Underwater sound DSSS signal identification method combining correlation peak and secondary power spectrum periodic impulse characteristics

Abstract

The invention discloses a correlation peak and secondary power spectrum periodic impulse characteristic combined underwater sound DSSS signal identification method, which comprises the following steps of: carrying out self-correlation on a received suspected communication signal to obtain a signal self-correlation waveform R (m); searching a second peak point from the signal autocorrelation waveform R (m); extracting a recognition feature f1 through the second peak point; performing signal processing on the suspected communication signal to obtain a signal secondary power spectrum; extracting spectrum peaks in the signal secondary power spectrum to form a spectrum peak set; extracting an impulse point position set in the spectrum peak set and a difference value set of impulse point positions; performing linear fitting on the difference value set of the impulse point positions; extracting a recognition feature f2 according to a linear fitting result; and realizing underwater sound DSSS signal identification by combining the identification feature f1 and the identification feature f2. According to the method, the defect of insufficient reliability of single feature recognition can be made up by combining multiple feature recognition, and meanwhile, misjudgment on DSSS signals can be greatly reduced by the newly added secondary power spectrum features.

Description

Technical Field

[0001] This invention belongs to the field of underwater communication and relates to underwater acoustic DSSS signal identification technology, specifically to an underwater acoustic DSSS signal identification method that combines correlation peaks and periodic impulse characteristics of the secondary power spectrum. Background Technology

[0002] Underwater acoustic spread spectrum (DSSS) utilizes pseudo-random codes and other spread spectrum techniques to achieve significant processing gain. It possesses excellent resistance to multipath interference, anti-interference capabilities, and stealth, giving it a clear advantage over other modulation methods in long-distance underwater communication and leading to its widespread application in various underwater communication scenarios. Identifying underwater acoustic DSSS signals is a crucial aspect of underwater information warfare. Based on the identification results, and by leveraging the characteristics of the DSSS signal, it is possible to further estimate typical parameters such as intermediate frequency, bandwidth, and pseudo-code period, providing the technical prerequisites for conducting underwater acoustic countermeasures.

[0003] Compared to other types of communication signals, such as underwater acoustic orthogonal frequency division multiplexing (OFDM), underwater acoustic DSSS signals are more difficult to identify, and there are fewer identification methods. Because underwater acoustic DSSS signals are typically modulated using pseudo-random codes, autocorrelation of the DSSS signal will generate multiple correlation peaks. Existing technologies for identifying underwater acoustic DSSS signals usually extract features from the autocorrelation waveform of the DSSS signal. However, in underwater acoustic communication modulation schemes, the autocorrelation waveform of underwater acoustic OFDM signals may also contain multiple correlation peaks. This can lead to misclassification of signals with multiple autocorrelation peaks, such as underwater acoustic OFDM signals, as DSSS. Therefore, underwater acoustic DSSS signal identification methods based on a single correlation peak feature lack robustness. Summary of the Invention

[0004] Purpose of the invention: In order to overcome the shortcomings of the existing technology, this invention provides a method for identifying underwater acoustic DSSS signals by combining correlation peaks and periodic impulse characteristics of the secondary power spectrum. This method can effectively reduce the false judgment rate of DSSS signals, improve the reliability of underwater acoustic DSSS signal identification, and provide a new approach for feature extraction in underwater acoustic DSSS identification.

[0005] Technical Solution: To achieve the above objectives, this invention provides a method for identifying underwater acoustic DSSS signals by combining correlation peaks and periodic impulse characteristics of the secondary power spectrum, comprising the following steps:

[0006] S1: Perform autocorrelation on the received suspected communication signal to obtain the signal autocorrelation waveform R(m);

[0007] S2: Search for the second peak point from the signal autocorrelation waveform R(m);

[0008] S3: Extract and identify feature f1 through the second peak point;

[0009] S4: Perform signal processing on the suspected communication signal to obtain the secondary power spectrum of the signal;

[0010] S5: Extract the spectral peaks from the secondary power spectrum of the signal to form a set of spectral peaks;

[0011] S6: Extract the set of impulse point positions and the set of differences between impulse point positions from the set of spectral peaks;

[0012] S7: Perform linear fitting on the set of differences in the position of the impulse point;

[0013] S8: Based on the linear fitting results, extract the recognition feature f2;

[0014] S9: Combine the identification features f1 and f2 to realize underwater acoustic DSSS signal identification.

[0015] Furthermore, the calculation of the signal autocorrelation waveform R(m) in step S1 is expressed as follows:

[0016]

[0017] Where x(n) is the suspected communication signal and N is the signal length.

