Frequency hopping carrier signal detection method, system and device based on time-frequency analysis and storable medium

Abstract

The invention discloses a frequency hopping carrier signal detection method, system and device based on time-frequency analysis and a storable medium, and belongs to the technical field of frequency hopping carrier signal detection. According to the detection method, firstly, a sampling signal is subjected to combined time-frequency analysis transformation (STFTamp; according to the method, a to-be-detected time-frequency graph is obtained through a WVD method, structural elements of different scales are constructed through self-adaptive morphological filtering to process the time-frequency graph, noise and burst signals are removed, influences on to-be-detected frequency hopping signals are avoided, and the detection accuracy is improved. And then, aiming at the residual fixed-frequency signal and sweep-frequency signal interference, Hough transform is used for carrying out feature extraction on a straight line, and a frequency hopping signal is judged according to different time-frequency features of different signals, so that the detection process of the frequency hopping signal is completed. The detection method can be applied to a wireless sensor network and is used for realizing data communication and information transmission between sensor nodes. Through detection and analysis of the frequency hopping signal, the communication stability and reliability of the sensing network in a complex electromagnetic environment can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of frequency hopping carrier signal detection, and in particular to a frequency hopping carrier signal detection method, system, device and storable medium based on time-frequency analysis. Background Art

[0002] Communication frequency hopping technology is an effective method to prevent radio interference. It has been paid attention to by many countries and is widely used in mobile communications and military communications. In the shortwave frequency band, it can overcome various channel problems, such as channel leakage, multipath delay and Doppler shift. Shortwave frequency hopping communication is crucial in military communications. It can maintain effective communication in complex electromagnetic environments and resist various interferences. The new generation of radio stations has faster hopping speeds and better performance. These technological advances have brought challenges and opportunities to shortwave frequency intelligent technology. Fast and accurate locking of frequency hopping signals has become a key task for shortwave frequency measurement. In complex communication environments, especially in the field of military communications, inaccurate detection results will not only lead to difficulties in obtaining information and waste of resources, but more seriously, it may be at a disadvantage in confrontation and have a fatal impact on the entire military environment. Therefore, it is of great significance to accurately detect frequency hopping signals in actual complex communication environments.

[0003] The main difficulty in detecting frequency-hopping carrier signals is that their instantaneous frequency is unstable and the hopping is irregular. Accurate detection under unknown hopping speed and hopping range is a key challenge. The time-frequency analysis method analyzes the signal characteristics by constructing a joint distribution function in the time-frequency domain, converting electromagnetic waves into time-frequency information, and better reflecting the signal characteristics. This method has become a hot topic in frequency-hopping carrier detection.

[0004] In recent years, image processing and deep learning technologies have developed rapidly in the field of target detection and have begun to be applied to frequency hopping signal analysis in military communications. These technologies provide new methods for solving the blind detection problem of frequency hopping carrier signals.

[0005] Xie Peng, Liang Rui and others proposed an algorithm for detecting the time domain parameters of underwater frequency hopping signals. They used the inverse time-frequency transform to obtain the signal in the time domain, and finally extracted the time domain ridge to calculate the energy, thereby obtaining the parameters of the frequency hopping signal and completing the detection process of the frequency hopping signal. Lü Lequn, Yi Yunqing and others judged the state of the signal, and then used a mechanism to correct the parameters of the judged state, improving the measurement accuracy of the time-frequency parameters in the process of the frequency hopping signal, completing the detection of the frequency hopping signal and the parameter estimation of the frequency hopping signal.

[0006] Zhu Wengui, Xie Liyun and others provided a blind detection method for frequency hopping signals based on array signal processing. First, the power spectrum density matrix of the frequency hopping signal is calculated, and then the characteristic parameters of the frequency hopping signal to be detected are obtained from the matrix, thereby realizing the detection of the frequency hopping signal. This method can effectively overcome the interference of fixed-frequency signals and swept-frequency signals on the signals to be detected, but the computational complexity of the entire algorithm is relatively high. Gu Pei, Wang Bin and others used short-time Fourier transform for time-frequency conversion, converted the complex electromagnetic signal to be detected into a time-frequency graph through time-frequency analysis, and then processed the time-frequency graph through image processing methods, but this method also has certain problems, that is, the time-frequency resolution is relatively low and the anti-noise performance is relatively poor. Liu Xubo divided the power spectrum into two sections for analysis, used mathematics to derive the detection threshold, and completed the detection of the frequency hopping signal, but if there is a fixed-frequency signal, misjudgment will occur. Summary of the invention

