A differential chaotic shift keying communication signal selection method based on feature fusion

Abstract

The application provides a differential chaotic shift keying communication signal selection method based on feature fusion, and relates to the field of chaotic secure communication; in order to improve the low detection and low interception probability of the differential chaotic shift keying communication signal, the concept of covert communication is introduced, including differential chaotic shift keying communication signal feature analysis and differential chaotic shift keying communication signal selection method; the application respectively analyzes the signal features of the differential chaotic shift keying signal in the time domain, the frequency domain and the statistical domain, constructs multiple differential chaotic shift keying signal selection evaluation models, selects the differential chaotic shift keying signal with the highest similarity of signal features and Gaussian white noise from the given differential chaotic shift keying signal, establishes the optimal communication signal selection method, and improves the security of the differential chaotic shift keying communication system signal.

Description

Technical Field

[0001] This invention belongs to the field of chaotic secure communication, specifically relating to a differential chaotic shift keying communication signal selection method based on feature fusion. Background Technology

[0002] Chaotic communication, a research hotspot in the field of digital communication, has attracted much attention due to its nonlinear dynamic characteristics and potential to provide high security under certain conditions. However, in practical applications, due to the difficulty in implementing carrier synchronization technology, research focus has gradually shifted to differential chaotic shift keying (DCSK) systems for incoherent communication.

[0003] The unique combination of chaotic dynamics and differential modulation in differential chaotic shift keying (DCS) communication systems offers potential advantages in secure communication and interference resistance. However, several challenges remain, one major issue being the lack of a systematic evaluation method for DCS communication signals.

[0004] At present, comprehensive research and development of performance evaluation methods for DCSK communication system signals have not yet been established. This poses a challenge to the security and stability of DCSK systems in practical communication, especially in scenarios facing detection and interception by non-cooperative parties. To better understand and optimize differential chaotic shift keying (DCSK) communication systems, in-depth research on their signal characteristics is urgently needed, along with the development of comprehensive and reliable evaluation methods to ensure their reliability and security in practical applications. This requires both a deep dive into the interaction mechanism between chaotic dynamics and differential chaotic shift keying modulation technology, and the development of more advanced analysis tools and algorithms to improve the accuracy of DCSK communication evaluation. Related algorithmic research will contribute to further promoting the development of differential chaotic shift keying communication systems and expanding their practical applications in the field of digital communication signal detection. Developing a selection evaluation method based on differential chaotic shift keying communication signals is particularly important for improving the low detection and low interception probability of DCSK communication signals. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a differential chaotic shift keying communication signal selection method based on feature fusion, which is based on covert communication and chaotic signal feature analysis. This method aims to solve the problems of low detection probability and low interception probability in the current lack of analysis of differential chaotic shift keying communication signals.

