Broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis
Through the broadband frequency hopping pulse detection method of two-dimensional energy spectrum analysis, the problems of large resource occupancy and high power consumption in the existing technology are solved, and efficient detection of broadband frequency hopping pulse signals is realized, suitable for various known pulse widths and pulse detection of unknown frequencies.
Patent Information
- Application Number
- CN202411105804.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-11-22
AI Technical Summary
The existing broadband frequency hopping pulse detection method has large resource occupancy and high power consumption, making it difficult to implement on a single FPGA chip.
Using a broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis, the steps of signal preprocessing, signal frequency division processing, pulse waveform detection and parameter estimation include filtering, short-time Fourier transform, smoothing processing, dynamic thresholding and non- Maximum suppression, estimate the pulse frequency and incoming wave orientation, and reduce resource usage.
It realizes detection of broadband frequency hopping pulses, reduces logical resource usage and power consumption, and is suitable for broadband frequency hopping pulse signal detection of unknown frequencies, and has strong practical value.
Smart Images

Figure CN119001620A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of signal processing, and in particular to a broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis. Background Art
[0002] From radar theory, we know that broadband signal system radar has high distance resolution ability, supports high-precision imaging of targets, can extract target frequency domain information, provide basis for signal processing such as target identification, and significantly improve the overall performance of radar. It can be seen that broadband frequency hopping pulse signals occupy a prominent position in the field of modern radar. Therefore, in modern electronic information warfare, the research on detection technology of broadband frequency hopping pulse signals is also very critical.
[0003] For the problem of broadband frequency hopping pulse detection, the existing technology uses narrowband filtering plus digital beam synthesis, and obtains the frequency, azimuth and other information of the signal through directional filters, narrowband filters and sum-difference ratio direction finding algorithms. This method is relatively simple and can inherit some narrowband signal digital beam synthesis algorithms, but this method requires multiple directional filters and multiple narrowband filters for signals in each direction, which consumes a lot of overall resources and is difficult to achieve on a single FPGA chip. Summary of the invention
[0004] The present invention aims to provide a broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis, which analyzes the energy spectrum of broadband signals, searches for two-dimensional peaks, estimates pulse frequencies, obtains baseband data, and estimates the pulse direction based on the phase difference between fitting channels. The detection of broadband frequency hopping pulses is realized, and the problems of large logic resource occupation and high power consumption in existing conventional processing methods are solved.
[0005] The present invention provides a broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis, comprising the following steps:
[0006] (1) Signal preprocessing: Receive multi-channel broadband frequency-hopping pulse signals, divide the frequency band of each channel’s broadband frequency-hopping pulse signals into several even segments through filtering, and reduce the sampling rate and processing speed;
[0007] (2) Signal frequency division processing: For each signal segment, a windowed short-time Fourier transform is used to calculate the time-frequency two-dimensional energy spectrum;
[0008] (3) Pulse waveform detection: Smooth the time-frequency two-dimensional energy spectrum, use dynamic thresholds, and adopt non-maximum suppression methods to complete two-dimensional spectrum peak search, detect pulse arrival time, and estimate pulse frequency;
[0009] (4) Parameter estimation: Based on the estimated pulse frequency, the baseband sequence of the multi-channel broadband frequency-hopping pulse signal is obtained by down-conversion processing. The phase difference between channels is calculated using the baseband phase fitting method to estimate the direction of the incoming wave. The phase position confidence is calculated to reduce false alarms.
[0010] In some preferred embodiments, the frequency band of the broadband frequency hopping pulse signal of each channel is divided into several uniform segments by filtering in step (1), specifically: the frequency band of the received broadband frequency hopping pulse signal is F L ~F H , the broadband frequency hopping pulse signal band of each channel is evenly divided into N segments, and the center frequency of the N segment signal is F i , i = 1, 2, ... N, filter each signal, F i Down-conversion, low-pass filtering and decimation down-sampling operations are performed to obtain N segments of zero intermediate frequency signals.
