A method and system for analyzing intra-pulse characteristics of aliased radar signals

Abstract

The present invention discloses a method and system for analyzing intra-pulse features of aliased radar signals, relating to the field of radar countermeasure detection. The method comprises: Step 1: separating a mixed signal with overlapping time and frequency domains into single signals; Step 2: detecting the start and end times of each separated signal; Step 3: identifying the modulation type of each separated signal; and Step 4: estimating the modulation parameters of each separated signal. The present application also discloses a computer system based on the intra-pulse feature analysis technology for aliased radar signals. The system is capable of automatically, accurately, and rapidly analyzing the intra-pulse features of aliased radar signals.

Description

Technical Field

[0001] The present invention belongs to the field of radar countermeasure reconnaissance technology, and in particular relates to a method and system for analyzing intra-pulse characteristics of aliased radar signals. Background Art

[0002] Today, new radar systems are developing rapidly and gradually becoming dominant. Conventional pulse radar signals are decreasing in these systems, while linear frequency modulation (LFM), nonlinear frequency modulation (NFM), and phase-coded radar signals are increasing. This places higher demands on radar signal sorting and recognition, necessitating analysis of the intra-pulse modulation characteristics of radar signals.

[0003] Radar signal intrapulse feature analysis is a new technology that began to be researched in the mid-1980s. It primarily uses subtle intrapulse features to distinguish and identify complex modulation signals, guiding timely and accurate responses, and playing a crucial role in radar countermeasures. Radar signal intrapulse features are the most concentrated and important manifestation of subtle radar signal characteristics. Compared to the basic characteristics of radar signals, they provide a deeper and more nuanced mathematical and physical description of the signal. Currently, there are two main categories of radar signal modulation identification methods: decision-making methods based on hypothesis testing theory and statistical pattern recognition methods that do not rely on assumptions. Radar modulation parameter estimation mainly includes high-order statistics, cyclic spectral correlation, time-frequency distribution, wavelet transform, fractional Fourier transform, chaotic signal processing, Chirplet transform, and modern spectral estimation methods.

[0004] For intermediate frequency signals, multiple signals will arrive at the electronic reconnaissance receiver simultaneously during the interception process, especially in the low frequency band. During broadband reception, communication signals and other signals that we are not interested in will be mixed, which increases the difficulty of identifying the signals of interest and brings great trouble to the analysis of the radar signal pulse characteristics, resulting in errors in the modulation mode identification and modulation parameter estimation results.

[0005] Therefore, when analyzing the intra-pulse characteristics of radar signals, it is necessary to first effectively separate the overlapping signals in the time and frequency domains within the pulse. In addition, how to correctly identify the modulation type of the separated signal and estimate its modulation parameters to facilitate signal processing and application is also a technical problem that needs to be solved urgently. Summary of the Invention

[0006] To solve the above technical problems, the present invention proposes a method and system for analyzing intra-pulse characteristics of aliased radar signals, which are technical solutions for effectively analyzing intra-pulse characteristics caused by multi-signal aliasing.

[0007] A first aspect of the present invention discloses a method for analyzing intra-pulse characteristics of an aliased radar signal. The method comprises the following steps:

[0008] Step 1: Separate the mixed signal with overlapping time and frequency domains into independent signals and store them;

[0009] Step 2: Detecting the start and end time of each separated signal based on the envelope of each separated independent signal;

[0010] Step 3: Identify the modulation type of each of the separated signals;

[0011] Step 4: Based on the obtained modulation type of each of the signals, estimate the modulation parameters of each of the separated signals.

[0012] According to the method of the first aspect of the present invention, in step 1, the process of separating the aliased signal includes:

[0013] Step 11), perform fast Fourier transform on the original I / Q channel signal to obtain the spectrum of the signal to be processed;

[0014] Step 12) finding the position where the maximum amplitude of the spectrum of the signal to be processed is located;

[0015] Step 13), determine whether the position where the maximum amplitude of the spectrum of the signal to be processed is located is a third-order polynomial phase signal; if so, calculate the spectrum width that should be retained near the maximum amplitude, and enter step 16); if not, enter step 14);

[0016] Step 14), calculating the kurtosis and skewness values ​​of the position where the maximum value of the spectrum amplitude of the signal to be processed is located, and judging whether the position where the maximum value of the spectrum amplitude is located is a signal or noise based on the kurtosis and skewness values; if it is a signal, proceeding to step 15), otherwise exiting the aliasing signal separation process;

[0017] Step 15), calculating the spectrum width that should be retained near the maximum value of the spectrum amplitude;

[0018] Step 16) retain the signal of the spectrum width and store it, set the other parts of the spectrum of the signal to be processed as noise, and return to step 12).

[0019] According to the method of the first aspect of the present invention, in step 15), calculating the spectrum width that should be retained near the maximum spectrum amplitude specifically includes:

[0020] Determine whether the signal with the maximum spectrum amplitude is a sinusoidal frequency modulation signal. If so, calculate the number of bandwidth points k to be retained around the maximum frequency point of the sinusoidal frequency modulation signal, set the number of points on the left and right sides to be the same, and determine the spectrum width to be retained. If not, calculate the similarity coefficient between the signal spectrum and the rectangular spectrum and the triangular spectrum, and determine the spectrum width to be retained based on the obtained similarity coefficient between the signal spectrum and the rectangular spectrum and the triangular spectrum.

[0021] According to the method of the first aspect of the present invention, in step 2, the start and end time of each signal separated by envelope detection includes:

[0022] Step 21: normalize and smooth the envelope of each signal;

[0023] Step 22: Calculate the smoothed normalized envelope and calculate its mean and standard deviation;

[0024] Step 23: Determine the optimal ratio between the rising edge and the falling edge when intercepting the signal; if the standard deviation is less than 0.1, the optimal ratio is the sum of the mean and the standard deviation, and proceed to step 24; if the standard deviation is greater than or equal to 0.25, the optimal ratio is 0.4, and proceed to step 24; if the standard deviation is in the interval [0.1, 0.25], compare the mean values, and if the mean is greater than 0.7, remove the signal at the small value point; if the mean is less than 0.25, remove the signal at the large value point; if the mean is in the interval [0.25, 0.7], retain all signals, calculate the start and end times of the signals based on the subscript serial numbers of the retained signals, and skip step 24;

[0025] Step 24: Calculate the start and end times of the signal based on the determined optimal ratio, determine the points in the smoothed normalized signal envelope that are greater than the optimal ratio, calculate the subscript numbers of the rising and falling edges of the signal envelope, and calculate the start and end times of the signal based on the subscript numbers.

[0026] According to the method of the first aspect of the present invention, in step three, the modulation type of each signal after identification and separation includes: estimating the signal-to-noise ratio using the second-order fourth-order moment method; performing short-time filtering on the signal; calculating the instantaneous phase of the signal after dewarping; using a 20-fold phase difference method to calculate the instantaneous frequency of the signal; performing a first-order linear fit on the time-frequency curve to calculate the regression coefficient B, residual r and complex correlation coefficient R; and judging the modulation type of each signal based on the obtained regression coefficient B, residual r and complex correlation coefficient R.

[0027] According to the method of the first aspect of the present invention, in step four, the estimation of the modulation parameters of each separated signal includes: dividing the modulation parameters into common parameters and unique parameters for estimation, the common parameters include center frequency, bandwidth, amplitude and pulse width, and the unique parameters are related to the modulation type. For different modulation types, different estimation methods are set to obtain different unique parameters.

