Method for extracting intra-pulse characteristics of composite modulation signal in real time based on FPGA (Field Programmable Gate Array)

Abstract

The invention provides a method for extracting intra-pulse characteristics of a composite modulation signal in real time based on an FPGA (Field Programmable Gate Array). The method comprises the following steps: firstly, extracting phase information of a signal, identifying phase encoding information of the signal, extracting a code element value and code element position information, then carrying out decorrelation processing on the code element information to obtain the phase information only with frequency modulation, then identifying the frequency encoding information, further extracting frequency modulation characteristics, and finally carrying out frequency modulation. And finally, comprehensively judging whether the signal is an FSK-PSK or FLM-PSK composite modulation signal or not through phase modulation and frequency modulation results. The method has the advantages of high anti-noise capability, high recognition rate and less resource consumption, so that the method can be effectively applied to actual equipment.

Description

Technical Field

[0001] This technology belongs to the field of signal detection technology. Background Art

[0002] With the development of modern technology, the modulation mode of radar signals has changed from single modulation to composite modulation. Composite modulation signals combine the characteristics of multiple modulation modes and have strong concealment, which increases the difficulty of detection. Traditional passive detection methods sort targets by pulse width, carrier frequency, power, arrival time, and azimuth, which can no longer meet the detection needs of modern radars. Therefore, the extraction of characteristic parameters in radar pulses and the increase of the dimension of target recognition have become urgent problems to be solved.

[0003] The research on the intra-pulse analysis method of radar signals started late in China, but in recent years, the research in this area has been relatively active. At present, some relatively mature algorithms for intra-pulse feature analysis include instantaneous autocorrelation, zero-crossing detection algorithm, short-time Fourier transform, wavelet analysis, etc., which have certain effects on single carrier frequency, linear frequency modulation, phase coding, and frequency coding. However, they are relatively limited in practical application, mainly due to the following reasons:

[0004] 1. The pipeline processing mode based on FPGA cannot complete complex algorithms;

[0005] 2. Parallel processing of multiple algorithms significantly increases resource consumption;

[0006] 3. Low signal-to-noise ratio signals may lead to serious misjudgment;

[0007] 4. Composite modulation signals will increase the correlation between judgments and make judgments more difficult;

[0008] In terms of real-time intra-pulse feature extraction, the patent "A real-time extraction and classification system for intra-pulse modulation features based on FPGA" is mainly based on FFT operation, extracting real-time intra-pulse features through statistics of bandwidth, which consumes a lot of resources and does not involve feature extraction of composite modulated signals. In terms of composite modulated signal parameter evaluation, the patent "A radar composite modulated pulse signal identification method" is mainly based on constructing a time-frequency data matrix to obtain wavelet ridge data, and calculating the signal square sum, residual square sum and other methods to obtain composite modulation parameters, which cannot be completed in real time on FPGA. Summary of the invention

[0009] In response to the needs of real-time identification of FSK-PSK and FLM-PSK composite modulated pulse signals and modulation parameter evaluation, the present invention proposes a method for real-time extraction of composite modulated signal pulse features based on FPGA through FPGA calculation of phase mutation points, phase smoothing and multi-order phase derivative methods.

[0010] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0011] Step 1: Extract the phase information p of the signal, calculate the phase difference Δp between adjacent points, perform M-point average processing on Δp, and calculate the N-point spacing difference to obtain the second-order phase difference Δ(Δp);

[0012] Step 2: Determine whether it is a PSK signal by Δ(Δp): Search for the mutation point of Δ(Δp) and determine the size of the mutation point. arrive As The basis for judging the phase jump is that the phase jump from π-θ to π+θ is the basis for the phase jump of π. The number and position of mutation points of π, determine whether it is a PSK signal and the corresponding code element information;

[0013] Step 3: Remove the phase modulation information in the phase difference and obtain the phase difference with only frequency modulation characteristics: delayed phase difference Δp, alignment The mutation point position of π is calculated by subtracting the code element mutation value corresponding to the phase difference from the phase jump position. The calculation formula is ΔP f =Δp-θ i ;

[0014] Step 4: Determine whether it is a simple pulse signal: Phase difference ΔP f Perform the average processing at point L and calculate the spacing difference at point K to obtain the second-order phase difference Δ(ΔP f ), if Δ(ΔP f ) All values ​​are less than the threshold mod0, which means it is a simple pulse signal, otherwise go to step 5;

[0015] Step 5: Determine whether it is an FSK signal: If Δ(ΔP f ) is greater than the threshold mod0 and is discrete and evenly spaced, then the number and value range of this discrete point are determined. If the number is less than A, the value range is F min ~F max , then it is a frequency hopping signal, and the size and position of the frequency hopping value are stored, otherwise go to step 6;

[0016] Step 6: Determine whether it is an FLM signal: Further perform L-point average processing and calculate the K-point spacing difference to obtain the third-order phase difference Δ(Δ(ΔP f )), if Δ(Δ(ΔP f )) All values ​​are less than mod1, the signal is a linear frequency modulation signal, and the frequency modulation slope is Δ(ΔP obtained in step 4 f ) average value, otherwise it is a nonlinear frequency modulation signal;

[0017] Step 7: Comprehensive judgment: If step 2 is established and step 4 is not established, further judge that step 5 is established, then the pulse is an FSK-PSK composite modulation signal, and the code element information and frequency hopping information are extracted. If step 2 is established and steps 4 and 5 are not established, further judge that step 6 is established, then the pulse is an FLM-PSK composite modulation signal, and the code element information and frequency modulation slope information are extracted.