[0018] Further, step S2 includes: In order to reduce the influence of the main autocorrelation peak on the search for the second correlation peak, the main peak needs to be removed when searching for the second correlation peak. An empirical value can be set for the search starting point in the autocorrelation waveform. Assuming the starting point is set to t_start, the signal sampling rate is fs, and the starting point for searching the signal autocorrelation waveform R(m) is m_start = floor(t_start * fs), the position n_max of the second peak point and the peak value C_max are obtained from this starting point.

[0019] Further, step S3 includes: based on the characteristics of the signal autocorrelation peak, setting a relatively large interval [n_max-τ] centered on the position n_max of the second peak point. max ,n_max+τ max ] and a smaller interval [n_max-τ min ,n_max+τ min The principle for setting larger intervals is τ. max The shorter interval is less than the length between two related peaks; the principle for setting a smaller interval is τ. min If the autocorrelation waveform R(m) is greater than half the duration of the correlation peak, the signal autocorrelation waveform R(m) will be in a smaller interval [n_max-τ]. min ,n_max+τ min The peak value of ] is set to zero, and then the maximum peak value C_new_max is searched again in a larger interval to identify the feature f1 = C_max / C_new_max.

[0020] Further, step S4 specifically includes: first, segmenting the suspected communication signal, and then calculating the secondary power spectrum; first, performing a first power spectrum calculation on each signal segment, and then performing a second power spectrum calculation on the obtained power spectrum to obtain the secondary power spectrum of each signal segment; superimposing the secondary power spectra of each signal segment to obtain the final obtained signal secondary power spectrum PSD, with a length of M.

[0021] Further, step S5 specifically includes: setting the sliding window size N_win according to the signal sampling frequency, sliding the window sequentially from low frequency to high frequency in the secondary power spectrum in a non-overlapping manner, the total number of sliding is num_win = floor(M / N_win), and extracting the maximum spectral value sequentially in each window to form a spectral peak set PSD_max.

[0022] Further, step S6 specifically includes: setting a multiple threshold, the extracted impulse point needs to satisfy that the peak value of the point is higher than the average value of the interval near the point multiplied by the multiple threshold, the impulse point positions that satisfy the condition are saved to form an impulse point position set dL, assuming that the length of dL is K, then the impulse point position difference set DL=[dL(2)-dL(1),dL(3)-dL(1),...,dL(K)-dL(1)], and the length of DL is K-1.

[0023] Furthermore, step S8 specifically includes:

[0024] The coefficient of determination R 2 As a recognition feature f2, the recognition coefficient If the linearity of DL_fit is higher, R 2 The closer it is to 1, the closer it is to 0. Since the secondary power spectrum of DSSS has periodic impulse characteristics, ideally DL_fit should be close to 1, while other signals are much smaller than 1.

[0025] Furthermore, in step S9, if both identification feature f1 and identification feature f2 simultaneously satisfy a set threshold, the identification signal can be judged as a DSSS signal.

[0026] Furthermore, if no suitable impulse point location is found in step S6, or if the number of impulse point locations is too small, the identification result is directly determined to be a non-DSSS signal.

[0027] Beneficial Effects: Compared with existing technologies, this invention extracts two recognition features from signal correlation and secondary power spectrum, and combines these two features to achieve underwater acoustic DSSS signal recognition. By combining multiple feature recognition methods, the reliability of single feature recognition can be compensated for, and the newly added secondary power spectrum feature can significantly reduce the misjudgment of DSSS signals. Specifically, this invention has the following advantages:

[0028] (1) Compared with existing single feature recognition methods, the present invention adopts a combination of multiple features to achieve recognition, which is more reliable.

[0029] (2) Based on the periodic impulse characteristics of the secondary power of the DSSS signal, the present invention cleverly extracts the position of the impulse point through peak search, and obtains the set of impulse point position differences. The recognition features are set by using linear fitting, which improves the robustness of the periodic impulse characteristics of the DSSS signal, effectively reduces the misjudgment rate of the DSSS signal, and improves the reliability of underwater acoustic DSSS signal recognition. Attached Figure Description

[0030] Figure 1 This is a flowchart of the method of the present invention;

[0031] Figure 2 This is a channel impulse response diagram used in the test during this invention;

[0032] Figure 3 The image shows the autocorrelation waveform of the simulated DSSS signal.

[0033] Figure 4 The autocorrelation waveform of the simulated OFDM signal is shown.