[0028] A second aspect of the present invention discloses a system for analyzing intra-pulse characteristics of aliased radar signals. The system stores computer program instructions, and by executing the computer program instructions, a method for analyzing intra-pulse characteristics of aliased radar signals is implemented. The system can operate according to default parameters, and also allows users to modify some parameters, and the operating results are displayed visually.

[0029] A third aspect of the present invention discloses an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the methods for analyzing intra-pulse characteristics of aliased radar signals according to the first aspect of the present disclosure are implemented.

[0030] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the methods for analyzing intra-pulse characteristics of aliased radar signals according to the first aspect of the present disclosure.

[0031] In summary, the solution proposed in the present invention has the following technical effects:

[0032] 1. This application proposes a method for separating mixed signals with overlapping time and frequency domains, which can automatically, accurately and quickly separate aliased signals, laying a solid foundation for the intra-pulse feature analysis of aliased signals.

[0033] 2. This application can effectively identify common signal modulation methods with high accuracy and strong adaptability.

[0034] 3. This application can quickly and accurately estimate the modulation parameters of common signals with high parameter estimation accuracy.

[0035] 4. This application is applicable to both the analysis of intra-pulse characteristics of aliased signals and the analysis of intra-pulse characteristics of non-aliased signals.

[0036] 5. This application can be run automatically with default parameters, greatly reducing the workload and professional requirements of the operator, or some key parameters can be manually modified to achieve detailed analysis of key signals.

[0037] 6. The computer software provided in this application has the characteristics of simple operation, fast running speed, and visual analysis results, which is convenient for operators to use and analyze. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 Schematic diagram of the signal processing flow of the method for analyzing intra-pulse characteristics of aliased radar signals of the present invention.

[0040] Figure 2 This is a software business flow chart of the method for analyzing intra-pulse characteristics of aliased radar signals of the present invention.

[0041] Figure 3 This is a flowchart of the signal decomposition business in the method for analyzing intra-pulse characteristics of aliased radar signals of the present invention. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0043] See also Figure 1 The method for analyzing the intra-pulse characteristics of aliased radar signals includes the following steps:

[0044] Step 1: After obtaining the aliased signal, separate the mixed signal with overlapping time and frequency domains into independent signals and store them;

[0045] Step 2: Detect the start and end time of each separated signal according to the envelope;

[0046] Step 3: Identify the modulation type of each separated signal;

[0047] Step 4: Estimate the modulation parameters of each separated signal;

[0048] Step 5: Output the results.

[0049] Specifically, step 1, separating the aliased radar signals includes the following steps:

[0050] Step 11), perform fast Fourier transform on the original I / Q channel signal to obtain the spectrum of the signal to be processed;

[0051] Step 12) finding the position where the maximum amplitude of the spectrum of the signal to be processed is located;

[0052] Step 13) Because the spectrum width of a third-order polynomial phase signal is relatively wide, and its kurtosis and skewness are close to those of a normal distribution, it is easily judged as noise. Therefore, first determine whether it is a third-order polynomial phase signal. Determine whether the position where the maximum amplitude of the spectrum of the signal to be processed is a third-order polynomial phase signal; if so, only calculate the spectrum width that should be retained near the maximum value (point k1 on the left and point k2 on the right), and proceed to step 16); if not, proceed to step 14);

[0053] Step 14), calculating the kurtosis and skewness values ​​of the position where the maximum value of the spectrum amplitude of the signal to be processed is located, and judging whether the position where the maximum value of the spectrum amplitude is located is a signal or noise based on the kurtosis and skewness values; if it is a signal, proceeding to step 15), otherwise exiting the aliasing signal separation process;

[0054] Step 15), calculating the spectrum width that should be retained near the maximum value of the spectrum amplitude (left k1 point, right k2 point or left and right k points);

[0055] Step 16) retain the spectrum of a certain bandwidth (left k1 point, right k2 point, or left and right k points) near the maximum frequency point, and set the rest as noise, pending further modulation type identification and modulation parameter estimation; and set the extracted part of the original signal as noise, and return to step 12).

[0056] Furthermore, in step 12), the process of finding the location of the spectrum maximum includes:

[0057] Only the spectrum maximum point is found each time. In the subsequent steps, if the conditions for exiting the aliased radar signal separation cannot be met, step 2 is executed at most 15 times and then automatically exits.

[0058] Furthermore, in step 13), the process of determining whether it is a third-order polynomial phase signal includes:

[0059] The modulus of the frequency domain signal is smoothed using the moving average method, and the relationship between the minimum, maximum, and median of the smoothed modulus is compared. If twice the difference between the median and the minimum is less than the difference between the maximum and the median, the instantaneous frequency of the signal is calculated and a second-order linear fit is performed on the instantaneous frequency. If the complex correlation coefficient is greater than 0.85 and the residual standard deviation is less than 3, the signal is considered to be a third-order polynomial phase signal. Otherwise, it is not considered a third-order polynomial phase signal.

[0060] Furthermore, in step 13), the process of calculating the spectrum width that should be retained near the maximum value includes:

[0061] Let med be the median of the smoothed modulus, and let th = 0.5 × med. Starting from the maximum value, move leftward in steps of 20, calculating the average of these 20 points. If the difference between the average and med is less than th for two consecutive times, stop moving and obtain the number of points to be retained on the left, k1. Use the same method to calculate the number of points to be retained on the right, k2.

[0062] Furthermore, in step 15), the process of calculating the spectrum width to be retained near the maximum value (left k1 point, right k2 point, or left and right k points) includes:

[0063] 1. First, determine whether the signal is a sinusoidal FM signal. If so, calculate the number of bandwidth points k to be retained around the maximum frequency of the sinusoidal FM signal (the number of points on both sides is the same). The judgment and calculation method is:

[0064] 1) After normalizing the signal spectrum, calculate the modulus F of the normalized spectrum;

[0065] 2) Set the maximum value ratio threshold th = max(0.22, mean(F) + 3×std(F)) and select consecutive sequence numbers greater than the threshold;

[0066] 3) Each consecutive sequence number should theoretically have the same maximum frequency point. If the number of maximum frequency points is less than or equal to 2, further determine whether it is a sinusoidal FM signal with a relatively small modulation frequency. The judgment method is as follows. Otherwise, proceed to step 4);

[0067] a. Calculate the instantaneous frequency of the signal;

[0068] b. Calculate the maximum and minimum points of the instantaneous frequency;

[0069] c. Calculate the period of the instantaneous frequency from the positions of the maximum and minimum points;

[0070] d. Divide the instantaneous frequency into segments by period and perform a sine fit on each segment. The idea of ​​sine fitting is based on the principle of least squares method, which obtains a nonlinear equation system about the parameters to be fitted. Then, the Gauss-Newton iteration method is used to solve the nonlinear equation system, and finally the fitting coefficient, fitting residual, and residual standard deviation can be obtained.

[0071] e. If all residual standard deviations are less than 0.6, it is considered to be a sinusoidal FM signal; otherwise, it is not.

[0072] 4) Eliminate the consecutive numbers in each set that have more than 5 consecutive numbers, and repeat the previous step;

[0073] 5) Determine whether the large value point is in the large value point sequence set. If not, it means it is not a sinusoidal FM signal. If so, proceed to the next step.

[0074] 6) Calculate the spacing between the large frequency points based on the maximum value in each segment, set the tolerance to 3, and find the large frequency points with spacing less than the tolerance (if the spacing is less than the tolerance, the spacing is considered equal). If the number of large frequency lines with equal spacing is greater than 20, it is considered not a sinusoidal FM signal, otherwise proceed to the next step;

[0075] 7) If the proportion of equally spaced large frequency points exceeds 80%, it is considered a sinusoidal FM signal and the points with unequal spacing are deleted; otherwise, the difference between adjacent large values ​​is removed. The judgment basis is that the difference between adjacent large values ​​is greater than 5 times the median of the spacing between the large frequency points;

[0076] 8) If the maximum frequency point is located within 1 / 3 of the maximum frequency point, it is considered not an SFM signal. Otherwise, continue the determination.

