A method and system for joint sorting and frequency set estimation of asynchronous multi-hop frequency signals

Abstract

The application provides a joint sorting and frequency set estimation method and system of asynchronous multi-frequency hopping signals, and relates to the technical field of radio signal analysis and processing. The method comprises the following steps: preprocessing the collected asynchronous multi-frequency hopping signals to obtain a time-frequency matrix; performing ridge line slicing on the time-frequency matrix to obtain a ridge line slicing line segment set; constructing a frequency hopping signal ridge line graph according to the ridge line slicing line segment set; and performing ridge line path searching on the frequency hopping signal ridge line graph based on the equal-length change constraint to obtain a frequency hopping signal joint sorting result and a frequency set parameter, thereby solving the joint sorting and frequency set estimation problem of asynchronous multi-frequency hopping signals under complex conditions.

Description

Technical Field

[0001] This application relates to the field of radio signal analysis and processing technology, and in particular to a method and system for joint sorting and frequency set estimation of asynchronous multi-frequency hopping signals. Background Technology

[0002] With the rapid development of wireless communication technology, electronic countermeasures technology, and dynamic spectrum resource management, frequency-agile communication, due to its strong anti-interference capability and high confidentiality, has been widely used in private network communication and signal transmission in complex electromagnetic environments. Among them, asynchronous frequency-hopping communication, as a typical frequency-agile method, has advantages such as no need to maintain synchronization in hopping time and high degree of freedom in network deployment, and has become an important technical form in current complex networking. Under complex conditions such as unknown number of signals, consistent length of multiple frequency-hopping signals, and possible time-frequency overlap, how to quickly and reliably sort multiple frequency-hopping signals and accurately estimate the frequency hopping set of each signal is the core challenge in asynchronous multi-frequency-hopping signal analysis.

[0003] Currently, the analysis of multi-frequency hopping signals typically employs methods such as multi-channel spatial filtering, single-channel blind source separation, or signal length-based clustering to first separate and sort the signals, and then uses time-frequency energy detection or frequency change detection to estimate the frequency set. However, most existing methods first perform multi-signal separation and sorting, and then estimate the frequency set for each separated single-frequency hopping signal, generally assuming that the number of frequency-hopping signals is known. Meanwhile, the multi-signal separation process usually requires multi-channel reception, or, under single-channel conditions, assumes different signal lengths, incoherence between signals, or the use of different modulation schemes. This limits the practical application of existing multi-signal separation methods, making it impossible to achieve asynchronous multi-frequency hopping signal sorting and frequency set estimation under complex conditions such as unknown number of frequency-hopping signals, multiple frequency-hopping signals of the same length, and potential time-frequency overlap. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and system for joint sorting and frequency set estimation of asynchronous multi-frequency hopping signals to address the above problems, thereby solving the problem that joint sorting and frequency set estimation of asynchronous multi-frequency hopping signals cannot be achieved under complex conditions.

[0005] To achieve the above-mentioned technical effects, the technical solution of the present invention is as follows:

[0006] A joint sorting and frequency set estimation method for asynchronous multi-frequency hopping signals includes:

[0007] S1. Preprocess the acquired asynchronous multi-frequency hopping signal to obtain the time-frequency matrix;

[0008] S2. Perform ridge slicing on the time-frequency matrix to obtain a set of ridge slice segments;

[0009] S3. Construct a frequency-hopping signal ridge map based on the set of ridge slices;

[0010] S4. Perform a ridge path search based on equal-length jump constraints on the ridge map of the frequency hopping signal to obtain the joint sorting results and frequency set parameters of the frequency hopping signal.

[0011] Preferably, the preprocessing of the acquired asynchronous multi-frequency hopping signal to obtain a time-frequency matrix includes:

[0012] S11. Decompose the asynchronous multi-frequency hopping signal into IQ signals;

[0013] S12. Perform time-frequency transformation on the IQ signal to obtain an initial time-frequency matrix;

[0014] S13. Binarize the initial time-frequency matrix to obtain the final time-frequency matrix.

