Effective information extraction method for communication signals in low signal-to-noise ratio complex environment

Abstract

A method for extracting effective information of a communication signal in a low signal-to-noise ratio complex environment relates to the field of communication signal processing, and comprises the following steps: firstly, obtaining a plurality of intrinsic mode functions mixed with effective signals with different frequencies and complex noise interference by using an adaptive noise complete empirical mode decomposition method, and carrying out wavelet denoising and reconstruction on the intrinsic mode functions to obtain a plurality of intrinsic mode functions; the method comprises the following steps of: obtaining a de-noised signal with a high signal-to-noise ratio, pre-processing the de-noised signal, calculating a separation matrix of an observation signal matrix after pre-processing according to a blind source separation algorithm, and extracting an effective communication signal from the original observation signal by utilizing the separation matrix so as to complete the effective information extraction of the communication signal. According to the method, noise and invalid signal interference in the signal transmission process can be suppressed under the conditions of strong environmental noise and low signal-to-noise ratio, signals containing effective information can be accurately extracted, and the target detection efficiency can be improved under the condition of high signal-to-noise ratio.

Description

Technical Field

[0001] The present invention relates to the field of communication signal processing, and in particular to an effective information extraction method for communication signals in a low signal-to-noise ratio complex environment. Background Art

[0002] A complex electromagnetic environment generally refers to an environment within a certain spatial range that is created by the interaction of man-made electromagnetic emissions and various electromagnetic phenomena. This complex electromagnetic environment encompasses four fundamental elements: space, time, spectrum, and energy, encompassing phenomena such as electromagnetic radiation, electromagnetic induction, and electromagnetic interference. The factors that contribute to a complex electromagnetic environment are diverse. For example, in the military, electronic countermeasures, high-density and high-intensity electromagnetic wave emissions, and naturally occurring electromagnetic waves are all significant factors. In the civilian sector, industrial electromagnetic interference, radiation from civilian equipment, and electromagnetic interference within and between systems are all key factors. In such environments with strong environmental noise and low signal-to-noise ratios, effectively improving the signal-to-noise ratio of communication signals and detecting valid information is an urgent issue.

[0003] Currently, in terms of communication signal processing methods, common methods mainly include signal smoothing methods based on time domain filtering, filtering based on transform domain, adaptive filtering technology, and algorithms based on artificial intelligence and machine learning.

[0004] Signal smoothing methods based on time-domain filtering calculate the average value within a sliding window as a filtering threshold, then use this threshold to filter the signal, thereby improving the detection of valid signal information. This method is effective when the aliasing of valid and noise signals in communication signals is not obvious, but it is less effective when it comes to signal extraction in complex electromagnetic environments.

[0005] Transform-domain filtering, based on mathematical tools such as the Fourier transform, first decomposes the signal into a superposition of harmonics of different frequencies. Based on known prior information, linear filtering techniques such as Wiener filtering and Kalman filtering are then used to suppress noise, or signal reduction is achieved by subtracting the noise spectrum from the original spectrum. However, when the statistical characteristics of the noise are unknown and the filter design phase lacks accurate knowledge of parameters such as signal bandwidth and center frequency, the performance of manually designed transform domains and filters may be limited or even rendered inappropriate.

[0006] Adaptive filtering techniques automatically adjust parameters, allowing the noise reduction system to better adapt to the processing requirements of stationary signals. Adaptive filtering techniques include least mean square adaptive filtering and recursive least squares adaptive filtering. Each of these adaptive algorithms has advantages and limitations, and the appropriate choice should be based on the specific application scenario. Furthermore, while adaptive filtering techniques excel at processing periodic stationary signals, they present challenges when processing non-stationary signals.

[0007] In recent years, algorithms based on artificial intelligence and machine learning have been introduced into communication signal processing. Using methods such as deep learning and convolutional neural networks, they can intelligently identify and classify aliased communication signals and automatically optimize signal processing strategies. These methods demonstrate strong adaptability in handling complex, noisy, and dynamic environments. However, their drawbacks include the large amount of data and computational resources required, and the effectiveness of model training may be limited by the quality of the sample data.

[0008] Among these methods, blind source separation (BSS) algorithm has gradually become a potential solution, especially when facing unknown signal sources. The blind source separation algorithm separates the received multi-channel communication signals without relying on prior information about the signal source, and can adaptively separate the signal containing effective information from the mixed signal. This makes blind source separation more flexible and robust in dynamically changing noise environments, and can effectively cope with complex environments.

