A communication signal recognition system and method based on complex similarity

Abstract

The application discloses a communication signal recognition system and method based on complex similarity, and belongs to the field of signal processing. The system comprises the following steps: obtaining the maximum value of zero-centered normalized instantaneous amplitude spectrum density of a signal; classifying the signal into an AM signal and several other signals according to the maximum value of zero-centered normalized instantaneous amplitude spectrum density; classifying the remaining signals into a CW signal, a 2FSK signal, an FM signal, a BPSK signal, a QPSK signal and an 8PSK signal according to the proportion of over-threshold spectral peak values; distinguishing the CW signal and the 2FSK signal according to the spectral peak value and the spectrum center difference value of the signal; distinguishing the 2FSK signal through the number of over-threshold left and right spectrums; and distinguishing the FM signal, the BPSK signal, the QPSK signal and the 8PSK signal through phase similarity, and the BPSK signal, the QPSK signal and the 8PSK signal can also be distinguished through phase similarity. The system and method solve the problems of the existing algorithm, such as a small number of recognized signal types, low recognition accuracy and high complexity, and improve the recognition effect of the communication signal recognition under the influence of noise and under the condition of more modulation signals.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and in particular to a communication signal recognition system and method based on complex similarity. Background Technology

[0002] Modulation identification (MDI) technology is a crucial component of communication systems, aiming to identify the modulation scheme used by analyzing the characteristics of received signals. This process provides essential information for subsequent signal demodulation, ensuring accurate data recovery. MDI is widely used in complex communication environments, such as battlefield communications, satellite telemetry and control, and radio communications. In these scenarios, signals may be subject to various interferences, including the Doppler effect and nonlinear noise; therefore, MDI technology needs to possess high robustness and anti-interference capabilities. Improvements in modulation algorithms not only enhance the accuracy of signal identification but also adapt to diverse communication needs, improving the overall efficiency and reliability of the system, making it an indispensable part of modern communication technology.

[0003] In existing technologies, the methods used to distinguish between BPSK, QPSK, and 8PSK typically involve evaluating higher-order accumulators. These higher-order accumulators primarily reflect the phase information within the higher-order moments. Therefore, down-conversion of the signal and subsequent higher-order accumulator transformation of the baseband signal are necessary to accurately obtain the higher-order moments. However, due to its low robustness to non-Gaussian noise, inaccurate down-conversion can lead to severe distortion of the higher-order accumulator values. Thus, evaluating higher-order accumulators is a complex process.

[0004] While traditional modulation recognition methods perform poorly in complex electromagnetic environments, they offer several advantages: high stability, simple and easy-to-implement algorithms, and low computational resource requirements. In contrast, deep learning methods suffer from drawbacks: they require extensive labeled data for training, resulting in long training times, high computational resource consumption, and are prone to overfitting or decreased accuracy when data is insufficient or noise is prevalent. Therefore, traditional methods remain crucial in scenarios with limited resources and high real-time requirements. Summary of the Invention

[0005] The purpose of this invention is to provide a communication signal recognition system and method based on complex similarity to solve the problems existing in the background technology.

[0006] To achieve the above objectives, the present invention provides a communication signal recognition method based on complex similarity, comprising the following steps:

[0007] S1. Generate the baseband I and Q signals of each communication signal and obtain the modulation signal to be identified;

[0008] S2. Obtain the maximum value of the zero-center normalized instantaneous amplitude spectral density γmax of the signal through calculation;

[0009] S3. Set an appropriate threshold that can distinguish AM signals, so that the signal can be classified into AM signals and other signals;

[0010] S4. Obtain the percentage of peak values ​​of the remaining signals that cross the threshold spectrum;

[0011] S5. Based on the peak ratio of the spectrum above the threshold, the remaining major categories of signals are divided into wideband signals and narrowband signals. Setting an appropriate threshold can distinguish between the two categories. Narrowband signals are CW and 2FSK, while wideband signals are FM, BPSK, QPSK, and 8PSK.

[0012] S6. Calculate the peak value and center value of the spectrum of the narrowband signal, and the number of left and right spectrum crossing the threshold;

[0013] S7. Set a threshold that can distinguish the peak value and center value difference of the spectrum of the CW signal in the narrowband signal, and then determine the 2FSK modulated signal by the number of left and right spectrum crossing the threshold.