[0015] Preferably, the IQ signal is subjected to time-frequency transformation to obtain an initial time-frequency matrix. as follows:

[0016]

[0017] in, The IQ signal, , The length of the IQ signal, , The number of frequency points, For Fourier synchronous compression transform operators;

[0018] For the initial time-frequency matrix Binarization is performed to obtain the final time-frequency matrix. as follows:

[0019]

[0020] in, for Maximum and minimum value normalized matrix The threshold is used for binarization. Preferably, the step of ridge slicing the time-frequency matrix to obtain a set of ridge slice segments includes:

[0021] S21. Perform morphological processing on the time-frequency matrix to obtain a morphologically enhanced time-frequency map matrix;

[0022] S22. Perform edge detection on the morphologically enhanced time-frequency map matrix to obtain an edge pixel map;

[0023] S23. Fit the edge pixel map to obtain a set of ridge slice line segments.

[0024] Preferably, the step of performing morphological processing on the time-frequency matrix to obtain a morphologically enhanced time-frequency map matrix includes:

[0025] S211. Perform dilatation and erosion processing on the time-frequency matrix to obtain an edge-enhanced time-frequency matrix. as follows:

[0026]

[0027] Where ⊙ represents expansion operation, ⊕ represents corrosion operation, and the calculation kernel used for expansion corrosion treatment... use The identity matrix, The time-frequency matrix;

[0028] S212. The edge-enhanced two-dimensional amplitude time-frequency matrix is... Median filtering is performed to obtain the morphologically enhanced time-frequency map matrix. as follows:

[0029]

[0030] in, To perform the operation of finding the median of a set of numbers, .

[0031] Preferably, the edge detection is performed on the morphologically enhanced time-frequency map matrix to obtain an edge pixel map. as follows:

[0032]

[0033] in, This is the Canny edge detection function. and These are the low and high thresholds for Canny edge detection, respectively.

[0034] Preferably, fitting the edge pixel map to obtain the ridge slice line segment set includes:

[0035] S231. Map the edge pixels Perform shape contour fitting to obtain a contour set. as follows:

[0036]

[0037] in, This is the contour detection fitting function. To retrieve the outer contour, To preserve all outline pixels;

[0038] S232. Set the contours Each group of contours is fitted into ridge line slices, the position information of the ridge line slices is calculated, and the ridge line slices with the calculated position information are used as a set to form the ridge line slice set.

[0039] Preferably, the frequency hopping signal ridge map includes a set of nodes. Sum of edges ,in This represents the i-th node. Represents a node and nodes By performing a ridge path search based on equal-length jump constraints on the ridge map of the frequency hopping signal between the edges, the joint sorting results and frequency set parameters of the frequency hopping signal are obtained, including:

[0040] S41. Based on the node set Build a node set Search evaluation function When node When there is no parent node, the search evaluation function The calculation expression is as follows:

[0041]

[0042] in, Represents a node The cost function, Represents a node Heuristic functions;

[0043] When it is a node Has a parent node At that time, the search evaluation function The calculation expression is as follows:

[0044]

[0045] in, Indicates the parent node The cost function;

[0046] S42. Using the aforementioned search evaluation function The search for ridge paths is performed with the minimum target as the objective, until the search evaluation function is reached. The current node corresponding to the minimum The endpoint is reached when there are no more horizontal edges on the right for further progress.

[0047] S43 starts from the endpoint and reverses through its parent node until it reaches the starting point to obtain the frequency hopping path;

[0048] S44. Based on the frequency hopping path, extract the frequency hopping signal sorting results and frequency set parameters.

[0049] Preferably, obtaining the joint sorting result and frequency set parameters of the frequency hopping signal based on the frequency hopping path includes:

[0050] S441. Extract the joint sorting result from the frequency hopping path;

[0051] S442. Multiply the frequency pixel ordinate of the node in the frequency hopping path by the sampling frequency of the asynchronous multi-frequency hopping signal, and then divide by the length of the IQ signal to obtain the frequency set parameters. :

[0052]

[0053] in, K is the number of asynchronous multi-frequency hopping signals. , , The sampling frequency of the IQ signal. The length of the IQ signal.