[0009] Compared with traditional methods, blind source separation algorithms have higher noise immunity and can significantly improve the communication reliability of communication systems. Signal blind source separation algorithms mainly include independent component analysis, joint diagonalization, optimal global decomposition, and maximum signal-to-noise ratio separation. Joint diagonalization and maximum signal-to-noise ratio separation are the main methods used in communication signal processing.

[0010] However, in practice, communication signals are often in a complex electromagnetic environment during transmission, and the space is full of various noises, resulting in the signal-to-noise ratio of the received communication signal often being at a low level; and the blind source separation method can only obtain the information of the observed signal, and cannot obtain other prior information, so there are certain requirements for the signal when separating the signal. First, it is required that there can only be one Gaussian source in the source signal. During the transmission of the communication signal, the Gaussian white noise in the communication channel is the Gaussian source. Second, it is required that the source signals are uncorrelated with each other. In a complex electromagnetic environment, the signal transmitted in the channel is a mixture of signals emitted by multiple transmitting sources in the environment, among which there may be signals with high correlation, which directly affects the final effective signal extraction effect.

[0011] Therefore, due to the algorithm characteristics of blind source separation itself, the effect of signal separation will be greatly reduced when the received signal is at a low signal-to-noise ratio, resulting in the inability to effectively detect signals containing valid information. Summary of the Invention

[0012] The purpose of the present invention is to provide a method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio. The method can suppress noise and invalid signal interference during signal transmission under strong environmental noise and low signal-to-noise ratio conditions, and accurately extract signals containing valid information.

[0013] The technical solution adopted by the present invention to achieve the above technical objectives is: a method for effectively extracting information from communication signals in a complex environment with a low signal-to-noise ratio, comprising the following steps: 1) The original observation signal sequence is constructed by obtaining the signal received in each channel during actual transmission. The original observation signal sequence is subjected to adaptive noise complete empirical mode decomposition to obtain several intrinsic mode functions that are mixed with effective signals of different frequencies and complex noise interference; 2) Perform wavelet denoising on each intrinsic mode function, and then reconstruct the processed component data to obtain a denoised signal with a high signal-to-noise ratio; 3) Preprocess the denoised signal to obtain the preprocessed observation signal matrix; 4) Calculate the separation matrix of the preprocessed observation signal matrix based on the blind source separation algorithm, and then use the separation matrix to extract the effective communication signal from the original observation signal, thus completing the effective information extraction of the communication signal.

[0014] As an optimization scheme for the effective information extraction method of communication signals in the above-mentioned low signal-to-noise ratio complex environment, the adaptive noise complete empirical mode decomposition in step 1) generates adaptive white noise by calculating the noise standard deviation of the previous-order residual in real time. Then, single-order focused decomposition and noise component pre-screening are used to achieve bidirectional transmission of the decomposed residual, so that the high-frequency noise energy is concentrated in a specific frequency band, facilitating better decomposition.

[0015] As another optimization scheme for the above-mentioned effective information extraction method of communication signals in a complex environment with low signal-to-noise ratio, the specific operation of obtaining several intrinsic mode functions by adaptive noise complete empirical mode decomposition in step 1) is as follows: 1.1) Set the signal received in each channel is the original communication signal, is the number of channels, and the first White noise added , the original communication signal in each channel is repeatedly superimposed with white noise to construct the original observation signal sequence ,Right now: ; Where, is the standard deviation of the noise, It represents the kth order component obtained after performing M-order empirical mode decomposition on the signal, where M is given by the empirical value; is generated by adaptive noise standard deviation control times of white noise added, the number of times ; 1.2) Constructing the original observation signal sequence Perform the first layer of empirical mode decomposition to obtain the first IMF components of the M original observation signals, and then average the M first IMF components obtained as the first IMF1 decomposed by the empirical mode decomposition method, that is: ; 1.3) Use original communication signals Subtract the value obtained in step 1.2) , get the first residual of the original signal ,Right now: ; 1.4) White noise Perform M-order empirical mode decomposition to obtain M IMF components of the noise, and convert the first residual Plus the sub-white noise The M IMF components of the first-order residual sequence are constructed , the number of times ,Right now: ; 1.5) For the first-order residual sequence Repeat steps 1.2) to 1.4) to finally obtain the M IMF components of the original communication signal and the The residual amount , thereby obtaining the M intrinsic mode functions of the original observation signal, namely: .