[0006] A differential chaotic shift keying communication signal selection method based on feature fusion specifically includes the following steps: Step 1: Obtain the results under the conditions of the same half-spreading factor, same mapping, and same sampling frequency. NDifferential chaotic shift keying communication signals are categorized into several types, each with the same signal length. a Group samples, calculate each signal separately n The sample groups are based on the characteristics of the statistical domain, time domain, and frequency domain, and the normalized values ​​of these characteristics are used for comparative analysis. in N , a It is a positive integer; The statistical domain feature analysis includes: mean, variance, standard deviation, interquartile range, kurtosis, and skewness; The mean is the average of a set of data, which is the sum of all data points divided by the number of data points; its expression is: (1) in, This represents the mean of the data; This represents each observation in the data. n Represents the number of data points, a positive integer; The variance is a measure of the dispersion of a set of data; it is the average of the sum of the squares of the differences between each data point and the mean. Its expression is: (2) in, Indicates variance; This represents each observation in the data; This represents the average value of the data. The standard deviation is a measure of the dispersion of a set of data; it measures the distance between data points and the mean. Its expression is: (3) in, Indicates standard deviation; This represents each observation in the data; This represents the average value of the data. The interquartile range (IQR), also known as the quartile difference, is the difference between the upper quartile and the lower quartile. The quartiles are defined as the values ​​at each dividing point after sorting all values ​​from smallest to largest and dividing them into four equal parts. The skewness, which measures the direction and degree of skewness in the distribution of statistical data, is expressed as follows: (4) in, Indicates skewness, This represents each observation in the data; This represents the average value of the data. The standard deviation of the data; This represents the mean. E Represents the mathematical expectation. X This refers to the current data observation value; Kurtosis is used to describe the peak value of a probability density distribution curve at its mean; the calculation formula is: (5) in, Indicates kurtosis. This represents each observation in the data; This represents the average value of the data. Kurtosis represents the standard deviation of the data; a negative kurtosis indicates a more angular data distribution; a zero kurtosis indicates a data distribution similar to a normal distribution; a positive kurtosis indicates a flatter data distribution. The time-domain feature analysis includes: mean difference, mean absolute difference, median difference, median absolute difference, sum of absolute differences, entropy, curve coverage area, and number of peaks. The Mean Differences refers to the average difference between two adjacent data points in a time series; the calculation formula is: (6) in, For the first in the time series i One element; The mean absolute differences refer to the average of the absolute values ​​of the differences between two adjacent data points in a time series; its expression is: (7) in, The average of the differences between two adjacent elements is also called the difference mean. The average of the absolute values ​​of the differences between two adjacent elements is also called the mean of the absolute values ​​of the differences. This represents the i-th element in the time series; Represents the first in a time series i -1 element; n This represents the number of elements in the time series; the Median Differences refers to the median difference between two adjacent data points in the time series. The formula for calculating this difference is: (8) Where median represents the median of the differences between all adjacent elements; The median absolute difference refers to the median of the absolute values ​​of the differences between two adjacent data points in a time series. The formula for calculating this median is: (9) The sum of absolute differences refers to the sum of the absolute values ​​of the differences between two adjacent data points in the time series; the calculation formula is: (10) Entropy, a metric for measuring signal complexity, is calculated using the following formula: (11) in, In representing probability distribution x The probability of occurrence; Represents the logarithm to the base 2; The area under the curve refers to the area between the signal curve and the horizontal axis; the signal curve is generated by differential chaotic shift keying communication signal through MATLAB simulation. The peak values ​​are divided into maximum peak values ​​and minimum peak values; the number of maximum peak values ​​refers to the number of the highest peak values ​​in the signal, which describes the number and height of the signal peaks; the number of minimum peak values ​​refers to the number of the lowest peak values ​​in the signal, which describes the number and depth of the signal troughs. The frequency domain feature analysis includes: the fundamental frequency and the intermediate frequency of the spectrum; The fundamental frequency of the spectrum refers to the lowest fundamental frequency component of a signal in the frequency domain, and is often referred to as the lowest frequency component. In the frequency domain, the fundamental frequency of the spectrum is determined by calculating the power spectral density function (PSD) of the signal in the frequency domain, which usually corresponds to the lowest non-zero frequency component of the signal in the frequency domain. The power spectral density function is a function that describes the energy distribution of a signal in the frequency domain, representing the density of signal power at different frequencies. The intermediate frequency (IF) is the frequency at the center of the signal spectrum, used to describe the frequency distribution of the signal. The IF is obtained by calculating the center position in the signal spectrum, and can also be understood as the average value of the signal's frequency distribution. It is determined by calculating the weighted average value of the signal in the frequency domain, which is usually based on the power spectral density function of the signal in the frequency domain. The spectral center of a signal spectrum with symmetry is equal to the fundamental frequency. The calculation formula is: (12) in The frequency of the signal's spectrum, which is the frequency domain form of the time-domain data, P ( ) represents the power spectral density function of the signal in the frequency domain. Represents frequency; represents the power spectral density function of the signal at all frequency points. P ( ) and frequency The integral of the product of the products, with the denominator representing the power spectral density function of the signal. P ( f Integral over all frequency points; Step 2: Based on the fuzzy evaluation method, select the DCSK signal that is closest to the noise characteristics for each signal feature obtained in Step 1. Step 2.1, determine the fuzzy comprehensive evaluation index; Sixteen feature dimensions were extracted through signal feature analysis. It is indicated that their biological attitudes are used as evaluation indicators. The feature dimension parameters are explained below: -Mean; -Standard deviation; -variance; - Interquartile range; -Kushin; - Skewness; - Difference mean; - Mean of absolute difference; -difference median; -Median of the absolute value of the difference; - The sum of the absolute values ​​of the differences; -entropy; -Curve coverage area; -Number of peaks; - Spectrum fundamental frequency; -Intermediate frequency spectrum; Step 2.2: Establish a comprehensive evaluation set; Five rating levels were assigned to the 16 signal characteristic parameters mentioned above. That is, there are 5 level sets; each level corresponds to a fuzzy subset, respectively. Five levels of classification; Step 2.3: Perform single-factor fuzzy evaluation and obtain the evaluation matrix; Based on the selected data features, a single-factor fuzzy evaluation of the DCSK signal parameters is performed, resulting in an evaluation matrix: (13) Step 2.4: Determine the weight vectors of the evaluation factors; Based on the data features extracted from the data source, the weights of the 16 feature dimensions are evaluated as follows: (14) Step 2.5: Synthesize the fuzzy comprehensive evaluation result vector; (15) in It is composed of the weight matrix With the evaluation matrix The Obtained through columnar operations, it represents the overall view of... The degree of membership of a hierarchical fuzzy subset; Step 2.6: Analyze the fuzzy comprehensive evaluation result vector; Based on the principle of maximum membership, for , , , , The maximum value in the range is used to determine the degree of mimicry of the signal, which is obtained from formulas (13)-(15): = (16) in The DCSK signal with the closest noise characteristics is determined by the evaluation results of formula (16). The optimal differential chaotic shift keying communication signal is determined.

