Frequency hopping signal parameter identification method and system

Abstract

The application discloses a frequency hopping signal parameter identification method, comprising the following steps: optimizing a signal connected region; extracting parameters of the signal after the connected region optimization under multi-target signal aliasing; and identifying the signal with the extracted parameters. The application is characterized in that: firstly, the connected region is optimized, including: connected region segment association under the condition of signal connected region fracture and connected region interference suppression; and secondly, a multi-target connected region reconstruction method based on amplitude information difference is designed for multi-target frequency hopping signal aliasing, which greatly improves the identification rate of the frequency hopping signal.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of signal estimation and identification, and particularly relates to a frequency hopping signal parameter identification method and system. BACKGROUND

[0002] Under a low signal-to-noise ratio and interference environment, the estimation of the signal connected region by the traditional connected region marking method is inaccurate, resulting in a large error in subsequent parameter extraction. Firstly, under the condition of a low signal-to-noise ratio, the self-adaptive threshold noise removal method based on energy statistics will make the amplitudes of part of the signal region zero, thereby causing the breakage of the same type of signal connected region. Secondly, due to the existence of fixed frequency and slope signal interference in the time-frequency distribution of the signal, collision with the target signal (frequency hopping signal) is prone to occur, resulting in the distortion of the shape of the connected region.

[0003] When multiple target frequency hopping signals occur time-frequency aliasing, the connected region is severely deformed, resulting in a large error in time-frequency parameter extraction. SUMMARY

[0004] In order to solve the problems existing in the prior art, the application provides a frequency hopping signal parameter identification method, which can improve the frequency hopping signal identification rate.

[0005] The technical problem to be solved by the application is solved by the following technical scheme:

[0006] In a first aspect, a frequency hopping signal parameter identification method is provided, comprising:

[0007] optimizing the signal connected region;

[0008] performing parameter extraction on the signal after the optimization of the connected region under multiple target signal aliasing;

[0009] identifying the signal of the extracted parameters.

[0010] In combination with the first aspect, further, the optimization of the signal connected region comprises:

[0011] associating and splicing the broken fragments of the signal connected region;

[0012] performing interference suppression on the spliced signal connected region.

[0013] In combination with the first aspect, further, the associating and splicing of the broken fragments of the signal connected region comprises:

[0014] judging whether the broken fragments of the signal connected region meet the association condition;

[0015] performing amplitude filling on the broken regions between the broken fragments of the signal connected region that meet the association condition.

[0016] With reference to the first aspect, further, the judging whether the signal connected area broken piece meets the correlation condition comprises:

[0017] If the formula (1) is established, it means that the correlation condition is met

[0018]

[0019] Wherein, x i is the starting time of the i-th signal piece, y i is the starting frequency of the i-th signal piece, L i is the duration of the i-th signal piece, H i is the frequency bandwidth of the i-th signal piece, Δt represents the number of resolution units of time difference, ΔH represents the number of frequency units of bandwidth difference, and Δf represents the number of frequency units of center frequency difference.

[0020] With reference to the first aspect, further, the amplitude filling of the broken area between the signal connected area broken pieces meeting the correlation condition comprises:

[0021] The time-frequency amplitude of the broken area between the signal connected area broken pieces meeting the correlation condition is set as the mean value of the amplitudes of the two broken pieces, and the mean value is represented by formula (2)

[0022]

[0023] With reference to the first aspect, further, the interference suppression of the spliced signal connected area comprises:

[0024] On the connected area label distribution map, the fixed frequency signal interference is suppressed, comprising:

[0025] The frequency hopping period L i of the i-th connected area is extracted, if L i ≥ L max , it is determined that the connected area is interfered by the fixed frequency, and the area is set to zero;

[0026] The ramp signal interference is suppressed, comprising:

[0027] On the connected area label distribution map, it is judged whether the overall zero ratio Δn of the connected area is less than the overall zero ratio threshold Δn1, if not, it is further judged whether the zero ratio of each row of the connected area is not less than the row zero ratio threshold Δn2, if not, it is determined that the connected area is interfered by the ramp signal, and the non-zero position signal of the row is set to zero;

[0028] Wherein, L max is the maximum frequency modulation period,

[0029] With reference to the first aspect, further, the parameter extraction of the signal after the optimization of the connected region comprises:

[0030] determining a minimum rectangular boundary according to the connected region marking distribution map, and extracting a parameter vector (x a ,y a ,L a ,H a ) according to the minimum rectangular boundary;