[0077] 9) Since large value points may be missed, it is necessary to consider points with multiple spacing. If the difference between the spacing and the minimum spacing or twice the minimum spacing is within the tolerance (set to 3) and the large value frequency points exceed 80%, it is judged to be a sinusoidal FM signal;

[0078] 10) If it is a sinusoidal FM signal, recalculate the self-differential of the maximum value point and shift the maximum value point boundary by 5 intervals to the left and right to check if there are any missing maximum value points. If so, add the missing maximum value points to obtain the number of points k1 and k2 to be retained. Finally, add the 0.5MHz bandwidth to the calculated k1 and k2. If it is not a sinusoidal FM signal, proceed to step 2.

[0079] 2. If it is not a sinusoidal FM signal, calculate the similarity coefficient between the signal spectrum and the rectangle and triangle. The steps to calculate the similarity coefficient are as follows:

[0080] Calculate the center frequency and effective bandwidth of the signal spectrum and normalize the bandwidth to retain only the spectrum within the effective bandwidth, thus eliminating the influence of out-of-band noise. The preprocessed frequency domain signal is {G(j), j = 1, 2, ..., N}, where N is the length of the preprocessed signal.

[0081] 1) Construct a rectangular sequence: mx is the maximum value of G(j);

[0082] 2) Calculate the similarity coefficient of G(j) and U(k). The calculation formula is as follows:

[0083]

[0084] 3) Construct a triangle sequence:

[0085]

[0086] Where mx is the maximum value of G(j);

[0087] 4) Calculate the similarity coefficient of {G(j)} and {T(k)}. The calculation formula is as follows:

[0088]

[0089] 5) When calculating the similarity coefficient, in addition to obtaining the similarity coefficients of the signal spectrum with the rectangle and triangle respectively, the number of points B greater than the set bandwidth threshold and the bandwidth ratio threshold th<1 of the large value points are also obtained;

[0090] 6) If B≤5, the number of points retained on both sides is k, and the value is 0.1N / f s The integer part of s is the sampling frequency of the signal, and N is the number of signal points. To prevent k from being too small or too large, the minimum value of k is limited to 7 and the maximum value is 11. When B=1, the minimum value of k is limited to 5.

[0091] 7) If B>5, and the similarity coefficient between the signal spectrum and the rectangle is greater than that between the signal spectrum and the triangle, it is considered to be an FM signal. The calculation method for the number of bandwidth points retained near the maximum frequency point is as follows: first fix the number of points to the left of the maximum frequency point to 10, and then select the frequency domain signal S on the right side of the maximum frequency point in 2MHz steps. f Calculate S according to the above method f The similarity coefficient with the rectangular sequence is calculated until the similarity coefficient no longer changes, and the current number of points is used as the number of points p2 to be retained on the right side of the maximum frequency point; the signal on the right side of the maximum frequency point is fixed according to p2, and the frequency domain signal S is selected in 2MHz steps on the left side of the maximum frequency point. f , calculate S f Calculate the similarity coefficient with the rectangular sequence until the similarity coefficient no longer changes. Use the current number of points as the number of points p1 to be retained to the left of the maximum frequency point. Based on p1 and p2, reduce the step size for more precise calculations. For example, using the number of points on the right as an example, gradually reduce the number of points retained in small steps of 0.5MHz starting from the rough value p2. Calculate the similarity coefficient with the rectangular sequence until the similarity coefficient changes. Use the current number of points as the number k2 to be retained to the right of the maximum frequency point. Use the same method to calculate the number k1 to be retained to the left of the maximum frequency point.

[0092] 8) If B>5, and the signal spectrum has a smaller similarity coefficient with the rectangle than with the triangle, it is considered a coded signal. Since its spectrum is symmetrical, the bandwidth reserved on both sides is the same. The number of bandwidth points reserved near the maximum frequency is calculated as follows: select the frequency domain signal S on both sides of the maximum frequency in 2MHz steps. f , calculate S f The similarity coefficient with the triangular sequence is calculated until the similarity coefficient no longer changes. The current number of points is used as the rough value p of the number of points to be retained on both sides of the maximum frequency point. Starting from p, the number of retained points is gradually reduced in small steps of 0.5MHz. The similarity coefficient with the triangular sequence is calculated until the similarity coefficient changes. The current number of points is used as the number k of points to be retained on the right side of the maximum frequency point.

[0093] Furthermore, in step 2, the process of detecting the start and end time of the separated signals according to the envelope includes:

[0094] 1) Calculate the envelope of the signal, normalize it, and smooth it;

[0095] 2) Calculate the smoothed normalized envelope and calculate its mean and standard deviation;

[0096] 3) Determine the optimal ratio of rising and falling edges when intercepting the signal. If the standard deviation is less than 0.1, the optimal ratio is the sum of the mean and the standard deviation, and proceed to step 4). If the standard deviation is greater than or equal to 0.25, the optimal ratio is 0.4, and proceed to step 4). If the standard deviation is in the range [0.1, 0.25], consider the mean. If the mean is greater than 0.7, remove the small value points. If the mean is less than 0.25, remove the large value points. If the mean is in the range [0.25, 0.7], retain all signals, calculate the start and end times based on the retained signal subscript numbers, and skip step 4.

[0097] 4) Calculate the start and end times of the signal based on the determined optimal ratio and extract the signal. Find the points in the smoothed normalized envelope that are greater than the optimal ratio. If there are multiple segments, process them segment by segment. Calculate the sequence numbers of the rising and falling edges of the envelope and divide them by the sampling frequency to obtain the start and end times of the signal.

[0098] In step 3, the method for identifying multiple intra-pulse modulation types specifically includes the following steps:

[0099] Step 31) First, determine whether it is a third-order polynomial phase signal or a sinusoidal frequency modulation signal. This step has already been done in the aliasing signal separation and detection. If so, re-enter step 31) for the next separated signal; otherwise, proceed to step 32) for the current signal.

[0100] Step 32), using the second-order fourth-order moment method to estimate the signal-to-noise ratio;

[0101] Step 33), short-time filtering is performed on the signal;

[0102] Step 34), calculating the instantaneous phase of the signal after deconvolution;

[0103] Step 35), using 20-fold phase difference method to calculate the instantaneous frequency of the signal;

[0104] Step 36), perform a first-order linear fit on the time-frequency curve to calculate the regression coefficient B, residual r and complex correlation coefficient R;

[0105] Step 37) determines whether it is a normal signal, if so, proceeds to step 315), otherwise proceeds to step 38);

[0106] Step 38), determine whether it is a linear frequency modulation signal, if so, go to step 315), otherwise go to step 39);

[0107] Step 39), determine whether it is a dual linear frequency modulation signal, if so, proceed to step 315), otherwise proceed to step 310);

[0108] Step 310), detect the frequency jump point, find the sequence number of the jump point, if the number of jump points is less than or equal to 3, it is considered to be an unknown modulation type signal, and proceed to step 315), otherwise proceed to step 311);

[0109] Step 311), calculate the kurtosis of the instantaneous frequency, if the kurtosis is greater than 4, proceed to step 312), otherwise proceed to step 313);

[0110] Step 312), if the complex correlation coefficient is greater than 0.45, it is determined to be a linear frequency modulation-two-phase coding composite modulation signal, otherwise it is determined to be a phase coding signal. The specific coding type needs to be further determined. After the determination is completed, step 315 is entered;

[0111] Step 313), remove the jump points detected in step 10, and re-detect the jump points. If the number of jump points detected for the second time exceeds 60% of the number of jump points detected for the first time, it is determined to be a frequency coded signal and the process proceeds to step 315), otherwise, the process proceeds to step 314);

[0112] Step 314), perform a first-order linear fit on the instantaneous frequency. If the complex correlation coefficient is greater than 0.85 and the residual standard deviation is less than 6, it is determined to be a linear frequency modulation-two-phase coding composite modulation signal; otherwise, it is determined to be a frequency coding-two-phase coding composite modulation signal;

[0113] Step 315) Check whether all separated signals have completed modulation type identification. If so, proceed to step 316). Otherwise, proceed to step 31) for signals that have not completed modulation type identification.

