Unmanned aerial vehicle communication frequency band monitoring method and interference method

Abstract

The invention relates to the field of unmanned aerial vehicle communication, in particular to an unmanned aerial vehicle communication frequency band monitoring method and interference method.The method comprises the following steps that 1, background noise signals of a full frequency band are collected, Fourier transform is conducted on the collected background noise signals, and a signal frequency band with energy exceeding a preset threshold value is extracted; step 2, acquiring each signal frequency band with energy, and performing adaptive IMF component decomposition on each signal frequency band with energy; 3, screening the IMF components according to the impact energy of the IMF components, and performing signal reconstruction on the screened IMF components; and step 4, inputting the reconstructed signal into the neural network model for rule analysis, and generating a risk coefficient of existence of communication in a signal frequency band according to the rule analysis.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) communication technology, and more specifically, to a method for monitoring UAV communication frequency bands. Background Technology

[0002] The content in this section provides only background information related to this application and may not constitute prior art.

[0003] With the rapid popularization of drone technology, its applications in logistics inspection, agricultural plant protection, and aerial filming are becoming increasingly widespread. However, the disorderly use of drones has also brought a series of security risks, including invasion of privacy areas, interference with civil aviation flights, illegal surveying, and even terrorist attacks. As a core component of countermeasures systems, drone communication monitoring technology can effectively achieve early warning, threat assessment, and control response for unauthorized drone flights by detecting, identifying, and locating the communication signals between the drone and its remote controller in real time. This is of great strategic significance for maintaining airspace security, protecting critical infrastructure, and safeguarding citizens' privacy.

[0004] Currently, mainstream UAV communication monitoring systems primarily employ single-band signal scanning and analysis technology. The underlying principle involves using a narrowband or wideband receiver to scan the spectrum of commonly used UAV communication frequency bands (such as the 2.4GHz and 5.8GHz ISM bands). When the detected signal energy exceeds a threshold, the system extracts the time-domain characteristics (such as modulation scheme and code rate), frequency-domain characteristics (such as spectrum waveform and bandwidth), or protocol fingerprints (such as WiFi, Bluetooth, and DJI proprietary protocol characteristics) of the signal in that frequency band to determine whether it is a UAV communication signal and attempt to identify the UAV model. Such systems typically rely on fixed monitoring stations or mobile monitoring equipment, using a single-channel receiver to achieve continuous monitoring of the target frequency band. Summary of the Invention

[0005] The summary section of this application is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this application propose a method for monitoring the communication frequency band of unmanned aerial vehicles (UAVs) to solve the technical problems mentioned in the background section above.

[0007] As a first aspect of this application, some embodiments of this application provide a method for monitoring the communication frequency band of an unmanned aerial vehicle (UAV), comprising the following steps:

[0008] Step 1: Collect background noise signals across the entire frequency band, perform Fourier transform on the collected background noise signals, and extract the frequency bands where the energy exceeds a preset threshold;

[0009] Step 2: Obtain the signal frequency bands of each energy source and perform adaptive IMF component decomposition on each signal frequency band of each energy source;

[0010] Step 3: Based on the impact energy of the IMF components, the IMF components are screened, and the screened IMF components are reconstructed.

[0011] Step 4: Input the reconstructed signal into the neural network model for pattern analysis, and generate a risk coefficient for communication in the signal frequency band based on the pattern analysis.

[0012] The UAV communication frequency band monitoring method provided by this invention effectively overcomes the shortcomings of traditional single-band scanning, which is susceptible to environmental noise interference and prone to missing non-preset frequency bands or low-power signals, by collecting background noise across the entire frequency band and using adaptive IMF component decomposition and screening. Combined with IMF component screening and reconstruction based on impulse energy, it significantly improves the ability to extract and enhance weak, transient, or overlapping UAV communication signals in complex electromagnetic environments. Finally, by using neural networks to perform regularity analysis on the reconstructed signals and generate risk coefficients, it not only achieves more accurate UAV signal identification and UAV type discrimination, but also quantitatively assesses the signal threat level, greatly reducing the false alarm rate and missed alarm rate. This provides a more reliable and intelligent basis for subsequent early warning, positioning, and countermeasure decisions, thereby significantly improving the UAV communication monitoring system's all-domain perception capability, anti-interference capability, and proactive early warning effectiveness in complex real-world scenarios, and effectively ensuring airspace safety and the protection of critical facilities.

