Signal Recognition Method Based on Wideband Spectrum Detection

Abstract

The present invention discloses a signal recognition method based on broadband spectrum detection. The FPGA is used to perform power spectrum estimation and signal detection on broadband IQ data. After obtaining the signal frequency point parameters, the DDC module is configured to complete digital downconversion. The power spectral density estimation and noise floor estimation of narrowband IQ signals are performed on the upper computer. Preliminary classification is carried out by calculating the ratio parameter C1 of the maximum value to the second maximum value of the spectrum. When C1 is greater than the threshold, the AM and single-carrier signals are discriminated by combining the spectrum feature parameter C2 and the amplitude spectrum feature parameter C3. When C1 is less than or equal to the threshold, a double-threshold algorithm is used for signal detection and the merging and screening in the multi-signal scenario are completed. Classification is carried out based on the normalized zero-centered instantaneous amplitude spectrum density maximum value parameter C4. When C4 is less than the threshold, the FM, MSK, and FSK signals are discriminated. When C4 is greater than or equal to the threshold, the BPSK, QPSK, 8PSK, and 16QAM signals are discriminated. The present invention adopts a multi-parameter joint decision strategy, which significantly improves the accuracy and real-time performance of signal recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of signal recognition, and in particular to a signal recognition method based on broadband spectrum detection. Background Art

[0002] In the field of modern wireless communications and electronic countermeasures, signal detection and modulation recognition technology has important military and civilian value. Existing technologies mainly include time domain analysis based on high-order cumulants, frequency domain analysis based on power spectrum density, and time-frequency joint analysis based on wavelet transform. The performance of these methods drops sharply in low signal-to-noise ratio environments, and the computational complexity is high, making it difficult to meet real-time processing requirements. Especially in complex electromagnetic environments, problems such as multiple signal overlaps leading to spectrum aliasing, significant power differences causing weak signal drowning, and difficulty in estimating the number of signals seriously restrict the detection and recognition performance. At the same time, traditional methods for identifying signal modulation types mainly rely on single feature judgments, have poor robustness, and are difficult to simultaneously process multiple modulation modes such as AM, FM, PSK, and QAM.

[0003] The common technical difficulties in engineering implementation at present include: the contradiction between ADC sampling bandwidth and signal bandwidth restricts system performance, limited FPGA resources affect parallel processing capabilities, insufficient data throughput affects real-time performance; at the algorithm level, there are problems such as high-order statistical features are sensitive to noise, the accuracy of instantaneous feature extraction is insufficient, and there is a lack of theoretical guidance for feature dimension selection; in terms of system integration, multi-board synchronization, limited data interface bandwidth, power consumption and heat dissipation and other engineering problems also need to be solved urgently. In particular, in terms of the measurement accuracy of key parameters such as carrier frequency estimation, signal bandwidth estimation, and signal-to-noise ratio estimation, existing methods are difficult to meet the requirements of high-performance signal recognition.

[0004] To address the above issues, new signal detection and recognition methods need to be developed. Summary of the invention

[0005] In order to solve the problems existing in the prior art, the present invention provides a signal recognition method based on broadband spectrum detection, which adopts the following technical solutions:

[0006] A signal recognition method based on broadband spectrum detection, the method comprising the following steps:

[0007] Step S1, use FPGA to perform power spectrum estimation on broadband IQ data and then perform signal detection to obtain the starting frequency and ending frequency of the signal, and use the median of the starting frequency and the ending frequency as the carrier frequency; use the carrier frequency as the DDC frequency parameter, and use 8 times the detection bandwidth as the DDC bandwidth parameter, configure DDC and perform digital DDC on the broadband IQ to obtain narrowband IQ.

[0008] Step S2: Based on the host computer, the narrowband IQ signal is subjected to power spectrum estimation using the periodogram method to obtain the power spectrum density psd, and the noise floor is estimated using the power spectrum histogram method to obtain the noise floor estimation value. .

[0009] Step S3, calculate the ratio of the maximum value to the second largest value of the power spectrum density psd, recorded as parameter C1, when parameter C1 is greater than the threshold C1th, calculate parameters C2 and C3; if parameter C2 is greater than the threshold C2th, and parameter C3 is greater than the threshold C3th, it is determined to be an AM signal, otherwise it is determined to be a single carrier signal; wherein parameter C2 represents the ratio of the spectrum mean of the first k largest values ​​within a preset interval around the maximum value of the spectrum to the noise floor, and parameter C3 represents the maximum value of the spectrum density of the normalized zero-centered instantaneous amplitude.

