Unmanned aerial vehicle individual countering method and system based on radio frequency identification and dynamic interference

Abstract

The invention discloses an unmanned aerial vehicle individual countering method and system based on radio frequency identification and dynamic interference, and belongs to the technical field of unmanned aerial vehicles, and the method comprises the steps: collecting wide-band IQ data, extracting hidden features such as instantaneous frequency offset fluctuation and envelope entropy abnormity, and achieving high-dimensional embedded expression through a variational auto-encoder. Structured recognition is carried out on signals in combination with a graph neural network, the discrimination capability of frequency hopping, spread spectrum or Wi-Fi / LTE simulation signals is remarkably improved, a dynamic interference strategy is finally generated according to a recognition result, accurate recognition and intelligent suppression of illegal unmanned aerial vehicle communication are achieved, and the method is suitable for popularization and application. The technical defect that an existing system is insufficient in recognition robustness and countering effectiveness in a complex environment is effectively overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and particularly to an individual countermeasure method and system for unmanned aerial vehicles based on radio frequency identification and dynamic interference. Background Art

[0002] Individual countermeasures for unmanned aerial vehicles refer to the process of taking technical measures such as detection, identification, interference, control or forced landing against a single or a small number of unauthorized unmanned aerial vehicles that may pose a security threat, in order to achieve effective management and control or drive them away. This means is commonly used in scenarios such as airports, sensitive areas, and major event venues to ensure airspace safety and public order.

[0003] The prior art has the following deficiencies: In the discrimination of unmanned aerial vehicle communication based on radio frequency identification, malicious unmanned aerial vehicles can use signal camouflage means to make their remote control or video transmission signals mimic legitimate communications (such as Wi-Fi, LTE, etc.) in the frequency spectrum, or use frequency hopping, spread spectrum, etc. to deliberately weaken the feature stability, thereby bypassing the feature library matching and pattern recognition, resulting in the system misjudging or failing to judge them as ordinary devices. Such signal deception is particularly concealed in complex environments, which may cause the countermeasure system to fail and form a security vulnerability. Summary of the Invention

[0004] The purpose of the present invention is to provide an individual countermeasure method and system for unmanned aerial vehicles based on radio frequency identification and dynamic interference to solve the deficiencies in the background art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: An individual countermeasure method for unmanned aerial vehicles based on radio frequency identification and dynamic interference, including: Collecting communication signals in the airspace by using a wideband radio frequency receiving device to obtain complex IQ data streams; Performing band-pass filtering, window framing and synchronization processing on the IQ data streams to extract the target signal segment; Extracting radio frequency high-order micro-features based on the target signal segment, and the micro-features include instantaneous frequency deviation fluctuation features and envelope entropy anomaly features; Inputting the instantaneous frequency deviation fluctuation features and envelope entropy anomaly features into a representation learning model for high-dimensional coding processing, and the representation learning model is a variational autoencoder, and outputting an embedding vector for identification and discrimination; Inputting the embedding vector into a signal classification model to discriminate whether the signal belongs to unmanned aerial vehicle communication, and the signal classification model is selected from graph neural networks; If the recognition result shows that the signal is an illegal unmanned aerial vehicle communication signal, generate an interference strategy according to its frequency, modulation characteristics and timing rules, and perform dynamic interference on the target signal segment.

[0006] Preferably, the acquisition of the airspace communication signal by using the wide-band radio frequency receiving device specifically includes: using a software-defined radio device with a frequency coverage range of 300 MHz to 6 GHz to collect wireless signals in the target airspace, and saving the collected original signals in the form of complex IQ sampling. The IQ signal is expressed as ; where I[t] and Q[t] are the in-phase and quadrature components respectively, and j is the imaginary unit.

[0007] Preferably, the method for obtaining the instantaneous frequency deviation fluctuation feature is: calculating the instantaneous frequency , which represents the derivative of the signal phase with respect to time, and is defined as: ; where is the instantaneous phase of the IQ signal, and the calculation formula is: ; by calculating the phase change amount for consecutive sampling points and dividing by the sampling period T, a frequency offset sequence is obtained: ; in the formula, T is the sampling period; represents the instantaneous frequency value of each sampling point; further calculating the standard deviation of the frequency sequence, the frequency deviation fluctuation intensity value is obtained, and its formula is: ; in the formula, SG is the frequency deviation fluctuation intensity value; where is the average frequency offset, and M is the number of samples.