[0013] Furthermore, step 1 includes the following steps:

[0014] Step 11: Collect background noise information x[n] and perform a fast Fourier transform on x[n];

[0015] n=0,1…N-1; N represents the number of sampling points, k represents the frequency index, j represents the imaginary unit, and π represents pi.

[0016] Step 12: Calculate the power spectral density for each frequency;

[0017] ;

[0018] Step 13: Pre-set the screening threshold u0, and combine any u frequencies whose sum of power spectrum densities exceeds u0 into a suspicious frequency band;

[0019] Step 14: Based on the lowest communication power within the surrounding visual range, screen the suspicious frequency bands. If the signal power of the suspicious frequency band exceeds the minimum power required for the UAV to communicate outside the visual range, then the suspicious frequency band is used as the signal frequency band.

[0020] This method achieves objective quantification of the full-spectrum energy distribution through accurate calculation of the Fast Fourier Transform (FFT) and Power Spectral Density (PSD) of the full-band background noise. By using multi-frequency energy aggregation screening based on statistical thresholds, it effectively identifies "suspicious frequency bands" with abnormal energy concentrations, significantly reducing false alarms caused by instantaneous noise interference at a single frequency point. Furthermore, a secondary screening mechanism based on the physical layer communication limit (the lowest communication power outside the line of sight) is introduced to filter out pseudo signals (such as environmental electromagnetic noise and interference from non-target devices) that do not meet the basic power requirements for UAV communication, based on the physical nature of signal propagation characteristics. This significantly improves the accuracy and reliability of capturing real potential UAV communication signals in complex electromagnetic environments.

[0021] Furthermore, in step 14, the minimum frequency for the drone to communicate outside the line of sight is as follows:

[0022] Step 141a: Determine the minimum ground clearance h, minimum communication range, and beam range of the UAV based on the line of sight.

[0023] Step 142a: Establish the signal power model J between the UAV and the ground-based hidden base station based on the beam range;

[0024] ;

[0025] Where B is the bandwidth of the suspected frequency band, N0 is the noise power spectral density, and P is the minimum signal power received by the ground-based hidden base station.

[0026] ;

[0027] in, Indicates the transmit antenna gain. P represents the user's antenna gain (a constant). t Let λ be the output power of the drone, and λ be the wavelength of the suspected frequency band. This is the overall power attenuation factor;

[0028] Step 143a: Calculate the UAV's output power P based on the minimum signal power received by the set ground-based hidden base station. t The output power P of the drone t This serves as the minimum power required for drones to communicate outside of visual range.

[0029] This method, for the first time in the field of UAV monitoring, establishes a precise signal power model that integrates spatial geometric constraints (minimum ground clearance h, beam range) and communication channel physical characteristics (free-space path loss, antenna gain, bandwidth, noise) to calculate the dynamic minimum communication power threshold based on the physical layer propagation limit. This model comprehensively considers the core limiting factors faced by UAVs conducting covert communication outside the visual range (such as highly correlated path loss, directional beam gain, environmental noise, and receiver sensitivity). Through rigorous derivation of Shannon's formula and transmission equations, it calculates the theoretical minimum transmit power necessary to maintain a covert communication link. Using this physically reliable and environmentally adaptive value as the power threshold for screening suspicious frequency bands in step 14 fundamentally ensures the scientific rigor and scenario adaptability of the signal screening criteria.

[0030] Furthermore, in step 14, the signal frequency band is generated as follows:

[0031] Step 141b: Obtain the minimum power required for the drone to communicate outside of visual range;

[0032] Step 142b: Calculate the band power of the suspected frequency band. ;

[0033] ;

[0034] ;

[0035] ;

[0036] in, denoted as unilateral power spectrum, k represents the frequency index within the suspected frequency band, and Na represents the total number of sampling points within the suspected frequency band;

[0037] Step 143b: Determine the band power of the suspected frequency band. If the frequency band is greater than the minimum power required for the drone to communicate outside the line of sight, the suspicious frequency band will be used as the signal frequency band; otherwise, it will be deleted.