[0031] calculating the amplitude difference between the frequency point at the next moment and the frequency point at the previous moment row by row, setting the (x+1, y) point satisfying formula (3) as the aliasing column boundary point, and obtaining all points of the aliasing column boundary after the calculation of each row is completed;

[0032]

[0033] wherein σ A is the root mean square error of the non-zero amplitude value in the minimum rectangular boundary of the connected region, k is a scale factor, F(x+1, y) and F(x, y) are the amplitudes of the frequency point at the next moment and the frequency point at the previous moment respectively, x a and y a are the horizontal coordinate and the vertical coordinate of the lower left corner of the minimum rectangular boundary respectively, L a and H a are the horizontal length and the vertical length of the minimum rectangular boundary respectively;

[0034] calculating the amplitude difference between the frequency point at the next moment and the frequency point at the previous moment column by column, setting the (x, y+1) point satisfying formula (4) as the aliasing row boundary point, and obtaining all points of the aliasing row boundary after the calculation of each column is completed;

[0035]

[0036] wherein F(x, y+1) and F(x, y) are the amplitudes of the frequency point at the next moment and the frequency point at the previous moment respectively;

[0037] re-marking the connected region according to the obtained aliasing column boundary point and the aliasing row boundary point, separating two connected regions, and extracting the time-frequency parameter according to the marked connected region information.

[0038] With reference to the first aspect, further, the identification of the extracted parameter comprises:

[0039] firstly clustering the signal according to the frequency hopping period L and the bandwidth H in the extracted parameter in the time-frequency signal by using the DBSCAN algorithm, and then re-clustering the clustered signal according to the signal occurrence moment by using the DBSCAN algorithm;

[0040] The frequency hopping period L, the bandwidth H and the center frequency f0 of the signal sorted by the secondary clustering are used to form a signal feature vector, and the signal category is determined according to the signal feature vector and the signal feature vector in the unmanned aerial vehicle remote control signal feature database.

[0041] In combination with the first aspect, further, the determination of the signal category comprises:

[0042] The feature vector [L, H, f0] of the signal sorted by the secondary clustering is calculated according to formula (5), and the Euclidean distance between the feature vector [L i ,H i ,f 0i ](i=1, 2, …N) of the signal feature vector in the unmanned aerial vehicle remote control signal feature database is calculated.

[0043]

[0044] If the Euclidean distance R i is the smallest when i=k, the target signal is determined as the kth signal.

[0045] The second aspect provides a frequency hopping signal parameter identification system, comprising:

[0046] A connected region optimization module is configured to optimize the signal connected region.

[0047] A parameter extraction module is configured to extract parameters from the signal after the connected region optimization under the condition of multi-target signal mixing.

[0048] A signal identification module is configured to identify the signal with the extracted parameters.

[0049] The present application has the following advantages: firstly, the connected region is optimized, including: the connection of the connected region fragments under the condition of the broken connected region of the signal, and the interference suppression of the connected region; secondly, a multi-target connected region reconstruction method based on the amplitude information difference is designed for the multi-target frequency hopping signal mixing, which greatly improves the identification rate of the frequency hopping signal. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The flowchart of the present application;

[0051] Figure 2 The interference and frequency modulation signal connected region marking distribution diagram in the present application;

[0052] Figure 3 The signal mixing schematic diagram in the present application;

[0053] Figure 4 The parameter extraction schematic diagram after the connected region reconstruction in the present application;

[0054] Figure 5Time-frequency distribution diagram of signal after denoising in traditional method;

[0055] Figure 6 Connected region marking diagram of signal after optimization in traditional method;

[0056] Figure 7 Time-frequency distribution diagram of signal after denoising in the method of the application;

[0057] Figure 8 Connected region marking diagram of signal after optimization in the method of the application;

[0058] Figure 9 Comparison diagram of frequency hopping signal recognition probability between the application and traditional method. DETAILED DESCRIPTION

[0059] In order to further describe the technical features and effects of the application, the application is further described below in combination with the drawings and specific embodiments.

[0060] Embodiment 1

[0061] As shown in the drawings, the frequency hopping signal parameter recognition method provided by the application comprises the following steps: Figure 1

[0062] Step 1, signal connected region optimization

[0063] The optimization mainly comprises correlating and splicing the broken fragments of the signal connected region and suppressing interference of the spliced signal connected region.

[0064] First, the broken fragments of the signal connected region are correlated and spliced.