[0114] Step 316) When separating the aliased signals, the frequency coded signal may be split into a regular signal, or the frequency coded-two-phase coded composite modulated signal may be split into multiple two-phase coded signals. Therefore, it is necessary to judge and process these two situations, and after the processing is completed, enter the modulation parameter estimation.

[0115] Furthermore, in step 37), the process of determining whether it is a normal signal includes:

[0116] 1) If the standard deviation of the residual r is greater than 5, it is considered not a regular signal, otherwise further judgment is made;

[0117] 2) If the standard deviation of the residual r is less than 0.8 and R is less than 0.5 at the same time, it is considered a normal signal, otherwise further judgment is made;

[0118] 3) Use 3dB bandwidth to estimate the center frequency of the signal. If the center frequency is less than 0.05, it is considered that there is no modulation in the pulse. Otherwise, calculate the normalized spectrum of the signal and find the maximum value point of the spectrum. If there is only one maximum value point, it is considered to be a normal signal. Considering the influence of noise, if there is more than one maximum value point, further determine the number and continuity of the maximum value points and set the threshold L represents the signal length, f s Indicates the sampling frequency, round indicates rounding to the nearest integer, and is limited to 3≤p≤5. If the number of large-value points is less than p and the number spacing of large-value points is greater than 3, the proportion of the total number of large-value points is less than 10%, then it is considered a normal signal.

[0119] Furthermore, in step 38), the process of determining whether it is a linear frequency modulation signal includes:

[0120] 1) If the standard deviation of the residual r is greater than 6 or the complex correlation coefficient R is less than 0.2, it is considered not a linear frequency modulation signal, otherwise further judgment is made;

[0121] 2) Estimate the signal center frequency using 3dB bandwidth;

[0122] 3) Estimate the instantaneous phase of the signal after removing the carrier frequency;

[0123] 4) Perform a second-order linear fit on the phase and calculate the regression coefficient B2, residual r2 and complex correlation coefficient R2;

[0124] 5) If the standard deviation of r is less than 2.5, the standard deviation of r2 is less than 0.45, and R2 is greater than 0.95, it is considered to be a linear frequency modulation signal;

[0125] 6) If R2 is less than 0.5 or the standard deviation of r and r2 is greater than 5, it is determined that it is not a linear frequency modulation signal and further judgment is required;

[0126] 7) If the above two conditions are not met, further differentiation is performed based on the spectrum. Calculate the amplitude of the signal spectrum, take the frequency point with a maximum amplitude exceeding half of the maximum amplitude as the maximum value, and calculate the difference δ between the frequency points with the maximum value f , if R is greater than 0.45 and the maximum value point spectrum width is greater than 3, all δ f If it is less than 0.3, it is considered to be a linear frequency modulation signal; if R is greater than 0.7 and R2 is greater than 0.95, the number of points whose frequency difference exceeds the threshold is calculated (when the bandwidth of the large-value frequency point exceeding the threshold is greater than 20, the threshold is 1.5, and in other cases it is 1). If the number of large-value frequency points is greater than 3 and the proportion of the points whose frequency difference exceeds the threshold is less than 10%, it is considered to be a linear frequency modulation signal, otherwise further judgment is made.

[0127] Furthermore, in step 39), the process of determining whether it is a dual-linear frequency modulation signal includes:

[0128] 1) Find the peak point of the instantaneous frequency data and divide the data into two segments based on the peak point;

[0129] 2) Perform first-order linear fitting on the two segments of data respectively;

[0130] 3) If the correlation coefficient of the two segments is less than 0.5, it is determined not to be a DLFM signal and further evaluation is required;

[0131] 4) If the first-order coefficients (slopes) at both ends are opposite to each other and any correlation coefficient is greater than 0.85, it is determined to be a DLFM signal. Otherwise, further judgment is required.

[0132] Furthermore, in step 310), the method for detecting the frequency hopping point includes:

[0133] 1) The instantaneous frequency phiN is subtracted every 40 points, that is, δ = phiN i+40 -phiN i ;

[0134] 2) Calculate the jump point detection threshold pjump according to the following formula;

[0135]

[0136] Among them, f s is the sampling frequency, N = 20 is the multiplicity of phase difference, p = 0.45;

[0137] 3) If the absolute value of δ is greater than pjump, it is determined to be a jump point.

[0138] Furthermore, in step 312), the process of further determining the specific coding type includes:

[0139] 1) After calculating the fourth power of the time domain signal, perform a fast Fourier transform, calculate the square of the frequency domain amplitude, and move the zero frequency point to the center of the frequency domain to obtain x n ;

[0140] 2) Set the threshold mx = 0.5 × max(x n ), find the point that is greater than the threshold, which is called the large value point;

[0141] 3) If there are more than one large value point and the sequence numbers of the large value points are discontinuous, it is determined to be a polyphase coded signal, otherwise further determination is performed;

[0142] 4) Use 3dB bandwidth for coarse carrier frequency estimation;

[0143] 5) Estimate the signal symbol period based on multi-scale Haar wavelet transform;

[0144] 6) If the symbol period is greater than 1 μs, extract data from one symbol period and estimate the carrier frequency again;

[0145] 7) After converting the signal to baseband, perform autocorrelation operation to obtain its imaginary part and normalize it;

[0146] 8) If the imaginary part of the autocorrelation is all 0, it is determined to be a two-phase coded signal, otherwise further judgment is required;

[0147] 9) Perform a fast Fourier transform on the square of the time-domain signal, calculate the amplitude of the spectrum, normalize it, and find the maximum value. Denote the spectrum 5 MHz before and after the maximum value by xfm. Given a coefficient th, calculated as the mean of xfm plus twice its standard deviation plus 0.1, with th limited to 0.5, use the maximum value of the spectrum multiplied by the coefficient th as the threshold. Find points in the spectrum that are greater than the threshold. If there are fewer than three points, identify the signal as a two-phase coded signal; otherwise, identify the signal as a four-phase coded signal.

[0148] Furthermore, in step 316), the process of determining and processing whether the frequency coded signal or the frequency coded-two-phase coded composite modulated signal is split includes:

[0149] 1. Frequency coded signals are split into regular signals

[0150] 1) Record the start and end times and time domain signals of all signals identified as regular signals;

[0151] 2) Find all possible combinations of frequency-coded signal segments, based on the principle that each segment of the frequency-coded signal does not overlap and the time interval between two adjacent segments is within 0.15 μs;

[0152] 3) Splice together the remaining regular signal fragments that may have been split apart by frequency coding, calculate the signal-to-noise ratio according to the method described above, and identify the modulation type. If the identification result is a frequency coded signal, store the identification result.