[0010] Step S4, when the parameter C1 is less than or equal to the threshold C1th, the signal is merged and screened according to the number of detected signals to obtain the target signal, and the power spectrum density psd of the target signal is detected using a double threshold signal detection algorithm to obtain the signal start frequency, end frequency and signal-to-noise ratio parameters.

[0011] Step S5, frequency offset compensation and low-pass filtering are performed on the narrowband IQ data according to the center frequency and bandwidth to obtain IQ_filter; the maximum spectral density of the normalized zero-center instantaneous amplitude is calculated and recorded as parameter C4; when parameter C4 is less than threshold C4th, modulation type discrimination is performed on IQ_filter to identify FM, MSK and FSK signals; when parameter C4 is greater than or equal to threshold C4th, modulation type discrimination is performed on IQ_filter to identify BPSK, QPSK, 8PSK and 16QAM signals.

[0012] Furthermore, the calculation method of the parameter C1 is: search for the maximum value psd_max of psd and its position max_i in the range of Nfft / 4 to 3Nfft / 4, set the psd value in the interval of max_i-2 to max_i+2 to 0, search for the maximum value of the remaining psd again and mark it as the second largest value psd_max2, and calculate C1=psd_max / psd_max2.

[0013] Furthermore, in step S4, signal merging and screening are performed according to the number of detected signals, including: for the kth signal, k is a natural number, and the starting frequency of the signal is , the end frequency is ,bandwidth , the in-band power is , the signal-to-noise ratio is , the estimated noise floor is , power similarity factor , indicating the The signal and The signal power ratio.

[0014] Bandwidth similarity factor: ; indicates the The signal and The signal bandwidth ratio;

[0015] Distance factor between signals: ; indicates the The signal and The spectral distance of the signal;

[0016] The power spectral density between the two signals ; Represents the power spectral density value of the middle frequency point between the starting frequency point of signal i+1 and the ending frequency point of signal i.

[0017] When two signals are detected, the two signals are recorded as the first signal and the second signal;

[0018] like ;

[0019] or ;

[0020] Then the two signals are merged into one signal, and the starting frequency point after merging is recorded as the starting frequency point of the first signal before merging, and the ending frequency point is recorded as the ending frequency point of the second signal before merging.

[0021] like ,but If the above merging conditions are not met, the distance from the signal to zero frequency is calculated. ,in represents the number of Fourier transform points, k represents the kth signal, and Smaller signal.

[0022] If not satisfied , the signal with larger power is retained.

[0023] Further, signal merging and screening are performed according to the number of detected signals, including: when three signals are detected, the three signals are recorded as the first signal, the second signal and the third signal;

[0024] like , then the starting frequency and ending frequency of the combined signal are , indicating that the starting frequency point extends to the left by a small signal width.

[0025] like , then the starting frequency and ending frequency of the combined signal are , indicating that the end frequency point extends to the right by a small signal width.

[0026] like , then the three signals are combined and expressed as .

[0027] Furthermore, in step S4, signal merging and screening are performed according to the number of detected signals, including: when more than 3 signals are detected, the merging and screening method, assuming that the number of detected signals is N, performs the following steps:

[0028] Step a), find the signal with the highest power among all signals, assuming it is , compare the bandwidth and power factor of the remaining signals with the kmax-th signal in turn, and find the smallest Signals:

[0029] ;

[0030] If the above formula If it is less than the preset value, the The signal and signal and discard other signals, otherwise proceed to the next step;

[0031] Step b), if the above merging conditions are not met, The adjacent signals outside the range are merged one by one, and the merged signals are used as the first Repeat step a) for each signal. If the signal still cannot be merged after repeating step a), proceed to step c);

[0032] Step c), all N signals are sorted by power, the top three signals are taken out, the distances of these three signals to zero frequency are compared, the signal with the smallest zero frequency distance is taken as the final signal and the other signals are discarded.