[0016] As another optimization scheme for the above-mentioned effective information extraction method of communication signals in a complex environment with low signal-to-noise ratio, the wavelet denoising and reconstruction in step 2) refers to performing wavelet transform on the signal by selecting the wavelet basis and the number of decomposition layers to obtain the wavelet decomposition coefficients. , then set the threshold Wavelet decomposition coefficients Processing is performed to obtain the estimated value of the decomposition coefficient , and then As the optimization criterion, the decomposition coefficient estimate obtained is By performing wavelet reconstruction, the signal after wavelet denoising can be obtained.

[0017] As another optimization scheme for the above-mentioned effective information extraction method of communication signals in a complex environment with low signal-to-noise ratio, the specific operations of wavelet denoising and reconstruction in step 2) are as follows: 2.1) Construct the soft threshold function of wavelet threshold denoising as follows: ; Where, is a sign function, is the estimated value of the decomposition coefficient, is the wavelet decomposition coefficient, is the set threshold, and its calculation formula is: ; Where, is the length of the signal, is the estimate of the noise variance, ,in To perform wavelet decomposition The median of the effective part of the layer wavelet coefficients; 2.2) Use the soft threshold function in step 2.1) to perform wavelet soft threshold denoising on each intrinsic mode function obtained in step 1). After denoising, reconstruct the component data to obtain a denoised signal with a high signal-to-noise ratio: ; in is the denoised signal for each channel, is the wavelet decomposition and reconstruction operator.

[0018] As another optimization scheme for the above-mentioned effective information extraction method of communication signals in a complex environment with a low signal-to-noise ratio, the preprocessing of the denoised signal in step 3) refers to performing de-averaging and pre-whitening operations on the denoised signal.

[0019] As another optimization solution for the above-mentioned method for effectively extracting information from communication signals in a complex environment with a low signal-to-noise ratio, the specific operation of preprocessing the denoised signal in step 3) is as follows: 3.1) De-mean the denoised signal; The function to remove the mean is: ; Where, represents the signal matrix after removing the mean, is the denoised signal matrix, Represents the denoised signal matrix Each channel signal Take the mean; 3.2) Pre-whitening the signal after de-averaging; Whitening matrix for: ; Where, is the covariance matrix of the received signal after removing the mean of The matrix composed of the eigenvectors corresponding to the largest eigenvalues, the covariance matrix of the signal after removing the mean The formula is: ; Where, is the covariance matrix of the source communication signal, is the mixing matrix, for The transpose of is the noise power, is the identity matrix. The identity matrix is ​​a square matrix with all elements on the main diagonal being 1 and all elements in the rest of the positions being 0. = is the diagonal matrix corresponding to these eigenvalues, is the eigenvalue of the matrix after calculating the autocorrelation of the signal matrix; The observation signal matrix after whitening is: .

[0020] As another optimization scheme for the above-mentioned effective information extraction method of communication signals in complex environments with low signal-to-noise ratio, in step 3.1), the denoised signal matrix is the original communication signal in all channels The denoised signal matrix formed after denoising is Each row represents the denoised signal in one channel .

[0021] As another optimization solution for the above-mentioned method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio, the specific operations of step 4) are as follows: 4.1) Substituting into the maximum signal-to-noise ratio blind source separation algorithm, we get: ; Where, represents the transpose calculation, , is the correlation matrix, , ; 4.2) Solve the separation matrix for the formula in step 4.1) The gradient of , we get: ; 4.3) Based on the fact that the gradient value at the extreme point of the maximum signal-to-noise ratio is equal to 0, rewrite the formula in step 4.2) as follows: ; 4.4) Solve the formula in step 4.3) and we know that the matrix The eigenvectors of are arranged in rows to obtain the final optimized separation matrix ; 4.5) Using the Separation Matrix From the original received signal matrix The effective communication signal is extracted from the formula: ; Where, The signal in each channel row That is the effective communication signal in the channel.

[0022] As another optimization scheme for the above-mentioned effective information extraction method of communication signals in a complex environment with low signal-to-noise ratio, the step 4.1) The original received signal matrix composed of the original communication signals received by all communication channels, Each row represents the original communication signal in a channel .

