Frequency hopping signal parameter estimation method based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s
Through the combination of maximum entropy binarization and ABF-YOLOv5s model, the problem of insufficient frequency hopping signal period and frequency estimation accuracy under low signal-to-noise ratio is solved, and higher robustness and accuracy are achieved, and errors are reduced.
Patent Information
- Application Number
- CN202411147771.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-21
AI Technical Summary
In a low signal-to-noise ratio environment, it is difficult for the prior art to improve the period estimation and frequency estimation accuracy of the frequency hopping signal at the same time, and it is easily disturbed by noise, resulting in large errors.
Using the maximum entropy binarized time-frequency graph and improved ABF-YOLOv5s model, the frequency hopping signal is extracted through the combination of short-time Fourier transform, maximum entropy binarization, multi-scale morphological filtering and ABF-YOLOv5s model. The position information and time frequency parameters are used to estimate the period and frequency. Through the ASPP module, BOT3 module and BiFPN module, the ABF-YOLOv5s model increases the receptive field and pays attention to the frequency hopping signal characteristics, and improves positioning accuracy and edge detection accuracy.
The error of frequency hopping signal period and frequency estimation is effectively reduced, and the robustness at low signal-to-noise ratio is improved. Compared with traditional methods, the error is reduced by 2 orders of magnitude and the detection accuracy is improved.
Smart Images

Figure CN119093966A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of frequency hopping signal parameter estimation and relates to a frequency hopping signal period estimation and frequency estimation method based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s. Background Art
[0002] Frequency hopping communication is a communication technology that transmits data by quickly switching between different frequencies. It has the characteristics of strong anti-interference, anti-interception and anti-fading. Obtaining the parameters of the frequency hopping signal is the prerequisite for implementing frequency hopping communication interference, but with the increasing complexity of the electromagnetic environment, how to obtain accurate frequency hopping signal parameters has become a problem.
[0003] Since the spectrum of frequency hopping signal changes with time, it is a typical non-stationary signal. Time-frequency analysis can characterize different frequencies and different signal energy intensities under different time conditions. Therefore, time-frequency analysis is the main method for frequency hopping signal parameter estimation. Hui DI et al. published a paper titled "Research on parameter blind estimation algorithm of costas frequency hopping signals" using short-time Fourier transform (STFT) to represent frequency hopping signals in time and frequency. However, it is affected by the window function and has time-frequency uncertainty. It cannot take into account both time resolution and frequency resolution, resulting in the failure to achieve the best accuracy for period estimation and frequency estimation of frequency hopping signals at the same time. Wang Zhongren et al. proposed using Wigner-Ville distribution (WVD) to estimate the parameters of frequency hopping signals. Compared with STFT, which is a single linear transformation, WVD is a bilinear time-frequency transformation method, which solves the problem of STFT window function and improves the accuracy of frequency hopping signal parameter estimation. However, for frequency hopping signals, which contain multiple components with different frequencies in the spectrum, WVD transformation is prone to generate cross terms, which interferes with the frequency estimation of frequency hopping signals. Chen T proposed to use smoothed pseudo-Wigner-Wiley distribution (SPWD) to analyze frequency hopping signals. SPWD suppresses the cross-interference terms in WVD and reduces the frequency estimation error of frequency hopping signals by windowing in both time and frequency domains for smoothing, but also reduces the time-frequency resolution to a certain extent. Chen Chao, Fu Weihong and others combined STFT with WVD or SPWVD to suppress the cross-interference terms while maintaining a high time-frequency resolution, so that the parameter estimation accuracy of frequency hopping signals is further improved under high signal-to-noise ratio conditions. However, at low signal-to-noise ratios, the time-frequency distribution is greatly affected by noise, resulting in low accuracy of parameter estimation of frequency hopping signals under low signal-to-noise ratios. In order to improve the parameter estimation accuracy of frequency hopping signals under low signal-to-noise ratios, Xi Youyou and others combined binarization processing with morphological filtering, and effectively removed noise by binarization processing using the global threshold method. However, the global threshold method only considers maximizing the background and object segmentation when segmenting the image, which is easy to lose the edge information of the frequency hopping signal, while morphological filtering can restore part of the edge information of the frequency hopping signal. With the development of deep learning, applying deep learning to the field of wireless communications has become a research hotspot. Wang Yuyang and others used the SSD model to estimate the parameters of frequency hopping signals. The experimental results show that deep learning can estimate the hopping rate and frequency of frequency hopping signals with high accuracy. However, the SSD model uses a single network on multiple feature maps, and the representation ability of shallow feature maps is not strong, which makes it easy to miss or misdetect frequency hopping signals, resulting in large errors in the estimation of the frequency hopping signal period. Summary of the invention