[0014] S8. Obtain the phase similarity of the broadband signal. Based on the phase similarity set in the broadband signal, set a threshold for the phase similarity to distinguish between FM, BPSK, QPSK, and 8PSK.

[0015] Preferably, the signal modulation schemes identified using this method in S1 include: AM, FM, CW, 2FSK, BPSK, QPSK, and 8PSK.

[0016] Preferably, in S2, the maximum value of the zero-center normalized instantaneous amplitude spectral density γmax of the signal is obtained, and the specific steps are as follows:

[0017] S21. First, calculate the normalized instantaneous amplitude of the signal:

[0018]

[0019] Where N is the length of the input signal, μ A The average amplitude of the signal is represented by n, where n represents the index of the sampling point of the signal.

[0020] S22. Calculate the square spectrum of the normalized instantaneous amplitude:

[0021]

[0022] S23. Calculate the normalized maximum value:

[0023]

[0024] Preferably, in S3, a suitable threshold value for distinguishing AM signals is set. The specific steps include:

[0025] The calculated γ max >γ AM_max , then the signal modulation mode is AM modulation signal;

[0026] If the calculated γ max <γ AM_max , then the signal modulation mode is other modulation signals;

[0027] where γ AM_max is set to 2000.

[0028] Preferably, in S4, the proportion of the spectrum peak value exceeding the threshold for other signals is obtained. The specific steps include:

[0029] S41. Perform FFT on the input signal and shift the spectrum of the result so that the spectrum center is in the middle;

[0030] S42. Calculate the logarithm of the spectrum amplitude, convert the spectrum amplitude to decibel dB units for subsequent threshold processing;

[0031] S43. Find the maximum value of the spectrum amplitude and subtract 10 dB from it as the threshold for the peak level; this threshold is used to identify the peak points in the spectrum;

[0032] S44. Find the spectrum points exceeding the peak level and calculate the proportion of these points:

[0033]

[0034] Preferably, in S5, the steps for distinguishing other major signals into wideband signals and narrowband signals according to the proportion of the spectrum peak value exceeding the threshold are as follows:

[0035] S51. If spec_peak >= spec_level, the signal is a narrowband signal;

[0036] S52. If spec_peak < spec_level, the signal is a wideband signal;

[0037] where spec_level is set to 0.0002.

[0038] Preferably, the steps in S6 are as follows:

[0039] S61. Calculate the maximum value within the center spectrum range:

[0040] center_fre = max(abs_fft_sig[st:ed]);

[0041] S62. Calculate the difference threshold between the spectral peak and the spectral center:

[0042] sub_peak = ceter_fre - max(abs_fft_sig);

[0043] S63. Find the position of the spectral peak and count the number of spectral peaks in the left half and the right half respectively. The formula is:

[0044]

[0045] Preferably, in S7, the steps are as follows:

[0046] S71. Set the difference threshold between the spectral peak and the spectral center:

[0047] If sub_peak < sub_level, the narrowband signal is a CW signal; otherwise, determine whether it is a 2FSK signal. Here, sub_level is set to 10.

[0048] S72. Make a decision based on the counted number of spectral peaks in the left half and the right half:

[0049] If left num > 0 and right num > 0, then determine that the signal is a 2FSK signal.

[0050] Preferably, the specific steps of S8 are as follows:

[0051] S81. First, calculate the unwrapped phase angle and the phase difference of the signal;

[0052] S82. Set the cumulative sum of the phase difference within a window, accumulate the phase change rate within each window, represent the integral value of the phase change within the window, and take its absolute value;

[0053] S83. Set a window with a length twice as long. For each point, determine whether it is a peak within the sliding window:

[0054] If the value of the current point is less than any value within its neighborhood or the value of the current point is less than the threshold, it is not a peak; otherwise, it is a peak; [[ID=%44]]

[0055] S84. Calculate the Euclidean distance between the detected peak and the expected peaks of different PSK modulations (BPSK, QPSK, 8PSK). Among them, the smaller the distance, the higher the similarity;

[0056] S85. Set the threshold of the phase similarity to distinguish FM, BPSK, QPSK, 8PSK. The specific steps are as follows:

[0057] S851. Determine the PSK modulation type based on similarity;

[0058] S852. Further determine based on similarity. Set a threshold phdiff_level for further confirming the recognition result. For the similarity discrimination of the broadband signal, if the similarity phdiff_peak < phdiff_level, then determine the PSK modulation type according to the above similarity discrimination, otherwise the broadband signal is an FM signal, where phdiff_level is set to 0.33.