[0007] In view of the above problems, the present invention aims to provide a method, system, device and storable medium for detecting frequency hopping carrier signals based on time-frequency analysis. First, the signal data is converted into a time-frequency diagram to improve the image feature resolution of the algorithm and reduce the influence of cross-term interference; then, structures of different scales are constructed, and adaptive morphological filtering is used to filter the converted time-frequency diagram to remove Gaussian noise and burst signals; finally, the Canny operator and Hough transform method are used to obtain the signal segment coordinates, and each signal is distinguished according to the different characteristics of the signal on the time-frequency diagram to complete the required frequency hopping signal detection. This method can improve the detection performance of frequency hopping signals and effectively improve the communication stability and reliability of sensor networks in complex electromagnetic environments.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0009] The frequency hopping carrier signal detection method based on time-frequency analysis includes the following steps:

[0010] S1: Use the combined time-frequency analysis transform STFT&WVD method to obtain the time-frequency diagram to be tested on the sampled signal;

[0011] S2: Binarization and adaptive morphological filtering are performed on the time-frequency image to be inspected to filter out the burst signal;

[0012] S3: Use edge detection to smooth the remaining image obtained in step S2, and use Hough transform to extract features of the remaining image;

[0013] S4: According to the image signal features extracted in step S3, the signal form is determined and the frequency hopping signal is detected.

[0014] Furthermore, the specific operation of step S1 includes the following steps:

[0015] S101: Perform short-time Fourier transform on the sampled signal to obtain STFT x (t,f);

[0016] S102: Perform Welch transformation on the sampled signal to obtain WVD x (t,f);

[0017] S103: Combine the time-frequency analysis transformation results to obtain the time-frequency diagram to be tested. The STFT&WVD combination result is recorded as

[0018] z(t,f)=WVD x (t,f)⊙STFT x (t,f)

[0019] Among them, ⊙ is the Hadamard product operation symbol.

[0020] Furthermore, the specific operation of step S2 includes the following steps:

[0021] S201: binarizing the time-frequency graph using the maximum variance threshold method;

[0022] S202: performing adaptive morphological filtering on the binarized time-frequency image.

[0023] Furthermore, the specific operation of step S202 includes the following steps:

[0024] S2021: define a rectangular element Z1, perform an expansion operation ⊕ on the image F(x, y) after the binarization processing in step S201, and then perform an erosion operation Θ, which is specifically expressed as

[0025]

[0026] S2022: Construct straight line element X1 to F 1 (x,y) is opened and we get

[0027]

[0028] Among them, · represents the opening operation;

[0029] S2023: To fill the holes in the target area, use the straight line element X2 to F 2 (x,y) is closed, and we get

[0030]

[0031] in, Represents the closing operation.

[0032] Furthermore, the specific operation of step S3 includes the following steps:

[0033] S301: Use the Canny operator to perform edge detection on the remaining image;

[0034] S302: extracting features of the remaining image after edge detection using Hough transform, including width, height, length and angle of the image;

[0035] Among them, the width and height of each signal segment are obtained through the [m,n]=size(D(t,f)) function, and then the length of each signal segment is calculated by Len=abs(x2-x1), and the angle

[0036] Ang = abs((y2-y1) / (x2-x1)).

[0037] Furthermore, the specific operation of step S301 includes the following steps:

[0038] S3011: preprocessing the remaining image after processing in step S2 using Gaussian filtering;

[0039] S3012: Calculate the components of each pixel in the x direction and the y direction, expressed as g x and g y ,

[0040]

[0041] Then the pixel gradient size g θ and direction θ are respectively expressed as

[0042]

[0043] Where S(x,y) represents the pixel intensity of the image at position (x,y), S(x+1,y) is the intensity of the pixel on the right, and S(x-1,y) is the intensity of the pixel on the left; S(x,y+1) is the intensity of the pixel below, and S(x,y-1) is the intensity of the pixel above;

[0044] S3013: After gradient calculation, edge detection is performed using the Canny edge detection algorithm.