[0065] To splice successfully, we first need to correlate the broken fragments of the signal connected region. Under the condition of low signal-to-noise ratio, the self-adaptive threshold noise removal method based on energy statistics will make the amplitudes of part of the signal region zero, thereby causing the breakage of the signal connected region of the same type. Therefore, the broken fragments need to be correlated according to the characteristic similarity, and then spliced. The correlation conditions of the broken regions of the same type of signal include the similarity of the center frequency, the similarity of the bandwidth and the time continuity (i.e. the end time of the previous signal connected region and the start time of the next signal connected region are within a small range, i.e. formula (1) is satisfied)

[0066] When the broken fragments of the signal connected region satisfy the following conditions, it is indicated that they need to be correlated, and the correlation condition is shown as formula (1)

[0067]

[0068] Let the center frequency of the i-th signal segment be (y i +H i ​ / 2), the ending time is (x i +L i ), where x i is the starting time of the i-th signal segment, y i is the starting frequency of the i-th signal segment, L i is the duration of the i-th signal segment, H i is the frequency bandwidth of the i-th signal segment, Δt represents the number of resolution units of the time difference, ΔH represents the number of frequency units of the bandwidth difference, and Δf represents the number of frequency units of the center frequency difference. Here, Δt=3, ΔH=1, and Δf=1, which means that the time difference is less than 3 time resolution units, the bandwidth difference is less than 1 frequency unit, and the center frequency difference is less than 1 frequency unit.

[0069] In order to complete the splicing, the amplitude of the broken area between the broken segments of the signal connected region that meets the correlation condition needs to be filled.

[0070] That is, the time-frequency amplitude of the broken area between the broken segments of the signal connected region that meets the correlation condition is set to the average of the amplitudes of the two broken segments. At this point, the splicing is completed, and the average is represented by formula (2)

[0071]

[0072] Next, interference suppression needs to be performed on the spliced signal connected region to remove the connected region formed by the interference signal.

[0073] Due to the existence of fixed-frequency and ramp signal interference in the signal time-frequency distribution, it is easy to collide with the target signal (frequency hopping signal), resulting in distortion of the connected region shape and ultimately producing a large time-frequency parameter estimation error. The distribution of the interference and frequency hopping signal on the connected region marker map is shown in Figure 2 .

[0074] Observing the distribution characteristics of the fixed-frequency and ramp signal interference, it can be found that there is a large difference between the two types of interference and the connected region marker distribution of the target signal. Among them, the fixed-frequency signal interference is continuously distributed in a fixed frequency range for a long period of time, i.e., the time axis ratio is very high, while the frequency hopping signal has a relatively low time axis ratio. The ramp signal interference has a very high zero ratio in the smallest rectangular boundary formed by the connected region, while the frequency hopping signal has a relatively low zero ratio. Based on the above analysis of the characteristic differences, interference suppression can be performed on the connected region marker distribution map.

[0075] First, the fixed-frequency signal interference is suppressed:

[0076] For all connected regions, the frequency hopping period is extracted. The specific method is as follows: for the i-th connected region, the starting time x s and the ending time x e are extracted, and the frequency hopping period Li = x e - x s , extract the frequency hopping period L of the ith connected region i , if L i ≥ L max ((L max is the maximum frequency hopping period in the known frequency hopping signal database)) is satisfied, it is determined that the connected region is interfered by a fixed frequency signal, and the region is set to zero;

[0077] Next, the ramp signal interference is suppressed, specifically:

[0078] On the connected region marking distribution map, it is determined whether the overall zero ratio of the connected region is less than the overall zero ratio threshold An1, where An1 is taken as 20%, if it is not less than, it is further determined whether the zero ratio of each row of the connected region is not less than the row zero ratio threshold An2, if it is not less than, it is determined that the behavior is a ramp signal, and the signal at the non-zero position of the row is set to zero;

[0079] wherein,

[0080] When the suppression is completed, the entire optimization process is ended, and next, parameter extraction needs to be performed.

[0081] Step two, parameter extraction under multi-target signal aliasing

[0082] Due to the differences in the occurrence time (starting time), frequency hopping period, center frequency and bandwidth of different frequency hopping signals, there may be aliasing in the time and frequency dimensions on the time-frequency distribution, causing deformation of the connected region, and causing very large errors in parameter extraction. The signal aliasing is shown in Figure 3 .