[0034] Figure 5 The second power spectrum of the simulated DSSS signal;

[0035] Figure 6 The second power spectrum of the simulated OFDM signal;

[0036] Figure 7 The peak-to-peak values ​​of the simulated DSSS signal under all sliding windows are shown.

[0037] Figure 8 Peak-to-peak values ​​of the simulated OFDM signal under all sliding windows;

[0038] Figure 9 This is a linear fit plot of the set of impulse point position differences for the simulated DSSS signal. Detailed Implementation

[0039] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0040] Example 1:

[0041] like Figure 1 As shown, this embodiment provides a method for identifying underwater acoustic DSSS signals by combining correlation peaks and periodic impulse characteristics of the secondary power spectrum, including the following steps:

[0042] S1: Perform autocorrelation on the received suspected communication signal to obtain the signal autocorrelation waveform R(m);

[0043] The calculation of the signal autocorrelation waveform R(m) is expressed as follows:

[0044]

[0045] Where x(n) is the suspected communication signal and N is the signal length.

[0046] S2: Search for the second peak point from the signal autocorrelation waveform R(m);

[0047] To reduce the impact of the main autocorrelation peak on the search for the second correlation peak, the main peak needs to be removed when searching for the second correlation peak. Considering that the OFDM cyclic prefix length is generally related to the multipath delay to be resisted, an empirical value can be set for the search starting point in the autocorrelation waveform. This empirical value needs to be determined based on the channel scenario; an empirical value (e.g., 30ms) can be set. Figure 4 The high amplitude of the signal waveform in the middle and early part is due to channel limitations, which affects the extraction of subsequent peaks. It is necessary to set a starting point as an empirical value to filter out the part with high waveform amplitude.

[0048] Assuming the starting point is set to t_start, the signal sampling rate is fs, and the starting point for searching the signal autocorrelation waveform R(m) is m_start = floor(t_start * fs), the position n_max of the second peak point and the peak value C_max are obtained from this starting point.

[0049] S3: Extract and identify feature f1 through the second peak point;

[0050] Based on the characteristics of the signal autocorrelation peak, a relatively large interval [n_max-τ] is set with the position n_max of the second peak point as the center. max ,n_max+τ max ] and a smaller interval [n_max-τ min ,n_max+τmin The principle for setting larger intervals is τ. max The shorter interval is less than the length between two related peaks; the principle for setting a smaller interval is τ. min If the autocorrelation waveform R(m) is greater than half the duration of the correlation peak, the signal autocorrelation waveform R(m) will be in a smaller interval [n_max-τ]. min ,n_max+τ min The peak value of ] is set to zero, and then the maximum peak value C_new_max is searched again in a larger interval to identify the feature f1 = C_max / C_new_max.

[0051] S4: Perform signal processing on the suspected communication signal to obtain the secondary power spectrum of the signal;

[0052] First, the suspected communication signal is segmented, and then the secondary power spectrum is calculated.

[0053] Since the secondary power spectrum of DSSS signals not only has periodic peaks but also interference peaks, segmented processing can reduce the influence of interference peaks, thereby improving algorithm performance.

[0054] For each signal segment, a first power spectrum calculation is performed, and then a second power spectrum calculation is performed on the obtained power spectrum to obtain the second power spectrum of each signal segment. The second power spectra of each signal segment are superimposed to obtain the final obtained signal second power spectrum PSD, with a length of M.

[0055] S5: Extract the spectral peaks from the secondary power spectrum of the signal to form a set of spectral peaks;

[0056] The sliding window size N_win is set according to the signal sampling frequency. The window is slid sequentially from low frequency to high frequency in the secondary power spectrum in a non-overlapping manner. The total number of slides required is num_win = floor(M / N_win). The maximum spectral value is extracted sequentially in each window to form the spectral peak set PSD_max.

[0057] S6: Extract the set of impulse point positions and the set of differences between impulse point positions from the set of spectral peaks;

[0058] Set a multiple threshold. The extracted impulse points need to satisfy the condition that the peak value of the point is higher than the mean value of the interval near the point multiplied by the multiple threshold. Save the impulse point positions that meet this condition to form the impulse point position set dL. Assuming that the length of dL is K, the impulse point position difference set DL=[dL(2)-dL(1),dL(3)-dL(1),...,dL(K)-dL(1)], and the length of DL is K-1;

[0059] It should be noted that if no suitable impulse point location is found, or if the number of impulse point locations is too small (the number of impulse points should be greater than or equal to 4 to ensure that the difference set DL has at least 3 numbers for subsequent fitting to be effective), then the identification result is directly determined to be a non-DSSS signal, and the subsequent steps will not be executed.