[0045] Furthermore, the specific operation of step S4 includes the following steps:

[0046] S401: According to the characteristics of sweep frequency signal transmission, the threshold angle zs_d is set to 5 degrees. When Ang>5, it proves that the sweep frequency signal exists;

[0047] S402: according to the characteristics of the fixed frequency signal during transmission, the length threshold is set to zs_m, where m is the length of the time-frequency graph. When Len>zs_m=m / 3, it is determined that a fixed frequency signal exists;

[0048] S403: judging the remaining signal segments, setting the threshold value of the frequency hopping signal segment to zs_n=6, where n represents the number of signal segments of the frequency hopping signal, and when n>zs_n, it is judged that the frequency hopping signal exists.

[0049] Furthermore, the present invention also includes a frequency hopping carrier signal detection system based on time-frequency analysis, wherein the signal detection system includes a signal conversion module, a burst signal filtering module, a signal feature extraction module and a judgment module;

[0050] The signal transformation module is used to obtain the time-frequency diagram to be tested by using the combined time-frequency analysis transformation STFT&WVD method on the sampled signal;

[0051] The burst signal filtering module is used to perform binarization processing and adaptive morphological filtering on the time-frequency graph to be inspected to filter out the burst signal;

[0052] The signal feature extraction module is used to smooth the remaining image after filtering out the burst signal, and use Hough transform to extract the features of the remaining image;

[0053] The judgment module is used to judge the signal type;

[0054] The signal conversion module, burst signal filtering module, signal feature extraction module and judgment module are all implemented by the signal detection method as described above.

[0055] Furthermore, the present invention also includes an electronic device, which includes at least one processor; and a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, and the instructions are executed by the processor so that the processor can execute the signal detection method as described above.

[0056] Furthermore, the present invention also includes a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the signal detection method as described above.

[0057] The beneficial effects of the present invention are:

[0058] The present invention proposes a frequency hopping carrier signal detection method based on wireless sensor network and time-frequency analysis. The method first performs STFT&WVD combined time-frequency analysis on the signal to be tested to ensure the suppression of interference rate and improve the time-frequency resolution, and uses adaptive morphological filtering to eliminate the interference of the time-frequency graph noise and burst signal, so that the time-frequency graph is clearer and more specific. Finally, the remaining fixed-frequency signal, sweep frequency signal and frequency hopping signal are judged by Hough transform, and the result of detecting the frequency hopping signal is finally achieved. The simulation results show that the frequency hopping detection method has a high detection probability in a complex electromagnetic environment with multiple interference signals, and has good anti-noise performance despite the low signal-to-noise ratio. It is a feasible time-frequency detection method. The frequency hopping signal detection algorithm based on combined time-frequency analysis, multi-scale morphological filtering and Hough transform can help the sensor network better avoid electromagnetic interference and improve the detection accuracy and anti-interference ability of the signal. This is crucial for information transmission in military operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a flow chart of the frequency hopping carrier signal detection method in the present invention.

[0060] Figure 2 It is the time-frequency diagram of the signal corresponding to different window function sizes in the present invention.

[0061] Figure 3 It is a contour map obtained after STFT transformation in the present invention.

[0062] Figure 4 It is the contour map obtained after WVD transformation in the present invention.

[0063] Figure 5 This is the time-frequency diagram after binarization and the first closed-loop operation in the present invention.

[0064] Figure 6 This is the time-frequency diagram after the burst signal is filtered out in the present invention.

[0065] Figure 7 It is the detection result of the Canny operator in the present invention.

[0066] Figure 8 It is the frequency hopping signal detection result in the present invention.

[0067] Fig. 9 This is a diagram showing the accuracy of frequency hopping signal detection with different window lengths in the first simulation experiment of the present invention.

[0068] Fig.10 This is a spectrum diagram of the simulated frequency hopping signal in the second simulation experiment of the present invention.

[0069] Fig.11These are the frequency hopping signal detection success probability curves of the three algorithms under different signal-to-noise ratio conditions in the second simulation experiment of the present invention. DETAILED DESCRIPTION

[0070] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0071] Embodiment 1:

[0072] Embodiment 1 provides a frequency hopping carrier signal detection method based on time-frequency analysis, as shown in the attached Figure 1 As shown, the specific steps include:

[0073] S1: Use the combined time-frequency analysis transform STFT&WVD method to obtain the time-frequency diagram to be tested on the sampled signal;

[0074] Specifically, S101: perform short-time Fourier transform (STFT) on the sampled signal to obtain STFT x (t,f);

[0075] A Hamming window function is selected to smooth the signal and reduce edge effects; the signal is framed into multiple overlapping parts, with the length of each frame matching the window function; the window function is applied to each frame; a Fourier transform is performed on each processed frame; the transform results of all frames are summarized into a two-dimensional matrix, showing how the signal changes with time and frequency.