[0023] Compared with the prior art, the present invention has the following beneficial effects: The present invention first obtains several intrinsic mode functions (IMFs) of mixed valid signals of different frequencies and complex noise interference through adaptive noise complete empirical mode decomposition, then performs wavelet denoising and preprocessing on each IMF. Through these operations, the signal-to-noise ratio of the received signal in strong ambient noise and low signal-to-noise ratio environments is significantly improved. The processed signal is then used to replace the original signal to optimize the separation matrix coefficients in the blind source separation method, thereby solving the problem of poor decomposition effect of the blind source separation method in strong noise and low signal-to-noise ratio environments, and at the same time, it can also improve target detection efficiency under high signal-to-noise ratio conditions. 2) The noise intensity added in the existing traditional noise-complete empirical mode decomposition method is a fixed value. Adding noise of fixed intensity to the IMF of different signal frequencies to generate a signal sequence can easily cause noise residue in the reconstructed signal. The present invention improves on this by using a dynamic noise intensity adjustment strategy to optimize the signal decomposition effect. It uses the calculation results based on the previous order residual to maintain a high noise intensity in the high-frequency stage of signal decomposition, while gradually attenuating the noise energy in the low-frequency stage. This achieves the goal of suppressing modal aliasing while effectively improving the noise residue problem in the signal reconstruction process. 3) Based on the communication signal received in a complex electromagnetic environment, the present invention uses adaptive noise complete empirical mode decomposition to perform modal decomposition on the communication signal, obtains the IMF components containing various frequency components, and offsets the superimposed noise influence by superimposing adaptive white noise on each component and solving the IMF for each signal component respectively and averaging them, while improving the modal aliasing phenomenon and providing a signal basis for subsequent filtering effects; then, based on the extracted IMF components, each IMF component is filtered by introducing a wavelet denoising algorithm. The noise component in each IMF component has been weakened compared to the original signal, so a better denoising effect can be obtained by using soft threshold denoising. The threshold value method obtains the threshold value, and the signal reconstruction of the denoised IMF component can obtain a signal with a higher signal-to-noise ratio. Compared with other methods for improving the signal-to-noise ratio, this method has better effect and higher efficiency; finally, the present invention adopts the maximum signal-to-noise ratio blind source separation algorithm to extract effective information from the communication signal. In the case of strong environmental noise and low signal-to-noise ratio, the adaptive noise complete empirical mode decomposition-wavelet algorithm is used to improve the signal-to-noise ratio, thereby improving the maximum signal-to-noise ratio blind source separation algorithm. By optimizing the separation matrix coefficients, the separation effect is improved. In the case of lower signal-to-noise ratio, the effective information in the communication signal can be effectively detected, and the detection effect can also be improved in the case of higher signal-to-noise ratio. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The effective pulse signal and the superimposed noise interference signal in the communication signal of the embodiment; Figure 2 The IMF components obtained by performing adaptive noise complete empirical mode decomposition on the communication signal; Figure 3 It is the signal reconstructed after wavelet denoising of each IMF component; Figure 4 This is the result of pulse pressure detection of each channel signal separated by the existing maximum signal-to-noise ratio blind source separation algorithm at low signal-to-noise ratio (SNR=-5db); Figure 5 The results of pulse pressure detection of each channel signal separated by the method of the present invention when the signal-to-noise ratio is low (SNR=-5db); Figure 6 This is the result of pulse pressure detection of each channel signal separated by the existing maximum signal-to-noise ratio blind source separation algorithm at a higher signal-to-noise ratio (SNR=10db); Figure 7 The results of pulse pressure detection of each channel signal separated by the method of the present invention at a higher signal-to-noise ratio (SNR=10db) are shown. DETAILED DESCRIPTION

[0025] The scheme of the present invention is further elaborated in detail below with reference to specific embodiments. The parts not explained in the following embodiments of the present invention are deemed to be existing technologies known or should be known to those skilled in the art, such as conventional blind source separation methods, conventional adaptive noise complete empirical mode decomposition, etc.