[0060] S7: Perform linear fitting on the set of differences DL at the position of the impulse point;

[0061] Select a linear fitting method to perform linear fitting on DL. The result of the linear fitting is DL_fit.

[0062] S8: Based on the linear fitting results, extract the recognition feature f2;

[0063] The coefficient of determination R 2 As a recognition feature f2, the recognition coefficient If the linearity of DL_fit is higher, R 2 The closer it is to 1, the closer it is to 0. Since the secondary power spectrum of DSSS has periodic impulse characteristics, ideally DL_fit should be close to 1, while other signals are much smaller than 1.

[0064] S9: Combine identification feature f1 and identification feature f2 to realize underwater acoustic DSSS signal identification;

[0065] Because underwater acoustic DSSS signals have multiple autocorrelation peaks and the secondary power spectrum exhibits periodic impulse characteristics, identification features f1 and f2 must simultaneously meet a set threshold (the threshold for identification feature f1 is usually greater than 2, and the threshold for identification feature f2 is usually greater than 0.95) for the signal to be identified as a DSSS signal.

[0066] Example 2:

[0067] To verify the effectiveness and efficacy of the method of the present invention, the following simulation experiments and analyses were conducted in this embodiment:

[0068] The signals to be identified are OFDM and DSSS signals, with a sampling rate of 96kHz, an intermediate frequency of 6kHz, a bandwidth of 4kHz, and a pulse width of 1s. The simulated signal-to-noise ratio is set to 5dB within the [0.5kHz, 10kHz] frequency band, and the noise is based on actual environmental data collected during sea trials. The channel impulse response used in the simulation is as follows: Figure 2 As shown, the received signal x(n) after passing through the channel and noise is obtained.

[0069] The specific identification process includes the following steps:

[0070] 1) Perform autocorrelation on the received signal x(n). The autocorrelation plots for the DSSS and OFDM signals are shown below. Figure 3and Figure 4 As shown;

[0071] 2) Set t_start to 30ms, and search for the position n_max and peak value C_max of the second peak point from the starting point m_start = floor(t_start * fs), i.e. m_start = 2880. The second peak point position n_max of the DSSS signal is 13489 and C_max is 0.14. The second peak point position n_max of the OFDM signal is 9409 and C_max is 0.15.

[0072] 3) By reviewing relevant literature regarding parameter settings such as pseudocode length, the larger interval τ_max was set to 800 sampling points, and the smaller interval τ_min to 250 sampling points. A re-search yielded a maximum peak value C_new_max of 0.04 for the DSSS signal and a recognition feature f1 of 3.5. Conversely, the maximum peak value C_new_max for the OFDM signal was found to be 0.05, and the recognition feature f1 was 3.0.

[0073] 4) Calculate the second power spectrum of the DSSS and OFDM signals respectively, as shown below. Figure 5 and 6 As shown;

[0074] 5) Based on empirical values, set the sliding window size N_win to 200. Extract the maximum spectral peak value within the sliding window to form PSD_max. The PSD_max values ​​for the DSSS and OFDM signals are as follows: Figure 7 and 8 As shown;

[0075] 6) Based on the obtained DSSS signal and OFDM signal PSD_max, set the multiple threshold to 5 according to the empirical value, extract the peak points, form the impulse point position set dL, and further obtain the impulse point position difference set DL based on dL. Since no impulse point position set that meets the requirements is found in the OFDM signal PSD_max, the OFDM signal will not continue to be executed.

[0076] 7) Perform linear fitting on the set of differences DL between the impulse point positions of the DSSS signal to obtain the fitted curves, as shown below. Figure 9 As shown;

[0077] 8) Based on the fitted curve, calculate the recognition feature f2, i.e., the coefficient of determination R. 2 The coefficient of determination of the DSSS fitted curve was calculated to be 0.9999.

[0078] 9) The identification feature f1 of the DSSS signal is 3.5 (significantly higher than the nearby peak value), and the identification feature f2 is 0.9999 (consistent with the characteristics of periodic impulse). Therefore, the suspected communication signal can be identified as a DSSS signal. However, the identification feature f1 of the OFDM signal is 3.0 (significantly higher than the nearby peak value), but it does not conform to the periodic impulse characteristics of the secondary power spectrum. Therefore, the suspected communication signal can be identified as another signal.