[0054] This invention also provides a joint sorting and frequency set estimation system for asynchronous multi-frequency hopping signals, comprising:

[0055] The signal processing module is used to preprocess the acquired asynchronous multi-frequency hopping signals to obtain a time-frequency matrix;

[0056] The ridge slicing module is used to slice the time-frequency matrix into ridge segments to obtain a set of ridge slice lines.

[0057] The ridge map construction module is used to construct a frequency hopping signal ridge map based on the set of ridge slices.

[0058] The ridge path search module is used to perform ridge path search on the ridge map of the frequency hopping signal to obtain the joint sorting result and frequency set parameters of the frequency hopping signal.

[0059] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0060] This invention proposes a method and system for joint sorting and frequency set estimation of asynchronous multi-frequency hopping signals. The method involves preprocessing the asynchronous multi-frequency hopping signals to obtain a time-frequency matrix; then, ridge slicing is performed on the time-frequency matrix to obtain a set of ridge slice segments; a ridge map of the frequency hopping signals is constructed based on the set of ridge slice segments; finally, a ridge path search based on equal-length jump constraints is performed on the ridge map of the frequency hopping signals to obtain the joint sorting results and frequency set parameters of the frequency hopping signals. This system does not require the assumption that the number of frequency hopping signals is known, and can achieve effective joint sorting and frequency set estimation of asynchronous multi-frequency hopping signals even in complex situations where the frequency hopping signals have the same length and overlapping time and frequency, improving signal recognition accuracy and processing efficiency, and providing reliable technical support for spectrum analysis in complex electromagnetic environments. Attached Figure Description

[0061] Figure 1 This is a flowchart of the joint sorting and frequency set estimation method for asynchronous multi-frequency hopping signals in an embodiment of the present invention;

[0062] Figure 2 This is a schematic diagram of the process for extracting and estimating the ridge slices of multi-frequency hopping signals in an embodiment of the present invention;

[0063] Figure 3 This is a diagram of frequency-hopping signal ridge slices drawn based on a set of ridge slice segments in an embodiment of the present invention;

[0064] Figure 4 This is a frequency hopping signal ridge map generated from a frequency hopping signal ridge slice map in an embodiment of the present invention;

[0065] Figure 5 This is a visualization of all edges in the frequency hopping signal ridge diagram in an embodiment of the present invention;

[0066] Figure 6 This is the original time-domain signal diagram in an embodiment of the present invention;

[0067] Figure 7 This is a time-frequency diagram of the IQ signal after passing through FSST in an embodiment of the present invention;

[0068] Figure 8 This is the time-frequency diagram after morphological filtering of the time-frequency matrix in this embodiment of the invention;

[0069] Figure 9 This is a ridge slice line segment diagram obtained by fitting and extracting the ridge slice line segment from the contour set in an embodiment of the present invention;

[0070] Figure 10 This is a graph showing the A* path search results based on equal-length jump constraints in an embodiment of the present invention.

[0071] Figure 11This is a block diagram of the joint sorting and frequency set estimation system for asynchronous multi-frequency hopping signals in an embodiment of the present invention. Detailed Implementation

[0072] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent.

[0073] It is understandable to those skilled in the art that some well-known details may be omitted from the accompanying drawings;

[0074] The positional relationships depicted in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent.

[0075] To better illustrate this embodiment, some parts of the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions;

[0076] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.

[0077] Example 1

[0078] like Figure 1 As shown, this embodiment provides a joint sorting and frequency set estimation method for asynchronous multi-frequency hopping signals, including:

[0079] S1. Preprocess the acquired asynchronous multi-frequency hopping signal to obtain the time-frequency matrix;

[0080] In S1, the preprocessing of the acquired asynchronous multi-frequency hopping signal to obtain a time-frequency matrix includes:

[0081] S11. Decompose the asynchronous multi-frequency hopping signal into IQ signals;

[0082] S12. Perform time-frequency transformation on the IQ signal to obtain an initial time-frequency matrix;

[0083] S13. Binarize the initial time-frequency matrix to obtain the final time-frequency matrix.