[0026] Example 1 A method for effectively extracting information from communication signals in a complex environment with a low signal-to-noise ratio comprises the following steps: 1) The original observation signal sequence is constructed by obtaining the signal received in each channel during actual transmission. The original observation signal sequence is subjected to adaptive noise complete empirical mode decomposition to obtain several intrinsic mode functions that are mixed with effective signals of different frequencies and complex noise interference; The adaptive noise complete empirical mode decomposition in this step generates adaptive white noise by calculating the noise standard deviation of the previous order residual in real time. It then uses single-order focused decomposition and noise component pre-screening to achieve two-way transmission of the decomposed residual, concentrating the high-frequency noise energy in a specific frequency band for better decomposition. The specific operations of obtaining several intrinsic mode functions by the adaptive noise complete empirical mode decomposition in this step are as follows: 1.1) Set the signal received in each channel is the original communication signal, is the number of channels, and the first White noise added , the original communication signal in each channel is repeatedly superimposed with white noise to construct the original observation signal sequence ,Right now: ; Where, is the standard deviation of the noise, It represents the kth order component obtained after performing M-order empirical mode decomposition on the signal, where M is given by the empirical value; is generated by adaptive noise standard deviation control times of white noise added, the number of times ; 1.2) Constructing the original observation signal sequence Perform the first layer of empirical mode decomposition to obtain the first IMF components of the M original observation signals, and then average the M first IMF components obtained as the first IMF1 decomposed by the empirical mode decomposition method, that is: ; 1.3) Use original communication signals Subtract the value obtained in step 1.2) , get the first residual of the original signal ,Right now: ; 1.4) White noise Perform M-order empirical mode decomposition to obtain M IMF components of the noise, and convert the first residual Plus the sub-white noise The M IMF components of the first-order residual sequence are constructed , the number of times ,Right now: ; 1.5) For the first-order residual sequence Repeat steps 1.2) to 1.4) to finally obtain the M IMF components of the original communication signal and the The residual amount , thereby obtaining the M intrinsic mode functions of the original observation signal, namely: ; After the original signal undergoes adaptive noise complete empirical mode decomposition, the IMF components obtained by decomposing the signal in descending order of frequency are selected. Selecting specific IMF components to reconstruct the signal can achieve different frequency filtering of the original signal. In addition, this method determines the decomposition criterion based on the input signal, which is data-driven and has strong adaptability. The communication signal transmitted in the channel is subjected to adaptive noise complete empirical mode decomposition. Each IMF component obtained contains the effective components of the original signal. If some of the IMFs are discarded, the reconstructed signal will be distorted and the target signal cannot be detected. Therefore, a wavelet algorithm is needed for filtering to improve the signal-to-noise ratio of the signal. Each intrinsic mode function is subjected to wavelet denoising, and then the processed component data is reconstructed to obtain a denoised signal with a high signal-to-noise ratio; In this step, wavelet denoising and reconstruction means performing wavelet transform on the signal by selecting the wavelet basis and the number of decomposition layers to obtain the wavelet decomposition coefficients. , then set the threshold Wavelet decomposition coefficients Processing is performed to obtain the estimated value of the decomposition coefficient , and then As the optimization criterion, the decomposition coefficient estimate obtained is By performing wavelet reconstruction, we can get the signal after wavelet denoising. The specific operation is as follows: 2.1) Construct the soft threshold function of wavelet threshold denoising as follows: ; Where, is a sign function, is the estimated value of the decomposition coefficient, is the wavelet decomposition coefficient, is the set threshold, and its calculation formula is: ; Where, is the length of the signal, is the estimate of the noise variance, ,in To perform wavelet decomposition The median of the effective part of the layer wavelet coefficients; 2.2) Use the soft threshold function in step 2.1) to perform wavelet soft threshold denoising on each intrinsic mode function obtained in step 1). After denoising, reconstruct the component data to obtain a denoised signal with a high signal-to-noise ratio: ; in is the denoised signal for each channel, is the wavelet decomposition and reconstruction operator; Preprocess the denoised signal to obtain the preprocessed observation signal matrix; In this step, preprocessing the denoised signal means performing de-averaging and pre-whitening operations on the denoised signal. The specific operations are as follows: 3.1) De-mean the denoised signal; The function to remove the mean is: ; Where, represents the signal matrix after removing the mean, is the denoised signal matrix, Represents the denoised signal matrix Each channel signal Take the mean; Denoised signal matrix is the original communication signal in all channels The denoised signal matrix formed after denoising is Each row represents the denoised signal in one channel ; 3.2) Pre-whitening the signal after de-averaging; Whitening matrix for: ; Where, is the covariance matrix of the received signal after removing the mean of The matrix composed of the eigenvectors corresponding to the largest eigenvalues, the covariance matrix of the signal after removing the mean The formula is: ; Where, is the covariance matrix of the source communication signal, is the mixing matrix, for The transpose of is the noise power, is the identity matrix. The identity matrix is ​​a square matrix with all elements on the main diagonal being 1 and all elements in the rest of the positions being 0. = is the diagonal matrix corresponding to these eigenvalues, is the eigenvalue of the matrix after calculating the autocorrelation of the signal matrix; The observation signal matrix after whitening is: ; 4) Calculate the separation matrix of the preprocessed observation signal matrix based on the blind source separation algorithm, and then use the separation matrix to extract the effective communication signal from the original observation signal, that is, to complete the effective information extraction of the communication signal. The specific operations are as follows: 4.1) Substituting into the maximum signal-to-noise ratio blind source separation algorithm, we get: ; Where, represents the transpose calculation, , is the correlation matrix, , ; The original received signal matrix composed of the original communication signals received by all communication channels, Each row represents the original communication signal in a channel ; 4.2) Solve the separation matrix for the formula in step 4.1) The gradient of , we get: ; 4.3) Based on the fact that the gradient value at the extreme point of the maximum signal-to-noise ratio is equal to 0, rewrite the formula in step 4.2) as follows: ; 4.4) Solve the formula in step 4.3) and we know that the matrix The eigenvectors of are arranged in rows to obtain the final optimized separation matrix ; 4.5) Using the Separation Matrix From the original received signal matrix The effective communication signal is extracted from the formula: ; Where, The signal in each channel row That is the effective communication signal in the channel.