[0033] Furthermore, in step S5, when C4 is less than a threshold C4th, the modulation type determination method includes:

[0034] Perform frequency demodulation on IQ_filter and perform sign judgment on the demodulation result to obtain a ±1 sequence, take the absolute value after taking the difference of the sequence, calculate the absolute value spectrum of the differential frequency, and detect the single spectrum prominence parameter C5 of this spectrum;

[0035] When C5 is less than the threshold C5th, it is determined to be an FM signal;

[0036] When C5 is greater than or equal to the threshold C5th, the distance D1 from the single spectrum position to the middle position is recorded, and the square spectrum modulus of IQ_filter is taken to calculate the distance D2 between the maximum value and the second largest value of the spectrum;

[0037] When the difference between D1 and D2 is within the preset error range, it is considered that D1=D2 and the signal is an MSK signal, otherwise the signal is determined to be an FSK signal.

[0038] Furthermore, in step S5, when C4 is greater than or equal to the threshold C4th, the modulation type determination method includes:

[0039] The square spectrum of IQ_filter is modulo and recorded as C7. When there is a single spectrum component, that is, C7>C7th, it is determined to be a BPSK signal.

[0040] When there is no single spectrum component, the calculated fourth power spectrum is recorded as C8. When there is no single spectrum component in the fourth power spectrum, that is, C8 < C8th, it is determined to be an 8PSK signal. When there is a single spectrum component in the fourth power spectrum, the ratio parameter R of the square of the amplitude to the square of the mean of the square of the amplitude is calculated.

[0041] The corresponding decision threshold is selected according to the current signal-to-noise ratio. The decision threshold is a broken line function of the signal-to-noise ratio. When the parameter R is less than the threshold Rth, it is determined to be a QPSK signal, otherwise it is determined to be a 16QAM signal.

[0042] Furthermore, the thresholds C1th, C2th, C3th, C4th, C5th, C7th, and C8th are set to 8, 4, 40, 50, 10, 8, and 8, respectively.

[0043] Furthermore, the threshold Rth is a function of the signal-to-noise ratio. When the SNR is between 8 dB and 18 dB, the threshold Rth is a diagonal line, and the thresholds at the end points of the diagonal line are 0.55 and 0.37 respectively. When the SNR is greater than 18 dB, the threshold Rth is a constant 0.37.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The invention combines the signal time domain and spectrum characteristics to complete the recognition and parameter estimation of 9 common signals, including single carrier, Am, fm, MSK, FSK, BPSK, QPSK, 16QAM and 8PSK; adopts a joint detection scheme of coarse spectrum detection and fine spectrum detection to improve the accuracy of frequency, bandwidth and signal-to-noise ratio estimation; Am signals and single carrier signals are identified based on the joint judgment of amplitude change characteristics and spectrum characteristics, and different detection thresholds are set based on different signal-to-noise ratio intervals to improve recognition accuracy.

[0046] Since rectangular pulse QPSK-like spectrum and FSK signals with different modulation indexes show different irregular spectrum characteristics under high signal-to-noise ratio, the present invention intelligently merges spectrum according to spectrum characteristics under different modulation indexes, avoiding the problem of FSK signals with large modulation indexes being mistakenly detected as multiple signals; QPSK, BPSK, 16QAM and 8PSK consider the influence of rectangular, raised cosine and root raised cosine shaped pulses on recognition parameters, and different shaped pulses do not affect the recognition performance of the nine signals. The practicality and accuracy of the recognition algorithm under a wide range of signal-to-noise ratios are improved, and the engineering practicality is strong. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a flowchart of a signal recognition method based on broadband spectrum detection according to an embodiment of the present invention;

[0048] Figure 2 It is a schematic diagram of Am spectrum and C2 parameter calculation according to an embodiment of the present invention;

[0049] Figure 3 This is a spectrum diagram of a rectangular pulse QPSK with an SNR of 20 dB according to an embodiment of the present invention;

[0050] Figure 4 This is a spectrum diagram of an FSK signal with an SNR of 20 dB and a modulation index of 1.3 according to an embodiment of the present invention;

[0051] Figure 5 Graphs of R parameters for QPSK and 16QAM according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0053] like Figure 1 FIG. 1 is a flowchart of a signal recognition method based on broadband spectrum detection according to the present invention. The signal recognition method based on broadband spectrum detection comprises the following steps:

[0054] Step S1, use FPGA to perform power spectrum estimation on broadband IQ data and then perform signal detection to obtain the starting frequency and ending frequency of the signal, and use the median of the starting frequency and the ending frequency as the carrier frequency; use the carrier frequency as the DDC frequency parameter, and use 8 times the detection bandwidth as the DDC bandwidth parameter, configure DDC and perform digital DDC on the broadband IQ to obtain narrowband IQ.