[0076] Since the noise in the time-frequency graph interferes with the signal to be detected, a threshold value ε is set. If the pixel value in the time-frequency graph is lower than the set threshold value, it is set to 0, and if it is higher than the set threshold value, it is set to 1, that is,

[0077]

[0078] The expression for setting the threshold value is:

[0079]

[0080] Where σ is the threshold factor, the signal-to-noise ratio affects the value of the threshold factor, and STFT x (n,k) is the discretization representation of STFT, N represents the number of sampling points in time, n represents the time point, and k represents the frequency point.

[0081] The window function sizes are set to 65, 129, 256, and 512 respectively, and the sliding step is 25% of the window length. The time-frequency diagram of the signal is shown in the attached figure. Figure 2 As shown. Figure 2It can be clearly seen that the frequency hopping signal is more blurred and the time-frequency resolution is relatively low when the window length is low. As the window length increases, the clarity of the frequency hopping signal and the degree of energy concentration are improved, but the detection time is also increasing during the detection process, which means that the amount of calculation is also increasing. When the window length is set to 512, not only the time-frequency diagram after STFT has a better effect, but also the process of detecting the frequency hopping signal has a better effect. Set the window length to 512, and the contour map obtained after STFT transformation is shown in the attached figure. Figure 3 shown.

[0082] S102: Perform Welch transformation (WVD) on the sampled signal to obtain WVD x (t,f);

[0083] Similarly, we first select the window function (window length is 512) and divide the signal into frames; calculate the autocorrelation function for each frame; perform Fourier transform on the autocorrelation function; combine the transformation results of all frames into a two-dimensional matrix to display the signal characteristics in terms of time and frequency. The contour map obtained after WVD transformation is shown in the attached figure. Figure 4 shown.

[0084] S103: Combine the time-frequency analysis transformation results to obtain the time-frequency diagram to be tested. The STFT & WVD combination result is recorded as z(t,f)=WVD x (t,f)⊙STFT x (t,f);

[0085] Among them, ⊙ is the Hadamard product operation symbol.

[0086] Further, step S2: performing binarization processing and adaptive morphological filtering on the time-frequency graph to be inspected to filter out the burst signal;

[0087] Specifically, S201: using the maximum variance threshold method to perform binarization processing on the time-frequency graph;

[0088] First, the original grayscale values ​​of the time-frequency spectrum are divided into two different classes from large to small and from the middle value, and then the average value of the number of pixels and the average value of the grayscale of the two classes are calculated. After calculating the average value of the number of pixels and the average value of the grayscale, these values ​​are used to calculate the inter-class variance. After completing this process, the maximum value of the inter-class variance is selected as the corresponding grayscale threshold. The calculation formula of the inter-class variance is as follows:

[0089] σ(i) 2 =N1(i)×N2(i)×[V1(i)-V2(i)] 2

[0090] Where i is the grayscale threshold, N1(i) is the number of pixels less than the grayscale threshold, N2(i) represents the number of pixels greater than or equal to the grayscale threshold, V1(i) and V2(i) are the average values ​​of N1(i) and N2(i), respectively, and σ(i) 2 It is a function with gray value as independent variable, find the position of the maximum value of the threshold: t = argMaxσ(i) 2 .

[0091] After obtaining the threshold, the original spectrum z(t,f) is binarized according to the following formula to obtain the binary image F(x,y):

[0092]

[0093] In actual operation, first select the middle value T to distinguish black and white pixels. Then, compare the gray value G(x,y) of each pixel (x,y) of the spectrum graph with the threshold T. Pixels with large gray values ​​are set to white in the binary graph F(x,y), and pixels with small gray values ​​are set to black. The binarized time-frequency graph is shown in the attached figure. Figure 5 As shown in (a).

[0094] S202: performing adaptive morphological filtering on the binarized time-frequency image.