Claims

1. A method for identifying underwater acoustic DSSS signals by combining correlation peaks and periodic impulse characteristics of the secondary power spectrum, characterized in that, Includes the following steps: S1: Perform autocorrelation on the received suspected communication signal to obtain the signal autocorrelation waveform R(m); S2: Search for the second peak point from the signal autocorrelation waveform R(m); S3: Extract and identify feature f1 through the second peak point; S4: Perform signal processing on the suspected communication signal to obtain the secondary power spectrum of the signal; S5: Extract the spectral peaks from the secondary power spectrum of the signal to form a set of spectral peaks; S6: Extract the set of impulse point positions and the set of differences between impulse point positions from the set of spectral peaks; S7: Perform linear fitting on the set of differences in the position of the impulse point; S8: Based on the linear fitting results, extract the recognition feature f2; S9: Combine identification feature f1 and identification feature f2 to realize underwater acoustic DSSS signal identification.

2. The underwater acoustic DSSS signal identification method according to claim 1, characterized in that, The calculation of the signal autocorrelation waveform R(m) in step S1 is expressed as follows: Where x(n) is the suspected communication signal and N is the signal length.

3. The underwater acoustic DSSS signal identification method according to claim 2, which combines correlation peaks and periodic impulse characteristics of the secondary power spectrum, is characterized in that... Step S2 includes: assuming the starting point is set to t_start, the signal sampling rate is fs, and the starting point for searching the signal autocorrelation waveform R(m) is m_start = floor(t_start * fs), the position n_max of the second peak point and the peak value C_max are obtained from this starting point.

4. The underwater acoustic DSSS signal identification method according to claim 3, which combines correlation peaks and periodic impulse characteristics of the secondary power spectrum, is characterized in that... Step S3 includes: based on the characteristics of the signal autocorrelation peak, setting a relatively large interval [n_max-τ] centered on the position n_max of the second peak point. max ,n_max+τ max ] and a smaller interval [n_max-τ min ,n_max+τ min The principle for setting larger intervals is τ. max The shorter interval is less than the length between two related peaks; the principle for setting a smaller interval is τ. min If the autocorrelation waveform R(m) is greater than half the duration of the correlation peak, the signal autocorrelation waveform R(m) will be in a smaller interval [n_max-τ]. min ,n_max+τ min The peak value of ] is set to zero, and then the maximum peak value C_new_max is searched again in a larger interval to identify the feature f1 = C_max / C_new_max.

5. The underwater acoustic DSSS signal identification method according to claim 1, characterized in that, Step S4 specifically includes: first, segmenting the suspected communication signal, and then calculating the secondary power spectrum; first, performing a first power spectrum calculation on each signal segment, and then performing a second power spectrum calculation on the obtained power spectrum to obtain the secondary power spectrum of each signal segment; superimposing the secondary power spectra of each signal segment to obtain the final obtained signal secondary power spectrum PSD, with a length of M.

6. The underwater acoustic DSSS signal identification method according to claim 5, which combines correlation peaks and periodic impulse characteristics of the secondary power spectrum, is characterized in that... Step S5 specifically includes: setting the sliding window size N_win according to the signal sampling frequency, sliding the window sequentially from low frequency to high frequency in the secondary power spectrum in a non-overlapping manner, the total number of slidings required is num_win = floor(M / N_win), and extracting the maximum spectral value sequentially in each window to form a spectral peak set PSD_max.

7. The underwater acoustic DSSS signal identification method according to claim 6, which combines correlation peaks and periodic impulse characteristics of the secondary power spectrum, is characterized in that... Step S6 specifically includes: setting a multiple threshold, the extracted impulse point needs to satisfy that the peak value of the point is higher than the mean of the interval near the point multiplied by the multiple threshold, the impulse point positions that meet the condition are saved to form an impulse point position set dL, assuming the length of dL is K, then the impulse point position difference set DL=[dL(2)-dL(1),dL(3)-dL(1),...,dL(K)-dL(1)], and the length of DL is K-1.

8. The underwater acoustic DSSS signal identification method according to claim 7, characterized in that, Step S8 specifically includes: The coefficient of determination R 2 As a recognition feature f2, the recognition coefficient 9. The underwater acoustic DSSS signal identification method according to claim 8, which combines correlation peaks and periodic impulse characteristics of the secondary power spectrum, is characterized in that... In step S9, if both the identification feature f1 and the identification feature f2 are greater than the set threshold, the identification signal is a DSSS signal.

10. The underwater acoustic DSSS signal identification method according to claim 7, characterized in that, If no suitable impulse point location is found in step S6, or if the number of impulse point locations is too small, the identification result is directly determined to be a non-DSSS signal.

Images