[0055] FPGA is used to use the periodogram method to estimate the power spectrum of broadband IQ data, and a double threshold algorithm is used to detect the signal on the spectrum to obtain the start frequency and end frequency of the signal. The median of the start frequency and the end frequency is used as the carrier frequency. A power spectrum density estimation algorithm based on the Welch method is used to segment the input broadband IQ data and add windows for FFT transformation. The variance of the spectrum estimation is reduced by averaging the segmented data, thereby improving the accuracy of spectrum estimation.

[0056] On this basis, the CFAR detection algorithm is used to perform adaptive threshold detection on the power spectrum to obtain the starting frequency f_start and the ending frequency f_end of the signal, and (f_start + f_end) / 2 is used as the signal carrier frequency f_c; the carrier frequency f_c is used as the center frequency parameter of DDC, and considering the spectrum extension and modulation sideband of the signal, the detection bandwidth B = 8 times of f_end - f_start is set as the DDC bandwidth parameter, that is, B_ddc = 8B, which can effectively avoid signal distortion and aliasing; accordingly, the NCO, CIC filter and FIR filter parameters of the programmable DDC module are configured, and the input broadband IQ data is efficiently digitally down-converted and extracted and filtered, and finally the narrowband IQ data of the required frequency band is output.

[0057] Among them, the reasonable configuration of DDC bandwidth parameters is crucial to subsequent modulation recognition. Too narrow bandwidth will cause signal distortion, and too wide bandwidth will introduce additional noise. The 8x bandwidth coefficient is the optimal empirical value obtained through a large number of simulations and experiments.

[0058] Step S2: Based on the host computer, the narrowband IQ signal is subjected to power spectrum estimation using the periodogram method to obtain the power spectrum density psd, and the noise floor is estimated using the power spectrum histogram method to obtain the noise floor estimation value. .

[0059] The power spectrum histogram method estimates the noise floor and obtains the noise floor estimate The following steps are involved:

[0060] Step 1: Use the periodogram method to estimate the power spectrum of the baseband IQ signal, and the power spectrum estimation result is psd;

[0061] Step 2: Calculate the mean of the power spectrum psd, recorded as psd_mean; find the minimum value of psd, recorded as psd_min, and normalize the values ​​of psd falling into the interval [psd_min,psd_mean] to the interval [0,1];

[0062] Step 3: For the above normalized [0,1] interval, according to the number of divided intervals P, count the number of Ki falling into each interval, i = 1, 2, 3...P, and obtain the histogram of the normalized power spectrum;

[0063] Step 4: Find the interval corresponding to the maximum value of the power spectrum histogram, denormalize the interval to obtain the noise floor estimate Snoise;

[0064] Step 5: Based on the noise power spectrum estimation obtained in the previous step, calculate th1 and th2 as follows: th1=K1*Snoise+C1*(psd_mean-psd_min); th2=K2*Snoise+C2*(psd_mean-psd_min); Among them, K1 and K2 are threshold multiplication factors, for example, K1=6, K2=12, C1 and C2 are DC bias coefficients much less than 1, and the role of C1 and C2 is to avoid the problem of too small threshold calculation when the signal-to-noise ratio is extremely high.

[0065] Step S3, calculate the ratio of the maximum value to the second largest value of the power spectrum density psd, recorded as parameter C1, when parameter C1 is greater than the threshold C1th, calculate parameters C2 and C3; if parameter C2 is greater than the threshold C2th, and parameter C3 is greater than the threshold C3th, it is determined to be an AM signal, otherwise it is determined to be a single carrier signal; wherein parameter C2 represents the ratio of the spectrum mean of the first k largest values ​​within a preset interval around the maximum value of the spectrum to the noise floor, and parameter C3 represents the maximum value of the spectrum density of the normalized zero-centered instantaneous amplitude.

[0066] The calculation method of the parameter C1 is: search for the maximum value psd_max of psd and its position max_i in the range of Nfft / 4 to 3Nfft / 4, set the psd value in the interval of max_i-2 to max_i+2 to 0, search for the maximum value of the remaining psd again and mark it as the second largest value psd_max2, and calculate C1=psd_max / psd_max2.