[0004] In order to improve the accuracy of frequency hopping signal period estimation and frequency estimation under low signal-to-noise ratio, this paper proposes a frequency hopping signal parameter estimation method based on maximum entropy binarization time-frequency graph and ABF-YOLOv5s, which includes the following steps:
[0005] S1, obtaining a frequency hopping signal, and performing short-time Fourier transform on the frequency hopping signal to obtain a time-frequency diagram;
[0006] The time-frequency graph is then subjected to maximum entropy binarization processing; and then processed using multi-scale morphological filtering to fill the holes inside the frequency-hopping signal after the maximum entropy binarization processing, remove the points or lines corresponding to the noise signal, better highlight the contour and edge of the frequency-hopping signal, thereby eliminating the influence of noise and obtaining a clear maximum entropy binarization time-frequency graph;
[0007] S2. Input the maximum entropy binarization time-frequency graph into the ABF-Yolov5s model to obtain the frequency hopping signal location information:
[0008] The ABF-Yolov5s model includes an input part, an ABF-Backbone part, an ABF-Neck part, and a Detection part; the ABF-Yolov5s model is improved on the basis of the Yolov5s model;
[0009] The ABF-Backbone part and the ABF-Neck part correspond to the Backbone part and the Neck part of the Yolov5s model respectively; the Detection part of the ABF-Yolov5s model has the same structure as the Detection part of the Yolov5s model; the input part of the ABF-Yolov5s model has the same structure as the input part of the Yolov5s model;
[0010] The SPP module of the Backbone part of the Yolov5s model is replaced with an ASPP module; the last C3 module of the Backbone part of the Yolov5s model is replaced with a BOT3 module to obtain the ABF-Backbone part; the BOT3 module is set after the ASPP module; the four PANET modules of the Neck part of the Yolov5s model are replaced with BiFPN modules respectively to obtain the ABF-Neck part;
[0011] The Detection part obtains the frequency hopping signal location information through three 1×1 convolutions;
[0012] S3 estimates the period and frequency of the frequency hopping signal according to the comparison relationship between the frequency hopping signal position information and the time-frequency parameters, wherein the time-frequency parameters are extracted from the time-frequency diagram obtained in step S1.
[0013] Preferably, in S1, the maximum entropy binarization processing includes: binarizing the time-frequency graph using maximum entropy threshold segmentation; the maximum entropy threshold segmentation refers to finding the optimal threshold to maximize the sum of entropies of the background and the object.
[0014] Preferably, in S1, the mathematical model of the frequency hopping signal is:
[0015]
[0016] Among them, T represents the frequency hopping period, rect T represents a rectangular window with a width of T, τ represents the jump time, and f k represents the hopping frequency, M represents the total number of hopping frequencies within the observation time, j is an imaginary unit, and v(t) represents Gaussian white noise;
[0017] The method of performing short-time Fourier transform on the frequency hopping signal to obtain a time-frequency diagram comprises the following steps:
[0018] The STFT spectrum is obtained according to the following formula:
[0019]
[0020] Wherein, h(n) represents the window function, k=1,2,3,…,K represents the discrete frequency point, n=1,2,3,…,N represents the discrete time point, and s represents the frequency hopping signal;
[0021] The time-frequency diagram is obtained from the STFT spectrum diagram according to the following formula:
[0022] SPCE s (k,n)=|STFT s (k,n)| 2
[0023] Preferably, in S1, the calculation of the maximum entropy threshold comprises the following steps:
[0024] The entropy in a digital image is defined as:
[0025]
[0026] Where I represents an image with N pixels; u represents the average grayscale value of pixels below the threshold t, corresponding to the grayscale value of the background area; v represents the average grayscale value of pixels above the threshold t, corresponding to the grayscale value of the object area; p represents probability; g = I(u, v); log b(x) represents the logarithm with base b. Assuming the threshold is t, A is the grayscale distribution of {0, 1, 2, ..., t}, and B is the grayscale distribution of {t+1, t+2, ..., L-1}, then the entropies of A and B are H(A) and H(B), respectively, and are calculated as follows;
[0027]
[0028]
[0029] Among them, p n and 1-p n denote the cumulative probabilities of background and object pixels segmented by t threshold, respectively;
[0030] Define the objective function is the sum of H(A) and H(B):
[0031]
[0032] Then the maximum entropy threshold is calculated as:
[0033]
[0034] Preferably, in S2, the ASPP module is used to increase the receptive field while keeping the resolution size of the maximum entropy binarization time-frequency graph unchanged, and fully extract the global features and local features of the maximum entropy binarization time-frequency graph; the global features include the shape contour features of the maximum entropy binarization time-frequency graph; the local features include the corner points and edge features of the maximum entropy binarization time-frequency graph.