[0007] Beneficial technical effects of the present invention: This invention analyzes the signal characteristics of DCSK signals in the time, frequency, and statistical domains, constructs a pseudo-attitude evaluation model for DCSK signals, selects the differential chaotic shift keying signal with the highest similarity to Gaussian white noise from various DCSK signals, establishes an optimal communication signal selection method, and improves the security of the physical layer signal of the differential chaotic shift keying communication link. Attached Figure Description

[0008] Figure 1 Block diagram for analyzing the characteristics of chaotic signal parameters in an embodiment of the present invention; Figure 2 Mean feature map in the embodiments of the present invention; Figure 3 Variance feature map in the embodiments of the present invention; Figure 4 Standard deviation feature map in the embodiments of the present invention; Figure 5 Lower quartile feature map in the embodiments of the present invention; Figure 6 Median feature map in the embodiments of the present invention; Figure 7 Upper quartile feature map in the embodiments of the present invention; Figure 8 Skewness value feature diagram in the embodiments of the present invention; Figure 9Kurtosis feature map in the embodiments of the present invention; Figure 10 Differential mean feature map in the embodiments of the present invention; Figure 11 In this embodiment of the invention, the mean of the absolute difference is characterized. Figure 12 Differential median feature map in the embodiments of the present invention; Figure 13 The median feature map of the absolute value of the difference in the embodiments of the present invention; Figure 14 Feature diagram of the sum of absolute differences in embodiments of the present invention; Figure 15 Entropy feature map in the embodiments of the present invention; Figure 16 Lower coverage area feature diagram in the embodiment of the present invention; Figure 17 Minimum number of peaks characteristic map in the embodiments of the present invention; Figure 18 The maximum number of peaks characteristic map in the embodiments of the present invention; Figure 19 Spectral fundamental frequency characteristic diagram in the embodiments of the present invention; Figure 20 Intermediate frequency characteristic diagram of the spectrum in the embodiments of the present invention; Figure 21 A schematic diagram illustrating the DCSK signal modeling research based on pseudo-attitude in an embodiment of the present invention. Detailed Implementation

[0009] The present invention will be further described below with reference to the accompanying drawings and embodiments; A method for selecting differential chaotic shift keying (DFT) communication signals based on feature fusion includes: feature analysis of DFT communication signals based on the statistical domain, including mean, variance, standard deviation, interquartile range, kurtosis, and skewness; feature analysis of DFT communication signals based on the time domain, including differential mean, differential absolute mean, differential median, differential absolute median, sum of differential absolute values, entropy, curve coverage area, and number of peaks; and feature analysis of DFT communication signals based on the frequency domain, mainly including the fundamental frequency and intermediate frequency of the spectrum.

[0010] The similarity index for differential chaotic shift keying (DPSK) communication signals can be measured by calculating the differences in characteristic parameters between the DPSK communication signal and Gaussian white noise in the time, frequency, and statistical domains. First, the DPSK communication signal is selected as the detection signal, and parameter features are extracted in the statistical, frequency, and time domains, respectively. These features are then compared and analyzed with those of the noise.

[0011] Based on the analysis of covert mimicry and chaotic signal characteristics, this invention proposes a differential chaotic shift keying communication signal selection method based on feature fusion to solve the problem of high low detection probability and low interception probability in differential chaotic shift keying.