[0018] Preferably, in step 1: when calculating the phase difference Δp, phase unwinding is required at the point where the phase p crosses zero, and the calculation formula is: Find the M-point smoothing and N-point spacing difference, assuming that the impact of M on the mutation value is π / M, and the impact on the noise is Noise / M. The N-point spacing difference is selected based on being greater than 2*M and less than half of the code element spacing;

[0019] In step 2: The phase jump boundary of and π is and The magnitude of the phase-jump deviation value θ is determined by the signal-to-noise ratio, and the calculation formula is:

[0020] In step 4: the selection basis of L and K, L is the positive and negative mutation value of the smoothed noise and the symbol, and the smoothed phase noise is K is the phase difference change caused by frequency change, the phase difference change is 4π*f var *K*t n , L and K should be chosen so that the phase noise is less than about 5% of the phase variation;

[0021] In step 5: the number of discrete points and the range of values ​​are determined based on: A is the number of frequency hopping times, which is a single digit by default, and F min is the parameter corresponding to the minimum frequency hopping interval, F max is the parameter corresponding to the maximum frequency hopping spacing, and the calculation formula is 2π*(F i -F i-1 )*t n .

[0022] The beneficial effects of the present invention are as follows: in order to solve the problem of difficulty in sorting complex radar radiation source signals, when the intercepted pulse groups cannot be effectively clustered by frequency, azimuth, period, and pulse width, a clustering and merging method based on intra-pulse features is added to improve the accuracy of sorting complex radar radiation sources. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a preferred flow chart of the present invention;

[0024] Figure 2 It is the flow chart of phase extraction of FSK-PSK signal;

[0025] Figure 3 It is the flow chart of FLM-PSK signal phase extraction;

[0026] Figure 4 It is a flow chart of extracting phase coding information and de-phase coding of FSK-PSK signal;

[0027] Figure 5 It is a flow chart of extracting phase coding information and de-phase coding of FLM-PSK signal;

[0028] Figure 6 It is a flow chart of extracting frequency hopping information from FSK-PSK signal;

[0029] Figure 7 It is the flow chart of extracting linear frequency modulation information from FLM-PSK signal. DETAILED DESCRIPTION

[0030] The embodiments of the present invention are further described below with reference to the accompanying drawings and specific examples.

[0031] For FSK-PSK and FLM-PSK composite modulated pulse signals, the present invention proposes a method for extracting composite modulated signal pulse features in real time based on FPGA. The preferred implementation process includes:

[0032] 1. Find the modulus and phase of FSK-PSK and FLM-PSK composite modulation signals:

[0033] The FSK-PSK signal formula is:

[0034] s(t)=esp[j(2πf i t+πθ(m)]+snr*esp[j*randn(t)];

[0035] where f i is the frequency hopping frequency, θ(m) is the binary or quadrature symbol value, snr is the signal-to-noise ratio, and randn(t) is the random noise. The FLM-PSK signal formula is:

[0036] s(t)=esp[i(2πf 0 t+πkt 2 +πθ(m)]+snr*esp[j*randn(t)];

[0037] where f 0is the starting frequency, k is the frequency modulation slope, θ(m) is the binary or quadrature symbol value, snr is the signal-to-noise ratio, and randn(t) is the random noise. This example uses a 100MHz clock, a 12dB signal-to-noise ratio, a 13-bit Barker code for phase encoding, a 20us pulse width, a 1ms period, a 5MHz FSK frequency hopping size, a 4us interval, and a 10MHz FLM bandwidth. The IQ output modulus and phase are calculated. The IQ signal waveform output and the results of the modulus and phase calculation are shown in the following table. Figure 2 , Figure 3 shown.

[0038] 2. Calculate the phase difference Δp for the phase information, perform M-point average processing on Δp and N-point spacing difference to calculate Δ(Δp). In this example, M point is selected as 4 and N is selected as 8. Search whether the singular point of Δ(Δp) falls on or π, the judgment boundary θ in this example is set to It is used to determine the size and position of the phase jump, and to remove the phase modulation information of the signal by aligning the phase difference delay to obtain new phase difference information with only frequency modulation. Figure 4 As shown in the figure, after the 4-point average processing of the FSK-PSK signal, the noise is significantly reduced. After the 8-point spacing difference processing, two mutation points with positive and negative correlation appear at each code element value change, which are used to judge the code element value and code element position. After the phase jump point removes the phase modulation information in the phase difference, the new phase difference ΔP f Some of the mutation points are eliminated, and each of the remaining mutation points has a corresponding opposite mutation point that can offset each other. Figure 5 Shown is the processing process of FLM-PSK signal.