[0035] Preferably, in S2, the BOT3 module is used to focus the ABF-Yolov5s model on the area where the frequency hopping signal is located in the maximum entropy binary time-frequency graph through a multi-head self-attention mechanism. Paying attention to the information of different positions of the frequency hopping signal in the maximum entropy binary time-frequency graph helps the model focus on the area where the frequency hopping signal is located to improve the positioning accuracy of the frequency hopping signal.
[0036] Preferably, in S2, the BiFPN module introduces a bidirectional connection to fuse the multi-scale shallow graphic features output by the c3 module in the ABF-Backbone and the deep semantic features output by the BOT3 module for distinguishing the frequency hopping signal from the background to obtain a feature map, thereby improving the edge detection accuracy of the frequency hopping signal while improving the positioning accuracy of the frequency hopping signal, and reducing the errors in the period estimation and frequency estimation of the frequency hopping signal.
[0037] Preferably, the S3 specifically includes the following steps:
[0038] The frequency hopping signal position information is specifically the normalized coordinates (x1, y1) (x2, y2) of the frequency hopping signal in the maximum entropy binary time-frequency diagram;
[0039] Wherein, x1 is the starting position of the frequency hopping signal, and x2 is the ending position of the frequency hopping signal; y1 is the upper boundary of the frequency of the frequency hopping signal, and y2 is the lower boundary of the frequency of the frequency hopping signal; the period estimation T and frequency estimation value F of the frequency hopping signal are calculated as follows:
[0040] T=(x3-x1)×t_size
[0041] F=(y1+y2) / 2×f_size
[0042] Among them, x3 refers to the starting position of the next hop signal, t_size refers to the maximum value of the time axis in the time-frequency graph, and f_size refers to the maximum value of the frequency in the time-frequency graph;
[0043] Then, the tail mean of the results calculated in each time-frequency diagram is calculated, and finally, the mean of the time-frequency diagram results under the same signal-to-noise ratio is calculated again to obtain the period and frequency of the frequency-hopping signal estimated by the ABF-Yolov5s network.
[0044] Preferably, the value of t_size is 0.02s, and the value of f_size is 2.5×10 5 Hz.
[0045] Preferably, the value of b is 2.
[0046] The beneficial effects of the present invention are as follows:
[0047] 1. This paper proposes the ABF-Yolov5s model, which uses deep learning to solve the problem of large errors in period estimation and frequency estimation of frequency hopping signals in low signal-to-noise ratio environments.
[0048] 2. The maximum entropy threshold segmentation method is used to binarize the time-frequency graph and combine it with morphological filtering to eliminate the problem of reduced resolution of the time-frequency graph caused by noise.
[0049] 3. The ASPP module is introduced in the ABF-Backbone part to increase the receptive field while keeping the resolution of the maximum entropy binary time-frequency graph unchanged, and fully extract the global features such as the shape contour and local features such as corners and edges of the frequency hopping signal; the BOT3 module is introduced to focus on the information of the frequency hopping signal at different positions in the maximum entropy binary time-frequency graph through the multi-head self-attention mechanism, helping the model focus on the area where the frequency hopping signal is located, so as to improve the positioning accuracy of the frequency hopping signal.
[0050] 4. The BiFPN module is introduced in the ABF-Neck part. Through bidirectional connection, the multi-scale shallow graphic features and the deep semantic features used to distinguish the frequency hopping signal and the background are better integrated. The learnable weights are introduced to determine the importance of different input features, thereby enhancing the effect of the frequency hopping signal feature fusion, improving the edge detection accuracy of the frequency hopping signal while improving the positioning accuracy of the frequency hopping signal, and reducing the errors of the frequency hopping signal period estimation and frequency estimation.