[0067] The C2 parameter represents the ratio of the mean of the first k largest spectrum to the noise floor within a certain interval around the maximum value of the spectrum. The certain interval around is a preset interval and can be set according to the waveform of the signal. As shown in Figure 2, it is a schematic diagram of the calculation of the Am spectrum and C2 parameters in an embodiment of the present invention. The Am spectrum is characterized by a certain interval around the spectrum peak, where spectrum points higher than the noise floor are distributed, but the specific location of the higher spectrum points is related to the specific speech signal.

[0068] Therefore, we search for the first k maximum values ​​within a certain frequency band, and calculate the ratio of the average of the first k maximum values ​​to the noise floor to evaluate whether there is a spectrum higher than the noise floor near the maximum value. The frequency point where the spectrum maximum value is located is used as the carrier frequency point and multiplied by the local carrier to correct the frequency deviation, and low-pass filtering is performed with a single-sided passband of 6kHz (considering that the width of the Am single-sided spectrum is about 6kHz). After frequency deviation correction and low-pass filtering, the parameter C3 is calculated. The parameter C3 represents the maximum value of the spectral density of the normalized zero-centered instantaneous amplitude, which represents the amplitude change degree of the signal. The formula is expressed as:

[0069] ;

[0070] ;

[0071] ;

[0072] in represents the baseband signal amplitude at the i-th point, N represents the number of signal points involved in the calculation, and FFT represents Fourier transform.

[0073] Step S4, when the parameter C1 is less than or equal to the threshold C1th, the signal is merged and screened according to the number of detected signals to obtain the target signal, and the power spectrum density psd of the target signal is detected using a double threshold signal detection algorithm to obtain the signal start frequency, end frequency and signal-to-noise ratio parameters.

[0074] In step S4, signal merging and screening are performed according to the number of detected signals, including: for the kth signal, k is a natural number, and the starting frequency of the signal is , the end frequency is ,bandwidth , the in-band power is , the signal-to-noise ratio is , the estimated noise floor is , power similarity factor , indicating the The signal and The signal power ratio.

[0075] Bandwidth similarity factor: ; indicates the The signal and The signal bandwidth ratio;

[0076] Distance factor between signals: ; indicates the The signal and The spectral distance of the signal;

[0077] The power spectral density between the two signals ; Represents the power spectral density value of the middle frequency point between the starting frequency point of signal i+1 and the ending frequency point of signal i.

[0078] When two signals are detected, the two signals are recorded as the first signal and the second signal;

[0079] like ;

[0080] or ;

[0081] Then the two signals are merged into one signal, and the starting frequency point after merging is recorded as the starting frequency point of the first signal before merging, and the ending frequency point is recorded as the ending frequency point of the second signal before merging.

[0082] like ,but If the above merging conditions are not met, the distance from the signal to zero frequency is calculated. ,in represents the number of Fourier transform points, k represents the kth signal, and Smaller signal.

[0083] If not satisfied , the signal with larger power is retained.

[0084] Further, signal merging and screening are performed according to the number of detected signals, including: when three signals are detected, the three signals are recorded as the first signal, the second signal and the third signal;

[0085] like , that is, "two large and one small", then the starting frequency and ending frequency of the combined signal are , indicating that the starting frequency point extends to the left by a small signal width.

[0086] like , that is, "one small and two large", then the starting frequency and ending frequency of the combined signal are , indicating that the end frequency point extends to the right by a small signal width.

[0087] like , that is, "two large and one small", then the three signals are combined and expressed as .

[0088] In summary, when the power and bandwidth difference of the two high-power signals is within the preset range, and the ratio of the spectral density mean of the large signal to the spectral density mean of the small signal exceeds the preset threshold:

[0089] If the small signal is between two high-power signals, the three signals are combined into one signal;

[0090] If the small signal is in front and close to the high-power signal, then a small signal spectrum width will be extended backwards;

[0091] If the small signal is at the back and close to the high-power signal, then the spectrum width of the small signal will be extended forward;

[0092] If the above conditions are not met and the two high-power signals are close to each other, the two high-power signals are merged and the low-power signal is discarded;

[0093] If none of the above conditions are met, the signal with the highest power is retained.

[0094] In step S4, signal merging and screening are performed according to the number of detected signals, including: when more than 3 signals are detected, the merging and screening method, assuming that the number of detected signals is N, performs the following steps:

[0095] Step a), find the signal with the highest power among all signals, assuming it is , compare the bandwidth and power factor of the remaining signals with the kmax-th signal in turn, and find the smallest Signals:

[0096] ;

[0097] If the above formula If it is less than the preset value, the The signal and signal and discard other signals, otherwise proceed to the next step;

[0098] Step b), if the above merging conditions are not met, The adjacent signals outside the range are merged one by one, and the merged signals are used as the first Repeat step a) for each signal. If the signal still cannot be merged after repeating step a), proceed to step c);

[0099] Step c), all N signals are sorted by power, the top three signals are taken out, the distances of these three signals to zero frequency are compared, the signal with the smallest zero frequency distance is taken as the final signal and the other signals are discarded.