[0059] A communication signal recognition system based on complex similarity includes:

[0060] A wide / narrowband signal discrimination module that obtains the γ max value of the signal, discriminates between AM signals and other signals, calculates the proportion of the spectral peak above the threshold for the other signals, and obtains the wide / narrowband signal;

[0061] A narrowband signal extraction module that calculates the maximum value within the central spectral range of the narrowband signal, as well as the spectral peak and the spectral center difference threshold, and identifies the obtained narrowband signal based on the calculated results;

[0062] A broadband signal extraction module that sets a sliding window, calculates the phase similarity of the broadband signal, and sets a threshold based on the obtained phase similarity to distinguish and identify the signal.

[0063] Therefore, the present invention adopts the above communication signal recognition system and method based on complex similarity, and has the following beneficial effects:

[0064] (1) It can accurately identify various communication signals in a complex environment;

[0065] (2) It is applicable to signal processing and communication systems in complex electromagnetic environments. It can still maintain high recognition performance in the case of mixed signals and strong noise interference, providing important support for the development of communication technology;

[0066] (3) For BPSK, QPSK, and 8PSK modulation methods, a recognition method based on phase similarity is proposed, which improves the recognition accuracy of the above several communication signals to a certain extent.

[0067] The technical solution of the present invention will be further described in detail below through the accompanying drawings and embodiments. Description of the Drawings

[0068] Figure 1 It is a specific flowchart of a communication signal recognition method based on complex similarity implemented in an embodiment of the present invention;

[0069] Figure 2 This is a schematic diagram of the structure of a communication signal recognition system based on complex similarity according to an embodiment of the present invention;

[0070] Figure 3 This is a graph showing the calculated maximum value of the zero-center normalized instantaneous amplitude spectral density of the communication signal as a function of the signal-to-noise ratio in an embodiment of the present invention.

[0071] Figure 4 This is a graph showing the spectral peak ratio (PSRR) of narrowband and wideband signals as a function of signal-to-noise ratio, as shown in an embodiment of the present invention.

[0072] Figure 5 This is an embodiment of the present invention, showing a curve illustrating the change in phase similarity values ​​of several PSKs with signal-to-noise ratio. Detailed Implementation

[0073] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0074] Please see Figure 1 A communication signal recognition method based on complex similarity includes the following steps:

[0075] S1. Generate the I and Q baseband signals of each communication signal to obtain the modulation signal to be identified. The signal modulation modes identified using the method of this embodiment include: AM, FM, CW, 2FSK, BPSK, QPSK, and 8PSK.

[0076] S2. Obtain the maximum value of the zero-center normalized instantaneous amplitude spectral density γmax of the signal by calculation.

[0077] The specific steps to obtain the maximum value of the zero-center normalized instantaneous amplitude spectral density γmax of the signal are as follows:

[0078] S21. First, calculate the normalized instantaneous amplitude of the signal:

[0079]

[0080] Where N is the length of the input signal, μ A The average amplitude of the signal is represented by n, where n represents the index of the sampling point of the signal.

[0081] S22. Calculate the square spectrum of the normalized instantaneous amplitude:

[0082]

[0083] S23. Calculate the normalized maximum value:

[0084]

[0085] S3. Set an appropriate threshold that can distinguish AM signals, so that signals can be classified into AM signals and other signals.

[0086] Setting a suitable threshold that can distinguish AM signals involves the following steps:

[0087] Calculated γ max >γ AM_max The signal modulation method is AM modulation.

[0088] If the calculated γ max <γ AM_max If so, the signal modulation method is the other modulation signal;

[0089] Where γ AM_max Set it to 2000.

[0090] S4. Obtain the percentage of peak values ​​of the spectrum of the remaining signals that cross the threshold.

[0091] To obtain the percentage of peak values ​​of the remaining signals that cross the threshold, the specific steps include:

[0092] S41. Perform FFT on the input signal and then perform spectral shift on the result so that the spectral center is in the middle.

[0093] S42. Calculate the logarithm of the spectral amplitude and convert the spectral amplitude to decibels (dB) for subsequent threshold processing.