[0083] Although aliasing occurs in the time and frequency dimensions of the two frequency hopping signals, since the denoised time-frequency distribution matrix retains the signal amplitude information, and the amplitudes of the two frequency hopping signals in the aliasing area are obviously larger than the amplitude of the single signal in the non-aliasing area, therefore, the abruptness of the amplitude can be used to reconstruct the connected regions of different frequency hopping signals, so as to separate different frequency hopping signals and more accurately extract signal parameters. The parameter extraction steps under multi-target signal aliasing are as follows:

[0084] According to the connected region marking distribution map, the minimum rectangular boundary is determined, and the parameter vector is extracted according to the minimum rectangular boundary, specifically: the connected region is marked, the time-frequency two-dimensional amplitude value is 0 / 1 binary, and the marking matrix is formed, the minimum rectangular boundary is determined according to the marking matrix, the left lower corner point horizontal coordinate x a , vertical coordinate y a of the minimum rectangular boundary represent the starting time and starting frequency of the signal respectively; the rectangular horizontal axis width L a , vertical axis width Ha respectively signal time width and frequency band width, so the parameter vector (x a ,y a , L a , H a ) can be extracted according to the minimum rectangular border.

[0085] The amplitude difference of the frequency points of the next time and the previous time is calculated row by row, the (x+1, y) point satisfying formula (3) is set as the aliasing column boundary point, and after the calculation of each row is completed, all the points of the aliasing column boundary are obtained.

[0086]

[0087] Wherein, σ A is the root mean square error of the non-zero amplitude value in the minimum rectangular border of the connected region, k is the scale factor, F(x+1, y) and F(x, y) are the amplitudes of the frequency points of the next time and the previous time respectively, x a and y a are the horizontal coordinate and vertical coordinate of the lower left corner of the minimum rectangular border respectively; L a and H a are the horizontal length and vertical length of the minimum rectangular border respectively.

[0088] The amplitude difference of the frequency points of the next time and the previous time is calculated row by row, the (x+1, y) point satisfying formula (3) is set as the aliasing column boundary point, and after the calculation of each row is completed, all the points of the aliasing column boundary are obtained.

[0089]

[0090] Wherein, F(x, y+1) and F(x, y) are the amplitudes of the frequency points of the next time and the previous time respectively.

[0091] According to the obtained aliasing column boundary points and aliasing row boundary points, the connected region marking is re-performed (according to the connected characteristics of the signal distribution presenting regular region, the 8-neighbor connected method based on the region growing idea is used to re-perform the connected region marking, the marking refers to marking the connected region of the connected graph to form a marking matrix, which can separate different connected regions and extract the time-frequency parameters of each connected region), two connected regions are separated, and the time-frequency parameters are extracted according to the marked connected region information, as shown in formula (5). Figure 4

[0092] Step three, recognizing the signal of the extracted parameters

[0093] ​Since the remote control signals (frequency hopping signals) adopted by different types of unmanned aerial vehicles have different frequency hopping periods and modulation bandwidths, first, the frequency hopping period L and the bandwidth H in the extracted parameters in the time-frequency signal are clustered according to the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, and then the clustered signals are further clustered according to the time of occurrence of the signals by the DBSCAN algorithm;

[0094] The frequency hopping period L, the bandwidth H and the center frequency f0 of the signals sorted by the secondary clustering are used to form a signal feature vector, and the signal category is determined according to the signal feature vector and the signal feature vector in the unmanned aerial vehicle remote control signal feature database.

[0095] The DBSCAN algorithm is a classic density clustering algorithm, which can divide any shape outside a region with sufficiently high density into a class by describing the class as a maximum set of tightly connected nodes, and the advantage is that it does not need to preset the class.

[0096] The frequency hopping period L, the bandwidth H and the center frequency f0 of the signals sorted by the secondary clustering are used to form a signal feature vector [L, H, f0], and the signal feature vector [L i ,H i ,f 0i ] (i=1, 2, …N) in the unmanned aerial vehicle remote control signal feature database is compared, and the normalized Euclidean distance between the two is calculated, which is expressed as follows:

[0097]

[0098] When i=k, the Euclidean distance R i is the smallest, and the target signal is determined as the kth signal.

[0099] Embodiment 2

[0100] The application also provides a frequency hopping signal parameter identification system, comprising:

[0101] A connected region optimization module is configured to optimize the signal connected region.

[0102] A parameter extraction module is configured to extract parameters from the signal after the connected region optimization under multi-target signal mixing.