[0084] Specifically, the initial time-frequency matrix is ​​a two-dimensional amplitude spectrum time-frequency matrix, and the final time-frequency matrix is ​​a binarized two-dimensional amplitude spectrum time-frequency matrix.

[0085] For the IQ signal Perform time-frequency transformation to obtain the two-dimensional amplitude spectrum time-frequency matrix:

[0086]

[0087] in , The length of the IQ signal. , Indicates the number of frequency points. This represents the Fourier Synchro-Squeezed Transform (FSST) operator.

[0088] For the two-dimensional amplitude spectrum time-frequency matrix Binarization preprocessing is performed to obtain a binarized two-dimensional amplitude spectrum time-frequency matrix. :

[0089]

[0090] in, for Maximum and minimum value normalized matrix This is the binarization threshold, with a default value of 0.6.

[0091] S2. Perform ridge slicing on the time-frequency matrix to obtain a set of ridge slice segments;

[0092] In S2, the binarized two-dimensional amplitude spectrum time-frequency matrix is... Treat it as a pixel image and extract it using image processing methods. The ridge slices of the frequency-hopping signal are analyzed, and the number and length of the frequency-hopping signal are estimated, such as... Figure 2 As shown, it includes:

[0093] (1) Morphological preprocessing

[0094] First, the expansion-corrosion-expansion method is used for treatment. ,improve Non-zero pixel connectivity, removal of non-zero noise pixels, and edge enhancement:

[0095]

[0096] Where ⊙ represents expansion operation, ⊕ represents corrosion operation, and the calculation kernel used for expansion corrosion treatment... use The identity matrix.

[0097] Then, to A 3-pixel median filter is applied to remove interference signals such as frequency sweeps and bursts, as well as residual noise pixels. At the same time, cracks are repaired and holes are filled to obtain the morphologically preprocessed time-frequency map matrix. :

[0098]

[0099] in, This indicates the calculation of the median of a set of numbers. .

[0100] (2) Edge detection

[0101] Time-frequency plot matrix after morphological preprocessing Perform edge detection to obtain a continuous and clear edge pixel map. ,Right now

[0102]

[0103] in This is the Canny edge detection function in OpenCV. and These are the low and high thresholds for Canny edge detection, typically set to... and .

[0104] (3) Slice fitting

[0105] First, fitting The possible shape outlines are:

[0106]

[0107] in This is the OpenCV contour detection fitting function; the contour retrieval mode adopts... That is, only the outer contour is retrieved, and the contour approximation method adopts... That is, all outline pixels are preserved. For detected A set of pixels representing a contour, where each contour contains the coordinates of several pixels along its boundary. ,in , For the first Number of pixels in the group outline.

[0108] Then, Each group of contours is fitted into ridge line slices, and the position information of each ridge line slice segment is calculated to obtain a set of ridge line slice segments. .

[0109] No. The first set of contour points calculated Position information vector of ridge line slice segments ],in , and The first The start time point, end time point, and frequency hopping point position of the ridge line slice segment:

[0110]

[0111]

[0112]

[0113] Specifically, the number and length of the frequency-hopping signals are estimated as follows:

[0114] 1) Estimate the number of frequency hopping signals:

[0115]

[0116] in, Indicates rounding up;

[0117] .

[0118] 2) Estimate the length of the frequency hopping signal:

[0119]

[0120] Among them, the set of lengths of all ridge slice segments , For K-means clustering function, For cluster estimation The signal length of a frequency hopping signal.

[0121] S3. Construct a frequency-hopping signal ridge map based on the set of ridge slices;

[0122] In S3, based on the ridge slice line segment diagram of the frequency hopping signal (from the ridge slice line segment set) Obtain, such as Figure 3 As shown), the equal length characteristics of the signal ridge slices and the vertical frequency hopping characteristics are used to construct a frequency hopping signal ridge map. .