[0011] In some preferred embodiments, the short-time Fourier transform with windowing in step (2) specifically includes the following sub-steps:
[0012] Step (2.1): If the pulse is intra-pulse modulated and the chip width is L, the window size is L and the step is L / 3; if the pulse is not intra-pulse modulated and the pulse width is W, the window size is W / 10 and the step is W / 10;
[0013] The windowed signal in step (2.2) is subjected to Fourier transform;
[0014] Step (2.3) calculates the energy of the Fourier transformed result, accumulates the energy of multiple channels at the same frequency point at the same time, cuts out the invalid frequency band, and obtains a two-dimensional energy spectrum of time-frequency.
[0015] In some preferred schemes, when Fourier transform is performed in step (2.2), if L does not satisfy the relationship of 2 to the power of N, a coprime factor algorithm is used to divide L into coprime small sequences, and the Fourier transforms of the small sequences are first calculated separately, and then the Fourier transforms of the L sampling points are obtained.
[0016] In some preferred embodiments, the smoothing process of the time-frequency two-dimensional energy spectrum in step (3) is specifically to filter the time domain signal of each frequency band separately using the pulse width as the width of the mean filter to obtain the smoothing result.
[0017] In some preferred schemes, the dynamic threshold in step (3) is specifically, for the result of smoothing, the data at the Mth sampling point position before and after the time axis and half of the value, multiplied by the threshold factor x, as the dynamic threshold; wherein M is the number of samples corresponding to the pulse width.
[0018] In some preferred schemes, the non-maximum suppression method is adopted in step (3), specifically, in the range of left and right M sampling points, if the value exceeding the threshold is not the maximum value in the range, it is suppressed, otherwise it is a detection peak; the time corresponding to the peak is the pulse arrival time, and the corresponding channel is the estimated pulse frequency.
[0019] In some preferred embodiments, the baseband phase fitting method in step (4) comprises the following specific steps:
[0020] Step (4.1), the multi-channel broadband frequency hopping pulse signal is used as input X;
[0021] Step (4.2), estimate the covariance matrix based on the input:
[0022]
[0023] Step (4.3), estimate the phase matrix based on the covariance matrix:
[0024] φ i =angle(mean(dia(R,i))),i=0,1,...7
[0025] Among them, dia(R,i) represents the vector composed of the data on the upper i diagonal lines of the covariance matrix R, mean represents the mean, and angle represents the phase angle of the complex number;
[0026] Step (4.4), perform phase fitting on the phase matrix to obtain the normalized spatial frequency value:
[0027] Step (4.5), convert the corresponding normalized spatial frequency value φ into the source arrival angle θ, and estimate the incoming wave direction through the source arrival angle θ:
[0028]
[0029] Where D is the multiple of the array element spacing relative to the source wavelength.
[0030] In some preferred embodiments, the least squares method is used in step (4.4) to calculate the phase matrix φ i The first fitting result of is taken as the normalized spatial frequency value φ.
[0031] In some preferred embodiments, the phase position reliability calculated in step (4) is used to reduce false alarms. Specifically, according to the phase matrix φ i The adjacent angle difference: Δφ j =φ j -φ j ; j = 1, 2, ..., J, J is the number of channels, calculate Δφ jThe confidence level P is determined based on the mean square error. When the confidence level P is lower than the required value, the pulse is suppressed:
[0032]
[0033] Among them, down The lower threshold of the mean square error, th up Indicates the upper threshold of the mean square error.
[0034] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:
[0035] The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis proposed in the present invention can be used to detect broadband frequency hopping pulse signals of unknown frequency. The implementation of this function is based on signal frequency division processing and two-dimensional energy spectrum pulse detection. The calculation is simple and the resource occupancy rate is less than that of traditional methods. It can be used for pulse detection and frequency estimation of various pulses with known pulse widths and unknown frequencies, and has strong practical value. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 The figure is a flow chart of a broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis in an embodiment of the present invention.
[0037] Figure 2 It is a schematic diagram of three-dimensional data of a two-dimensional energy spectrum in an embodiment of the present invention.