[0153] 2. Frequency coding - two-phase coding composite modulation signal is split into two-phase coding signal

[0154] 1) Record the start and end times and time domain signals of all identified two-phase coded signals;

[0155] 2) Based on the principle that each segment of the signal does not overlap in time and the time interval between two adjacent segments is within 0.15us, find all possible segment combinations of the frequency coding-two-phase coding composite modulation signal;

[0156] 3) Analyze each segment combination separately, splice these segments together, calculate the signal-to-noise ratio according to the method described above, identify the modulation type, and if the identification result is a frequency coding-two-phase coding composite modulation signal, store the identification result.

[0157] In step 4, the method for estimating the modulation parameters of the signal whose modulation type is identified specifically includes the following steps:

[0158] Step 41), public parameter estimation. Time domain signal s n Perform a fast Fourier transform to the frequency domain and calculate the square of the frequency domain amplitude, recorded as x n , find x n Greater than 0.5×max(x n ) is located at the position where the value is located, recorded as ff, the signal length is recorded as L, and the sampling frequency is recorded as f s ,make The center frequency f is calculated by the following formula center , bandwidth B, amplitude A and pulse width T.

[0159]

[0160] A=10log 10 max(abs(s n ))

[0161]

[0162] Step 42), unique parameter estimation.

[0163] The specific estimation method varies according to the signal type and the estimation method. The specific settings are as follows:

[0164] 1. Conventional signal.

[0165] The parameters that need to be estimated for conventional signals include carrier frequency f c , initial phase For f c The main method for estimating f is the modified frequency interpolation algorithm (M-Rife algorithm), which works as follows: after performing discrete Fourier transform on the signal, the maximum spectrum line of the signal spectrum is used to interpolate with the larger spectrum line of the nearest neighbor to obtain an accurate estimate of the true frequency. c Then, the average value of the phase sequence after removing the carrier frequency can be obtained

[0166] 2. Linear frequency modulation signal.

[0167] The parameters that need to be estimated for the linear frequency modulation signal are the starting frequency f0, the frequency modulation slope k and the initial phase The estimation process includes:

[0168] 1) The linear frequency modulation signal sequence can be expressed as:

[0169] s (n)=A(n)exp{j(2πf0nΔt+πkn 2 Δt 2 +φ0)}

[0170] Where A(n) is the signal amplitude, Δt is the sampling interval, k, f0 and φ0 are the frequency modulation slope, starting frequency and initial phase of the signal respectively, and the center frequency is defined as f M =f0+kNΔt / 2, then the signal sequence can be rewritten as

[0171]

[0172] where φ1=φ0+πf M NΔt-πkN 2 Δt 2 / 4.

[0173] 2) Delay correlation of the linear frequency modulation signal to obtain a sine wave sequence:

[0174] y(n)=s(n+N / 2)·s * (n) = A(n) 2 exp{j(πNknΔt 2 +φ2)},n=0,…,N / 2

[0175] where φ2=πf M NΔt-πkN 2 Δt 2 / 4. y(n) is a conventional carrier frequency kNΔt / 2

[0176] Signal

[0177] Sequence, use M-Rife algorithm to perform conventional signal frequency estimation on the sequence and obtain the estimated frequency f k , and calculate the estimated value of the frequency modulation slope k = f k / (NΔt / 2);

[0178] 3) Construct the de-FM slope sequence:

[0179] z(n)=exp{-j(πkΔt 2 (nN / 2) 2 )}

[0180] Will s (n) Multiplying by z(n) yields a frequency modulation with a small slope and a center frequency f M Since the frequency modulation slope is very small, the sequence can be regarded as a linear frequency modulation sequence with a carrier frequency of f M Therefore, we can use the M-Rife algorithm to estimate the frequency of this sequence to get f MThe estimated value of the starting frequency f0=f M -kNΔt / 2;

[0181] 4) Calculate the phase after removing the carrier frequency and make a second-order linear fit to the phase. The constant term of the fitting coefficient is the initial phase

[0182] 3. Dual linear frequency modulation signal.

[0183] The parameters to be estimated for the dual-linear frequency modulation signal are the carrier frequency f c , frequency modulation slope k, initial phase and the FM midpoint f m It is only necessary to take the part of the signal before the frequency peak point and use the estimation method of linear frequency modulation signal to get the carrier frequency f c , frequency modulation slope k and initial phase The FM midpoint can be calculated as follows:

[0184]

[0185] Where L is the number of signal points, f s is the sampling frequency, and round() means rounding to the nearest integer.

[0186] 4. Sinusoidal frequency modulated signal.

[0187] The parameters to be estimated for the sinusoidal FM signal are the carrier frequency f c , modulation coefficient m f , modulation frequency f m and initial phase

[0188] First, use fast Fourier transform to obtain the signal spectrum, find all frequency points that are greater than 0.9 times the maximum value, and take the average value as the signal carrier frequency f c After removing the carrier frequency of the signal, the instantaneous frequency is obtained through 20 phase differences. The frequency corresponding to the maximum value of the instantaneous frequency spectrum and less than half of the sampling frequency is the modulation frequency f m ; Modulation coefficient m f The maximum amplitude of the instantaneous frequency spectrum is twice that of f m After the phase is removed from the carrier frequency, a first-order linear fit is performed, and the constant term of the fitting coefficient is the initial phase

[0189] 5. Third-order polynomial phase signal.

[0190] The parameters to be estimated for the third-order polynomial signal are the frequency modulation coefficient a i ,i=1,2,3, carrier frequency f0 and initial phase

[0191] The third-order delay difference is given below:

[0192] DP1[s(n),τ]=s(n)

[0193] DP2[s(n),τ]=s(n)×s(n-τ)

[0194] DP3[s(n),τ]=s(n)×[s(n-τ)] 2 s(n-2τ)

[0195] Where τ is the delay length, 0≤τ≤N-1.

[0196] m-order discrete polynomial phase transform DPT m Defined as DP m The discrete Fourier transform of [s(n),τ], the parameter estimation process includes:

[0197] 1) Initialization: Set m = 3;

[0198] 2) Select the delay τ m =N / m, estimate a m ;

[0199]

[0200] 3)s (m-1) (n) = s (m) (n)exp{-ja m (nΔt) m};

[0201] 4) m is decremented by 1, i.e. m = m - 1. If m ≥ 1, return to step 2), otherwise go to step 5);

[0202] 5) After the signal autocorrelation operation, the frequency corresponding to the first maximum spectrum that appears is the carrier frequency f0;

[0203] 6) Calculation The phase of

[0204] 6. Phase-coded signal

[0205] The modulation parameters to be estimated for the two-phase coded and four-phase coded signals are carrier frequency f0, symbol period T b , coding sequence {α j}, initial phase The main modulation parameters of the polyphase coded signal are carrier frequency f0, symbol period T b , the number of code elements N, the type of polyphase coding type, the process of parameter estimation of the phase coded signal is:

[0206] 1) First, the spectrum distribution of the radar signal is obtained by Fourier transform, the spectrum is smoothed, and the center of gravity of the power spectrum within the 3dB bandwidth is used as a rough estimate of the carrier frequency;

[0207] 2) Based on the estimated value of carrier frequency f0 obtained in the previous step, the original signal is de-carriered and the symbol period of the phase-coded signal is estimated using Haar wavelet transform. If the signal is a two-phase coded or four-phase coded signal, execute steps 3) and 4). If the signal is a polyphase coded signal, execute steps 5) and 6). The process of estimating the phase symbol period using Haar wavelet transform is as follows:

[0208] a. Remove the carrier frequency of the signal based on the estimated value of the carrier frequency and down-convert it to baseband;

[0209] b. Estimate the 3dB bandwidth of the baseband signal and select three scaling factors a1 = 0.4 / B ω ,a2=

[0210] 0.6 / B ω ,a3=0.75 / B ω ;

[0211] c. Calculate the square of the Haar wavelet transform amplitude of the baseband signal under the three scaling factors respectively, and superimpose the three results;

[0212] d. Perform discrete Fourier transform on the superimposed result, filter out the DC component, and estimate the frequency f corresponding to the peak spectrum line b ;

[0213] e. Obtain the estimated symbol period T of the phase-coded signal b =1 / f b :

[0214] 3) Restore the phase of the baseband modulated signal. The first phase value is the initial phase.