[0051] In summary, the present invention can effectively reduce the errors of frequency hopping signal period estimation and frequency estimation under low signal-to-noise ratio and has good robustness. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 Time-frequency image processing (SNR = -1dB) image, in which (a) is the original time-frequency image, (b) is the grayscale image, (c) is the maximum entropy binarization image, and (d) is the image after morphological filtering;
[0053] Figure 2 The network structure of ABF-Yolov5s;
[0054] Figure 3 ASPP structure;
[0055] Figure 4 BOT3 structure and Bottleneck Transformer structure;
[0056] Figure 5 Comparison chart of BiFPN and PANet structures;
[0057] Figure 6 This is the frequency hopping signal detection result based on Yolov5s;
[0058] Figure 7 This is the frequency hopping signal detection result based on ABF-Yolov5s;
[0059] Figure 8The experimental results of frequency hopping signal period estimation and frequency estimation are shown in the figure. (a) is the error of frequency hopping signal period estimation, and (b) is the error of frequency hopping signal frequency estimation. It can be seen from the experimental results that under low signal-to-noise ratio conditions, the frequency hopping signal period estimation and frequency estimation method based on maximum entropy binarization time-frequency diagram and ABF-Yolov5s proposed in this paper can effectively reduce the error. At -5dB, the error of the method proposed in this paper is reduced by about 2 orders of magnitude compared with the frequency hopping signal period estimation and frequency estimation method based on STFT; at -1dB, the error is also greatly reduced compared with the frequency hopping signal period estimation and frequency estimation method based on SPWVD; compared with the frequency hopping signal period estimation and frequency estimation method based on SSD, the method proposed in this paper also has a significant reduction in the error of frequency hopping signal period estimation; compared with the original Yolov5s model, the ABF-Yolov5s proposed in this paper has higher detection accuracy and smaller errors in frequency hopping signal period estimation and frequency estimation. DETAILED DESCRIPTION
[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0061] The present invention first uses the maximum entropy threshold segmentation method to binarize the time-frequency graph and combines it with morphological filtering, which can better highlight the edge information of the frequency hopping signal while eliminating noise, thereby obtaining a clear maximum entropy binary time-frequency graph. Secondly, this paper proposes the ABF-Yolov5s model. In ABF-Backbone, the ASPP module is used to increase the receptive field while keeping the resolution of the maximum entropy binary time-frequency graph unchanged, and fully extract the global features such as the shape contour and local features such as corners and edges of the frequency hopping signal; the multi-head self-attention mechanism is introduced through the BOT3 module to focus on the information of different positions of the frequency hopping signal in the maximum entropy binary time-frequency graph, helping the model to focus on the area where the frequency hopping signal is located, so as to improve the positioning accuracy of the frequency hopping signal. In ABF-Neck, the BiFPN module is used to introduce a bidirectional connection, and the multi-scale shallow graphic features and the deep semantic features used to distinguish the frequency hopping signal and the background are better integrated, so as to improve the edge detection accuracy of the frequency hopping signal while improving the positioning accuracy of the frequency hopping signal, and reduce the errors of the frequency hopping signal period estimation and frequency estimation. Finally, the period estimation and frequency estimation of the frequency hopping signal are completed by comparing the signal coordinate position obtained by ABF-Yolov5s with the coordinates in the maximum entropy binarization time-frequency diagram.
[0062] The technical solution adopted by the present invention to solve the technical problem comprises the following steps:
[0063] Step 1: Perform short-time Fourier transform on the frequency-hopping signal to obtain a time-frequency diagram, then perform maximum entropy binarization on the time-frequency diagram, and use multi-scale morphological filtering to fill the holes inside the binarized frequency-hopping signal, remove points or lines isolated from the useful signal, better highlight the contour and edge of the frequency-hopping signal, thereby eliminating the influence of noise and obtaining a clear maximum entropy binarization time-frequency diagram.
[0064] Step 2: Input the clear maximum entropy binarized time-frequency graph into the ABF-Yolov5s model to estimate the frequency hopping signal parameters. The ASPP module is used in the ABF-Backbone part to increase the receptive field while keeping the resolution of the maximum entropy binarized time-frequency graph unchanged, and fully extract the global features such as the shape and contour of the frequency hopping signal and the local features such as corners and edges; the BOT3 module is used at the end of the ABF-Backbone to focus on the information of the frequency hopping signal at different positions in the maximum entropy binarized time-frequency graph through the multi-head self-attention mechanism, helping the model focus on the area where the frequency hopping signal is located, so as to improve the positioning accuracy of the frequency hopping signal.
[0065] Step 3: Use the BiFPN module in the ABF-Neck part and introduce bidirectional connections to better integrate the multi-scale shallow graphic features and the deep semantic features used to distinguish the frequency hopping signal and the background. This improves the edge detection accuracy of the frequency hopping signal while improving the positioning accuracy of the frequency hopping signal, and reduces the errors of the frequency hopping signal period estimation and frequency estimation.
[0066] Step 4: Finally, the frequency hopping signal position information is obtained at the Head end through three 1×1 convolutions. According to the comparison relationship between the signal coordinate position and the time-frequency parameters, the period estimation and frequency estimation of the frequency hopping signal can be obtained.
[0067] Step 1 The frequency hopping signal processing process is as follows:
[0068] The frequency hopping signal is a non-stationary signal whose frequency keeps changing under the control of a pseudo-random sequence. Its mathematical model is:
[0069]
[0070] Among them, T represents the frequency hopping period, rect T represents a rectangular window with a width of T, τ represents the jump time, and f k represents the hopping frequency, M represents the total number of hopping frequencies within the observation time, j is an imaginary unit, and v(t) represents Gaussian white noise.
[0071] By performing short-time Fourier transform on the signal in different time windows, we can get the change of the signal in time and frequency. To facilitate computer calculation, we can perform discretization to get the STFT spectrum:
[0072]
[0073] Wherein, h(n) represents a window function, k=1,2,3,…,K represents a discrete frequency point, n=1,2,3,…,N represents a discrete time point, and s represents a frequency hopping signal.