[0039] 3. Frequency hopping judgment: phase difference ΔP f Perform the average processing at point L and calculate the spacing difference at point K to obtain the second-order phase difference Δ(ΔP f ), in this example, L and K select 64 points, mod0 selects 0.2M, and the frequency hopping range is selected between 2M and 10M. Find the frequency hopping point. By judging the size and position of the mutation point, the frequency hopping interval and position can be obtained to determine whether it belongs to a frequency hopping signal, such as Figure 6 As shown in the figure, after the phase difference after unwrapping is smoothed by 64 points, the positive and negative related mutation points cancel each other out, and only the jitter information of the frequency jump position is left. After the 64-point spacing difference is made, there are mutation points at the frequency jump position and they are evenly spaced. The rest of the values ​​are jittering around 0, and the jitter range is less than 0.2M. By judging the size and position of the mutation point, the frequency hopping interval and position can be obtained, and finally it is determined whether the output is an FSK signal:

[0040] 4. Linear frequency modulation judgment, by Δ(ΔP f) Further perform L point average processing and calculate the K point spacing difference to obtain the third-order phase difference Δ(Δ(ΔP f )), in this example, L and K select 128 points, mod0 selects 0.2M, and mod1 selects 0.2M. Figure 7 As shown, by judging Δ(ΔP f ) is greater than mod0, the proportion is greater than 80%, Δ(Δ(ΔP f )) is greater than mod1 by less than 10%, which can be determined as a linear frequency modulation signal with a modulation frequency of Δ(ΔP f )’s average value.

Claims

1. A method for extracting the intra-pulse features of a composite modulated signal in real time based on FPGA, characterized in that: Step 1: Extract the phase information p of the signal, calculate the phase difference Δp between adjacent points, perform M-point average processing on Δp, and calculate the N-point spacing difference to obtain the second-order phase difference Δ(Δp); Step 2: Determine whether it is a PSK signal by Δ(Δp): Search for the mutation point of Δ(Δp) and determine the size of the mutation point. arrive As The basis for judging the phase jump is that the phase jump from π-θ to π+θ is the basis for the phase jump of π. The number and position of mutation points of π, determine whether it is a PSK signal and the corresponding code element information; Step 3: Remove the phase modulation information in the phase difference and obtain the phase difference with only frequency modulation characteristics: delayed phase difference Δp, alignment The mutation point position of π is calculated by subtracting the code element mutation value corresponding to the phase difference from the phase jump position. The calculation formula is: ΔP f =Δp-θ i ; Step 4: Determine whether it is a simple pulse signal: Phase difference ΔP f Perform the average processing at point L and calculate the spacing difference at point K to obtain the second-order phase difference Δ(ΔP f ), if Δ(ΔP f ) All values ​​are less than the threshold mod0, which means it is a simple pulse signal, otherwise go to step 5; Step 5: Determine whether it is an FSK signal: If Δ(ΔP f ) is greater than the threshold mod0 and is discrete and evenly spaced, then the number and value range of this discrete point are determined. If the number is less than A, the value range is F min ~F max , then it is a frequency hopping signal, and the size and position of the frequency hopping value are stored, otherwise go to step 6; Step 6: Determine whether it is an FLM signal: Further perform L-point average processing and calculate the K-point spacing difference to obtain the third-order phase difference Δ(Δ(ΔP f )), if Δ(Δ(ΔP f )) All values ​​are less than mod1, the signal is a linear frequency modulation signal, and the frequency modulation slope is Δ(ΔP obtained in step 4 f ) average value, otherwise it is a nonlinear frequency modulation signal; Step 7: Comprehensive judgment: If step 2 is established and step 4 is not established, further judge that step 5 is established, then the pulse is an FSK-PSK composite modulation signal, and the code element information and frequency hopping information are extracted. If step 2 is established and steps 4 and 5 are not established, further judge that step 6 is established, then the pulse is an FLM-PSK composite modulation signal, and the code element information and frequency modulation slope information are extracted.

2. The method for extracting the intra-pulse features of a composite modulated signal based on FPGA in real time according to claim 1, characterized in that: In step 1: when calculating the phase difference Δp, phase unwinding is required at the point where the phase p crosses zero, and the calculation formula is: Find the M-point smoothing and N-point spacing difference, assuming that the impact of M on the mutation value is π / M, and the impact on the noise is Noise / M. The N-point spacing difference is selected based on being greater than 2*M and less than half of the code element spacing; In step 2: The phase jump boundary of and π is and The magnitude of the phase-jump deviation value θ is determined by the signal-to-noise ratio, and the calculation formula is: In step 4: the selection basis of L and K, L is the positive and negative mutation value of the smoothed noise and the symbol, and the smoothed phase noise is K is the phase difference change caused by frequency change, the phase difference change is 4π*f var *K*t n , L and K should be chosen so that the phase noise is less than about 5% of the phase variation; In step 5: the number of discrete points and the range of values ​​are determined based on: A is the number of frequency hopping times, which is a single digit by default, and F min is the parameter corresponding to the minimum frequency hopping interval, F max is the parameter corresponding to the maximum frequency hopping spacing, and the calculation formula is: 2π*(F i -F i-1 )*t n 。

Images