[0215] 4) Estimate the coding sequence {α based on the method of instantaneous autocorrelation polarity judgment j};

[0216] 5) Number of code elements round(·) means rounding to the nearest integer;

[0217] 6) Use the maximum signal detection method to identify the type of polyphase coding. The identification process is as follows:

[0218] a. When the number of code elements N is not a perfect square number, the signal can only be P3 code or P4 code; when

[0219] When it is an odd number, the signal is one of the Frank code, P1 code, P3 code or P4 code;

[0220] when When it is an even number, the signal may be any of the polyphase codes;

[0221] b. Construct a corresponding local polyphase code based on the number of code elements N, and conjugate-multiply the received complex signal by the constructed polyphase code;

[0222] c. Convert the product to the frequency domain and move the zero frequency point to the center;

[0223] d. Take 0.5 times the maximum value of the spectrum as the threshold and find the number of the point where the spectrum is greater than the threshold.

[0224] The difference between the last and the first in the set is denoted as Δ;

[0225] e. The local polyphase code corresponding to the smallest Δ is the polyphase code type;

[0226] f. If there is more than one minimum value, transpose the received complex signal and multiply it with the constructed polyphase code signal matrix, and calculate the absolute value. The polyphase code signal type corresponding to the maximum absolute value is the polyphase code type detection result.

[0227] 7. Frequency Coded Signal

[0228] The parameters to be estimated for the frequency coded signal include the symbol period T c , subcode frequency f c and code width w, the process of parameter estimation for frequency coded signal is:

[0229] 1) Perform 20 phase differences on the complex signal to obtain the instantaneous frequency pf, and then perform self-differentiation of pf in steps of 40 to obtain Δpf;

[0230] 2) Detect the jump point of Δpf according to the 3σ criterion;

[0231] 3) Divide the complex signal into segments according to the location of the hopping point and estimate the carrier frequency of each segment using a 3dB bandwidth;

[0232] 4) Use STFT transform (short-time Fourier transform) combined with wavelet transform to estimate the symbol period of the frequency-coded signal. The specific process is as follows:

[0233] a. Perform an STFT transform on the signal, setting the parameters to a window width of 256, the number of overlapping samples in each segment to 87.5% of the window width, rounded to the nearest integer, and a Fourier transform length of 512. The STFT return values ​​include the frequency vector F, the time point T at which the spectrum was calculated, and the energy spectral density P.

[0234] b. Find the center frequency of each sliding window. Assume that the maximum value of each energy spectrum density is located at t m, then the center frequency f in each sliding window center The calculation method is:

[0235] c. Set the scaling factor to 8, perform discrete wavelet transform on the center frequency to obtain a sequence, and square the amplitude to obtain the CWT; the transformation formula is (the center frequency is abbreviated as follows):

[0236] y i =y i-1 +2f i+3 -f i-1 -f i+7 i≠1.

[0237] d. Set the detection threshold to 0.01263f s , for the points corresponding to the CWT detection jump amplitude exceeding the threshold, these points are differentiated, and the points between the middle value of the first continuous point segment and the middle value of the last continuous point segment are selected. The CWT values ​​corresponding to these points are subjected to Hilbert transform and then fast Fourier transform;

[0238] e. After removing the DC component, find the maximum peak point of the fast Fourier transform result and take the inverse to obtain the symbol period T c .

[0239] 5) The jump point number is self-differentiated and divided by the sampling frequency f s Get the width of each frequency hopping sequence and set it to less than 0.3T c The frequency value of is removed;

[0240] 6) Detect two adjacent frequencies with a difference less than 0.5. If there is any, merge them. Their arithmetic average is used as the new frequency after the merger, and the new code width w is obtained;

[0241] 7) Modify the symbol period according to the coding width, Where n1 is the number of frequency points, which is calculated by rounding the quotient of the code width w divided by the minimum code width and then summing them up;

[0242] 8) Extend the deleted and merged frequency values ​​to all frequency points, thus obtaining an estimated value of each subcode frequency.

[0243] 8. Frequency coding-two-phase coding composite modulation signal

[0244] The parameters to be estimated for the frequency coding-two-phase coding composite modulation signal include the frequency coding symbol period T f , the number of frequency codes N f , coding frequency f f , phase encoding symbol period T p , the number of phase-encoded symbols Np , phase coding sequence {α j}、Initial phase The parameter estimation process is as follows:

[0245] 1) Square the signal to remove phase-coded modulation;

[0246] 2) Use the STFT transform combined with wavelet transform introduced above to estimate the symbol period T of the frequency coded signal f , the number of frequency codes L is the signal length, f s is the sampling frequency, rround(·) means rounding to the nearest integer;

[0247] 3) Divide the original signal into N equal parts f Segment, use 3dB bandwidth to estimate the carrier frequency of each segment to get the coding frequency f f estimated value of;

[0248] 4) Divide the original signal into N equal parts f Segment, use the parameter recognition method of the two-phase coded signal introduced above to identify each segment of the signal;

[0249] 5) The symbol period value that appears the most times is considered to be the phase coding symbol period T p estimated value of;

[0250] 6) For all code element period values, the value is T p Calculate the length of the corresponding coding sequence and select the length with the largest number of occurrences as the number of phase coding symbols N p The estimated value of N p The first one of the corresponding coding sequences is used as the phase coding sequence {α j The estimated value of}, the average value of the initial phase estimated for each signal segment is the final initial phase estimated value.

[0251] 9. Linear frequency modulation-two-phase coded composite modulation signal

[0252] The parameters to be estimated for linear frequency modulation-two-phase coding include the carrier frequency f c , frequency modulation slope k0, phase encoding code element period T p , the number of phase-encoded symbols N p , phase coding sequence {α j}、Initial phase The parameter estimation process is as follows:

[0253] 1) Square the signal and divide the result obtained by the method of identifying linear frequency modulation signals by 2 to obtain the carrier frequency f c , FM slope k0;

[0254] 2) Signal de-frequency modulation, using the method of estimating the parameters of the two-phase coded signal to estimate the symbol period and initial phase of the phase coded signal, that is, the phase coded symbol period T p and initial phase estimated value of;

[0255] 3) Use the method introduced above to estimate the code for each sampling point of the two-phase coded signal. Represent consecutive 1s in the code as 1 1 and consecutive 0s as 1 0, recorded as vector v, and calculate the number of consecutive 1s or 0s in each segment, expressed as l.

[0256] Calculating vectors where f s Indicates the sampling frequency, round(·) indicates rounding to the nearest integer, and ceil(·) indicates rounding to the nearest integer;

[0257] 4) Expand the 1 or 0 in v by the corresponding number in num to obtain the final phase coding sequence {α j} is an estimated value.