[0038] Figure 3 Schematic diagram of frequency point cutting in an embodiment of the present invention.
[0039] Figure 4 Schematic diagram of energy accumulation in an embodiment of the present invention.
[0040] Figure 5 Schematic diagram of pulse detection results in an embodiment of the present invention
[0041] Figure 6 Graph showing the relationship between detection probability and baseband signal-to-noise ratio in an embodiment of the present invention. DETAILED DESCRIPTION
[0042] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.
[0043] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0044] Example
[0045] like Figure 1 As shown, this embodiment proposes a broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis, comprising the following steps:
[0046] (1) Signal preprocessing: Receive multi-channel broadband frequency-hopping pulse signals, divide the frequency band of each channel’s broadband frequency-hopping pulse signals into several even segments through filtering, and reduce the sampling rate and processing speed;
[0047] (2) Signal frequency division processing: For each signal segment, a windowed short-time Fourier transform is used to calculate the time-frequency two-dimensional energy spectrum;
[0048] (3) Pulse waveform detection: Smooth the time-frequency two-dimensional energy spectrum, use dynamic thresholds, and adopt non-maximum suppression methods to complete two-dimensional spectrum peak search, detect pulse arrival time (noise), and estimate pulse frequency;
[0049] (4) Parameter estimation: Based on the estimated pulse frequency, the baseband sequence of the multi-channel broadband frequency-hopping pulse signal is obtained by down-conversion processing. The phase difference between channels is calculated using the baseband phase fitting method to estimate the direction of the incoming wave. The phase position confidence is calculated to reduce false alarms.
[0050] In step (1), the frequency band of the broadband frequency hopping pulse signal of each channel is divided into several uniform segments by filtering, specifically: the frequency band of the received broadband frequency hopping pulse signal is F L ~F H , the broadband frequency hopping pulse signal band of each channel is evenly divided into N segments, the bandwidth of each segment does not exceed 200M, and the center frequency of the N segment signal is F i , i = 1, 2, ... N, filter each signal, F i Down-conversion, low-pass filtering and decimation down-sampling operations are performed to obtain N segments of zero intermediate frequency signals.
[0051] The short-time Fourier transform using windowing in step (2) specifically includes the following steps:
[0052] Step (2.1): If the pulse is intra-pulse modulated and the chip width is L, the window size is L and the step size is L / 3; if the pulse is not intra-pulse modulated and the pulse width is W, the window size is W / 10 and the step size is W / 10; the window function can be selected according to the filtering characteristics and the amount of calculation, including rectangular window, Hanning window, and Chebyshev window.
[0053] The signal after windowing in step (2.2) is Fourier transformed. If L does not satisfy the relationship of 2 to the power of N, the coprime factor algorithm is used to divide L into coprime small sequences. The Fourier transforms of the small sequences are calculated separately, and then the Fourier transforms of L sampling points are obtained. For example, 60 is divided into coprime small sequences 60=3×5×4, the Fourier transforms of the small sequences are calculated separately, and then the Fourier transforms of 60 sampling points are obtained. Among them, the FFT transform results of 3, 5, and 8 sampling points are obtained by the Fourier transform formula. The schematic diagram of the energy spectrum data of the time-frequency-channel is as follows Figure 2 shown.
[0054] Step (2.3) calculates the energy of the Fourier transform result, accumulates the energy of multiple channels at the same frequency at the same time, cuts the invalid frequency band, and obtains the time-frequency two-dimensional energy spectrum, such as Figure 3 shown.
[0055] The smoothing process of the time-frequency two-dimensional energy spectrum in step (3) is specifically to filter the time domain signal of each frequency band separately with the pulse width as the width of the mean filter to obtain the smoothing result, such as Figure 4 shown.
[0056] The dynamic threshold in step (3) is specifically, for the result of the smoothing process, the data at the Mth sampling point position before and after the time axis and half of the value (M is the number of samples corresponding to the pulse width), multiplied by the threshold factor x, as the dynamic threshold. The larger the value of x, the smaller the false alarm probability of pulse detection. Figure 5 shown.