[0095] Specifically, S2021: In order to implement the closed operation, first define a rectangular element Z1, which is a small two-dimensional matrix. Then, perform an expansion operation ⊕ on the binary image F(x, y) to increase the white area in the image, which helps fill the defects. Then perform an erosion operation Θ to reduce the white area to restore the excessive enlargement that may be caused by the expansion, ensuring that the original structure of the image is not destroyed. The time-frequency diagram after the first closed-loop operation is shown in the attached figure. Figure 5 As shown in (b), it is specifically expressed as

[0096]

[0097] S2022: In order to remove the interference caused by the burst signal to the detection process, construct the straight line element X1 to F 1 (x,y) is opened and we get

[0098]

[0099] Here, · represents an opening operation.

[0100] S2023: To fill the holes in the target area, use the straight line element X2 to F 2 (x, y) is closed and we get

[0101]

[0102] in, Represents the closing operation.

[0103] The burst signal can be filtered out through steps S201 and S202. The time-frequency diagram after filtering out the burst signal is shown in the attached figure. Figure 6 shown.

[0104] Further, step S3: using edge detection to smooth the residual image obtained in step S2, and using Hough transform to extract features of the residual image;

[0105] After the operation of step S1 and step S2, the noise signal interference and the burst signal interference can be removed, and the remaining swept frequency signal interference and the fixed frequency signal interference are mostly uniformly displayed on the time-frequency diagram, which jump from low frequency to high frequency or keep the frequency unchanged for a long time. Therefore, the Hough transform method which is more sensitive to straight line detection can be used to achieve the purpose of detecting swept frequency signals and fixed frequency signals. Specifically,

[0106] S301: Use the Canny operator to perform edge detection on the remaining image;

[0107] Hough transform is used to identify straight lines, circles and other shapes in images. When Hough transform is performed on a binary image, because the binary image converts grayscale information into black and white pixels, the edge details may be lost in the Hough transform, thus affecting the detection of curves or lines. In addition, because the binary image is sensitive to noise, noise points may be misjudged as edge points, which may cause the curves or lines detected by the Hough transform to be inconsistent with the actual situation, affecting the accuracy of the results. Therefore, edge detection should be performed on the image first.

[0108] More specifically, S3011: Before edge detection is performed, the image must be processed first for good detection results in the subsequent detection process. In the present invention, a Gaussian filter is used to process the image. The Gaussian filter uses a two-dimensional Gaussian function as a weight to calculate the weighted average between each pixel and its adjacent pixels, and assigns the average to the original pixel, thereby achieving smoothing of image noise. This processing process is a prior art and will not be described in detail in the present invention.

[0109] S3012: Calculate the gradient of each pixel in the image through the edge detection filter. The components of each pixel in the x direction and y direction are represented as g and x and g y , expressed as

[0110]

[0111] Among them, S(x,y) represents the pixel intensity of the image at position (x,y), S(x+1,y) is the intensity of the pixel on the right, S(x-1,y) is the intensity of the pixel on the left; S(x,y+1) is the intensity of the pixel below, and S(x,y-1) is the intensity of the pixel above.

[0112] Then the pixel gradient size g θ and direction θ are respectively expressed as

[0113]

[0114] S3013: After the gradient is calculated, the Canny edge detection algorithm is used to perform edge detection. For each pixel in the image, its gradient size is compared with the gradient size of the adjacent pixel in its gradient direction. If the gradient size of the current pixel is not the maximum value in the neighborhood along its gradient direction, the gradient size of the pixel is set to 0. This process suppresses non-edge points and only retains the local maximum value of the gradient, thereby refining the edge. The Canny operator detection results are shown in the attached figure. Figure 7 shown.

[0115] S302: Using Hough transform to extract features of the remaining image after edge detection, including width, height, length and angle of the image.

[0116] More specifically, the width and height of each signal segment are obtained through the function [m,n]=size(D(t,f)). Then the length Len=abs(x2-x1) and the angle Ang=abs((y2-y1) / (x2-x1)) of each signal segment are calculated.

[0117] Further, step S4: according to the image signal features extracted in step S3, the signal form is determined and the frequency hopping signal is detected.

[0118] Specifically, S401: Generally, the mixed signal includes the following signal types: a frequency hopping signal, a fixed frequency signal, a swept frequency signal and a burst signal.