[0258] In addition, the present application also provides an aliased radar signal intra-pulse feature analysis system, wherein the software system service implementation includes the following steps:

[0259] Step 1: Select software system technology. Since QT is a secondary development platform based on C++, it has an excellent encapsulation mechanism, which makes QT highly modular and reusable. In addition, QT provides a safe type called signals / slots to replace callbacks, which makes it very simple for various components to work together. Therefore, we use the most common cross-platform, open source programming tool QT, which supports software running on the two most popular platforms, Windows and Linux, covering more than 90% of users. For the intra-pulse signal graphic display, we use QWT, an open source graphic display technology based on QT, which has good execution efficiency and can provide millisecond-level display response.

[0260] Step 2: Software system architecture design: This application follows the design principles of front-end and back-end separation and modularization in its architecture design.

[0261] The concept of front-end and back-end separation stems from decoupling, which reduces code complexity and facilitates development, management, and maintenance. This has been a proven success in software development over the years. This software also adopts this concept, effectively isolating the display and manipulation of graphics and data from the back-end algorithm analysis, connecting them with a stable set of call interfaces.

[0262] (1) The backend algorithm is encapsulated into a dynamic link library, providing a stable calling interface according to the design specifications;

[0263] (2) The front end uses the view concept in QT and forms a display interface that is decoupled from data based on graphics classes such as Qmainwindow and Qwidget;

[0264] (3) The front end calls the back end algorithm through a pre-agreed interface to obtain display data. The display effect is driven by the display data, but the business logic of the interface will not be affected by the calculation results.

[0265] In the design, this application decomposes the software front-end into a main window module that controls the entire window, a graphic display module that displays graphics, a tree diagram module that displays parameters, and a file list module that displays files. This basically ensures good cohesion and reduces coupling to a level that is easy to maintain.

[0266] Step 3: Implement software system functions.

[0267] See also Figure 2 This is a software flow chart for the method for analyzing intra-pulse characteristics of aliased radar signals according to the present invention. Position 0 represents the intermediate frequency data file to be analyzed, collected by a field surveillance device. Position 1 represents the required configuration parameters. Because different acquisition devices may have different parameter characteristics, this configuration table is provided for users to modify based on their actual needs.

[0268] The software opens the data file at execution location 0 and reads the configuration file at execution location 1. After both are successfully acquired, the signal separation module at execution location 2 decomposes the signal to form a signal data list. After the signal separation is completed, the data enters execution location 3 for modulation type identification.

[0269] The recognition algorithm identifies the signal list one by one. The recognition result is one of the following lists:

[0270] 0 Unknown

[0271] 5 No modulation in pulse

[0272] 12 Judged as normal (NS) signal

[0273] 1. Determine that it is a linear frequency modulation (LFM) signal

[0274] 2. Determine if it is a polynomial phase (PPS) signal

[0275] 3. Determine the signal to be a binary phase coded (BPSK) signal

[0276] 4. Determine the signal to be a Quaternary Phase Keying (QPSK) signal

[0277] 13 Determined to be a multi-phase coded (MPSK) signal

[0278] 9 Identified as frequency coded (FSK) signal

[0279] 11 Determined to be dual linear frequency modulation (DLFM)

[0280] 21 Determined to be a sinusoidal frequency modulation (SFM) signal

[0281] 81 is determined to be FSK_BPSK composite modulation signal

[0282] 82 Determined to be LFM_BPSK composite modulation signal

[0283] After the signal modulation type is identified, the modulation parameter estimation is performed at execution position 4. The parameters are estimated according to the identified type. Different modulation types have different parameters. The modulation type and parameter comparison table is listed as follows:

[0284] NS: contains carrier frequency f0 (floating point number), initial phase phi0 (floating point number)

[0285] LFM: includes carrier frequency f0 (floating point number), initial phase phi0 (floating point number), frequency modulation slope k0 (floating point number)

[0286] DLFM: includes carrier frequency f0 (floating point number), initial phase phi0 (floating point number), frequency modulation slope k0 (floating point number), frequency modulation midpoint fM (floating point number)

[0287] BPSK: contains carrier frequency f0 (floating point number), initial phase phi0 (floating point number), symbol period Tc (floating point number), and coding sequence cn (integer, row vector)

[0288] QPSK: includes carrier frequency f0 (floating point number), initial phase phi0 (floating point number), symbol period Tc (floating point number), and coding sequence cn (integer, row vector))

[0289] FSK: contains code element period Tc (floating point number), coding frequency fvalue (floating point number, row vector), coding width width (floating point number, row vector)

[0290] PPS: contains carrier frequency fc (floating point number), initial phase phi0 (floating point number), frequency modulation coefficient k (floating point number, column vector), signal amplitude estimate b (floating point number))

[0291] After parameter estimation is complete, the coordinates are converted into a set of horizontal and vertical coordinates for the signal waveform, spectrum, time-frequency, and phase diagrams. This set of coordinates is returned to the graphical display interface. The process then proceeds to execution position 5, where the graphical display module is executed.

[0292] The chart display module is the interface part of this software. In the present invention, according to the modularization principle, the chart display module is decomposed into 4 modules.

[0293] (1) Main window submodule: used to form the main window body and provide a container for other modules to run.

[0294] (2) File list submodule: used to display the data list for analysis. Click the "Add IF File" button to open the file selection window for users to add files to be analyzed from the disk storage; click the "Clear" button to clear all files in the list.

[0295] (3) Graphical display submodule: used to display the distribution of signal data on time and frequency coordinates. This includes waveform graphs (with arrival time as the horizontal axis and amplitude as the vertical axis), spectrum graphs (with frequency as the horizontal axis and amplitude as the vertical axis), time-frequency graphs (with arrival time as the horizontal axis and frequency as the vertical axis), and phase graphs (with arrival time as the horizontal axis and phase value as the vertical axis). Graphics also provide operations such as zooming, local data selection, screenshots, and resetting.

[0296] (4) Parameter Display Submodule: Displays the parameters of the IF signal in the form of an attribute table. Different display lists are available depending on the modulation type. This software divides the common parameters into one column, including center frequency, original signal-to-noise ratio, signal-to-noise ratio, amplitude, pulse width, bandwidth, etc.; and dynamically displays the parameters specific to each modulation type in another column.

[0297] The signal separation process is described in Figure 3 After the signal separation module reads the intermediate frequency data and configuration parameters, it starts the analysis step. Figure 3 In step 202, the signal frequency starting point is obtained from the intermediate frequency data, and then the signal ending point is obtained along this data segment. Next, step 203 is executed to calculate the signal length based on the start and end times. If the duration is within 0.5us, this segment of data is discarded and not analyzed. If it is greater than 0.5us, step 204 is executed to store the data in a table for signal modulation type identification.

[0298] This example also provides a method for analyzing the intra-pulse characteristics of an aliased radar signal to analyze simulation data. The method includes the following steps:

[0299] Step 1: Simulate and generate a linear frequency modulation signal with a pulse width of 10us and a conventional signal with a pulse width of 5us. Use a 2us delayed mixed signal as the input signal with a signal-to-noise ratio of 5dB.

[0300] The modulation parameters of the linear frequency modulation signal are the starting frequency f0 = 400 MHZ, the frequency modulation slope k = 5 MHZ / us and the initial phase The modulation parameter of the conventional signal is the carrier frequency f c =350MHZ, initial phase

[0301] Step 2: performing aliasing separation on the input signal according to the method described in this application to obtain two signals;

[0302] Step 3: Detect the start and end time of the two separated signals respectively, and obtain the start and end time of the first signal is 0.03us-9.85us, and the start and end time of the second signal is 2.00us-7.09us;

[0303] Step 4: Identify the modulation types of the two signals according to the method described in this application. The identification result is that the first signal is a linear frequency modulation signal and the second signal is a conventional signal.