[0012] The differential chaotic shift keying communication signal is the signal value of each time slot obtained by sampling the chaotic map at a certain time interval. The similarity index of the chaotic signal can be used to measure the similarity difference between the chaotic signal and Gaussian white noise by calculating the difference of characteristic parameters in the time domain, frequency domain, and statistical domain. First, the chaotic modulation signal is selected as the detection signal, and parameter features are extracted in the statistical domain, frequency domain, and time domain respectively. Feature comparison analysis is then performed with the noise. The signal feature analysis box is shown in the figure. Figure 1 As shown.

[0013] Step 1: Obtain 8 differential chaotic shift keying communication signals under the same half-spreading factor, the same mapping and the same sampling frequency. Divide each differential chaotic shift keying communication signal into 10 groups of samples according to the same signal length. Calculate the characteristics of each signal's 10 groups of samples based on the statistical domain, time domain and frequency domain. Normalize these characteristics and perform comparative analysis. In this study, an improved Logistic mapping was used for the chaotic signal, with a half-spreading factor of 100 and a signal-to-noise ratio of 13dB. Features of eight modulation signals—DCSK, SR-DCSK, RM-DCSK, ODBR-DCSK, SR-ODBR-DCSK, TR-RM-DCSK, RSST-DSCK, and RSST-SR-DCSK—were extracted in the time, statistical, and frequency domains. These features were then compared with those of the WGN signal. The modulation signal most similar to the WGN signal was selected as the optimal signal, applicable to secure communication. DCSK, the most fundamental differential chaotic communication signal, was first proposed by Kolumban, solving the problem of timing synchronization issues with CSK signals. The SR-DCSK communication system, proposed by Kaddoum, improved the information transmission rate. The RM-DCSK communication system, proposed by Yang Hua, achieved twice the information transmission rate compared to DCSK. The ODBR-DCSK, SR-ODBR-DCSK, TR-RM-DCSK, and RSST-DSSK communication systems were proposed by scholar Sui Tao. Each system improves information transmission rate and security to varying degrees. Since the RSST-DCSK system improves information transmission security compared to the DCSK system, but does not improve the information transmission rate, the SR-DCSK and RSST-DCSK systems were combined to form the RSST-SR-DCSK communication system, which significantly improves both information transmission security and information transmission rate compared to the DCSK communication system.

[0014] Based on the statistical domain chaotic signal feature analysis, the main features include mean, variance, standard deviation, interquartile range, kurtosis, and skewness.

[0015] Taking the mean analysis of chaotic signal characteristics in the statistical domain as an example, the mean is the average value of a set of data, typically used to describe the central location of the data. It is the sum of all data points divided by the number of data points. The mean can be used to estimate the overall trend of a dataset, but it is highly sensitive to outliers. Its expression is: (1) in, This represents the average value of the data. This represents each observation in the data. n Indicates the number of data items. Figure 2 The mean parameter feature maps of the chaotic modulation signal and the noise signal are shown. From the numerical analysis of the mean parameter feature tables of the five groups, the mean of the chaotic modulation signal and the mean of WGN are not significantly different and both have a certain degree of randomness. Among them, the mean of the SR-DCSK signal and the mean of the WGN signal have the smallest difference in mean parameter.

[0016] Variance is a measure of the dispersion of a set of data. It is the average of the squares of the differences between each data point and the mean. While variance can also describe the dispersion of data, unlike standard deviation, it does not involve taking the square root; therefore, its unit is the square of the original data. Variance characterizes the degree to which a sample deviates from the mean, and also indicates the degree of fluctuation within the sample.

[0017] (2) in, Indicates variance; This represents each observation in the data; This represents the average value of the data.

[0018] Standard deviation is a measure of the dispersion of a set of data; it measures the distance between data points and the mean. A smaller standard deviation indicates that the data in the dataset is closer to the mean, while a larger standard deviation indicates that the data in the dataset is more dispersed.

[0019] (3) in, Indicates standard deviation; This represents each observation in the data; This represents the average value of the data. Figure 3 and 4 The diagram shows a comparison of the variance and standard deviation characteristics of chaotic signals. Overall, the variance and standard deviation of chaotic modulated signals are not significantly different from those of WGN. Among them, TR-RM-ODBR is most similar to WGN.