[0094] S43. Find the maximum value of the spectral amplitude and subtract 10dB from it to obtain the threshold for the peak level; this threshold is used to identify peak points in the spectrum.

[0095] S44. Find the spectral points that exceed the peak level and calculate the proportion of these points:

[0096]

[0097] S5. Based on the peak ratio of the spectrum above the threshold, the remaining major categories of signals are divided into wideband signals and narrowband signals. Setting an appropriate threshold can distinguish between the two categories. Narrowband signals are CW and 2FSK, while wideband signals are FM, BPSK, QPSK, and 8PSK.

[0098] The steps to classify other major signal categories into wideband and narrowband signals based on the proportion of peak values ​​across the threshold spectrum are as follows:

[0099] S51. If spec_peak >= spec_level, then the signal is a narrowband signal;

[0100] S52. If spec_peak < spec_level, the signal is a wideband signal;

[0101] where spec_level is set to 0.0002.

[0102] The signals are classified into wideband signals and narrowband signals according to the proportion of the spectral peak crossing the threshold. The implementation of this method is based on the significant differences in spectral characteristics between the two types of signals. Since the spectral range of narrowband signals is relatively narrow, it is usually concentrated within a specific frequency band, with a fixed center frequency and a small spectral bandwidth. While the spectral range of wideband signals is wide, covering a wide range of frequencies and can carry multiple narrowband signals simultaneously.

[0103] Since the spectral energy of narrowband signals is concentrated at a few frequency points, while the spectral energy of wideband signals is distributed over a wide frequency range. Therefore, by reasonably setting the threshold value, the differences in the proportion of spectral peaks between the two types of signals can be accurately reflected.

[0104] S6. Calculate the difference between the spectral peak and the spectral center of the narrowband signal, and the number of spectral thresholds crossed on the left and right.

[0105] S61. Calculate the maximum value within the central spectral range:

[0106] cebter_fre = max(abs_fft_sig[st:ed]);

[0107] S62. Calculate the threshold for the difference between the spectral peak and the spectral center:

[0108] sub_peak = ceter_fre - max(abs_fft_sig);

[0109] S63. Find the position of the spectral peak, and separately count the number of spectral peaks in the left and right halves. The formula is:

[0110]

[0111] S7. Set the threshold for the difference between the spectral peak and the spectral center that can distinguish the CW signal in the narrowband signal, and then determine the 2FSK modulation signal based on the set number of spectral thresholds crossed on the left and right.

[0112] S71. Set the threshold for the difference between the spectral peak and the spectral center:

[0113] If sub_peak < sub_level, the narrowband signal is a CW signal, otherwise, determine whether it is a 2FSK signal. Where sub_level is set to 10;

[0114] S72. Make a judgment based on the number of spectral peaks in the left and right halves of the statistics:

[0115] If left num >0 and right num If the value is greater than 0, then the signal is determined to be a 2FSK signal.

[0116] For CW signals, because their spectrum is usually relatively concentrated, the spectral peaks are relatively high, and the spectral center difference may be small. Therefore, a small spectral center difference threshold can be set to identify CW signals.

[0117] For 2FSK signals, 2FSK signal modulation will generate two carrier frequencies (usually high frequency and low frequency), so there will be two obvious peaks in the spectrum.

[0118] S8. Obtain the phase similarity of the broadband signal. Based on the phase similarity set in the broadband signal, set a threshold for phase similarity to distinguish between FM, BPSK, QPSK, and 8PSK. The specific steps are as follows:

[0119] S81. First, calculate the unwrapping phase angle and phase difference of the signal;

[0120] S82. Set the cumulative sum of phase differences within a window, accumulate the phase change rate within each window, represent the integral value of the phase change within the window, and calculate its absolute value.

[0121] S83. Set a window with a length twice that of the point. For each point, determine whether it is a peak value within the sliding window:

[0122] If the value of the current point is less than any value in its neighborhood or less than the threshold, then it is not the peak value; otherwise, it is the peak value.

[0123] S84. Calculate the Euclidean distance between the detected peak and the expected peak of different PSK modulations (BPSK, QPSK, 8PSK), where the smaller the distance, the higher the similarity.