[0027] Experimental example Assume that the communication signal is in a strong electromagnetic environment, the communication signal is a linear frequency modulation signal, and there are a lot of noise signals in the environment. Figure 1 You can see the signal transmitted in the channel The effective communication signal is completely submerged by the invalid communication signal and the environmental thermal noise. At this time, the communication signal in the channel needs to be Processing is performed to reduce the impact of invalid signals and noise in order to detect valid communication signals; Adaptive noise complete empirical mode decomposition is performed using the method in step 1) of Example 1. In this experimental example, the number of decomposition layers is set to 6, so steps 1.2) to 1.4) in Example 1 are repeated 6 times in total. The IMF component of the first receiving channel signal decomposed is as follows: Figure 2 shown.

[0028] It has been theoretically proven that each IMF component contains the information of the original echo signal. Discarding any IMF component for signal reconstruction will cause signal distortion. Therefore, it is necessary to retain all IMF components and filter and reconstruct them. Therefore, on the basis of adaptive noise complete empirical mode decomposition of all receiving channel signals, the IMF component signals of each signal are subjected to wavelet denoising, and the filtered signals are reconstructed. Wavelet denoising mainly includes hard threshold denoising and soft threshold denoising. Based on the characteristics of signal processing and subsequent algorithm requirements of the present invention, the soft threshold method is selected for denoising. For specific operation steps, see steps 2.1)-2.2 of Example 1); the wavelet basis function is selected as sym8 wavelet, and the number of decomposition layers is selected as 5 layers; after the wavelet denoising is completed, all the IMF components that have undergone wavelet denoising are superimposed to reconstruct the denoised signal, as shown Figure 3 As shown; Obtaining the denoised signal Then, preprocessing is performed using the method of step 3.1) to step 3.2) of Example 1 to obtain a preprocessed observation signal matrix; Get the preprocessed observation signal matrix Then, the separation matrix is ​​calculated according to the blind source separation algorithm , and then use the separation matrix Extract the effective communication signal from the original observation signal. The maximum signal-to-noise ratio blind source separation algorithm uses the signal-to-noise ratio function as the calculation formula for minimizing the error, namely: ; In actual application, only the observation signal of the communication signal in the channel can be obtained. The observation signal is mixed with the effective communication signal, invalid communication signal and environmental noise signal, so the source signal in the above formula is is an unknown variable, and the original algorithm The sliding average As ,Right now: ; and , ,and ; This is equivalent to a simple low-pass filtering of the observed signal. Substituting the above formula into the maximum signal-to-noise ratio formula yields: ; in: , is the correlation matrix, , ; Find the separation matrix for the above formula Gradient: ; Because the gradient value is equal to 0 at the extreme point of the maximum signal-to-noise ratio, the above formula can be simplified to: ; Solving the above formula, we can get the separation matrix , and then find the original signal Estimated value of ; However, it has been shown in the literature that the separation matrix in the above formula The estimated solution is the matrix The eigenvector of , we can see that the final separation matrix is ​​consistent with the observed signal and its mean In strong ambient noise and low signal-to-noise ratio environments, due to The sliding average Can no longer replace the source signal , or by Alternative This will lead to larger errors, and The calculated separation matrix coefficients can no longer meet the requirements of the separation effect, so the separation effect of the algorithm is poor under low signal-to-noise ratio conditions. Therefore, the present invention uses the signal after adaptive noise complete empirical mode decomposition-wavelet denoising to replace , thereby optimizing the coefficients in the separation matrix and improving the final separation effect; executing steps 4.1) to 4.4) in Example 1), and finally solving the matrix The eigenvector of the final optimized separation matrix can be obtained , we can see that the optimized separation matrix is ​​related to the observed signal and its signal after adaptive noise complete empirical mode decomposition-wavelet filtering, so the separation matrix is ​​used in strong noise environment and low signal-to-noise ratio. It can also better separate the effective signal from the communication signal, namely: ; For each channel in the above formula Valid communication signals can be obtained by performing pulse pressure detection.