[0123] Frequency hopping signal ridge diagram It consists of a series of node elements and a series of directed edge elements The set formed, such as Figure 4 As shown. Among them, , and These are nodes The horizontal and vertical pixel coordinates in the ridge map. Node set. It includes two categories: one is the endpoints of all ridge line slice segments, such as... Figure 4 middle Another type is the intermediate connection point of a line segment, which is generated by any endpoint vertically jumping from its own ridge slice line segment to other possible ridge slice line segments, such as... Figure 4 middle Etc. Edge set It also includes two types of edge elements, one of which is a set of nodes. The transverse (unidirectional) edge obtained by dividing all ridge slice segments by points in the middle, such as... Figure 4 and Figure 5 As shown by the horizontal solid line segment, this edge is called the horizontal edge, denoted as . , among them correspond left endpoint , correspond right endpoint The other type is arbitrary nodes. Jump vertically to a certain node The resulting bidirectional edge, such as Figure 4 and Figure 5 As shown by the vertical dashed line segment, this edge is called the vertical edge, denoted as . . Figure 5 for Figure 4 A visualization of all edges in the frequency hopping signal ridge diagram.

[0124] S4. Perform a ridge path search based on equal-length jump constraints on the ridge map of the frequency hopping signal to obtain the joint sorting results and frequency set parameters of the frequency hopping signal.

[0125] In S4, the ridge path search is an A* path search algorithm based on equal-length jump constraints, which is performed by analyzing the ridge map of the frequency hopping signal. Ridge path search is performed to achieve joint sorting and frequency set estimation of ridge slices of multiple frequency hopping signals, even in cases where frequency repetition of frequency hopping signals is possible.

[0126] The A* path search algorithm based on equal-length jump constraints incorporates the actual cost function. and heuristic functions This allows for a comprehensive evaluation of node priorities, thereby guiding the search path efficiently toward the ridgeline of the same frequency hopping signal.

[0127] Cost function : Indicates starting from a given starting point To the current node The actual cost is defined as the node To the starting point The horizontal (lateral) distance, i.e.

[0128]

[0129] in, This represents the horizontal distance between nodes.

[0130] Heuristic functions : Represents the constraint cost of equal-length transitions along the ridge of a frequency-hopping signal, while also considering the special case of equal-length transitions where the frequency is the same before and after the transition. Specifically defined as...

[0131]

[0132] Among them, nodes It is a node The parent node, that is, there is an edge from point to ; To start from the beginning Passing through nodes along the way Reaching the node The average length of the ridge slice along the path; To find the modulo operation function.

[0133] Given the number of frequency hopping signals ,as well as The signal length of a frequency hopping signal The A* path search algorithm based on equal-length jump constraints is as follows:

[0134] First, initialize the path through the set of nodes. .

[0135] Then, execute The pathfinding task, for the second pathfinding task Second task, initialization , Repeat the following process:

[0136] 1) Select the ridgeline diagram of the frequency hopping signal. The endpoint of the leftmost ridge slice is used as the starting point. , will with Nodes that can be reached by a point are added to the open list. The point is its parent node. Add it to the close list so you can stop following it.

[0137] 2) Repeat the following process:

[0138] Iterate through the open list and calculate the value of each node in the list. Evaluation function ,when or When there is no parent node ,otherwise ,in yes The parent node. Find The node with the smallest value is selected as the current node to be processed, and denoted as node . .Will Move it to the off list so you no longer need to follow it. (Regarding the current node) For all other adjacent nodes, perform the following operations:

[0139] a) If the node (denoted as) If it's in the close list, ignore it. Otherwise, do the following.

[0140] b) If If it is not in the open list, add it to the open list for the current node. Its parent node. Calculate of .

[0141] c) If If it's already in the open list, check the node. To the node This path Is the value smaller? If A smaller value indicates a better path, so its parent node (denoted as node) is selected. Set it as the current node and recalculate. , and .

[0142] 3) The program terminates when one of the following conditions is met.

[0143] Path search fails when the open list is empty;

[0144] The endpoint is reached when there are no more horizontal edges to the right of the current node that allow for further progress.