[0119] According to the characteristics of swept frequency signal transmission, different frequencies are crossed during the transmission process. Therefore, among the various types of signals in the time-frequency diagram, only the swept frequency signal has an angle difference, and other signals such as fixed frequency signals and frequency hopping signals do not have angles. Therefore, based on this characteristic, it is judged whether there is a swept frequency signal. Set the threshold angle zs_d to 5 degrees. When Ang>5, it proves that there is a swept frequency signal;

[0120] S402: According to the characteristics of fixed frequency signals during transmission, the duration is relatively long during transmission and the transmission at the same frequency does not change. In comparison, the frequency of frequency hopping signals is not fixed and the duration at a certain frequency is short. Therefore, after extracting features by Hough transform, when the signal segment is longer than the set fixed frequency signal threshold, it is considered to be a fixed frequency signal. Set the length threshold to zs_m, where m is the length of the time-frequency graph. When Len>zs_m=m / 3, it is determined that there is a fixed frequency signal.

[0121] S403: The remaining signal segments are judged. The same length of these signal segments means the same hopping duration, and different vertical coordinates mean that the frequency hopping signal has hopping. The threshold value of the frequency hopping signal segment is set to zs_n=6, where n represents the number of signal segments of the frequency hopping signal. When n>zs_n, it is judged that a frequency hopping signal exists. Figure 8 shown.

[0122] Simulation experiment 1:

[0123] In simulation experiment 1, the frequency hopping signal detection method of the present invention is used to detect the frequency hopping signal under different window lengths and different signal-to-noise ratio environments.

[0124] The window lengths are set to 65, 129, 256, and 512 respectively. The accuracy of frequency hopping signal detection with different window lengths is shown in the attached figure. Fig. 9 As shown. Fig. 9 It can be seen that under different signal-to-noise ratio environments, when the window length is set to 512, the detection accuracy of the frequency hopping signal is the highest. Figure 2 It can be seen that when the window length is set to 512, not only the time-frequency diagram after STFT has a better effect, but also the frequency hopping signal detection process has a better effect.

[0125] Simulation experiment 2:

[0126] The simulation experiment is used to compare the frequency hopping signal detection method in the present invention with the detection method in the document "Zhao Fangchao, Jiang Jianzhong, Guo Junli, et al. Time-frequency detection algorithm based on adaptive morphological edge detection [J]" (hereinafter referred to as Comparison 1) and the detection method in the document "Xi Youyou, Li Zhigang, Wang Miao. Parameter estimation of direct spread / frequency hopping signals combined with image processing and time-frequency analysis [J]" (hereinafter referred to as Comparison 2). The specific simulation process includes the following contents:

[0127] The transmitter performs baseband modulation on the input information code sequence to obtain the modulated signal m(t). Then, the modulated signal m(t) modulates the random carrier frequency to obtain the frequency hopping signal

[0128] The modulated frequency hopping signal will be transmitted together with the noise signal n(t) and the interference signal J(t) at the transmitting end to obtain the signal h(t) = S at the receiving end. i (t)+n(t)+J(t). The signal h(t) at the receiving end is different from the local signal cos[(ω0+nω Δ )t+φ n ] After multiplication, we get In the formula, ω Δ is the frequency hopping interval, ω0 is the starting transmission frequency, is the initial phase. Then the integrated signal is passed through a low-pass filter to obtain

[0129] A simulation model of a frequency hopping carrier communication system is established on the Simulink simulation platform, and the parameters are as follows: data rate: 100bps; data modulation mode: BFSK; number of frequency hopping frequency points: 32; frequency hopping frequency interval: 1.2MHz; frequency hopping rate: 1000 hops / s; channel: AWGN channel. The signal transmission rate is 100bps, 100 bits are transmitted per second, and the frequency hopping rate is 1000 hops per second. This frequency hopping carrier communication system belongs to a fast frequency hopping system. In the process of system simulation using the Simulink simulation platform, five modules are mainly used: signal generation module, BFSK modulation module, channel module, and carrier generation module. The spectrum diagram of the simulated frequency hopping signal is attached. Fig.10 shown.

[0130] In the frequency hopping communication system, the signal generation module uses the Bernoulli Binary Generator of Simulink to generate a binary random sequence, and sets the data transmission rate to 100 bits / s and the sampling time to 0.01s.