[0074] The time-frequency diagram is defined as the square of the STFT modulus value, which has good time-frequency aggregation and its expression is:
[0075] SPCE s (k,n)=|STFT s (k,n)| 2 (3)
[0076] The maximum entropy threshold segmentation is used to binarize the time-frequency image. The principle of maximum entropy thresholding is similar to Otsu's method. Otsu's method is to find the optimal threshold to divide the image into background and object to obtain the maximum variance, while the maximum entropy is to find the optimal threshold to maximize the sum of the entropy of the background and object. The definition of entropy in digital images is:
[0077]
[0078] Where I represents an image with N pixels; u represents the average grayscale value of pixels below the threshold t, usually corresponding to the grayscale value of the background area; v represents the average grayscale value of pixels above the threshold t, usually corresponding to the grayscale value of the object area; p represents probability; g = I(u,v); log b (x) represents the logarithm with base b, where b is usually 2. Assuming the threshold is t, A is the grayscale distribution of {0, 1, 2, ..., t}, and B is the grayscale distribution of {t+1, t+2, ..., L-1}, the entropies associated with the two probability densities are H(A) and H(B), respectively, as shown in Equation 6 and Equation 5.
[0079]
[0080]
[0081] Among them, p n and 1-p n denote the cumulative probabilities of background and object pixels segmented by t threshold, respectively.
[0082] Define the function as the sum of H(A) and H(B):
[0083]
[0084] Then the maximum entropy threshold is:
[0085]
[0086] By binarizing the time-frequency graph using this threshold, the edge information of the frequency-hopping signal can be retained as much as possible while removing the noise. However, due to the influence of noise, the binarized time-frequency graph still has some problems, which can be further removed by multi-scale morphological filtering.
[0087] Due to the influence of noise, after the time-frequency graph is converted into a binary image, there may be points or lines isolated from the useful signal image; or there may be burrs around the useful signal image; or there may be holes inside the useful signal image. Therefore, morphological filtering is needed to repair the useful signal image. Erosion and dilation operations are the most basic operations in morphology. The erosion operation is mainly used to remove small isolated points and broken lines in the image; the dilation operation is mainly used to fill the holes in the image and connect the broken lines. The operation of first erosion and then dilation is called a closing operation, which is defined as follows, where X represents the image to be processed, B represents the structural element of the image, is the expansion operation, Θ is the erosion operation:
[0088]
[0089] Morphological filtering is to obtain local features of an image by operating the image with a set structural element. Therefore, the most important thing in morphological filtering is the selection of structural elements, including the setting of the shape and size of the structural elements. The shape of the element will affect the recognition and segmentation of targets of different shapes in the image. Selecting a suitable element shape can better highlight the contour or edge of the target. The size of the element determines the granularity of the filter. Smaller elements are usually used to refine the image and can better retain the detail information in the image. Larger elements are usually used to smooth the image, emphasize the target contour or fill the hole. Because the frequency hopping signal is generally rectangular, this paper first selects 10×10 rectangular elements for closing operations to make up for the holes left by the noise when the frequency hopping signal is binarized and better highlight the edge information of the frequency hopping signal, so as to improve the accuracy of locating the frequency hopping signal using ABF-Yolov5s; then create a horizontal straight line element with a length of 5 pixels to remove the points or lines isolated from the useful signal image after the time-frequency graph is converted into a binary image, and retain the detail information in the image. After binarization and morphological filtering, the influence of noise can be eliminated, a clear time-frequency image can be obtained, and the characteristics of the frequency-hopping signal can be retained, which is conducive to the subsequent use of ABF-Yolov5s to estimate the period and frequency of the frequency-hopping signal.
[0090] Step 2: The process of extracting the frequency hopping signal features in the ABF-Backbone part is as follows:
[0091] In the ABF-Backbone network structure, a 320×320×32 frequency hopping signal feature map is generated through the Focus module; then, through the deep convolutional network constructed by the C3 module and the Conv module, after each combination of Conv and C3 modules, the feature map will become 1 / 2 of the original size, thereby obtaining frequency hopping signal graphic features of different scales; then, the ASPP module is used to perform parallel sampling with dilated convolutions at different sampling rates to fully extract the global and local features of the frequency hopping signal and perform preliminary interaction; finally, the BOT3 module is used to extract the semantic features of the frequency hopping signal in the deep network.