[0038] This method ensures the mathematical rigor and physical accuracy of frequency band power calculations by accurately calculating the one-sided power spectral density of suspicious frequency bands and strictly adhering to the symmetry of the discrete Fourier transform for power correction (such as special handling for DC component k=0 and Nyquist frequency point k=N / 2). It utilizes dynamic band boundary indexing to achieve flexible and accurate power integration for suspicious frequency bands of arbitrary width and location. Finally, the actual frequency band power obtained by integration is directly compared with the minimum communication power threshold calculated based on the physical propagation model, forming a physically interpretable and environmentally adaptive final judgment criterion. This step not only effectively eliminates misjudgments caused by inaccurate power calculations (such as improper handling of bilateral spectra or endpoints) but also ensures that only signal frequency bands with sufficient energy levels to support a physically feasible and concealed UAV communication link are retained and sent to subsequent analysis stages. This maximizes the elimination of invalid interference signals at the source, significantly improving the quality and reliability of the input signals of the entire monitoring system.

[0039] Furthermore, step 2 includes the following steps:

[0040] Step 21: Obtain the suspicious frequency band h(t), set the initial residual r0(t), r0(t) = h(t), and set the IMF component index i;

[0041] Step 22: Set the screening signal h0(t) = r i-1 (t), set the iterative calculator g=0;

[0042] Step 22: Extract the current screening signal h g All local maxima and local minima of (t);

[0043] Step 23: Connect the maximum points using cubic spline interpolation to form the upper envelope. The lower envelope is formed by connecting the minimum points through cubic spline interpolation. Calculate the mean envelope ;

[0044] ;

[0045] Step 24: Update the screening signal Check the extreme point-zero crossing balance and the mean of the envelope of the sieved signal to be close to zero. If these conditions are met, the sieved signal is taken as the IMF component. i (t);

[0046] IMF i (t) = If the condition is not met, then set g equal to g+1 and repeat the above steps until all IMF components are obtained.

[0047] This method employs Empirical Mode Decomposition (EMD) to perform a fully data-driven adaptive decomposition of the suspicious frequency band signal h(t) selected in step 1, without requiring pre-defined basis functions or prior models. The process iteratively extracts local extrema, constructs cubic spline envelopes, and calculates the instantaneous mean envelope m_g(t), gradually separating the intrinsic oscillation modes (IMF components) at different time scales within the signal. Strict screening stopping criteria (balance between the number of extrema and zero-crossing points, and a mean envelope close to zero) ensure that each IMF component satisfies the physical requirements of the Hilbert transform for "intrinsic mode functions" (i.e., narrowband and symmetry), thereby accurately removing superimposed, non-stationary transient features (such as burst pulses and modulation jumps in UAV communication) and background noise from the original signal. This adaptive decomposition method is particularly adept at handling the common nonlinear and non-stationary characteristics of UAV communication signals.

[0048] Furthermore, step 3 includes the following steps:

[0049] Step 3 includes the following steps:

[0050] Step 31: Extract all IMFs i (t) component, calculate IMF i Kurtosis K of the (t) component i and peak factor C i ;

[0051] Step 32: Pre-set the threshold values ​​K0 and C0 for kurtosis and peak factor;

[0052] Step 33: If the kurtosis K i Greater than K0, and C i >C0 then retains the IMF i (t) component.

[0053] This method achieves intelligent identification and screening of transient impulse energy in IMF components by calculating the kurtosis and crazing factor of each IMF component and setting a joint judgment threshold. Kurtosis (measuring the sharpness of signal distribution) and crazing factor (measuring the ratio of peak value to RMS value), as key statistical indicators characterizing signal impulses, can effectively distinguish between high-impact components representing key transient features of UAV communication (such as protocol synchronization headers, burst data packets, and modulation transition edges) and low-impact components representing background noise or stable interference. This screening mechanism retains only IMF components that simultaneously meet the high thresholds for both kurtosis and crazing factor, precisely focusing on components containing core time-domain transient information of UAV communication and significantly suppressing noise-dominated or slowly changing useless components. This impulse energy-based screening strategy injects high-value threat feature information into subsequent signal reconstruction, significantly improving the identifiability and purity of target UAV communication patterns in the reconstructed signal. It lays a high-quality, strongly correlated feature foundation for high-precision regularity analysis and risk coefficient generation by neural networks, effectively enhancing the robustness of the system in detecting and identifying weak, transient UAV signals in complex electromagnetic environments.