[0020] In statistics, quartiles are calculated by sorting all values ​​from smallest to largest and dividing them into four equal parts. The value at each dividing point is called the quartile. In this way, the data is divided into four equal parts, each containing 25% of the data. Similar to the method for calculating the median, for ungrouped data, calculating quartiles requires first sorting the data and then determining the position of the quartile; the value at that position is the quartile. The interquartile range (IQR), also known as the quartile difference, is the difference between the upper quartile and the lower quartile. Figures 5-7 The figure shows the lower quartile, median, and upper quartile of the chaotic modulated signal, with no significant difference in the characteristic parameters.

[0021] Skewness is an indicator used to measure the direction and degree of skewness in the distribution of statistical data; it is also known as skewness or skewness coefficient.

[0022] (4) in, Indicates skewness, This represents each observation in the data; This represents the average value of the data. The standard deviation of the data. Figure 8 The image shows a comparison of the skewness values ​​of the chaotic modulated signal and Gaussian noise. It can be seen that the difference in skewness between the signals is not significant.

[0023] Kurtosis, also known as the kurtosis coefficient, describes the peak value of a probability density distribution curve at its mean. (See attached image.) Figure 9 As shown; the calculation formula is: (5) in, Indicates kurtosis. This represents each observation in the data; This represents the average value of the data. The standard deviation of the data is represented. A negative kurtosis indicates a more angular data distribution; zero kurtosis indicates a data distribution similar to a normal distribution; and positive kurtosis indicates a flatter data distribution. The table shows the kurtosis characteristics of chaotic signals, demonstrating that there is no significant difference in kurtosis between chaotic signals and noise.

[0024] Based on the analysis of time-domain chaotic signal characteristics, the main components include the mean of the difference, the mean of the absolute value of the difference, the median of the difference, the median of the absolute value of the difference, the sum of the absolute values ​​of the difference, the entropy, the area covered by the curve, and the number of peaks.

[0025] Mean differences refer to the average of the differences between two adjacent data points in a time series. It can be used to describe the overall trend of a signal and is commonly used for filtering and noise reduction. The calculation formula is: (6) Mean absolute differences (MAD) are the average of the absolute differences between two adjacent data points in a time series. They can also be used to describe the overall trend of a signal, but compared to the mean difference, they are more robust and less affected by outliers. Their expression is: (7) in, The average of the differences between two adjacent elements is also called the difference mean. The average of the absolute values ​​of the differences between two adjacent elements is also called the mean of the absolute values ​​of the differences. This represents the i-th element in the time series; Represents the first in a time series i-1 element; n This indicates the number of elements in the time series. Figure 10 , 11 The images show a comparison of the difference mean and the difference absolute mean characteristics of chaotic signals. Because chaotic modulated signals are bipolar signals, the calculated difference values ​​can be positive or negative, and the difference mean can also be positive or negative, exhibiting no regularity. Overall, the difference mean and difference absolute mean characteristics of chaotic modulated signals are not significantly different from Gaussian white noise. For the difference absolute mean, TR-RM-DCSK is closest to Gaussian white noise.

[0026] The median difference refers to the median of the differences between two adjacent data points in a time series. It can be used to describe the center position and trend of a signal, and is commonly used for anomaly detection and change point detection. The calculation formula is: (8) Here, `median` represents the median of the differences between all adjacent elements. The Median Absolute Differences (MAD) is the median of the absolute differences between two adjacent data points in a time series. It can be used to describe the center position and trend of a signal, but compared to the median difference, it is more robust and less affected by outliers. The calculation formula is: (9) Figure 12 , 13 The figures show a comparison of the median difference and the median absolute difference of chaotic modulated signals. Overall, the characteristics of the mean difference and the mean absolute difference of chaotic modulated signals are not significantly different from those of Gaussian white noise. For the mean absolute difference, TR-RM-DCSK is closest to Gaussian white noise.

[0027] The sum of absolute differences refers to the sum of the absolute values ​​of the differences between two adjacent data points in a time series. It can be used to describe the overall magnitude and complexity of the signal's variation. The calculation formula is: (10) Figure 14 The image shows a comparison of the sum of absolute differences of eight signals. Among them, the sum of absolute differences of TR-RM-DCSK is closest to Gaussian white noise.

[0028] Entropy is a metric for measuring the complexity of a signal. It describes the degree of randomness and disorder in a signal, also known as information entropy. Higher entropy indicates greater signal complexity, increasing the difficulty of signal prediction and analysis. Entropy can be used in signal classification, anomaly detection, and event recognition. The calculation formula is: (11) in, In representing probability distribution x The probability of occurrence; It represents the logarithm to the base 2. Figure 15 The image shows a comparison of the entropy characteristics of eight signals. Overall, the entropy of TR-RM-DCSK is closest to that of Gaussian white noise.