[0304] Step 5: Estimation of modulation parameters of the two signals according to the method described in this application is performed. The estimation result is a linear frequency modulation signal: starting frequency f0 = 400.159 MHZ, frequency modulation slope k = 4.999 MHZ / us and initial phase Conventional signal: carrier frequency f c =350.003MHZ, initial phase

[0305] This example processes an aliased signal and runs with default parameters in 0.275 seconds on a laptop computer running an AMD Ryzen 7 4700U processor. The aliased signal is correctly separated, the modulation type is correctly identified, the start and end time detection is highly accurate, and the parameter estimation results are highly precise.

[0306] Another example gives a pulse feature analysis technique for mixed radar signals. Figure 1 , analyzing the simulation data, the aliased radar signal pulse feature analysis method includes the following steps:

[0307] Step 1: Simulate and generate a conventional signal with a pulse width of 3us and a two-phase coded signal with a pulse width of 6us. Use a 3us delayed mixed signal as the input signal, and the signal-to-noise ratio is 0dB.

[0308] The modulation parameter of the conventional signal is f c =300MHZ, initial phase The modulation parameters of the two-phase coded signal are carrier frequency f0 = 300 MHZ, symbol period T b =1us, coding sequence {1,0,1,0,1,1}, initial phase

[0309] Step 2: performing aliasing separation on the input signal according to the method described in this application to obtain two signals;

[0310] Step 3: Detect the start and end time of the two separated signals respectively, and obtain the start and end time of the first signal as 0.00us-3.05us, and the start and end time of the second signal as 3.91us-9.97us;

[0311] Step 4: Identify the modulation types of the two signals according to the method described in this application. The identification result is that the first signal is a regular signal and the second signal is a two-phase coded signal.

[0312] Step 5: Estimating the modulation parameters of the two signals according to the method described in this application. The estimation result is a conventional signal: f c =299.992HZ, initial phase The modulation parameters of the two-phase coded signal are carrier frequency f0 = 300.047 MHz, symbol period T b =1.010us, coding sequence {1,0,1,0,1,1}, initial phase

[0313] This example processes an aliased signal with overlapping spectra but non-overlapping time. Using default parameters, the run time on a home laptop with an AMD Ryzen 74700U processor is 0.164 seconds. The aliased signal is correctly separated, the modulation type is correctly identified, the start and end time detection is highly accurate, and the parameter estimation results are highly precise.

[0314] This example presents an intermediate frequency signal analysis, implements the design flow stated in this application, and completes the analysis of polyphase coded data. The signal carrier frequency is 691 MHz, and the symbol period is 11.5 μs. The recognition result is consistent with the visualized signal characteristics in the figure.

Claims

1. A method for analyzing intra-pulse characteristics of an aliased radar signal, characterized in that: The aliased radar signal intra-pulse feature analysis method comprises the following steps: Step 1: Separate the mixed signal with overlapping time and frequency domains into independent signals and store them; Step 2: Detecting the start and end time of each separated signal based on the envelope of each separated independent signal; Step 3: Identify the modulation type of each of the separated signals; Step 4: estimating the modulation parameters of each of the separated signals based on the obtained modulation type of each of the signals; In step 1, the process of separating the mixed signal overlapping in the time and frequency domains includes: Step 11), perform fast Fourier transform on the original I / Q channel signal to obtain the spectrum of the signal to be processed; Step 12) finding the position where the maximum amplitude of the spectrum of the signal to be processed is located; Step 13), determine whether the position where the maximum amplitude of the spectrum of the signal to be processed is located is a third-order polynomial phase signal; if so, calculate the spectrum width that should be retained near the maximum amplitude, and proceed to step 16); if not, proceed to step 14); Step 14), calculating the kurtosis and skewness values ​​of the position where the maximum amplitude of the spectrum of the signal to be processed is located, and judging whether the position where the maximum amplitude of the spectrum is located is a signal or noise based on the kurtosis and skewness values; if it is a signal, proceeding to step 15), otherwise exiting the aliasing signal separation process; Step 15), calculating the spectrum width that should be retained near the maximum value of the spectrum amplitude; Step 16), retain the signal of the spectrum width and store it, set the other parts of the spectrum of the signal to be processed as noise, and return to step 12).

2. The method for analyzing intra-pulse characteristics of aliased radar signals according to claim 1, wherein: In step 15), calculating the spectrum width that should be retained near the maximum spectrum amplitude specifically includes: Determine whether the signal with the maximum spectrum amplitude is a sinusoidal frequency modulation signal. If so, calculate the number of bandwidth points k to be retained around the maximum frequency point of the sinusoidal frequency modulation signal, set the number of points on the left and right sides to be the same, and determine the spectrum width to be retained. If not, calculate the similarity coefficient between the signal spectrum and the rectangular spectrum and the triangular spectrum, and determine the spectrum width to be retained based on the obtained similarity coefficient between the signal spectrum and the rectangular spectrum and the triangular spectrum.

3. The method for analyzing intra-pulse characteristics of aliased radar signals according to claim 1, wherein: In step 2, detecting the start and end time of each separated signal according to the envelope of each separated independent signal includes: Step 21: normalize and smooth the envelope of each signal; Step 22: Calculate the smoothed normalized envelope and calculate its mean and standard deviation; Step 23: Determine the optimal ratio between the rising edge and the falling edge when intercepting the signal in the following manner: if the standard deviation is less than 0.1, the optimal ratio is the sum of the mean and the standard deviation, and proceed to step 24; if the standard deviation is greater than or equal to 0.25, the optimal ratio is 0.4, and proceed to step 24; if the standard deviation is in the interval [0.1, 0.25], compare the mean values. If the mean is greater than 0.7, remove the signal at the small value point for calculation; if the mean is less than 0.25, remove the signal at the large value point for calculation; if the mean is in the interval [0.25, 0.7], retain all signals. Calculate the start and end times of the signal based on the reserved signal subscript number and skip step 24; Step 24: Calculate the start and end times of the signal based on the determined optimal ratio, determine the points in the smoothed normalized signal envelope that are greater than the optimal ratio, calculate the subscript numbers of the rising and falling edges of the signal envelope, and calculate the start and end times of the signal based on the subscript numbers.

4. The method for analyzing intra-pulse characteristics of aliased radar signals according to claim 1, wherein: In step three, the modulation type of each separated signal is identified by: estimating the signal-to-noise ratio using the second-order fourth-order moment method; performing short-time filtering on the signal; calculating the instantaneous phase of the signal after deconvolution; calculating the instantaneous frequency of the signal using the 20-fold phase difference method; and performing a first-order linear fit on the time-frequency curve to determine the modulation type of each signal.

5. The method for analyzing intra-pulse characteristics of aliased radar signals according to claim 1, wherein: In step 4, the estimation of the modulation parameters of each separated signal includes: dividing the modulation parameters into common parameters and unique parameters for estimation, the common parameters include center frequency, bandwidth, amplitude and pulse width, and the unique parameters are related to the modulation type. Different modulation types require different estimation methods to obtain different unique parameters.

6. A system for analyzing intra-pulse characteristics of aliased radar signals, characterized in that: The system stores computer program instructions, and by executing the computer program instructions, implements the method for analyzing intra-pulse characteristics of aliased radar signals according to any one of claims 1 to 5, can run according to default parameters, allow users to modify some parameters, and visualize the running results.

7. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the method for analyzing intra-pulse characteristics of aliased radar signals according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the method for analyzing intra-pulse characteristics of aliased radar signals according to any one of claims 1 to 5 is implemented.

Images