[0057] The non-maximum suppression method is adopted in step (3). Specifically, in the range of left and right M sampling points, if the value exceeding the threshold is not the maximum value in the range, it is suppressed, otherwise it is a detection peak. The time corresponding to the peak is the pulse arrival time, and the corresponding channel is the estimated pulse frequency, such as Figure 4 shown.
[0058] Taking 8 channels as an example, the baseband phase fitting method in step (4) has the following specific steps:
[0059] Step (4.1), multi-channel broadband frequency hopping pulse signal as input:
[0060] The 8-channel complex vector input X=[x1,x2,x3,x4,x5,x6,x7,x8], n×8, is the amplitude and phase sampling of the 8-unit antenna corresponding to the signal pulse.
[0061] Step (4.2), estimate the covariance matrix based on the input:
[0062]
[0063] Step (4.3), estimate the phase matrix based on the covariance matrix:
[0064] φ i =angle(mean(dia(R,i))),i=0,1,...7
[0065] Among them, dia(R,i) represents the vector composed of data on the upper i diagonals of the covariance matrix R. mean represents the mean, and angle represents the phase angle of the complex number.
[0066] Step (4.4), perform phase fitting on the phase matrix to obtain the normalized spatial frequency value:
[0067] Using the least squares method, calculate the phase matrix φ i The first fitting result of is taken as the normalized spatial frequency value φ.
[0068] Step (4.5), convert the corresponding normalized spatial frequency value φ into the source arrival angle θ, and estimate the incoming wave direction through the source arrival angle θ:
[0069]
[0070] Where D is the multiple of the array element spacing relative to the source wavelength.
[0071] The calculation phase position reliability in step (4) is used to reduce false alarms. Specifically, according to the phase matrix φ i The adjacent angle difference: Δφ j =φ j -φ j ; j = 1, 2, ..., 7, calculate Δφ j The greater the mean square error, the lower the confidence level. If the confidence level is lower than a certain level, the pulse is suppressed. The confidence level P is calculated as follows:
[0072]
[0073] Among them, th down The lower threshold of the mean square error, th up It represents the upper threshold of the mean square error, and the upper and lower thresholds are obtained by a large number of experimental statistics.
[0074] In summary, the broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis proposed in the present invention can be used to detect broadband frequency hopping pulse signals of unknown frequency. The implementation of this function is based on signal frequency division processing and two-dimensional energy spectrum pulse detection. The calculation is simple and the resource occupancy rate is less than that of traditional methods. It can be used for pulse detection and frequency estimation of various pulses with known pulse widths and unknown frequencies, and has strong practical value. Figure 6 As shown, the SNR of the minimum detectable signal is related to the number of channels. The more channels there are, the smaller the minimum detectable pulse SNR is. For an 8-channel array signal, when the 360M intermediate frequency signal signal-to-noise ratio is ≥-23dB, the correct recognition probability is not less than 95%.
[0075] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A broadband frequency-hopping pulse detection method based on two-dimensional energy spectrum analysis, characterized in that: The steps include: (1) Signal preprocessing: Receive multi-channel broadband frequency-hopping pulse signals, divide the frequency band of each channel’s broadband frequency-hopping pulse signals into several even segments through filtering, and reduce the sampling rate and processing speed; (2) Signal frequency division processing: For each signal segment, a windowed short-time Fourier transform is used to calculate the time-frequency two-dimensional energy spectrum; (3) Pulse waveform detection: Smooth the time-frequency two-dimensional energy spectrum, use dynamic thresholds, and adopt non-maximum suppression methods to complete two-dimensional spectrum peak search, detect pulse arrival time, and estimate pulse frequency; (4) Parameter estimation: Based on the estimated pulse frequency, the baseband sequence of the multi-channel broadband frequency-hopping pulse signal is obtained by down-conversion processing. The phase difference between channels is calculated using the baseband phase fitting method to estimate the direction of the incoming wave. The phase position confidence is calculated to reduce false alarms.
2. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 1 is characterized in that: In step (1), the frequency band of the broadband frequency hopping pulse signal of each channel is divided into several uniform segments by filtering, specifically: the frequency band of the received broadband frequency hopping pulse signal is F L ~F H , the broadband frequency hopping pulse signal band of each channel is evenly divided into N segments, and the center frequency of the N segment signal is F i , i = 1, 2, ... N, filter each signal, F i Down-conversion, low-pass filtering and decimation down-sampling operations are performed to obtain N segments of zero intermediate frequency signals.
3. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 1 is characterized in that: The short-time Fourier transform using windowing in step (2) specifically includes the following sub-steps: Step (2.1): If the pulse is intra-pulse modulated and the chip width is L, the window size is L and the step is L / 3; if the pulse is not intra-pulse modulated and the pulse width is W, the window size is W / 10 and the step is W / 10; The windowed signal in step (2.2) is subjected to Fourier transform; Step (2.3) calculates the energy of the Fourier transformed result, accumulates the energy of multiple channels at the same frequency point at the same time, cuts out the invalid frequency band, and obtains a two-dimensional energy spectrum of time-frequency.
4. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 3 is characterized in that: When performing Fourier transform in step (2.2), if L does not satisfy the relationship of 2 to the power of N, the coprime factor algorithm is used to divide L into coprime small sequences, first calculate the Fourier transform of the small sequences separately, and then obtain the Fourier transform of L sampling points.
5. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 1 is characterized in that: The smoothing process of the time-frequency two-dimensional energy spectrum in step (3) is specifically to filter the time domain signal of each frequency band separately using the pulse width as the width of the mean filter to obtain the smoothing result.
6. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 1 is characterized in that: The dynamic threshold in step (3) is specifically, for the result of the smoothing process, the value of half of the data at the Mth sampling point position before and after the time axis is multiplied by the threshold factor x, as the dynamic threshold; wherein M is the number of samples corresponding to the pulse width.
7. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 6 is characterized in that: The non-maximum suppression method is adopted in step (3), specifically, in the range of left and right M sampling points, if the value exceeding the threshold is not the maximum value in the range, it is suppressed, otherwise it is a detection peak; the time corresponding to the peak is the pulse arrival time, and the corresponding channel is the estimated pulse frequency.
8. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 1 is characterized in that: The baseband phase fitting method in step (4) includes the following specific steps: Step (4.1), a multi-channel broadband frequency-hopping pulse signal is used as input X; Step (4.2), estimate the covariance matrix based on the input: Step (4.3), estimate the phase matrix based on the covariance matrix: φ i =angle(mean(dia(R,i))),i=0,1,...7 Among them, dia(R,i) represents the vector composed of the data on the upper i diagonal lines of the covariance matrix R, mean represents the mean, and angle represents the phase angle of the complex number; Step (4.4), perform phase fitting on the phase matrix to obtain the normalized spatial frequency value: Step (4.5), convert the corresponding normalized spatial frequency value φ into the source arrival angle θ, and estimate the incoming wave direction through the source arrival angle θ: Where D is the multiple of the array element spacing relative to the source wavelength.
9. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 8, characterized in that:In step (4.4), the least squares method is used to calculate the phase matrix φ i The first fitting result of is taken as the normalized spatial frequency value φ.
10. The broadband frequency hopping pulse detection method based on two-dimensional energy spectrum analysis according to claim 8, characterized in that: The calculation phase position reliability in step (4) is used to reduce false alarms. Specifically, according to the phase matrix φ i The adjacent angle difference: Δφ j =φ j -φ j ; j = 1, 2, ..., J, J is the number of channels, calculate Δφ j The confidence level P is determined based on the mean square error. When the confidence level P is lower than the required value, the pulse is suppressed: Among them, down The lower threshold of the mean square error, th up Indicates the upper threshold of the mean square error.
Citation Information
Cited By
Adaptive threshold spectrum peak detection and sequence prediction method for frequency hopping signal
CN121124854A