[0103] A signal identification module is configured to identify the signal with the extracted parameters.

[0104] Technical effects

[0105] In order to illustrate the technical effect of the present application, simulation is carried out by constructing two types of signal environments, and two indicators of signal parameter estimation error and frequency hopping signal recognition probability are selected to compare the traditional method and the method of the present application.

[0106] (1) Signal parameter estimation error

[0107] ① Effect of labeling connected regions in environment 1

[0108] The traditional method has two types of frequency hopping signals, fixed frequency interference, sweep frequency interference (ramp), and Gaussian noise in environment 1, wherein the frequency hopping signal parameters are: frequency hopping signal 1 (a certain type of unmanned aerial vehicle remote control signal): frequency hopping period is 1ms, frequency hopping frequency set, BPSK modulation bandwidth is 1.8MHz; frequency hopping signal 2 (other external unmanned aerial vehicles): frequency hopping period is 2ms, BPSK modulation bandwidth is 2MHz. In this environment, there is no multi-target frequency hopping signal aliasing. The advantage of the method of the present application lies in the correlation and splicing of the connected region segments, the suppression of the two types of interference, and the high quality of the obtained binary connected region label graph, as shown in Figure 5 and Figure 6 .

[0109] ② Effect of labeling connected regions in environment 2

[0110] The present application has two types of frequency hopping signals in environment 2 and time-frequency aliasing occurs, and also includes fixed frequency interference, sweep frequency interference (ramp), and Gaussian noise. In this environment, there is a multi-target frequency hopping signal aliasing situation. The advantage of the method of the present application lies in

[0111] The aliasing boundary signal amplitude difference feature is used to reconstruct the connected region boundary of the aliasing signal, thereby better separating the aliasing signals, and improving the quality of the connected regions, as shown in Figure 7 and Figure 8 .

[0112] ③ Comparison of parameter estimation error

[0113] The average frequency hopping period and bandwidth parameter estimation error are statistically calculated by 1000 times of Monte Carlo simulation, and the correct recognition probability of the target unmanned aerial vehicle remote control signal is statistically calculated under different signal-to-noise ratios. The parameter estimation errors of the two methods are shown in Table 1.

[0114] Table 1 Comparison of parameter estimation errors of two methods in different scenarios

[0115]

[0116] As shown in Table 1, under both Environment 1 and Environment 2, the relative error of the parameters extracted by the improved connectivity labeling method proposed in this patent is significantly smaller than that of the conventional connectivity labeling method. The performance improvement in Environment 1 is due to the improved connectivity labeling method's association of connected component segments and interference suppression, resulting in a more complete and cleaner connectivity graph of the entire time-frequency distribution. The performance improvement in Environment 2 is due to the use of connected component reconstruction based on time-frequency amplitude differences, which distinguishes the connected boundaries of the mixed frequency-hopping signals, thus making the parameter estimation more accurate.

[0117] (2) Frequency hopping signal sorting and recognition probability

[0118] like Figure 9 As shown, in a multi-target frequency hopping signal aliasing environment (Environment 2), the recognition probability of frequency hopping signal 2 (the remote control signal of an intrusive UAV) with different signal-to-noise ratios is obtained through 1000 Monte Carlo simulations. It can be seen that after adopting the improved connected component labeling method, the target recognition probability is improved due to the higher accuracy of parameter estimation, and the remote control signal of the intrusive UAV can be identified better.

[0119] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxesFigure 1 Figure 1 the functionality specified in the one or more blocks.

Claims

1. A method of identifying parameters of a frequency hopping signal, characterized by The method comprises the following steps: optimizing the signal connected region; extracting parameters of the signal after the connected region optimization under multi-target signal aliasing; identifying the signal after the parameter extraction; wherein the optimization of the signal connected region comprises: associating and splicing the broken pieces of the signal connected region; interference suppression on the spliced signal connected region; wherein the associating and splicing the broken pieces of the signal connected region comprises: judging whether the broken pieces of the signal connected region meet the association condition; amplitude filling in the broken region between the broken pieces of the signal connected region meeting the association condition; wherein the judging whether the broken pieces of the signal connected region meet the association condition comprises: if formula (1) is established, it means that the association condition is met （1）； wherein, is the start time of the th signal segment, is the start frequency of the th signal segment, is the duration of the th signal segment, is the frequency bandwidth of the th signal segment, is the number of resolution units representing the time difference, is the number of frequency units representing the bandwidth difference, is the number of frequency units representing the center frequency difference.