[0029] The method of the present invention is used to process the received signal of the final communication channel and detect the effective signal. The results are as follows: Figure 4-Figure 7 As shown in the figure, the waveforms detected by the method of the present invention and the waveforms detected by pulse compression using the existing blind source separation method are respectively shown under different signal-to-noise ratios. It can be seen that under the condition of a low signal-to-noise ratio (-5dB), the existing blind source separation algorithm cannot obtain effective communication signals after separating and pulse compressing the received signals of each communication channel, while the method of the present invention can effectively detect effective signals in each channel; Under normal signal-to-noise ratio (10dB), the method of the present invention can also effectively detect communication signals, and can suppress noise more effectively than the original blind source separation algorithm, which is conducive to subsequently decoding more effective information from the communication signals.

[0030] In summary, the method of the present invention can effectively extract useful communication signals in a strong noise and low signal-to-noise ratio environment, and can also improve the detection effect of effective communication signals in a high signal-to-noise ratio environment.

Claims

1. A method for extracting effective information from communication signals in a complex environment with low signal-to-noise ratio, characterized in that: The steps include: 1) The original observation signal sequence is constructed by obtaining the signal received in each channel during actual transmission. The original observation signal sequence is subjected to adaptive noise complete empirical mode decomposition to obtain several intrinsic mode functions that are mixed with effective signals of different frequencies and complex noise interference; 2) Perform wavelet denoising on each intrinsic mode function, and then reconstruct the processed component data to obtain a denoised signal with a high signal-to-noise ratio; 3) Preprocess the denoised signal to obtain the preprocessed observation signal matrix; 4) Calculate the separation matrix of the preprocessed observation signal matrix based on the blind source separation algorithm, and then use the separation matrix to extract the effective communication signal from the original observation signal, thus completing the effective information extraction of the communication signal.

2. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 1, wherein: The adaptive noise complete empirical mode decomposition in step 1) generates adaptive white noise by calculating the noise standard deviation of the previous order residual in real time. Then, single-order focused decomposition and noise component pre-screening are used to achieve bidirectional transmission of the decomposed residual, so that the high-frequency noise energy is concentrated in a specific frequency band, facilitating better decomposition.

3. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 1 or 2, characterized in that: The specific operation of obtaining several intrinsic mode functions by performing the adaptive noise complete empirical mode decomposition in step 1) is as follows: 1.1) Set the signal received in each channel is the original communication signal, is the number of channels, and the first White noise added , the original communication signal in each channel is repeatedly superimposed with white noise to construct the original observation signal sequence ,Right now: ； Where, is the standard deviation of the noise, It represents the kth order component obtained after performing M-order empirical mode decomposition on the signal, where M is given by the empirical value; is generated by adaptive noise standard deviation control times of white noise added, the number of times ; 1.2) Constructing the original observation signal sequence Perform the first layer of empirical mode decomposition to obtain the first IMF components of the M original observation signals, and then average the M first IMF components obtained as the first IMF1 decomposed by the empirical mode decomposition method, that is: ； 1.3) Use original communication signals Subtract the value obtained in step 1.2) , obtain the first residual of the original signal ,Right now: ； 1.4) White noise Perform M-order empirical mode decomposition to obtain M IMF components of the noise, and convert the first residual Plus the sub-white noise The M IMF components of the first-order residual sequence are constructed , the number of times ,Right now: ； 1.5) For the first-order residual sequence Repeat steps 1.2) to 1.4) to finally obtain the M IMF components of the original communication signal and the The residual amount , thereby obtaining the M intrinsic mode functions of the original observation signal, namely: 。 4. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 1, wherein: In step 2), the wavelet denoising and reconstruction is to perform wavelet transform on the signal by selecting the wavelet basis and the number of decomposition layers to obtain the wavelet decomposition coefficients. , then set the threshold Wavelet decomposition coefficients Processing is performed to obtain the estimated value of the decomposition coefficient , and then As the optimization criterion, the decomposition coefficient estimate obtained is By performing wavelet reconstruction, the signal after wavelet denoising can be obtained.

5. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 1 or 4, characterized in that: The specific operations of wavelet denoising and reconstruction in step 2) are as follows: 2.1) Construct the soft threshold function of wavelet threshold denoising as follows: ； Where, is a sign function, is the estimated value of the decomposition coefficient, is the wavelet decomposition coefficient, is the set threshold, and its calculation formula is: ； Where, is the length of the signal, is the estimate of the noise variance, ,in To perform wavelet decomposition The median of the effective part of the layer wavelet coefficients; 2.2) Use the soft threshold function in step 2.1) to perform wavelet soft threshold denoising on each intrinsic mode function obtained in step 1). After denoising, reconstruct the component data to obtain a denoised signal with a high signal-to-noise ratio: ； in is the denoised signal for each channel, is the wavelet decomposition and reconstruction operator.

6. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 1, wherein: The pre-processing of the denoised signal in step 3) refers to performing de-averaging and pre-whitening operations on the denoised signal.

7. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 1 or 6, characterized in that: The specific operations of pre-processing the denoised signal in step 3) are: 3.1) De-mean the denoised signal; The function to remove the mean is: ； Where, represents the signal matrix after removing the mean, is the denoised signal matrix, Represents the denoised signal matrix Each channel signal Take the mean; 3.2) Pre-whitening the signal after de-averaging; Whitening matrix for: ； Where, is the covariance matrix of the received signal after removing the mean of The matrix composed of the eigenvectors corresponding to the largest eigenvalues, the covariance matrix of the signal after removing the mean The formula is: ； Where, is the covariance matrix of the source communication signal, is the mixing matrix, for The transpose of is the noise power, is the identity matrix. The identity matrix is ​​a square matrix with all elements on the main diagonal being 1 and all elements in the rest of the positions being 0. = is the diagonal matrix corresponding to these eigenvalues, is the eigenvalue of the matrix after calculating the autocorrelation of the signal matrix; The observation signal matrix after whitening is: 。 8. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 7, wherein: In step 3.1), the denoised signal matrix is the original communication signal in all channels The denoised signal matrix formed after denoising is Each row represents the denoised signal in one channel .

9. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 1, wherein: The specific operations of step 4) are: 4.1) Substituting into the maximum signal-to-noise ratio blind source separation algorithm, we get: ； Where, represents the transpose calculation, , is the correlation matrix, , ; 4.2) Solve the separation matrix for the formula in step 4.1) The gradient of , we get: ； 4.3) Based on the fact that the gradient value at the extreme point of the maximum signal-to-noise ratio is equal to 0, rewrite the formula in step 4.2) as follows: ； 4.4) Solve the formula in step 4.3) and we know that the matrix The eigenvectors of are arranged in rows to obtain the final optimized separation matrix ; 4.5) Using the Separation Matrix From the original received signal matrix The effective communication signal is extracted from the formula: ； Where, The signal in each channel row That is the effective communication signal in the channel.

10. The method for extracting effective information from communication signals in a complex environment with a low signal-to-noise ratio according to claim 9, characterized in that: In step 4.1) The original received signal matrix composed of the original communication signals received by all communication channels, Each row represents the original communication signal in a channel .

Images