[0145] 4) If the destination has been found, backtrack the complete path: starting from the destination, trace each node backwards through its parent node until the starting point. This will give you the current path. Node vectors for secondary path search ,in For the first The number of nodes traversed in the secondary path search These are the nodes that the path passes through from left to right.

[0146] 5) Update the set of nodes traversed by the path. .

[0147] Finally, from Node vectors for secondary path search Extract The frequency pixel ordinate vector of the frequency hopping signal And then calculate The frequency hopping frequency set of a frequency hopping signal ,in , , The sampling frequency of the IQ signal. The length of the IQ signal.

[0148] This embodiment also proposes to perform FSST transformation, ridge slice extraction, and ridge path search based on equal-length jump constraints on asynchronous multi-frequency hopping signals. This enables effective joint sorting and frequency set estimation of asynchronous multi-frequency hopping signals under complex conditions such as unknown number of frequency hopping signals, multiple frequency hopping signals with the same length, and possible time-frequency overlap. This improves signal recognition accuracy and processing efficiency, and provides a reliable technical means for spectrum analysis in multi-signal environments.

[0149] Example 2

[0150] This embodiment uses real-world multi-frequency hopping signal data to verify the effectiveness of the invention.

[0151] The experimental parameters are set as follows: signal sampling frequency The frequency coverage range was 100~1000MHz, and the experimental signal-to-noise ratio (SNR) was 0dB. The test signal included two asynchronous multi-frequency hopping signals with the same length and time-frequency overlap. To ensure the statistical significance of the experimental results, each test was repeated 100 times.

[0152] The experiment uses mean square relative error (MSE) to quantify the accuracy of frequency hopping set estimation. The calculation formula is as follows:

[0153]

[0154] in This is the actual frequency value. To estimate the frequency values, N is the number of elements in the frequency set.

[0155] Figure 6 The original time-domain signal is a superposition of asynchronous multi-frequency hopping signals, resulting in time-frequency overlap. Figure 7 The image shows the time-frequency diagram of the signal after FSST transformation. It can be seen that the signal has good time-frequency clustering and the overlapping areas are clearly distinguishable. Figure 8 The time-frequency image after morphological preprocessing effectively suppresses noise and preserves signal edges; Figure 9 The image shows the ridge line segment diagram obtained by fitting and extracting the ridge line segment. The ridge line segment was successfully extracted without any omissions or errors. Figure 10 To use the A* path search algorithm based on equal-length jump constraints for ridge line search and sorting results, two paths correspond to two frequency-hopping signals, accurately traversing the overlapping region to achieve complete sorting. Table 1 shows the results based on... Figure 10The calculated frequency set estimates for the two frequency-hopping signals were compared with the actual values, and the mean square relative error of the estimation was within a certain range. Within a wide range of quantities, the sorting accuracy reaches 100%.

[0156] Table 1. Frequency point set estimation results for frequency hopping signals

[0157]

[0158] Example 3

[0159] like Figure 11 As shown, this embodiment provides a joint sorting and frequency set estimation system for asynchronous multi-frequency hopping signals, including:

[0160] The signal processing module is used to preprocess the acquired asynchronous multi-frequency hopping signals to obtain a time-frequency matrix;

[0161] The ridge slicing module is used to slice the time-frequency matrix into ridge segments to obtain a set of ridge slice lines.

[0162] The ridge map construction module is used to construct a frequency hopping signal ridge map based on the set of ridge slices.

[0163] The ridge path search module is used to perform ridge path search on the ridge map of the frequency hopping signal to obtain the joint sorting result and frequency set parameters of the frequency hopping signal.