[0100] Figure 3 The snr of the embodiment is a 20db rectangular pulse QPSK spectrum diagram. Before merging, the side lobes are detected as signals, and the number of signals is 5. After merging with this strategy, signal 3 is taken out and other side lobe signals are discarded; Figure 4 This is a spectrum diagram of an FSK signal with an SNR of 20 dB and a modulation index of 1.3 in the embodiment. Four signals are detected before merging. By applying this strategy, signals 2, 3, and 4 can be merged and other side lobes can be discarded.

[0101] Step S5, frequency offset compensation and low-pass filtering are performed on the narrowband IQ data according to the center frequency and bandwidth to obtain IQ_filter; the maximum spectral density of the normalized zero-center instantaneous amplitude is calculated and recorded as parameter C4; when parameter C4 is less than threshold C4th, modulation type discrimination is performed on IQ_filter to identify FM, MSK and FSK signals; when parameter C4 is greater than or equal to threshold C4th, modulation type discrimination is performed on IQ_filter to identify BPSK, QPSK, 8PSK and 16QAM signals.

[0102] In step S5, when C4 is less than the threshold C4th, the modulation type determination method includes:

[0103] Perform frequency demodulation on IQ_filter and perform sign judgment on the demodulation result to obtain a ±1 sequence, take the absolute value after taking the difference of the sequence, calculate the absolute value spectrum of the differential frequency, and detect the single spectrum prominence parameter C5 of this spectrum;

[0104] When C5 is less than the threshold C5th, it is determined to be an FM signal;

[0105] When C5 is greater than or equal to the threshold C5th, the distance D1 from the single spectrum position to the middle position is recorded, and the square spectrum modulus of IQ_filter is taken to calculate the distance D2 between the maximum value and the second largest value of the spectrum;

[0106] When the difference between D1 and D2 is within the preset error range, it is considered that D1=D2 and the signal is an MSK signal, otherwise the signal is determined to be an FSK signal.

[0107] In step S5, when C4 is greater than or equal to the threshold C4th, the modulation type determination method includes:

[0108] The square spectrum of IQ_filter is modulo and recorded as C7. When there is a single spectrum component, that is, C7>C7th, it is determined to be a BPSK signal.

[0109] When there is no single spectrum component, the calculated fourth power spectrum is recorded as C8. When there is no single spectrum component in the fourth power spectrum, that is, C8 < C8th, it is determined to be an 8PSK signal. When there is a single spectrum component in the fourth power spectrum, the ratio parameter R of the square of the amplitude to the square of the mean of the square of the amplitude is calculated.

[0110] The corresponding decision threshold is selected according to the current signal-to-noise ratio. The decision threshold is a broken line function of the signal-to-noise ratio. When the parameter R is less than the threshold Rth, it is determined to be a QPSK signal, otherwise it is determined to be a 16QAM signal.

[0111] Specifically, the above steps are described in detail as follows:

[0112] Frequency discrimination: ;

[0113] conj() means taking the conjugate, angle() means taking the phase angle of the complex number, N means the length of the data involved in the operation, sign means taking the sign, -1 is taken when it is less than 0, and 1 is taken when it is greater than or equal to 0.

[0114] Absolute value of difference .

[0115] Calculate the spectrum and take the modulus .

[0116] judge Whether there is a single spectrum, the discriminant is:

[0117] .

[0118] It represents the ratio of the current value to the mean value in the window, and W is the window length.

[0119] If the above formula is k traversed back If both are less than the threshold, there is no single spectrum and it is judged as FM signal.

[0120] Otherwise, record the first position kth greater than the threshold, and calculate the distance from kth to zero frequency ; Take the square spectrum modulus of IQ_filter, calculate the distance between the maximum value and the second maximum value of the spectrum and record it as D2. When looking for the second maximum value, bypass the two frequency points on the left and right of the maximum value. If the difference between D2 and D1 obtained in the previous step is within the error range, it means that the cpm modulation index is 0.5, and it is judged as MSK, otherwise it is FSK.