[0008] Preferably, the method for obtaining the envelope entropy anomaly feature is: let be the modulus value of the IQ signal, that is, the envelope, and the expression is: ; normalizing the signal envelope into a probability distribution form, and then calculating the entropy value based on the definition of Shannon entropy: ; H is the envelope entropy, represents the probability density of the i-th envelope amplitude interval; F is the number of segments. To detect anomalies, the envelope entropy anomaly deviation value is further defined, and the expression is: ; where is the standard envelope entropy of known legal unmanned aerial vehicles, represents the envelope entropy of the current signal.

[0009] Preferably, inputting the instantaneous frequency deviation fluctuation feature and the envelope entropy anomaly feature into a representation learning model for high-dimensional coding processing specifically includes: The micro features include the instantaneous frequency deviation fluctuation feature and the envelope entropy anomaly feature. The frequency deviation fluctuation intensity value and the envelope entropy anomaly deviation value are formed into a two-dimensional input vector and input into a variational autoencoder model for high-dimensional coding processing; The variational autoencoder model includes an encoder network for extracting the latent variable mean vector and standard deviation vector of the input feature vector, and calculating the latent variable embedding vector according to the latent variable mean and standard deviation; And use the embedded vector as the input feature for subsequent UAV signal discrimination.

[0010] Preferably, to discriminate whether the signal belongs to UAV communication, the signal classification model is selected from graph neural networks, specifically including: Use the embedded vector representing the output of the variational autoencoder as the node feature; each target signal segment corresponds to a node, and the embedded vector represents the attribute feature of the node, which is used to construct a graph structure; Construct a graph G=(V,E), where: V represents the set of nodes, and each node corresponds to a radio frequency signal; E represents the set of edges, indicating the correlation between nodes; construct a graph neural network model: each layer aggregates and updates the information of neighboring nodes, and the output of the last layer is the node label probability, indicating whether the target signal is UAV communication; Output a binary classification label for each node: Label 1: indicating that the target signal segment belongs to UAV communication; Label 0: indicating that the target signal segment is non-UAV communication or background interference.

[0011] Preferably, the interference strategy generation step includes: According to the center frequency of the target signal Calculate the interference center frequency, and the expression is: ; represents the center frequency of the interference signal, represents the center frequency of the target signal; Δf represents the frequency offset compensation coefficient; According to the target bandwidth Calculate the interference bandwidth, and the expression is: ; In the formula, represents the interference signal bandwidth, and α represents the safety multiple frequency factor; According to the target modulation method Determine the interference modulation type, and the expression is: ; In the formula, represents the interference modulation method, and Match is the modulation adaptation function; According to the target signal strength Calculate the interference power, and the expression is: ; In the formula, represents the interference transmission power, represents the target signal strength, and β represents the power margin coefficient; Z represents the minimum interference threshold of the device; According to the target communication period and the signal timing Generate the interference time window set, and the expression is: ; In the formula, represents the set of interference application time windows, Indicates the detected target signal active moment; Indicates the target communication cycle.

[0012] The present invention also provides an individual anti - interference system for unmanned aerial vehicles based on radio frequency identification and dynamic interference, including a radio frequency signal acquisition module, a signal pre - processing module, a micro - feature extraction module, a representation learning and encoding module, and a signal recognition and discrimination module; Radio frequency signal acquisition module: Utilize a wide - band radio frequency receiving device to collect communication signals in the airspace and obtain complex IQ data streams; Signal pre - processing module: Perform band - pass filtering, window framing, and synchronization processing on the IQ data streams to extract target signal segments; Micro - feature extraction module: Extract radio frequency high - order micro - features based on the target signal segments, and the micro - features include instantaneous frequency deviation fluctuation features and envelope entropy anomaly features; Representation learning and encoding module: Input the instantaneous frequency deviation fluctuation features and envelope entropy anomaly features into a representation learning model for high - dimensional encoding processing. The representation learning model is a variational auto - encoder, and output embedding vectors for recognition and discrimination; Signal recognition and discrimination module: Input the embedding vectors into a signal classification model to discriminate whether the signal belongs to unmanned aerial vehicle communication. The signal classification model is selected from graph neural networks; Dynamic interference decision - making and execution module: If the recognition result indicates that the signal is an illegal unmanned aerial vehicle communication signal, generate an interference strategy according to its frequency, modulation characteristics, and timing rules, and perform dynamic interference on the target signal segment.