[0164] This embodiment proposes a joint sorting and frequency set estimation system for asynchronous multi-frequency hopping signals. The system preprocesses the asynchronous multi-frequency hopping signals to obtain a time-frequency matrix; then, it slices the time-frequency matrix into ridge segments to obtain a set of ridge segment lines; a ridge map of the frequency hopping signals is constructed based on the set of ridge segment lines; finally, a ridge path search based on equal-length jump constraints is performed on the ridge map of the frequency hopping signals to obtain the joint sorting results and frequency set parameters of the frequency hopping signals. This system does not require the assumption that the number of frequency hopping signals is known, and it can achieve effective joint sorting and frequency set estimation of asynchronous multi-frequency hopping signals even in complex situations where the frequency hopping signals have the same length and overlapping time and frequency, improving signal recognition accuracy and processing efficiency, and providing reliable technical support for spectrum analysis in complex electromagnetic environments.

[0165] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. For those skilled in the art, other variations or modifications can be made based on the above description. It is impossible to exhaustively list all the implementation methods here. All obvious variations or modifications derived from the technical solutions of the present invention are still within the protection scope of the present invention.

Claims

1. A method of joint sorting and frequency set estimation of an asynchronous multi-hop frequency signal, characterized in that, The method comprises the following steps: S1. Preprocessing the collected asynchronous multi-frequency hopping signal to obtain a time-frequency matrix; S2. Ridge line slicing the time-frequency matrix to obtain a ridge line slicing segment set; S3. Constructing a frequency hopping signal ridge line graph according to the ridge line slicing segment set; S4. Performing ridge line path searching on the frequency hopping signal ridge line graph based on an equal-length hopping constraint to obtain a frequency hopping signal joint sorting result and a frequency set parameter; The frequency hopping signal ridge graph comprises a node set and an edge set , wherein represents an i-th node, represents a node and a node , and an edge between the node and the node , performing a ridge path search on the frequency hopping signal ridge graph based on an equal-length hopping constraint to obtain a frequency hopping signal joint sorting result and a frequency set parameter. S41. The search evaluation function of the node set is constructed according to the node set S42. The search evaluation function of the node set is constructed according to the node set S43. When the node has no parent node, the calculation expression of the search evaluation function is as follows: wherein, a cost function of a node a heuristic function of a node a heuristic function of a node​ When is a node has a parent node , the search evaluation function is computed as follows: wherein, represents a cost function of a parent node ; S42. evaluate the search with the search evaluation function ridge path search is targeted to be minimal until the search evaluation function the current node corresponding to the minimum the end is reached when there are no more lateral edges to the right on which to continue S43. Starting from the terminal point, passing through its parent node in reverse until the starting point to obtain a frequency hopping path; S44. Based on the frequency hopping path, extracting a frequency hopping signal sorting result and a frequency set parameter.

2. The method of joint sorting and frequency set estimation of asynchronous multi- frequency hopped signals according to claim 1, characterized in that, The preprocessing of the collected asynchronous multi-frequency hopping signal to obtain a time-frequency matrix comprises the following steps: S11. Decomposing the asynchronous multi-frequency hopping signal into an IQ signal; S12. Performing time-frequency transformation on the IQ signal to obtain an initial time-frequency matrix; S13. Binarizing the initial time-frequency matrix to obtain a final time-frequency matrix.

3. The method of joint sorting and frequency set estimation of asynchronous multi- frequency hopped signals according to claim 2, characterized in that, said IQ signal is time-frequency transformed to obtain an initial time-frequency matrix as follows: wherein, is the IQ signal, , is the length of the IQ signal, , is the number of frequency points, is a Fourier synchronous compressive transform operator; binarizing the initial time-frequency matrix to obtain a final time-frequency matrix as follows: wherein, is a maximum-minimum normalized matrix, is a binarization threshold.​ 4. The method of joint sorting and frequency set estimation of asynchronous multi- frequency hopped signals of claim 1, wherein, The ridge line slicing of the time-frequency matrix to obtain a ridge line slicing segment set comprises the following steps: S21. Performing morphological processing on the time-frequency matrix to obtain a morphologically enhanced time-frequency graph matrix; S22. Performing edge detection on the morphologically enhanced time-frequency graph matrix to obtain an edge pixel point graph; S23. Fitting the edge pixel point graph to obtain a ridge line slicing segment set.