[0121] Take the square spectrum modulo IQ_filter. If there is a single spectrum component, it is BPSK.

[0122] Otherwise, calculate the fourth power spectrum. If there is a single spectrum component in the fourth power spectrum, calculate the square of the amplitude and the square of the mean of the amplitude, and then use the threshold related to the signal-to-noise ratio. If it is less than the threshold, it is QPSK, otherwise it is 16QAM. Calculate the fourth power spectrum. If there is no single spectrum component in the fourth power spectrum, it is 8PSK.

[0123] The setting of the threshold is obtained according to the simulation results. For example, the simulation diagram of the C4 parameter is shown in the figure below. The C4 parameter of the constant envelope signal is close to 0, and the C4 parameter of the non-constant envelope is close to 100. Therefore, the threshold is set to 50 to distinguish the constant envelope signal from the non-constant envelope signal. A threshold value in an embodiment of the present invention is as follows: the thresholds C1th, C2th, C3th, C4th, C5th, C7th, and C8th are set to: 8, 4, 40, 50, 10, 8, and 8, respectively.

[0124] Figure 5The R parameter graph of QPSK and 16QAM in the embodiment, the threshold Rth is a function of the signal-to-noise ratio. When the snr is 8db to 18db, the threshold Rth is a diagonal line, and the thresholds at the end points of the diagonal line are 0.55 and 0.37 respectively. When the snr is greater than 18db, the threshold Rth is a constant 0.37.

[0125] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A signal recognition method based on broadband spectrum detection, characterized in that: The method comprises the following steps: Step S1, using FPGA to perform power spectrum estimation on broadband IQ data and then perform signal detection to obtain the start frequency and end frequency of the signal, and use the median of the start frequency and the end frequency as the carrier frequency; use the carrier frequency as the DDC frequency parameter, use 8 times the detection bandwidth as the DDC bandwidth parameter, configure DDC and perform digital DDC on the broadband IQ to obtain narrowband IQ; Step S2: Based on the host computer, the narrowband IQ signal is estimated by using the periodogram method to obtain the power spectrum density psd, and the noise floor is estimated by using the power spectrum histogram method to obtain the noise floor estimation value. ; Step S3, calculate the ratio of the maximum value to the second maximum value of the power spectrum density psd, recorded as parameter C1, when parameter C1 is greater than threshold C1th, calculate parameters C2 and C3; if parameter C2 is greater than threshold C2th, and parameter C3 is greater than threshold C3th, it is determined to be an AM signal, otherwise it is determined to be a single carrier signal; wherein parameter C2 represents the ratio of the spectrum mean of the first k largest values ​​within a preset interval around the maximum value of the spectrum to the noise floor, and parameter C3 represents the maximum value of the spectrum density of the normalized zero-centered instantaneous amplitude; Step S4, when the parameter C1 is less than or equal to the threshold C1th, the signal is merged and screened according to the number of detected signals to obtain the target signal, and the power spectrum density psd of the target signal is detected using a double threshold signal detection algorithm to obtain the signal start frequency, end frequency and signal-to-noise ratio parameters; In step S4, signal merging and screening are performed according to the number of detected signals, including: for the kth signal, k is a natural number, and the starting frequency of the signal is , the end frequency is ,bandwidth , the in-band power is , the signal-to-noise ratio is , the estimated noise floor is , power similarity factor , indicating the The signal and Signal power ratio; Bandwidth similarity factor: ; indicates the The signal and The signal bandwidth ratio; Distance factor between signals: ; indicates the The signal and The spectral distance of the signal; The power spectral density between the two signals ; represents the power spectral density value of the middle frequency point between the starting frequency point of signal i+1 and the ending frequency point of signal i; When two signals are detected, the two signals are recorded as the first signal and the second signal; like , or , Then the two signals are merged into one signal, and the starting frequency point of the merged signal is recorded as the starting frequency point of the first signal before the merge, and the ending frequency point is recorded as the ending frequency point of the second signal before the merge; like ,but If the above merging conditions are not met, the distance from the signal to zero frequency is calculated. ,in represents the number of Fourier transform points, k represents the kth signal, and Smaller signal; If not satisfied , then the signal with larger power is retained; In the step S4, signal merging and screening are performed according to the number of detected signals, including: when three signals are detected, the three signals are recorded as the first signal, the second signal and the third signal respectively; like , then the starting frequency and ending frequency of the combined signal are , indicating that the starting frequency point extends to the left by a small signal width; like , then the starting frequency and ending frequency of the combined signal are , indicating that the end frequency point extends to the right by a small signal width; like , then the three signals are combined and expressed as ; In step S4, signal merging and screening are performed according to the number of detected signals, including: when more than 3 signals are detected, the merging and screening method, assuming that the number of detected signals is N, performs the following steps: Step a), find the signal with the highest power among all signals, assuming it is , compare the bandwidth and power factor of the remaining signals with the kmax-th signal in turn, and find the smallest Signals: ； If the above formula If it is less than the preset value, the The signal and signal and discard other signals, otherwise proceed to the next step; Step b), if the above merging conditions are not met, The adjacent signals outside the range are merged one by one, and the merged signals are used as the first Repeat step a) for each signal. If the signal still cannot be merged after repeating step a), proceed to step c); Step c), all N signals are sorted by power, the top three signals are selected, the distances of the three signals to zero frequency are compared, the signal with the smallest zero frequency distance is selected as the final signal and the other signals are discarded; Step S5, frequency offset compensation and low-pass filtering are performed on the narrowband IQ data according to the center frequency and bandwidth to obtain IQ_filter; the maximum spectral density of the normalized zero-center instantaneous amplitude is calculated and recorded as parameter C4; when parameter C4 is less than threshold C4th, modulation type discrimination is performed on IQ_filter to identify FM, MSK and FSK signals; when parameter C4 is greater than or equal to threshold C4th, modulation type discrimination is performed on IQ_filter to identify BPSK, QPSK, 8PSK and 16QAM signals.