[0013] In the above technical solution, the technical effects and advantages provided by the present invention: 1. The present invention constructs an individual anti - interference method for unmanned aerial vehicles that integrates radio frequency signal high - order feature extraction, deep representation learning, graph - structure discrimination, and parameterized dynamic interference, breaking through the technical bottleneck that traditional radio frequency identification systems rely on static feature libraries and are easily evaded by signal camouflage. By introducing micro - features such as instantaneous frequency deviation fluctuation and envelope entropy anomaly, complex signal behaviors such as frequency hopping, spread spectrum, and camouflage modulation can be effectively characterized. With the help of variational auto - encoders, high - dimensional feature expression can be achieved, and then combined with graph neural networks to model the timing and correlation between signals, significantly improving the recognition robustness and accuracy of illegal unmanned aerial vehicle communication.

[0014] 2. The present invention further generates an interference strategy based on the recognition result, dynamically sets interference parameters according to the frequency, bandwidth, modulation method, power, and cycle structure of the target signal, and implements directional, synchronous, and adaptive interference suppression. This mechanism can achieve precise interference and guided suppression of camouflaged unmanned aerial vehicles, while reducing the risks of mis - interference and power consumption, and has the advantages of high recognition accuracy, high response flexibility, and strong environmental adaptability, and is suitable for actual combat deployment in scenarios such as airports, sensitive areas, and major events. Brief Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0016] Figure 1 It is the method mind map of the present invention.

[0017] Figure 2 It is the system module mind map of the present invention. Detailed Embodiments

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0019] Embodiment 1, please refer to Figure 1 As shown, the method for individual countermeasure of unmanned aerial vehicles based on radio frequency identification and dynamic interference in this embodiment includes: Collect communication signals in the airspace using a wideband radio frequency receiving device to obtain complex IQ data streams; Perform band-pass filtering, window framing, and synchronization processing on the IQ data stream to extract the target signal segment; Extract radio frequency high-order micro-features based on the target signal segment, and the micro-features include instantaneous frequency offset fluctuation features and envelope entropy anomaly features; Input the instantaneous frequency offset fluctuation features and envelope entropy anomaly features into a representation learning model for high-dimensional encoding processing. The representation learning model is a variational autoencoder, and an embedding vector for identification and discrimination is output; Input the embedding vector into a signal classification model to determine whether the signal belongs to unmanned aerial vehicle communication. The signal classification model is selected from a graph neural network; If the recognition result indicates that the signal is an illegal unmanned aerial vehicle communication signal, a jamming strategy is generated according to its frequency, modulation characteristics, and timing law, and dynamic jamming is performed on the target signal segment.

[0020] The radio frequency signal acquisition step of the present invention aims to obtain raw data of all possible unmanned aerial vehicle communication signals in the target airspace, providing a signal basis for subsequent identification and jamming.

[0021] During flight, drones rely on control signals and image transmission signals to communicate with their control terminals. Common frequency bands include, but are not limited to, 2.4 GHz, 5.8 GHz, 900 MHz, and some Wi-Fi, Bluetooth, LTE, or 5G frequency bands. To ensure wide coverage and adaptability of the system, this step uses a wide-band RF receiving device to synchronously or continuously scan and monitor multiple frequency bands, obtain wireless communication data streams, and save the original signals in the form of complex IQ (In-phase and Quadrature) sampling.

[0022] The wide-band RF receiving device may include one or a combination of the following components: a software-defined radio (SDR) with an adjustable frequency range covering 300 MHz to 6 GHz, such as USRP, HackRF, BladeRF, AirSpy, etc.; a front-end low-noise amplifier (LNA) to improve receiving sensitivity; a multi-band antenna or a controllable directional antenna array to enhance the signal receiving ability in a specific direction; a clock synchronization module (GPSDO or local frequency stabilizer) to ensure sampling consistency; and a high-speed data interface (USB 3.0, PCIe, Gigabit Ethernet, etc.) connected to the processing terminal to transmit the IQ data stream.

[0023] The device is set to cover the common communication frequency bands of drones (such as 2.4 GHz ± 100 MHz, 5.8 GHz ± 150 MHz, etc.), and can dynamically adjust the monitoring center frequency and bandwidth according to the scenario.