[0029] The area under the curve (AUC) is the area between the signal curve and the horizontal axis, usually calculated using numerical integration. It can be used to describe the overall size and amplitude of a signal, and also to compare the relative sizes of different signals. Figure 16 The image shows a comparison of the signal coverage area characteristics. Since the signal is a bipolar signal, the signal waveform is distributed in two parts above and below the horizontal axis, which shows the randomness of the signal and is not suitable as the main discrimination condition.

[0030] Peak values ​​are categorized into maximum and minimum peak values. The maximum number of peak values ​​refers to the number of the highest peaks in a signal, typically used to describe the number and height of peaks. It can be used to detect local extrema and peak locations, and is commonly used for peak detection and feature extraction. The minimum number of peak values ​​refers to the number of the lowest peaks in a signal, typically used to describe the number and depth of troughs. It can be used to detect local extrema and trough locations, and is commonly used for trough detection and feature extraction. The maximum and minimum peak values ​​for eight types of signals are shown below. Figure 17 , 18 As shown, there is no significant difference.

[0031] Based on the frequency domain chaotic signal characteristic analysis, it mainly includes the fundamental frequency and intermediate frequency of the spectrum.

[0032] Taking the fundamental frequency analysis of the characteristic spectrum of a chaotic signal in the frequency domain as an example, the fundamental frequency refers to the lowest fundamental frequency component of a signal in the frequency domain, often referred to as the lowest frequency component. Under frequency domain conditions, the fundamental frequency can be determined by calculating the power spectral density function (PSD) of the signal in the frequency domain, which usually corresponds to the lowest non-zero frequency component of the signal in the frequency domain.

[0033] The power spectral density function is a function that describes the energy distribution of a signal in the frequency domain, representing the density of signal power at different frequencies. When a signal is periodic, its frequency domain representation typically has a fundamental frequency component, which is the lowest non-zero frequency component, often referred to as the first harmonic component. The fundamental frequency refers to the lowest frequency component of a periodic signal, and can also be understood as the signal's dominant frequency or fundamental tone frequency. The spectral fundamental frequency is the lowest non-zero frequency component in the power spectral density function. The spectral fundamental frequency characteristic diagrams for eight signals are shown below. Figure 19 As shown.

[0034] The intermediate frequency (IF) is the center frequency of a signal's spectrum and is used to describe the signal's frequency distribution. The IF is typically obtained by calculating the center position of the signal's spectrum; it can also be understood as the average value of the signal's frequency distribution. It can be determined by calculating a weighted average of the signal's frequencies in the frequency domain, usually based on the signal's power spectral density function. For signals with symmetry, the spectral center is equal to the fundamental frequency. The calculation formula is: (12) in Indicates the center frequency of the signal's spectrum. P ( f ) represents the power spectral density function of the signal in the frequency domain. f The numerator represents the frequency. The numerator in the formula represents the power spectral density function of the signal at all frequency points. P ( f ) and frequency f The integral of the product of the products, with the denominator representing the power spectral density function of the signal. P ( f The integral over all frequency points. For example... Figure 20 The image shows the intermediate frequency (IF) characteristics of the spectrum of eight modulation signals. It can be seen that there is no significant difference in the IF of the spectrum of the eight signals.

[0035] Step 2: Based on the fuzzy evaluation method, select the DCSK signal that is closest to the noise characteristics for each signal feature obtained in Step 1. like Figure 8 As shown, a chaotic modulated signal is selected, and signal processing techniques are used to extract features such as variance, mean absolute value of difference, intermediate frequency of spectrum, and correlation. The obtained signal parameters are normalized. The normalized data information is analyzed to provide a reference for the relevant feature weights of the fuzzy comprehensive evaluation method, and also to provide a reference threshold for the parameters of the fuzzy evaluation matrix.