[0054] Furthermore, the neural network model in step 4 includes:

[0055] The input layer receives the raw signal sequence data;

[0056] The first convolutional block extracts local basic feature patterns;

[0057] The second convolutional block captures abstract high-level features;

[0058] The time-series modeling layer models long-term time dependencies;

[0059] Feature aggregation layer compresses temporal features into a global representation;

[0060] The classification output layer outputs the probability of the existence of a pattern.

[0061] The loss function of the neural network model is: :

[0062] ;

[0063] Where y represents the label indicating whether a pattern exists, x represents the sample, α represents the balance factor, γ represents the focusing parameter, and p represents the model prediction probability.

[0064] As a second aspect of this application, this application provides an interference method, comprising the following steps:

[0065] S1: The aforementioned UAV communication frequency band monitoring method is used to generate a risk factor for communication in the signal frequency band;

[0066] S2: Send noise signals to high-risk signal frequency bands.

[0067] Furthermore, S2 includes the following steps:

[0068] S21: Extract the signal length in the signal frequency band U(t) whose amplitude is higher than the background noise;

[0069] S22: Randomly generate a signal with an amplitude similar to the background noise and emit it in that frequency band.

[0070] This application generates a signal with an amplitude similar to the background noise, which can greatly increase the signal capacity of noise that is difficult to filter in the frequency band of the signal, severely reduce the signal-to-noise ratio of the channel, and make it impossible for the drone to communicate normally with the secret base station.

[0071] The technical solution provided in this application has the following technical effects:

[0072] Based on full-band scanning and physical propagation model (Friis equation + Shannon capacity), dynamic power threshold screening breaks through the limitations of traditional fixed frequency band monitoring. It achieves highly reliable capture of non-preset frequency band, low power and weak edge signals in complex electromagnetic environments, reducing the false detection rate by more than 60%.

[0073] By using EMD adaptive decomposition and kurtosis-peak factor dual-index impact screening, noise and stationary interference are accurately removed, and the intensity of UAV transient features (such as protocol synchronization header and jump pulse) in the reconstructed signal is improved by 3-5dB, providing a high signal-to-noise ratio input for subsequent analysis;

[0074] By leveraging neural networks to mine the time-frequency joint patterns of reconstructed signals, a physically interpretable risk coefficient is generated, enabling a leap from "signal presence or absence" to "threat level," reducing the false alarm rate by more than 40% and improving control and response efficiency by 50%. Attached Figure Description

[0075] Figure 1 This is a flowchart of a method for monitoring the communication frequency band of unmanned aerial vehicles (UAVs). Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments. The same reference numerals in the accompanying drawings represent the same components. It should be noted that the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the described embodiments of this application without creative effort are within the scope of protection of this application.

[0077] Compared to the embodiments shown in the accompanying drawings, feasible embodiments within the scope of this application may have fewer components, other components not shown in the drawings, different components, differently arranged components, or components with different connections, etc. Furthermore, two or more components in the drawings may be implemented in a single component, or a single component shown in the drawings may be implemented as multiple separate components.

[0078] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application pertains. The terms “first,” “second,” and similar terms used in this specification and claims do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms “an” or “a” and similar terms do not necessarily indicate a quantity limitation. Terms such as “upper” and “lower” are used only to indicate relative positional relationships, and these relative positional relationships may change accordingly when the absolute position of the described object changes.

[0079] refer to Figure 1 Example 1: The method for monitoring the communication frequency band of unmanned aerial vehicles includes the following steps:

[0080] Step 1: Collect background noise signals across the entire frequency band, perform Fourier transform on the collected background noise signals, and extract the frequency bands where the energy exceeds the preset threshold.

[0081] Step 1 includes the following steps:

[0082] Step 11: Collect background noise information x[n] and perform a fast Fourier transform on x[n];

[0083] n=0,1…N-1; N represents the number of sampling points, k represents the frequency index, j represents the imaginary unit, and π represents pi.

[0084] Background noise signals are signals that signal monitoring devices can directly collect. In the field of clandestine communication, the two parties transmitting signals hide them in background noise, making it difficult to detect by directly monitoring the background noise. However, by performing a Fourier transform, the time-domain signal can be converted into a frequency-domain signal, thus enabling the monitoring of the signal strength in each frequency domain.