[0131] The simulation directly uses the multi-tone interference model of the sine wave generator module in the Simulink module library to study the anti-interference performance analysis of the system. Multiple fixed-frequency signals in the transmission environment are simulated; the interference of broadband noise is simulated using the AWGN module in the Simulink simulation platform; the bandpass filter module is used to filter the Gaussian noise to generate partial frequency band interference;

[0132] BFSK modulation is implemented using the M-FSKModulatorBaseband module of Simulink, with M set to 2. Demodulation also uses a similar module, with M set to 2 for BFSK demodulation.

[0133] The carrier generation module uses the Subsystem PN Sequence to generate a pseudo-random frequency hopping carrier. The PN sequence generator generates a pseudo-random sequence to control the frequency hopping. After being processed by the to sample and Bit to Integer Converter modules, the bit information is converted into integer information to control parameters such as the frequency hopping rate.

[0134] In actual communication, the signal will be attenuated due to environmental influences, so the amplifier is needed to resist attenuation and ensure signal power for demodulation. At the same time, there is Gaussian white noise in the communication system, which is superimposed on the signal. In the simulation of the frequency hopping system, the additive Gaussian white noise channel module of Simulink is used to simulate this internal noise.

[0135] In different signal-to-noise ratio environments, the accuracy curves of the three detection methods for detecting frequency hopping signals are shown in the figure below: Fig.11 As shown in the figure, under different signal-to-noise ratio conditions, the frequency hopping signal detection frequency (P) of the three algorithms is P = J / L. Where: J is the number of trials to detect the frequency hopping signal in the observation data containing the frequency hopping signal, and L is the number of Monte Carlo experiments.

[0136] Fig.11 The frequency hopping signal detection success probability curves of the three algorithms under different signal-to-noise ratio conditions. It can be seen from the figure that in a high signal-to-noise ratio environment, the detection success rates of the three algorithms are relatively high, but in a low signal-to-noise ratio environment, the algorithm of the present invention is more adaptable than the other two algorithms. When the signal-to-noise ratio is -7dB, the detection success rate of the algorithm in the present invention can exceed 80%. However, when the fixed-frequency signal and the burst signal have the same frequency as the frequency hopping signal, and overlap with the frequency hopping signal, a missed alarm phenomenon will occur. However, if this situation only occurs in a small range, it will not have much impact on the correct detection of the signal. When the signal-to-noise ratio is too small, the interference to the signal is too large, resulting in a sudden drop in the probability of successful algorithm detection. Compared with the comparison algorithm, the algorithm in the present invention still has a higher detection rate under low signal-to-noise ratio conditions.

[0137] Embodiment 2:

[0138] Embodiment 2 provides a frequency hopping carrier signal detection system based on time-frequency analysis, the signal detection system comprising a signal conversion module, a burst signal filtering module, a signal feature extraction module and a judgment module;

[0139] The signal transformation module is used to obtain the time-frequency diagram to be tested by using the combined time-frequency analysis transformation STFT&WVD method on the sampled signal;

[0140] The burst signal filtering module is used to perform binarization processing and adaptive morphological filtering on the time-frequency graph to be inspected to filter out the burst signal;

[0141] The signal feature extraction module is used to smooth the remaining image after filtering out the burst signal, and use Hough transform to extract the features of the remaining image;

[0142] The judgment module is used to judge the signal type;

[0143] It should be noted that the signal conversion module, the burst signal filtering module, the signal feature extraction module and the judgment module in the present invention are all implemented by the method in the first embodiment.

[0144] Embodiment three:

[0145] Embodiment three provides an electronic device, which includes at least one processor; and a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, and the instructions are executed by the processor so that the processor can execute the signal detection method described in embodiment one.

[0146] Embodiment 4:

[0147] Embodiment 4 provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the signal detection method described in Embodiment 1.

[0148] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A frequency hopping carrier signal detection method based on time-frequency analysis, characterized in that: The following steps are included: S1: Use the combined time-frequency analysis transform STFT&WVD method to obtain the time-frequency diagram to be tested on the sampled signal; S2: Binarization and adaptive morphological filtering are performed on the time-frequency image to be inspected to filter out the burst signal; S3: Use edge detection to smooth the remaining image obtained in step S2, and use Hough transform to extract features of the remaining image; S4: According to the image signal features extracted in step S3, the signal form is determined and the frequency hopping signal is detected.