[0036] Chaotic modulation signals were selected as analysis samples, and fuzzy comprehensive evaluation method was used to obtain the evaluation of the pseudo-attitude, as shown in the appendix. Figure 21 As shown, the steps are as follows: Step 2.1, determine the fuzzy comprehensive evaluation index; Sixteen feature dimensions were extracted through signal feature analysis. It is indicated that their biological attitudes are used as evaluation indicators. The feature dimension parameters are explained below: -Mean; -Standard deviation; -variance; - Interquartile range; -Kushin; - Skewness; - Difference mean; - Mean of absolute difference; -difference median; -Median of the absolute value of the difference; - The sum of the absolute values ​​of the differences; -entropy; -Curve coverage area; -Number of peaks; - Spectrum fundamental frequency; -Intermediate frequency spectrum; Step 2.2: Establish a comprehensive evaluation set; Five rating levels were assigned to the 16 signal characteristic parameters mentioned above. That is, there are 5 level sets; each level corresponds to a fuzzy subset, respectively. Five levels of classification; Step 2.3: Perform single-factor fuzzy evaluation and obtain the evaluation matrix; The evaluation matrix is ​​obtained by performing a single-factor fuzzy evaluation of image parameters based on the selected data features: (13) Step 2.4: Determine the weight vectors of the evaluation factors; Based on the data features extracted from the data source, the weights of the 16 feature dimensions are evaluated as follows: (14) Step 2.5: Synthesize the fuzzy comprehensive evaluation result vector; (15) in It is composed of the weight matrix With the evaluation matrix The Obtained through columnar operations, it represents the overall view of... The degree of membership of a hierarchical fuzzy subset; Step 2.6: Analyze the fuzzy comprehensive evaluation result vector; Based on the principle of maximum membership, The maximum value is used to determine the degree of mimicry of the signal, which can be obtained from formulas (13)-(15): (16) The optimal differential chaotic shift keying communication signal can be determined by the evaluation results of formula (16).

Claims

1. A differential chaotic shift keying communication signal selection method based on feature fusion, characterized in that, Specifically, the following steps are included: Step 1: Obtain the results under the conditions of the same half-spreading factor, same mapping, and same sampling frequency. N Differential chaotic shift keying communication signals are categorized into several types, each with the same signal length. a Group samples, calculate each signal separately n The sample groups are based on the characteristics of the statistical domain, time domain, and frequency domain, and the normalized values ​​of these characteristics are used for comparative analysis. in N , a It is a positive integer; Statistical domain features include: mean, variance, standard deviation, interquartile range, kurtosis, and skewness; frequency domain features include: fundamental frequency and intermediate frequency; time domain features include: difference mean, difference absolute mean, difference median, difference absolute median, sum of difference absolutes, entropy, curve coverage area, and number of peaks. Step 2: For each signal feature obtained in Step 1, based on the fuzzy evaluation method, select the DCSK signal that most closely matches the noise characteristics; Step 2 specifically involves: Step 2.1, determine the fuzzy comprehensive evaluation index; Sixteen feature dimensions were extracted through signal feature analysis. It is indicated that their biological attitudes are used as evaluation indicators. The feature dimension parameters are explained below: -Mean; -Standard deviation; -variance; - Interquartile range; -Kushin; - Skewness; - Difference mean; - Mean of absolute difference; -difference median; -Median of the absolute value of the difference; - The sum of the absolute values ​​of the differences; -entropy; -Curve coverage area; -Number of peaks; - Spectrum fundamental frequency; -Intermediate frequency spectrum; Step 2.2: Establish a comprehensive evaluation set; Five rating levels were assigned to the 16 signal characteristic parameters mentioned above. That is, there are 5 level sets; each level corresponds to a fuzzy subset, respectively. Five levels of classification; Step 2.3: Perform single-factor fuzzy evaluation and obtain the evaluation matrix; Based on the selected data features, a single-factor fuzzy evaluation of the DCSK signal parameters is performed, resulting in an evaluation matrix: (13) Step 2.4: Determine the weight vectors of the evaluation factors; Based on the data features extracted from the data source, the weights of the 16 feature dimensions are evaluated as follows: (14) Step 2.5: Synthesize the fuzzy comprehensive evaluation result vector; (15) in It is composed of the weight matrix With the evaluation matrix The Obtained through columnar operations, it represents the overall view of... The degree of membership of a hierarchical fuzzy subset; Step 2.6: Analyze the fuzzy comprehensive evaluation result vector; Based on the principle of maximum membership, for , , , , The maximum value in the range is used to determine the degree of mimicry of the signal, which is obtained from formulas (13)-(15): = (16) in The DCSK signal with the closest noise characteristics is determined; the optimal differential chaotic shift keying communication signal is determined by the evaluation results of formula (16).

