0175] The multi-range segment distance-Doppler plane obtained by direct matched filter bank processing is as follows: Figure 10 As shown in (a), most of the range is obscured by the scattered energy of the strong clutter in the first range segment. The multi-range segment range-Doppler plane obtained by processing using the method proposed in this invention is as follows: Figure 10 As shown in (b), the target can be detected in the 11th distance segment. The top view and the magnified view of the target area are shown below. Figure 10 As shown in (c) and (d).

[0176] The results above show that the method proposed in this invention suppresses the fuzzy echo energy outside the local range, reduces the distortion and residual fuzziness of the desired signal, has low computational complexity, good robustness, and achieves fuzz-free imaging across multiple ranges.

[0177] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing, characterized in that, include: Step 1: The radar transmits phase-coded signals; Step two: Identify clutter target regions in each range segment using a range-Doppler two-dimensional nonlinear fuzziness suppression processing method; specifically: Two-dimensional coherent accumulation is performed for each distance segment. The two-dimensional coherent accumulation is as follows: the fast time dimension is processed by a joint reversible mismatch filter bank, and the slow time dimension is processed by windowed FFT. The joint reversible mismatch filter has low autocorrelation sidelobes and signal-to-noise ratio loss, as well as reversible filter spectrum and pulse compression sidelobe coherence. The clutter target region for this range is obtained sequentially based on the two-dimensional coherent accumulation results. and will After the region is nonlinearly zeroed and inversely transformed, the clutter target region I at each distance is iteratively updated by the residual components; Step 3: For each range segment, using the clutter target region I determined in Step 2, perform two-dimensional coherent accumulation on signals outside the current range segment sequentially to obtain the clutter target region for the suppressed range segment. ,Will Nonlinear zeroing and inverse transformation are used to obtain the echo signals after ambiguity suppression for each range segment; Step 4: Perform joint pulse-Doppler processing on the signals of each range segment, and then stitch them together sequentially to obtain a multi-range range-Doppler plane.

2. The distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing as described in claim 1, characterized in that, In step two, after all range segments have completed two-dimensional coherent accumulation, the residual components are subjected to two-dimensional coherent accumulation again. If the clutter target region is not updated, step three is executed; otherwise, two-dimensional coherent accumulation is performed again until the clutter target region is not updated, and then step three is executed.

3. The distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing as described in claim 1, characterized in that, The phase-encoded signal is: Within a CPI N A different pulse sequence, each pulse signal It is based on the code length of M Phase-coded modulation , No. n The time domain of each transmitted signal is: （1） （2） in For the first n The first pulse within the [number] pulses m Each phase code, For the phase modulation function, in Take any value, t Indicates a fast time. The width of the symbol. The pulse repetition interval; No. n The frequency domain of each transmitted pulse signal is: （3） in, f It indicates a fast time frequency.

4. The distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing as described in claim 3, characterized in that, In step two, the method for obtaining the joint reversible mismatch filter bank is as follows: S21, Construct the initialization mismatch filter: The complex vector representation of the input pulse signal sampling is as follows: , , The sampling frequency is [value], and the mismatch filter length is [value]. , K The oversampling coefficients are typically on the order of 2 to 4. The mismatch filter coefficient vector is represented as follows: The result of pulse signal mismatch filtering is: （20） Among them, when and hour, To represent equation (20) in matrix form, we first define the dimension as... Convolution matrix: （21） The output of pulse signal mismatch filtering can be expressed as: （22） First, the filter energy needs to be normalized, and the energy of the matched filter is defined as follows: ,in H The Hermitian operator is used to represent this; next, a mismatched filter with low sidelobes and signal-to-noise ratio loss is designed based on the criterion of minimizing the integral sidelobe level. The optimization problem can be expressed as: （23） in, Represents the L2 norm, It is the expected mismatch filter output; By solving equation (23), the time-domain coefficients of the mismatched filter are obtained. and frequency domain response ; S22, using the mismatch filter coefficients determined in S21 as initialization... Find its amplitude-frequency response: （25） Where F is P Point DFT matrix: （26） right Modulo is denoted as We determine whether it meets the requirement that an acceptable reversible mismatched filter has no zeros or minima. If it does, the required reversible mismatched filter is obtained; otherwise, we solve a constrained QP problem using a nested iterative algorithm to gradually approximate the required reversible mismatched filter. The optimization problem is then updated to: （27） Among them, threshold , ; The element in is less than The index is denoted as ,Will In Set the element at position to Recorded as ; Its function is to extract its Hadema product vector. The element at position, i.e. The element at position 1 is 0, and all other elements are 0. S23, based on the reversible mismatch filter obtained in S22, uses the dynamic alternating projection method to construct a joint reversible mismatch filter bank.

5. The distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing as described in claim 4, characterized in that, In S21, the desired mismatch filter output is... The criteria are set as follows: The main lobe value is set to the main lobe value of the energy-normalized matched filter result, while other values ​​are zero.

6. The distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing as described in claim 4, characterized in that, In S21, the least squares method is used or equation (23) is solved using the CVX toolbox.

7. The distance ambiguity suppression method based on joint reversible mismatch filtering and nonlinear processing as described in claim 4, characterized in that, Solve equation (27) using the CVX toolbox; check whether the cutoff condition is met in each loop: （28） If the conditions are not met, update the constraints. If the condition is met, the loop ends, and the desired joint reversible mismatch filter is obtained.

Images