2. The method of claim 1, wherein, the amplitude filling in the broken region between the broken pieces of the signal connected region meeting the association condition comprises: setting the time-frequency amplitude of the broken region between the broken pieces of the signal connected region meeting the association condition as the mean value of the amplitudes of the two broken pieces, and the mean value is represented by formula (2) （2）。 3. The method of claim 1, wherein, the interference suppression on the spliced signal connected region comprises: suppressing the fixed-frequency signal interference on the connected region label distribution map, comprising: Extract the frequency hopping period of the first connected region If the following conditions are met , it is determined that the connected region is subject to fixed frequency interference, and the region where it is located is set to zero. suppressing the slant signal interference, comprising: On the connected component label distribution map, determine the overall percentage of zero-occurrence of connected components. Is it less than the overall zero-occupancy threshold? If it is not less than, then further determine whether the zero-occupancy ratio of each row in the connected region is not less than the row zero-occupancy ratio threshold. If it is not less than, then the row is determined to be a slant signal, and the signal at the non-zero position of the row is set to zero. wherein is the maximum frequency modulation period, .

4. The method of claim 3, wherein, the parameter extraction of the signal after the connected region optimization under multi-target signal aliasing comprises: Determine the minimum rectangular boundary according to the connected region marking distribution, and extract the parameter vector according to the minimum rectangular boundary ; The amplitude difference of the frequency points of the next time and the previous time is calculated row by row, and all points satisfying formula (3) are obtained The points are set as the aliasing column boundary points, and all points of the aliasing column boundary are obtained after each row is calculated. （3）； wherein, is the root mean square error of non-zero amplitude values within the minimum rectangular boundary of the connected region, is a scale factor, and are the amplitudes of the frequency points at the next and previous time instant, respectively, and are the horizontal and vertical coordinates of the lower left corner of the minimum rectangular boundary, respectively; and are the horizontal and vertical lengths of the minimum rectangular boundary, respectively. The amplitude difference between the next frequency point and the previous frequency point is calculated column by column, and all points satisfying formula (4) are obtained The points are set as aliasing row boundary points, and all points of the aliasing row boundary are obtained after the calculation of each column is completed. （4）； wherein and are the amplitudes of the last and previous frequency points, respectively. re-performing the connected region labeling according to the obtained aliasing column boundary points and aliasing row boundary points, separating the aliasing connected regions, and extracting the time-frequency parameters according to the labeled connected region information for each separated connected region.

5. The method of claim 4, wherein, the signal identification after the parameter extraction comprises: First, the frequency hopping period in the extracted parameters in the time-frequency signal is determined by the DBSCAN algorithm , bandwidth The signals are clustered, and the clustered signals are further clustered according to the occurrence time of the signals by the DBSCAN algorithm; The frequency hopping period of the signal sorted by the secondary clustering , bandwidth , and center frequency constitute a signal feature vector, and the signal category is determined according to the signal feature vector and the signal feature vector in the unmanned aerial vehicle remote control signal feature database.

6. The method of claim 5, wherein, the determination of the signal category comprises: The feature vector of the signal selected by the quadratic clustering is calculated according to formula (5) The Euclidean distance between the feature vector of the signal and the feature vector of the signal in the feature database of the remote control signal of the unmanned aerial vehicle ​​ （5）； If when the Euclidean distance is minimum, the target signal is decided as the first class signal.

7. A frequency hopping signal parameter identification system characterized by, comprising: a connected region optimization module for optimizing the signal connected region; a parameter extraction module for extracting parameters of the signal after the connected region optimization under multi-target signal aliasing; a signal identification module for identifying the signal after the parameter extraction; wherein the optimization of the signal connected region comprises: associating and splicing the broken pieces of the signal connected region; interference suppression on the spliced signal connected region; wherein the associating and splicing the broken pieces of the signal connected region comprises: judging whether the broken pieces of the signal connected region meet the association condition; amplitude filling in the broken region between the broken pieces of the signal connected region meeting the association condition; wherein the judging whether the broken pieces of the signal connected region meet the association condition comprises: if formula (1) is established, it means that the association condition is met （1）； wherein, is the start time of the th signal segment, is the start frequency of the th signal segment, is the duration of the th signal segment, is the frequency bandwidth of the th signal segment, is the number of resolution units representing the time difference, is the number of frequency units representing the bandwidth difference, is the number of frequency units representing the center frequency difference.

Images