[0024] Sampling parameter settings: Sampling rate: The preferred range is 10 MHz to 40 MHz, determined according to the signal bandwidth and device performance; Resolution bits: Commonly 8-bit or 12-bit complex sampling, corresponding to the I / Q two channels respectively; Data structure: Each frame of data is a complex vector IQ of length N, where . I(t) represents the in-phase component, that is, the part with the same frequency and phase as the reference signal (usually the carrier signal). Usually it represents the real part of the signal. Q(t) represents the quadrature component, that is, the part with a 90-degree phase deviation from the I component. Usually it represents the imaginary part of the signal. j is the imaginary unit, used to represent the phase difference between the quadrature component Q(t) and the in-phase component I(t). The collected IQ data is stored in binary or HDF5 format, and the accompanying metadata includes: timestamp (Unix time), center frequency, bandwidth, sampling rate, device number; current receiving antenna direction (if applicable).

[0025] The complex IQ data stream may include, but is not limited to, the following types of signals: The frequency-hopping control signal (such as FHSS) transmitted by the UAV remote controller; the continuous video stream (such as the OFDM-encoded video signal) transmitted by the UAV video transmission module; the control and image data transmitted using Wi-Fi, Bluetooth, or LTE protocols; other non-UAV background wireless signals (as control samples or environmental noise).

[0026] This step supports parallel deployment of multiple devices to improve spatial resolution and response speed; it can combine array antenna technology to achieve signal direction positioning; if it is necessary to cover the 5G UAV communication channel, the frequency band can be extended to the Sub-6GHz and mmWave (millimeter wave) ranges.

[0027] Through this step, high-quality UAV communication data can be stably obtained in a real and complex electromagnetic environment, which is especially suitable for unbiased raw capture of unknown or camouflaged communication signals, providing a basic guarantee for subsequent feature extraction and intelligent recognition, and having strong environmental adaptability and system scalability.

[0028] The present invention performs necessary signal preprocessing on the collected complex IQ data stream, including band-pass filtering, windowing and framing, and signal synchronization processing to ensure the signal quality, structural consistency, and processing efficiency of the input data.

[0029] Band-pass filtering is used to remove background interference signals irrelevant to the target frequency band, improve the signal-to-noise ratio, and isolate the effective energy region of the frequency band where UAV communication is located.

[0030] According to the center frequency configured in the signal acquisition stage and bandwidth B, a band-pass filter is constructed: The filter can adopt one of the following forms: Digital FIR filter (finite impulse response); IIR filter (such as Butterworth, Chebyshev); In the implementation process, the signal can be converted to the frequency domain based on FFT (Fast Fourier Transform), then frequency-domain filtering is performed and inverse-transformed back to the time domain; The filter design parameters are determined according to the sampling rate and the target signal characteristics. For example, the center frequency of the video transmission signal is set to 5.8 GHz and the bandwidth is 20 MHz.

[0031] Windowing and Framing is used to divide the continuous IQ signal sequence into equal-length signal frames for batch processing and feature analysis, and to enhance the temporal sensitivity.

[0032] Set the window length L (e.g., 1024 points, 2048 points), and set the overlap ratio r (e.g., 50%) to prevent loss of edge information; use a window function to weight each frame to reduce spectral leakage: common window functions include: Hamming window, Hann window, Blackman window, etc.; the windowed signal is: ; where, x[n] is the original signal sample, a real or complex sequence (such as IQ data). This is one frame of the signal, containing L consecutive samples, expressed as the sampled values in the time domain. Usually from one frame segmented from IQ data. w[n] represents the window function coefficients, with the domain [0, L−1]. The window function is a mathematical function for weighting processing. Common windows include Hamming window, Hann window, Blackman window, etc. The main role is to weaken the influence of the frame edge and suppress spectral leakage. xw[n] represents the windowed signal, which is the result after each sample is multiplied by the corresponding position of the window function, and is the input for subsequent feature extraction or spectral analysis. n represents the discrete time index, ranging from 0 to L−1, indicating the position index of the signal sample currently being processed.

[0033] The frame sequence is obtained by sliding window processing: ; where, ; in the formula, is the frame shift.

[0034] Synchronization processing means unifying the reference starting point of the signal, ensuring the phase, frequency, and structure alignment between different frames, and eliminating the errors caused by sampling offset or carrier drift.

[0035] Starting point synchronization: Use the energy threshold method to detect the starting position of the signal and judge whether the frame contains a valid signal segment: ; if > θ (preset energy threshold), then determine that this frame is a signal frame.