2. The signal recognition method based on broadband spectrum detection according to claim 1, characterized in that: The calculation method of the parameter C1 is: search for the maximum value psd_max of psd and its position max_i in the range of Nfft / 4 to 3Nfft / 4, set the psd value in the interval of max_i-2 to max_i+2 to 0, search for the maximum value of the remaining psd again and mark it as the second largest value psd_max2, and calculate C1=psd_max / psd_max2.

3. The signal recognition method based on broadband spectrum detection according to claim 2, characterized in that: In step S5, when C4 is less than the threshold C4th, the modulation type determination method includes: Perform frequency demodulation on IQ_filter and perform sign judgment on the demodulation result to obtain a ±1 sequence, take the absolute value after taking the difference of the sequence, calculate the absolute value spectrum of the differential frequency, and detect the single spectrum prominence parameter C5 of this spectrum; When C5 is less than the threshold C5th, it is determined to be an FM signal; When C5 is greater than or equal to the threshold C5th, the distance D1 from the single spectrum position to the middle position is recorded, and the square spectrum modulus of IQ_filter is taken to calculate the distance D2 between the maximum value and the second largest value of the spectrum; When the difference between D1 and D2 is within the preset error range, it is considered that D1=D2 and the signal is an MSK signal, otherwise the signal is determined to be an FSK signal.

4. The method according to claim 3, characterized in that In step S5, when C4 is greater than or equal to the threshold C4th, the modulation type determination method includes: The square spectrum of IQ_filter is modulo and recorded as C7. When there is a single spectrum component, that is, C7>C7th, it is determined to be a BPSK signal. When there is no single spectrum component, the calculated fourth power spectrum is recorded as C8. When there is no single spectrum component in the fourth power spectrum, that is, C8 < C8th, it is determined to be an 8PSK signal. When there is a single spectrum component in the fourth power spectrum, the ratio parameter R of the square of the amplitude to the square of the mean of the square of the amplitude is calculated. The corresponding decision threshold is selected according to the current signal-to-noise ratio. The decision threshold is a broken line function of the signal-to-noise ratio. When the parameter R is less than the threshold Rth, it is determined to be a QPSK signal, otherwise it is determined to be a 16QAM signal.

5. The method according to claim 4, characterized in that The thresholds C1th, C2th, C3th, C4th, C5th, C7th, and C8th are set to 8, 4, 40, 50, 10, 8, and 8, respectively.

6. The method according to claim 5, characterized in that The threshold Rth is a function of the signal-to-noise ratio. When the SNR is between 8db and 18db, the threshold Rth is a diagonal line, and the thresholds at the end points of the diagonal line are 0.55 and 0.37 respectively. When the SNR is greater than 18db, the threshold Rth is a constant 0.37.

Images