2. The method for detecting a frequency hopping carrier signal based on time-frequency analysis according to claim 1, characterized in that: The specific operation of step S1 includes the following steps: S101: Perform short-time Fourier transform on the sampled signal to obtain STFT x (t,f); S102: Perform Welch transformation on the sampled signal to obtain WVD x (t,f); S103: Combine the time-frequency analysis transformation results to obtain the time-frequency diagram to be tested. The STFT&WVD combination result is recorded as z(t,f)=WVD x (t,f)⊙STFT x (t,f) Among them, ⊙ is the Hadamard product operation symbol.

3. The method for detecting a frequency hopping carrier signal based on time-frequency analysis according to claim 2, characterized in that: The specific operation of step S2 includes the following steps: S201: binarizing the time-frequency graph using the maximum variance threshold method; S202: performing adaptive morphological filtering on the binarized time-frequency image.

4. The method for detecting a frequency hopping carrier signal based on time-frequency analysis according to claim 3, characterized in that: The specific operation of step S202 includes the following steps: S2021: Define a rectangular element Z1 and perform dilation operation on the image F(x, y) after the binarization processing in step S201 Then the corrosion operation Θ is performed, which is specifically expressed as S2022: Construct straight line element X1 to F 1 (x,y) is opened and we get Among them, · represents the opening operation; S2023: To fill the holes in the target area, use the straight line element X2 to F 2 (x,y) is closed, and we get in, Represents the closing operation.

5. The method for detecting a frequency hopping carrier signal based on time-frequency analysis according to claim 3, characterized in that: The specific operation of step S3 includes the following steps: S301: Use the Canny operator to perform edge detection on the remaining image; S302: extracting features of the remaining image after edge detection using Hough transform, including width, height, length and angle of the image; Among them, the width and height of each signal segment are obtained through the [m,n]=size(D(t,f)) function, and then the length of each signal segment is calculated by Len=abs(x2-x1), and the angle Ang = abs((y2-y1) / (x2-x1)).

6. The method for detecting a frequency hopping carrier signal based on time-frequency analysis according to claim 5, characterized in that: The specific operation of step S301 includes the following steps: S3011: preprocessing the remaining image after processing in step S2 using Gaussian filtering; S3012: Calculate the components of each pixel in the x direction and the y direction, expressed as g x and g y , Then the pixel gradient size g θ and direction θ are respectively expressed as Where S(x,y) represents the pixel intensity of the image at position (x,y), S(x+1,y) is the intensity of the pixel on the right, and S(x-1,y) is the intensity of the pixel on the left; S(x,y+1) is the intensity of the pixel below, and S(x,y-1) is the intensity of the pixel above; S3013: After gradient calculation, edge detection is performed using the Canny edge detection algorithm.

7. The method for detecting a frequency hopping carrier signal based on time-frequency analysis according to claim 6, characterized in that: The specific operation of step S4 includes the following steps: S401: According to the characteristics of sweep frequency signal transmission, the threshold angle zs_d is set to 5 degrees. When Ang>5, it proves that the sweep frequency signal exists; S402: according to the characteristics of the fixed frequency signal during transmission, the length threshold is set to zs_m, where m is the length of the time-frequency graph. When Len>zs_m=m / 3, it is determined that a fixed frequency signal exists; S403: judging the remaining signal segments, setting the threshold value of the frequency hopping signal segment to zs_n=6, where n represents the number of signal segments of the frequency hopping signal, and when n>zs_n, it is judged that the frequency hopping signal exists.

8. A frequency hopping carrier signal detection system based on time-frequency analysis, characterized in that: The signal detection system includes a signal conversion module, a burst signal filtering module, a signal feature extraction module and a judgment module; The signal transformation module is used to obtain the time-frequency diagram to be tested by using the combined time-frequency analysis transformation STFT&WVD method on the sampled signal; The burst signal filtering module is used to perform binarization processing and adaptive morphological filtering on the time-frequency graph to be inspected to filter out the burst signal; The signal feature extraction module is used to smooth the remaining image after filtering out the burst signal, and use Hough transform to extract the features of the remaining image; The judgment module is used to judge the signal type; The signal conversion module, burst signal filtering module, signal feature extraction module and judgment module are all implemented by the signal detection method described in any one of claims 1-7.

9. An electronic device, characterized in that: The electronic device includes at least one processor; and a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, and the instructions are executed by the processor so that the processor can execute the signal detection method described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to enable the computer to execute the signal detection method described in any one of claims 1-7.

Images