[0085] Step 12: Calculate the power spectral density for each frequency;

[0086] ;

[0087] Step 13: Pre-set the screening threshold u0, and combine any u frequencies whose sum of power spectrum densities exceeds u0 into a suspicious frequency band;

[0088] After converting a time-domain signal to a frequency-domain signal, if there are too many signal characteristics within a frequency band, it indicates a problem at that frequency. For example, there may be abnormal communication or an abnormal signal source. In practice, although there is some energy in the background noise, the signal power is very low because there are no signal transmitting devices nearby. After converting to the frequency domain for analysis, if there is energy accumulation in a certain frequency domain, it indicates that there are devices transmitting signals in that frequency domain.

[0089] In the field of clandestine communications, dual-channel or even multi-channel communication is generally used to reduce the probability of detection. This means that each frequency band transmits only a portion of the signal, combining multiple channels into a single channel. For this purpose, frequencies whose sum of power spectral densities exceeds u0 can be combined into a suspicious frequency band. The frequencies within a suspicious band do not need to be continuous; they can be monitored as a single signal. The number of suspicious frequency bands extracted is related to the monitoring accuracy; the more suspicious frequency bands extracted, the higher the monitoring accuracy. In practice, this should be flexibly set according to requirements.

[0090] Step 14: Based on the lowest communication power within the surrounding visual range, screen the suspicious frequency bands. If the signal power of the suspicious frequency band exceeds the minimum power required for the UAV to communicate outside the visual range, then the suspicious frequency band is used as the signal frequency band.

[0091] In step 14, the minimum frequency for the drone to communicate outside the line of sight is as follows:

[0092] Step 141a: Determine the minimum ground clearance h, minimum communication range, and beam range of the UAV based on the line of sight.

[0093] Step 142a: Establish the signal power model J between the UAV and the ground-based hidden base station based on the beam range;

[0094] ;

[0095] Where B is the bandwidth of the suspected frequency band, N0 is the noise power spectral density, and P is the minimum signal power received by the ground-based hidden base station.

[0096] ;

[0097] in, Indicates the transmit antenna gain. P represents the user's antenna gain (a constant). t Let λ be the output power of the drone, and λ be the wavelength of the suspected frequency band. This is the overall power attenuation factor;

[0098] Step 143a: Calculate the UAV's output power P based on the minimum signal power received by the set ground-based hidden base station.t The output power P of the drone t This serves as the minimum power required for drones to communicate outside of visual range.

[0099] In step 14, the signal frequency band is generated as follows:

[0100] Step 141b: Obtain the minimum power required for the drone to communicate outside of visual range;

[0101] The line-of-sight range refers to the area that the monitoring equipment can see. For example, if the monitoring equipment can ensure that there are no drones within 10 meters in altitude, then the line-of-sight range is set to 10. The smaller the line-of-sight range, the more difficult the monitoring becomes. Whether drone communication exists within the line-of-sight range is not specifically considered in this application.

[0102] Step 142b: Calculate the band power of the suspected frequency band. ;

[0103] ;

[0104] ;

[0105] ;

[0106] in, denoted as unilateral power spectrum, k represents the frequency index within the suspected frequency band, and Na represents the total number of sampling points within the suspected frequency band;

[0107] Step 143b: Determine the band power of the suspected frequency band. If the frequency band is greater than the minimum power required for the drone to communicate outside the line of sight, the suspicious frequency band will be used as the signal frequency band; otherwise, it will be deleted.

[0108] This communication frequency is the minimum power required for a drone to communicate outside of visual range. If the power is below this, long-distance communication is not possible.

[0109] Step 2: Obtain the signal frequency bands of each energy source and perform adaptive IMF component decomposition on each signal frequency band of each energy source.

[0110] The presence of abnormal energy accumulation within the signal frequency band, exceeding the minimum communication power requirement, does not necessarily indicate abnormal communication. Therefore, IMF component decomposition of the signal is still necessary.

[0111] Step 2 includes the following steps:

[0112] Step 21: Obtain the suspicious frequency band h(t), set the initial residual r0(t), r0(t) = h(t), and set the IMF component index i;

[0113] Step 22: Set the screening signal h0(t) = r i-1 (t), set the iterative calculator g=0;

[0114] Step 22: Extract the current screening signal h g All local maxima and local minima of (t);

[0115] Step 23: Connect the maximum points using cubic spline interpolation to form the upper envelope. The lower envelope is formed by connecting the minimum points through cubic spline interpolation. Calculate the mean envelope ;

[0116] ;

[0117] Step 24: Update the screening signal Check the extreme point-zero crossing balance and the mean of the envelope of the sieved signal to be close to zero. If these conditions are met, the sieved signal is taken as the IMF component. i (t);

[0118] IMF i (t) = If the condition is not met, then set g equal to g+1 and repeat the above steps until all IMF components are obtained.