2. The differential chaotic shift keying communication signal selection method based on feature fusion according to claim 1, characterized in that, The mean is the average of a set of data, which is the sum of all data points divided by the number of data points; Its expression is: （1） in, This represents the mean of the data; This represents each observation in the data. n Represents the number of data points, a positive integer; The variance is a measure of the dispersion of a set of data; it is the average of the sum of the squares of the differences between each data point and the mean. Its expression is: （2） in, Indicates variance; This represents each observation in the data; This represents the average value of the data. The standard deviation is a measure of the dispersion of a set of data; it measures the distance between data points and the mean. Its expression is: （3） in, Indicates standard deviation; This represents each observation in the data; This represents the average value of the data. The interquartile range is the difference between the upper quartile and the lower quartile. The quartile is the value at each dividing point after sorting all values ​​from smallest to largest and dividing them into four equal parts. The skewness, which measures the direction and degree of skewness in the distribution of statistical data, is expressed as follows: （4） in, Indicates skewness, This represents each observation in the data; This represents the average value of the data. The standard deviation of the data; This represents the mean. E Represents the mathematical expectation. X This refers to the current data observation value; Kurtosis is used to describe the peak value of a probability density distribution curve at its mean; the calculation formula is: （5） in, Indicates kurtosis, This represents each observation in the data; This represents the average value of the data. The standard deviation of the data is represented by kurtosis; a negative kurtosis indicates a angular data distribution; a zero kurtosis indicates a normal data distribution; and a positive kurtosis indicates a flat data distribution.

3. The differential chaotic shift keying communication signal selection method based on feature fusion according to claim 2, characterized in that, The Mean Differences refers to the average difference between two adjacent data points in a time series; the calculation formula is: （6） in, For the first in the time series i One element; The mean absolute differences refer to the average of the absolute values ​​of the differences between two adjacent data points in a time series; its expression is: （7） in, The average of the differences between two adjacent elements is also called the difference mean. The average of the absolute values ​​of the differences between two adjacent elements is also called the mean of the absolute values ​​of the differences. This represents the i-th element in the time series; Represents the first in a time series i -1 element; n This represents the number of elements in the time series; the Median Differences refers to the median difference between two adjacent data points in the time series. The formula for calculating this difference is: （8） Where median represents the median of the differences between all adjacent elements; The median absolute difference refers to the median of the absolute values ​​of the differences between two adjacent data points in a time series. The formula for calculating this median is: （9） The sum of absolute differences refers to the sum of the absolute values ​​of the differences between two adjacent data points in the time series; the calculation formula is: （10） Entropy, a metric for measuring signal complexity, is calculated using the following formula: （11） in, In representing probability distribution x The probability of occurrence; Represents the logarithm to the base 2; The area under the curve refers to the area between the signal curve and the horizontal axis; the signal curve is generated by differential chaotic shift keying communication signal through MATLAB simulation. The peak values ​​are divided into maximum peak value and minimum peak value; the number of maximum peak values ​​refers to the number of the highest peak values ​​in the signal, which is used to describe the number and height of the signal peaks; the number of minimum peak values ​​refers to the number of the lowest peak values ​​in the signal, which is used to describe the number and depth of the signal troughs.

4. The differential chaotic shift keying communication signal selection method based on feature fusion according to claim 3, characterized in that, The fundamental frequency of the spectrum refers to the lowest fundamental frequency component of a signal in the frequency domain. Under frequency domain conditions, the fundamental frequency of the spectrum is determined by calculating the power spectral density function (PSD) of the signal in the frequency domain, which corresponds to the lowest non-zero frequency component of the signal in the frequency domain. The power spectral density function is a function that describes the energy distribution of a signal in the frequency domain, representing the density of signal power at different frequencies. The intermediate frequency (IF) is the frequency at the center of the signal spectrum, used to describe the frequency distribution of the signal. The IF is obtained by calculating the center position in the signal spectrum, and can also be understood as the average value of the signal's frequency distribution. It is determined by calculating the weighted average value of the signal in the frequency domain, based on the power spectral density function. The spectral center of a signal spectrum with symmetry is equal to the fundamental frequency. The calculation formula is: （12） in The frequency of the signal's spectrum, which is the frequency domain form of the time-domain data, P ( ) represents the power spectral density function of the signal in the frequency domain. Represents frequency; represents the power spectral density function of the signal at all frequency points. P ( ) and frequency The integral of the product of the products, with the denominator representing the power spectral density function of the signal. P ( f The integral over all frequency points.

Images