5. The method of joint sorting and frequency set estimation of asynchronous multi- frequency hopped signals according to claim 4, characterized in that, The morphological processing of the time-frequency matrix to obtain a morphologically enhanced time-frequency graph matrix comprises the following steps: S211. dilate-erode the time-frequency matrix to obtain an edge-enhanced time-frequency matrix As follows: Wherein, ⊙ represents the dilation operation, ⊕ represents the erosion operation, and the calculation core used for the dilation and erosion processing Adopting The unit matrix, is the time-frequency matrix; S212. The edge-enhanced two-dimensional amplitude time-frequency matrix is obtained S214. The morphologically enhanced time-frequency matrix is obtained by performing median filtering on the edge-enhanced two-dimensional amplitude time-frequency matrix as follows: wherein median operation on a set of numbers, .

6. The method of joint sorting and frequency set estimation of asynchronous multi- frequency hopped signals according to claim 4, characterized in that, The edge detection is performed on the morphologically enhanced time-frequency diagram matrix to obtain an edge pixel point diagram As follows: wherein, is a Canny edge detection function, and are a low threshold and a high threshold, respectively, for Canny edge detection.

7. The method of joint sorting and frequency set estimation of asynchronous multi- frequency hopped signals according to claim 4, characterized in that, The fitting of the edge pixel point graph to obtain a ridge line slicing segment set comprises the following steps: S231. generating the edge pixel map performing shape contour fitting to obtain a contour set as follows: wherein, is a contour detection fitting function, is a search outer contour, is to keep all contour pixel points; S232. Forming the ridge slice line segment set ridge slice line segments, calculating the position information of the ridge slice line segments, and forming the ridge slice line segment set according to the ridge slice line segments with calculated position information as a set.

8. The method of joint sorting and frequency set estimation of asynchronous multi- frequency hopped signals of claim 1, wherein, The obtaining of a frequency hopping signal joint sorting result and a frequency set parameter based on the frequency hopping path comprises the following steps: S441. Extracting a joint sorting result from the frequency hopping path; S442. multiplying the frequency pixel ordinate of the node in the frequency hopping path by the sampling frequency of the asynchronous multi-hop frequency signal, and dividing by the length of the IQ signal to obtain a frequency set parameter : wherein, K is the number of asynchronous multi-hop frequency signals, , , is the IQ signal sampling frequency, is the IQ signal length.

9. A system for joint sorting and frequency set estimation of an asynchronous multi-hop frequency signal, characterized by The method comprises the following steps: A signal processing module is configured to preprocess a collected asynchronous multi-frequency hopping signal to obtain a time-frequency matrix; A ridge line slicing module is configured to perform ridge line slicing on the time-frequency matrix to obtain a ridge line slicing segment set; A ridge line graph construction module is configured to construct a frequency hopping signal ridge line graph according to the ridge line slicing segment set; A ridge line path searching module is configured to perform ridge line path searching on the frequency hopping signal ridge line graph to obtain a frequency hopping signal joint sorting result and a frequency set parameter; The frequency hopping signal ridge map includes a set of nodes. Sum of edges ,in This represents the i-th node. Represents a node and nodes The edges between them, the ridge path search based on equal-length jump constraints on the ridge map of the frequency hopping signal, to obtain the joint sorting result and frequency set parameters of the frequency hopping signal, include: S41. The search evaluation function of the set of nodes is constructed S42. The search evaluation function of the set of nodes is constructed S43. The search evaluation function of the set of nodes is constructed wherein, a cost function of a node a heuristic function of a node a heuristic function of a node​ When is a node has a parent node , the search evaluation function is computed as follows: wherein, represents the cost function of the parent node ; S42. evaluate the search with the search evaluation function a spine path search is targeted to be minimal until the search evaluation function the current node corresponding to the minimum the end is reached when there are no more lateral edges to the right on which to proceed S43. Starting from the terminal point, passing through its parent node in reverse until the starting point to obtain a frequency hopping path; S44. Based on the frequency hopping path, extracting a frequency hopping signal sorting result and a frequency set parameter.

Images