[0119] The IMF component is a portion of the signal after decomposition. By decomposing and filtering the IMF component, most of the background noise signal can be removed. If communication signals are present, kurtosis K can be used to further reduce noise. i and peak factor C i It was determined.

[0120] Step 3: Based on the impact energy of the IMF components, the IMF components are screened, and the screened IMF components are reconstructed.

[0121] Step 3 includes the following steps:

[0122] Step 31: Extract all IMFs i (t) component, calculate IMF i Kurtosis K of the (t) component i and peak factor C i ;

[0123] Step 32: Pre-set the threshold values ​​K0 and C0 for kurtosis and peak factor;

[0124] Step 33: If the kurtosis K i Greater than K0, and Ci >C0 then retains the IMF i (t) component;

[0125] Step 34: Retain the IMF i (t) components are reconstructed.

[0126] Step 4: Input the reconstructed signal into the neural network model for pattern analysis, and generate a risk coefficient for communication in the signal frequency band based on the pattern analysis.

[0127] The neural network model in step 4 includes:

[0128] The input layer receives the raw signal sequence data;

[0129] The first convolutional block, connected to the input layer, extracts local basic feature patterns based on the convolutional network;

[0130] The second convolutional block, connected to the first convolutional block, captures abstract high-level features based on the convolutional network;

[0131] The temporal modeling layer, connected to the second convolutional block, models long-range temporal dependencies based on the LSTM network.

[0132] The feature aggregation layer, connected to the temporal modeling layer, compresses temporal features into a global representation.

[0133] The classification output layer outputs the probability of regularity based on a linear activation function.

[0134] The loss function of the neural network model is :

[0135] ;

[0136] Where y represents the label indicating whether a pattern exists, x represents the sample, α represents the balance factor, γ represents the focusing parameter, p represents the model prediction probability, the input of the neural network model is the signal reconstructed in step 3, and the output is the probability (0,1) of whether a pattern exists, where 0 indicates no pattern and 1 indicates pattern exists.

[0137] The neural network model is an existing neural network model, and the specific model structure will not be further elaborated. The scheme in steps 1 to 3 provided in this application can remove a large amount of background noise from the signal, and then, using a relatively simple neural network model, it is possible to extract the regular information. The regular information here does not actually involve the neural network model deciphering the signal, but rather determining whether there are periodic amplitudes in the signal.

[0138] Example 2 provides an interference method based on Example 1. The signal interference method only requires emitting a large amount of noise signal outward in space. The key is to find the frequency band to be interfered with and the interference time to reduce the impact on one's own signal.

[0139] The interference method includes the following steps:

[0140] S1: The aforementioned UAV communication frequency band monitoring method is used to generate a risk factor for communication in the signal frequency band;

[0141] S2: Send noise signals to high-risk signal frequency bands.

[0142] Furthermore, S2 includes the following steps:

[0143] S21: Extract the signal length in the signal frequency band U(t) whose amplitude is higher than the background noise;

[0144] S22: Randomly generate a signal with an amplitude similar to the background noise and emit it in that frequency band.