[0124] S85. Set a threshold for phase similarity to distinguish between FM, BPSK, QPSK, and 8PSK. The specific steps are as follows:

[0125] S851. Determine the PSK modulation type based on similarity;

[0126] S852. Further judgment is made according to the similarity. A threshold value phdiff_level is set for further confirming the recognition result. For the similarity discrimination of the broadband signal, if the similarity phdiff_peak < phdiff_level, the PSK modulation type is discriminated according to the above similarity discrimination, otherwise the broadband signal is an FM signal, where phdiff_level is set to 0.33.

[0127] Such as Figure 2 , a communication signal recognition system based on complex similarity, including:

[0128] A wide / narrowband signal discrimination module, which obtains the γ max value of the signal, discriminates AM signals from the rest of the signals, and calculates the proportion of the spectral peak above the threshold of the rest of the signals to obtain the wide / narrowband signal.

[0129] A narrowband signal extraction module, which calculates the maximum value within the central spectrum range of the narrowband signal, as well as the threshold of the difference between the spectral peak and the spectral center, and identifies the obtained narrowband signal according to the obtained results.

[0130] A broadband signal extraction module, which sets a sliding window, calculates the phase similarity of the broadband signal, and sets a threshold value according to the obtained phase similarity to distinguish and identify the signal.

[0131] The communication signals generated by the solution of this embodiment, the set of modulation methods of the test signals includes {CW, AM, FM, 2FSK, BPSK, QPSK, 8PSK}, where the AM modulation index is 0.3.

[0132] From Figure 3 it can be seen that for the seven modulation signals studied in this embodiment, under the influence of noise within the range of 0 - 20 dB, by obtaining the maximum value of the zero-centered normalized instantaneous amplitude spectral density of the signal, the AM signal can be clearly identified from the rest of the communication signals.

[0133] At the same time, from Figure 4 it can be seen that the method proposed in this embodiment can identify the obtained broadband signal and narrowband signal under the same settings as above.

[0134] From Figure 5 it can be seen that even under poor conditions, the method proposed in this embodiment can still identify several PSK modulation signals with low noise ratio through the calculation of similarity eigenvalue.

[0135] The method proposed in this embodiment has achieved good results in the recognition rate of each modulation signal under all modulation methods for the seven modulation signals studied.

[0136] Therefore, this invention employs a communication signal recognition system and method based on complex similarity, which, by integrating multiple feature extraction techniques, can accurately identify various communication signals in complex environments. This technology is applicable to signal processing and communication systems in complex electromagnetic environments, maintaining high recognition performance even under conditions of mixed signals and strong noise interference, thus providing important support for the development of communication technology.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A communication signal recognition method based on complex similarity, characterized in that, It comprises the following steps: S1, generating each communication signal baseband I, Q two-way signal, obtaining the modulation signal to be identified; S2, obtaining the zero center normalized instantaneous amplitude spectrum density maximum value γmax of the signal by calculation; the specific steps are as follows: S21, first, calculate the normalized instantaneous amplitude of the signal: where N is the length of the input signal; μ A denotes the average amplitude of the signal; n denotes the sample point index of the signal; S22, calculate the square spectrum of the normalized instantaneous amplitude: S23, calculate the normalized maximum value: S3, set a suitable threshold value that can distinguish AM signals, so that the signal can be classified as AM signal and the remaining signal; S4, obtain the over-threshold spectral peak value ratio of the remaining signal; S5, according to the over-threshold spectral peak value ratio, distinguish the remaining large category signal into wideband signal and narrowband signal, set the appropriate threshold value to distinguish the two categories, narrowband signal is CW, 2FSK, wideband signal is FM, BPSK, QPSK, 8PSK; S6, calculate the spectral peak value and spectral center difference value of the narrowband signal, and the number of left and right spectral over-threshold values; S7, set the spectral peak value and spectral center difference value threshold value that can distinguish the CW signal in the narrowband signal, and then determine the 2FSK modulation signal through the set left and right spectral over-threshold number; S8, obtain the phase similarity of the wideband signal, and set the phase similarity threshold value according to the set phase similarity of the wideband signal, to distinguish FM, BPSK, QPSK, 8PSK.

2. The communication signal recognition method based on complex similarity according to claim 1, characterized in that, The signal modulation mode identified in S1 includes: AM, FM, CW, 2FSK, BPSK, QPSK, 8PSK.