[0092] ASPP Module:
[0093] ASPP uses dilated convolutions with different expansion rates for parallel sampling, including two 1×1 convolution layers, multiple 3×3 convolution layers with different expansion rates, and a pooling layer. The feature information after pooling is restored to the expected value through 1×1 convolution and upsampling to restore the maximum entropy binary time-frequency image size to the expected value. Finally, all the results are spliced to expand the number of channels, and then the number of channels is reduced to the expected value for output through 1×1 convolution, thereby increasing the receptive field while keeping the resolution of the feature map unchanged. Multiple parallel convolution layers are used to capture the characteristics of frequency hopping signals of different scales, fully extract the global and local features of the frequency hopping signal, and perform preliminary interaction. Since the global features can reflect the overall attributes of the frequency hopping signal such as the shape and contour, and the local features pay more attention to specific key points such as corners or edges, the interaction between them can complement each other's advantages, improve the detection accuracy of the frequency hopping signal, and thus reduce the errors of the frequency hopping signal period estimation and frequency estimation.
[0094] BOT3 module:
[0095] This paper applies BOT3 at the end of Backbone because the feature map of the frequency hopping signal here is of lower resolution and contains more high-level semantic information. The stronger the semantic information, the stronger the model's ability to distinguish the frequency hopping signal from the background. However, the deeper the network depth, the more vague the location information of the frequency hopping signal. The most important structure in the BOT3 module is the multi-head self-attention mechanism (MHSA), which can use multiple different attention heads to perform parallel calculations on the input image to obtain spatially sensitive similar features, making it easier for the model to focus on the area where the frequency hopping signal is located, avoiding the loss of the location information of the frequency hopping signal. At the same time, through the semantic features in the deep network, the frequency hopping signal and the background are distinguished to the greatest extent, and can be integrated with the graphic features output by the shallow network to improve the model's detection and positioning accuracy of the frequency hopping signal and reduce the errors in the period estimation and frequency estimation of the frequency hopping signal.
[0096] Step 3: The process of partially fusing the frequency hopping signal features of ABF-Neck is as follows:
[0097] The ABF-Neck network is mainly responsible for fusing the frequency hopping signal feature maps of different scales, and is composed of a C3 module, an upsampling module, a BiFPN module, and a Conv module. In the left path, the fusion of different levels of features is achieved by upsampling and fusing with a coarser-grained frequency hopping signal feature map. In the right path, the feature fusion of different levels is achieved by using a convolutional layer, and finally the frequency hopping signal feature map is obtained, which is used to detect and locate the position of the frequency hopping signal and realize the period estimation and frequency estimation of the frequency hopping signal. In the ABF-Neck network, BiFPN is used to realize bidirectional feature propagation and multi-resolution feature processing, which promotes more thorough and detailed cross-level information transmission and fusion, that is, the equivariant features (graphic features) used in the shallow network to identify the position and scale changes of the frequency hopping signal and the invariant features (semantic features) used in the deep network to separate the frequency hopping signal from the background are better integrated, and learnable weights are introduced to determine the importance of different input features, so as to enhance the effect of frequency hopping signal feature fusion, improve the accuracy of frequency hopping signal detection and positioning, and reduce the errors of frequency hopping signal period estimation and frequency estimation.
[0098] BiFPN:
[0099] In the neural network model, the semantic information of the low-level frequency hopping signal features is often less but the signal position is accurate, while the semantic information of the high-level frequency hopping signal features is rich but the signal position is rough. In order to improve the accuracy of frequency hopping signal target detection and recognition, the features of the two need to be fused. BiFPN first deletes some input edge nodes that contribute less to the performance of fusion of different frequency hopping signal features; secondly, a jump connection is added between the input node and the output node of the same scale, which can fuse more frequency hopping signal image features; and unlike PANet, which has only one top-down and one bottom-up path, each bidirectional (top-down and bottom-up) path is regarded as a feature network layer, and the same layer is repeated many times to achieve higher-level frequency hopping signal feature fusion; finally, learnable weights are introduced to determine the importance of different input features, strengthen the effect of frequency hopping signal feature fusion, improve the accuracy of frequency hopping signal positioning, and reduce the errors of frequency hopping signal period estimation and frequency estimation.
[0100] Step 4: The frequency hopping signal period estimation and frequency estimation method are as follows:
[0101] After the above ABF-Yolov5s network, the model confidence and the normalized coordinates of the frequency hopping signal (x1, y1)(x2, y2) can be obtained. Among them, x1 is the starting position of the frequency hopping signal, x2 is the ending position of the frequency hopping signal; y1 is the upper boundary of the frequency of the frequency hopping signal, and y2 is the lower boundary of the frequency of the frequency hopping signal. The period estimation and frequency estimation values of the frequency hopping signal can be obtained by calculating the comparison relationship between the signal coordinates and the time-frequency parameters.
[0102] T = (x3 - x1) × t_size (9)
[0103] F=(y1+y2) / 2×f_size (10)
[0104] In formula 9, x3 refers to the starting position of the next hop signal, t_size refers to the maximum value of the time axis in the time-frequency diagram, which is 0.02s in this paper, and f_size refers to the maximum value of the frequency in the time-frequency diagram, which is 2.5×10 5 Hz. Then, the tail mean of the results calculated in each image is calculated, and finally the mean of the results of the images under the same signal-to-noise ratio is calculated again to obtain the period and frequency of the frequency hopping signal estimated by the ABF-Yolov5s network.