[0145] The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for monitoring the communication frequency band of an unmanned aerial vehicle (UAV), characterized in that, Comprising: Step 1: Collect the full-band background noise signal, and perform Fourier transform on the collected background noise signal to extract the signal frequency band with energy exceeding a preset threshold; Step 2: Obtain each energy signal frequency band, and perform adaptive IMF component decomposition on each energy signal frequency band; Step 3: According to the impact energy of the IMF component, screen the IMF component, and reconstruct the signal of the screened IMF component; Step 4: Input the reconstructed signal into a neural network model for regularity analysis, and generate a risk coefficient of signal frequency band existence communication according to the regularity analysis. 2.The method of claim 1, wherein, Step 1 includes the following steps: Step 1 includes the following steps: Step 11: Collect background noise information x[n], and perform fast Fourier transform on x[n]; ; n = 0, 1...N - 1; N represents the number of sampling points, k represents the frequency index, j represents the imaginary unit, and p represents the circular constant; Step 12: Calculate the power spectral density of each frequency; ； Step 13: Pre-set a screening threshold u0, and combine the frequency whose sum of any u power spectral densities exceeds u0 as a suspicious frequency band; Step 14: Based on the minimum communication power in the surrounding visible range, screen the suspicious frequency band, if the signal power of the suspicious frequency band exceeds the minimum power of the unmanned aerial vehicle communicating outside the visible range, then the suspicious frequency band is regarded as a signal frequency band. 3.The method of claim 1, wherein, In step 14, the minimum frequency of the unmanned aerial vehicle communicating outside the visible range is as follows: Step 141a: Determine the minimum height h of the unmanned aerial vehicle, the minimum communication range of the unmanned aerial vehicle, and the beam range of the unmanned aerial vehicle based on the visible range; Step 142a: According to the beam range, establish a signal power model J of the unmanned aerial vehicle and the ground hidden base station; ； Wherein, B is the frequency band bandwidth of the suspicious frequency band, N0 is the noise power spectral density, and P is the minimum signal power received by the ground hidden base station; ； wherein, denotes the transmit antenna gain, denotes the user's antenna gain (constant value), P t is the output power of the drone, and λ is the wavelength of the suspicious frequency band, is the combined power attenuation factor; Step 143a: Calculate the output power P of the drone according to the set minimum signal power received by the ground hidden base station t , the output power P of the drone t is the minimum power of the drone to communicate outside the visual range. 4.The method of claim 3, wherein, In step 14, the signal frequency band is generated as follows: Step 141b: Obtain the minimum power of the unmanned aerial vehicle communicating outside the visible range; Step 142b: Calculate band power of suspicious band ; ； ； ； wherein represents a one-sided power spectrum, k represents a frequency index within the suspicious frequency band, and Na represents the total number of sampling points within the suspicious frequency band. Step 143b: judging the band power of the suspicious band whether greater than the minimum power of the UAV communicating outside the visual range, if greater, the suspicious band is regarded as the signal band, if less, the suspicious band is deleted. 5.The method of claim 4, wherein, Step 2 includes the following steps: Step 21: Obtain the suspicious frequency band h(t), set the initial residual error r0(t), r0(t)=h(t), and set the IMF component index i; Step 22: Set the sieving signal h0(t) = r i-1 (t), set the iteration counter g = 0; Step 22: Extracting the current sieving signal h g all local maximum points and local minimum points of h(t). Step 23: Connect the maximum points by cubic spline interpolation to form the upper envelope Connect the minimum points by cubic spline interpolation to form the lower envelope Calculate the mean envelope ; ; Step 24: update the sifting signal , check the extreme points of the sifting signal - the zero-crossing balance and the envelope mean value is close to zero, if satisfied, the sifting signal is taken as the IMF component IMF i (t); ; IMF i (t) = 0 ; if not, let g equal g + 1 and repeat the above steps until all IMF components are obtained. 6.The method of claim 5, wherein, Step 3 includes the following steps: Step 3 includes the following steps: Step 31 : Extract all the IMFs i (t) component, calculate the IMF i (t) component of the kurtosis K i and the peak factor C i ; Step 32: Pre-set the kurtosis and peak factor determination threshold K0 and C0; Step 33: If kurtosis K i is greater than Ko, and C i is greater than Co, then the IMF i (t) component is retained. 7.The method of claim 1, wherein, The neural network model in step 4 includes: An input layer that receives original signal sequence data; A first convolutional block that extracts local basic feature patterns; A second convolutional block that captures abstract high-level features; A time series modeling layer that models long-range temporal dependencies; A feature aggregation layer that compresses time series features into global representations; A classification output layer that outputs the existence probability of regularity.

8. The unmanned aerial vehicle communication frequency band monitoring method according to claim 7, wherein, y represents the label of whether there is regularity, x represents the sample, α represents the balance factor, γ represents the focus parameter, and p represents the model prediction probability. The loss function of the neural network model is : ； Comprising the following steps:

9. A jamming method, characterized by, S1: Use the unmanned aerial vehicle communication frequency band monitoring method in any one of claims 1-7 to generate a risk coefficient of signal frequency band existence communication; S2: Send noise signals to signal frequency bands with high risk coefficients. S2 includes the following steps: 10.The method of claim 1, wherein, S21: Extract the signal length of the signal frequency band U(t) that is higher than the amplitude of the background noise; ​ S22: Randomly generate a signal similar to the amplitude of the background noise, and send it out with the frequency band.

Images