3. The method of claim 2, wherein the method further comprises: determining a similarity between the first and second signals; and determining a complexity of the first and second signals. In S3, a suitable threshold value that can distinguish AM signals is set, and the specific steps include: Calculated γ max >γ AM_max If the signal modulation mode is AM modulation signal; If the calculated γ max <γ AM_max If the calculated γ then the signal modulation mode is the remaining modulation signal; where γ AM_max is set to 2000.

4. The communication signal recognition method based on complex similarity according to claim 3, characterized in that, In S4, the over-threshold spectral peak value ratio of the remaining signal is obtained, and the specific steps include: S41, perform FFT on the input signal, and perform spectral shift on the result, so that the spectral center is located in the middle; S42, calculate the logarithm of the spectral amplitude, and convert the spectral amplitude to decibel dB unit; S43, find the maximum value of the spectral amplitude, and subtract 10dB from it as the threshold value of the peak level; S44, find the spectral points exceeding the peak level, and calculate the proportion of the spectral points exceeding the peak level:

5. The method of claim 4, wherein the method further comprises: In S5, according to the over-threshold spectral peak value ratio, the remaining large category signal is distinguished into wideband signal and narrowband signal, and the steps are as follows: S51, if spec_peak >= spec_level, the signal is narrowband signal; S52, if spec_peak < spec_level, the signal is wideband signal; Wherein, spec_level is set to 0.0002.

6. The communication signal recognition method based on complex similarity according to claim 5, characterized in that: In S6, the steps are as follows: S61, calculate the maximum value in the center spectral range: center_fre = max(abs_fft_sig[st:ed]); S62, calculate the spectral peak value and spectral center difference threshold value: sub_peak = ceter_fre - max(abs_fft_sig); S63, find the position of the spectral peak value, and respectively count the number of spectral peak values in the left half and the right half, the formula is:

7. The method of claim 6, wherein the method further comprises: determining a similarity between the first and second signals; and determining a complexity of the first and second signals. In S7, the steps are as follows: S71, set the spectral peak value and spectral center difference threshold value: If sub_peak<sub_level, the narrowband signal is a CW signal, otherwise, further determine whether it is a 2FSK signal, wherein sub_level is set to 10; S72, determine according to the number of spectral peak values of the left half and the right half of the spectrum; If left num > 0 and right num > 0, then the signal is determined to be a 2FSK signal.

8. The communication signal recognition method based on complex similarity according to claim 4, characterized in that: In S8, the phase similarity of the wideband signal is obtained, and the specific steps are as follows: S81, first calculate the unwrapped phase angle and the phase difference value of the signal; S82, set a window for the cumulative sum of the phase difference, and accumulate the phase change rate in each window to represent the integral value of the phase change in the window and calculate its absolute value; S83, set a window with a length of twice, for each point, determine whether it is a peak value in the sliding window: If the value of the current point is less than any value in its neighborhood or the value of the current point is less than the threshold value, it is not a peak value, otherwise, it is a peak value; S84, calculate the Euclidean distance between the detected peak value and the expected peak value of different PSK modulation, wherein the smaller the distance, the higher the similarity; S85, set a threshold value of the phase similarity to distinguish FM, BPSK, QPSK and 8PSK, and the specific steps are as follows: S851, determine the PSK modulation type according to the similarity; S852, further determine according to the similarity, set a threshold value phdiff_level for further confirming the recognition result, and determine the similarity of the wideband signal, if the similarity phdiff_peak<phdiff_level, determine the PSK modulation type according to the similarity, otherwise, the wideband signal is an FM signal, wherein phdiff_level is set to 0.

33.

9. A system for applying a complex similarity based communication signal recognition method according to any one of claims 1 to 8, characterized in that, It includes: Wide / narrowband signal differentiation module, acquires the gamma of the signal. max The value is used to distinguish AM signals from other signals, and the percentage of peak values ​​of the spectrum of the other signals that cross the threshold is calculated to obtain the wide / narrow band signal. A narrowband signal extraction module, which calculates the maximum value in the center spectrum range of the narrowband signal and the spectral peak value and the spectral center difference threshold value, and identifies the narrowband signal according to the obtained results; A wideband signal extraction module, which sets a sliding window, calculates the phase similarity of the wideband signal, sets a threshold value according to the obtained phase similarity, and distinguishes and identifies the signal.

Images