[0105] Example:
[0106] 1. Dataset
[0107] Use MATLAB to generate frequency hopping signals, use spread spectrum code sequences for frequency shift keying modulation, and add Gaussian noise. The signal-to-noise ratios are -5db, -3db, -1db, 1db, 3db, and 5db, respectively. The signal pseudo code length is 31, the code rate is 1k, the sampling rate is 248kHz, the frequency hopping signal period is 1ms, and the frequency set is [160, 200, 148, 208, 172, 116, 100, 180, 204, 192, 124, 140, 104, 136, 188, 164, 156, 184, 112, 176, 152, 108, 220, 144, 132, 128, 212, 120, 216, 196, 168]kHz.
[0108] First, the time-frequency diagram of the frequency-hopping signal is obtained by short-time Fourier transform. 600 images are generated for each signal-to-noise ratio, and 3,600 images are generated for the six signal-to-noise ratios. Then, the images are binarized and morphologically filtered in batches. The images are divided into training set, test set, and validation set in a ratio of 4:1:1, that is, the training set for each signal-to-noise ratio is 400 images, the test set is 100 images, and the validation set is 100 images. Then, Labelme is used to make labels, and the .png format images are converted to .jason format. Finally, the code is used to convert them to .text format.
[0109] 2. Experimental Environment
[0110] The GPU of the experimental server is NVIDIATITAN RTX 3090, the deep learning framework used is Pytorch, the model is implemented in Python programming language, and programming is done using Sublime Text3. Since model training using CPU is relatively slow, we use GPU to train the network framework. The GPU uses NVIDIATITAN RTX 3090, the model training rounds are 100, and the loss function is the cross entropy loss function.
[0111] 3. Performance simulation
[0112] Figure 6 and Figure 7 The detection results of the original Yolov5s and ABF-Yolov5s for frequency-hopping signals under the same signal-to-noise ratio are shown. Compared with the original Yolov5s model, whose average confidence is 0.66, the average confidence of ABF-Yolov5s is 0.92, which greatly improves the recognition and detection capabilities of frequency-hopping signals, thereby effectively reducing the period estimation and frequency estimation errors of frequency-hopping signals.
[0113] This paper compares the frequency hopping signal period estimation and frequency estimation based on ABF-Yolov5s with the frequency hopping signal period estimation and frequency estimation based on STFT, the frequency hopping signal period estimation and frequency estimation based on SPWVD, the frequency hopping signal period estimation and frequency estimation based on Yolov5s and the frequency hopping signal period estimation and frequency estimation based on SSD. The experimental results are as follows: Figure 8 shown.
[0114] Finally, it should be noted that the purpose of publishing the embodiments is to help further understand the present invention, but those skilled in the art can understand that various substitutions and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the present invention should not be limited to the contents disclosed in the embodiments, and the scope of protection claimed by the present invention shall be subject to the scope defined in the claims.
Claims
1. A frequency hopping signal parameter estimation method based on maximum entropy binarization time-frequency graph and ABF-YOLOv5s, characterized in that: The following steps are involved: S1, obtaining a frequency hopping signal, and performing short-time Fourier transform on the frequency hopping signal to obtain a time-frequency diagram; Then the time-frequency graph is subjected to maximum entropy binarization processing; and then multi-scale morphological filtering is used for processing to fill the holes inside the frequency-hopping signal after the maximum entropy binarization processing, remove the points or lines corresponding to the noise signal, and obtain a clear maximum entropy binarization time-frequency graph; S2. Input the maximum entropy binarization time-frequency graph into the ABF-Yolov5s model to obtain the frequency hopping signal location information: The ABF-Yolov5s model includes an input part, an ABF-Backbone part, an ABF-Neck part, and a Detection part; The ABF-Backbone part and the ABF-Neck part correspond to the Backbone part and the Neck part of the Yolov5s model respectively; the Detection part of the ABF-Yolov5s model has the same structure as the Detection part of the Yolov5s model; the input part of the ABF-Yolov5s model has the same structure as the input part of the Yolov5s model; The SPP module of the Backbone part of the Yolov5s model is replaced with an ASPP module; the last C3 module of the Backbone part of the Yolov5s model is replaced with a BOT3 module to obtain the ABF-Backbone part; the BOT3 module is set after the ASPP module; the four PANET modules of the Neck part of the Yolov5s model are replaced with BiFPN modules respectively to obtain the ABF-Neck part; The Detection part obtains the frequency hopping signal location information through three 1×1 convolutions; S3 estimates the period and frequency of the frequency hopping signal according to the comparison relationship between the frequency hopping signal position information and the time-frequency parameters, wherein the time-frequency parameters are extracted from the time-frequency diagram obtained in step S1.
2. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 1, characterized in that: In S1, the maximum entropy binarization process includes: performing binarization process on the time-frequency graph by using maximum entropy threshold segmentation; the maximum entropy threshold segmentation refers to obtaining the optimal threshold value to maximize the sum of entropies of the background and the object.
3. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 2, characterized in that: In S1, the mathematical model of the frequency hopping signal is: Among them, T represents the frequency hopping period, rect T represents a rectangular window with a width of T, τ represents the jump time, and f k represents the hopping frequency, M represents the total number of hopping frequencies within the observation time, j is an imaginary unit, and v(t) represents Gaussian white noise; The method of performing short-time Fourier transform on the frequency hopping signal to obtain a time-frequency diagram comprises the following steps: The STFT spectrum is obtained according to the following formula: Wherein, h(n) represents the window function, k=1,2,3,…,K represents the discrete frequency point, n=1,2,3,…,N represents the discrete time point, and s represents the frequency hopping signal; The time-frequency diagram is obtained from the STFT spectrum diagram according to the following formula: SPCEs(k,n)=|STFTs(k,n)|2.
4. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 3, characterized in that: In S1, the calculation of the maximum entropy threshold comprises the following steps: The entropy in a digital image is defined as: Where I represents an image with N pixels; u represents the average grayscale value of pixels below the threshold t, corresponding to the grayscale value of the background area; v represents the average grayscale value of pixels above the threshold t, corresponding to the grayscale value of the object area; p represents probability; g = I(u, v); log b (x) represents the logarithm with base b. Assuming the threshold is t, A is the grayscale distribution of {0, 1, 2, ..., t}, and B is the grayscale distribution of {t+1, t+2, ..., L-1}, then the entropies of A and B are H(A) and H(B), respectively, and are calculated as follows; Among them, p n and 1-p n denote the cumulative probabilities of background and object pixels segmented by t threshold, respectively; Define the objective function is the sum of H(A) and H(B): Then the maximum entropy threshold is calculated as:
5. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 1, characterized in that: In S2, the ASPP module is used to extract the global features and local features of the maximum entropy binarized time-frequency graph; the global features include the shape contour features of the maximum entropy binarized time-frequency graph; the local features include the corner points and edge features of the maximum entropy binarized time-frequency graph.
6. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 1, characterized in that: In S2, the BOT3 module is used to focus the ABF-Yolov5s model on the area where the frequency hopping signal is located in the maximum entropy binarized time-frequency diagram through a multi-head self-attention mechanism.
7. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 1, characterized in that: In S2, the BiFPN module is used to fuse the multi-scale shallow graphic features output by the c3 module in the ABF-Backbone and the deep semantic features output by the BOT3 module for distinguishing the frequency hopping signal and the background to obtain a feature map.
8. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 1, characterized in that: The S3 specifically includes the following steps: The frequency hopping signal position information is specifically the normalized coordinates (x1, y1) (x2, y2) of the frequency hopping signal in the maximum entropy binary time-frequency diagram; Wherein, x1 is the starting position of the frequency hopping signal, and x2 is the ending position of the frequency hopping signal; y1 is the upper boundary of the frequency of the frequency hopping signal, and y2 is the lower boundary of the frequency of the frequency hopping signal; the period estimation T and frequency estimation value F of the frequency hopping signal are calculated as follows: T=(x3-x1)×t_size F=(y1+y2) / 2×f_size Among them, x3 refers to the starting position of the next hop signal, t_size refers to the maximum value of the time axis in the time-frequency graph, and f_size refers to the maximum value of the frequency in the time-frequency graph; Then, the tail mean of the results calculated in each time-frequency diagram is calculated, and finally, the mean of the time-frequency diagram results under the same signal-to-noise ratio is calculated again to obtain the period and frequency of the frequency-hopping signal estimated by the ABF-Yolov5s network.
9. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s as claimed in claim 8, characterized in that: The value of t_size is 0.02s, and the value of f_size is 2.5×10 5 Hz.
10. The method for estimating frequency hopping signal parameters based on maximum entropy binarization time-frequency diagram and ABF-YOLOv5s according to claim 4, characterized in that: The value of b is 2.
Citation Information
Patent Citations
Frequency hopping signal feature extraction and parameter estimation method
CN112929053A
Construction worker dangerous area intrusion detection method based on improved Yolov5 model
CN114997279A
Quick identification method for frequency hopping signals under multiple signals based on YOLOv5
CN115630275A
Frequency hopping signal parameter estimation method based on time-frequency transformation and waveform shaping
CN116722892A
Image defogging method and system based on improved dark channel prior
CN116797484A