[0036] Estimate the carrier frequency offset Δf and phase offset Δϕ of adjacent frames, and perform digital down-conversion on the IQ data: ; in the formula, represents the complex exponential rotation factor, represents the corrected IQ signal.

[0037] Use the maximum likelihood frequency offset estimator (ML Estimator) or the pilot-based frequency offset estimation method to finely adjust the spectral center; the degree of frequency offset can be judged by periodic pilot symbol matching and dynamically corrected.

[0038] Based on the processed frame sequence above, extract the frame set that may be for drone communication, as the input for the subsequent feature extraction and recognition steps.

[0039] Simple rules can be used for initial screening (such as continuous energy above the threshold for more than 3 consecutive frames); or primary classifiers (such as lightweight CNN, clustering algorithm) can be introduced to make preliminary classification of signal segments; after extraction, the target signal segment data structure is generated: , each segment is of length L; each segment is accompanied by meta-information: center frequency, timestamp, preliminary confidence score, etc.

[0040] Through the above preprocessing process, background noise can be effectively filtered, signal morphology can be regularized, and key communication events can be aligned to ensure the data quality and discrimination efficiency of subsequent feature extraction and recognition tasks, especially in the face of complex multi-source interference scenarios. Stability and accuracy.

[0041] After completing the extraction of the target signal segment, the present invention needs to perform RF high-order micro-feature analysis on the segment to explore the fine-grained properties of the signal that cannot be identified by traditional frequency domain features, and enhance the robustness of recognition of camouflaged, encrypted or frequency-hopping drone signals.

[0042] The instantaneous frequency deviation fluctuation feature is used to characterize the slight change law of the signal frequency fluctuation on the time axis. This feature can reveal the communication behavior characteristics generated by the frequency hopping control signal or the unstable carrier source.

[0043] Instantaneous frequency represents the derivative of the signal phase with respect to time and is defined as: ;in, is the instantaneous phase of the IQ signal, and the calculation formula is: ; By calculating the phase change of continuous sampling points , and divided by the sampling period T, the frequency offset sequence can be obtained: ;Wherein, T is the sampling period; Represents the instantaneous frequency value of each sampling point; further calculate the standard deviation of the frequency sequence to obtain the frequency deviation fluctuation intensity value, and the formula is: ; In the formula, SG is the frequency deviation fluctuation intensity value; Among them, is the average frequency deviation, and M is the number of samples.

[0044] High-frequency deviation fluctuation intensity value: commonly seen in frequency-hopping remote controls or low-quality transmission sources; low-frequency deviation fluctuation intensity value: corresponds to industrial-grade image transmission modules or static relay signals; can be used as a key indicator to distinguish camouflage signals from real drone signals.

[0045] The envelope entropy anomaly feature is used to evaluate the signal modulation depth and its randomness characteristics, capture modulation anomalies, mask modulation patterns or structural unnatural signals.

[0046] set up is the IQ signal modulus, i.e., envelope, and is expressed as: Normalize the signal envelope into a probability distribution form (by constructing a histogram or kernel density estimation), and then calculate the entropy value based on the definition of Shannon entropy: ; H is the envelope entropy, represents the probability density of the i-th envelope amplitude interval; F is the number of segments. If the signal has high certainty (such as fixed modulation image transmission), its entropy value is low; if the signal shows strong modulation jitter and irregular envelope changes, its entropy value increases. To detect anomalies, the envelope entropy anomaly deviation value is further defined , and the expression is: ; where is the standard envelope entropy of known legal drones, represents the envelope entropy of the current signal. A larger may indicate that the signal is artificially scrambled, mixed modulated or simulated and forged; it is applicable to identify the mixed modulation drone control signal forged by SDR; combined with the high-frequency deviation fluctuation intensity value, it can improve the classification accuracy of complex signal sources.

[0047] Input the instantaneous frequency deviation fluctuation feature and the envelope entropy anomaly feature into a representation learning model for high-dimensional coding processing. The representation learning model is a variational autoencoder, and an embedding vector for identification and discrimination is output, specifically including: The micro-features include the instantaneous frequency deviation fluctuation feature and the envelope entropy anomaly feature, and include the following representation learning coding steps: Construct a two-dimensional input vector from the frequency deviation fluctuation intensity value and the envelope entropy anomaly deviation value, and input it into the variational autoencoder model for high-dimensional coding processing, The variational autoencoder model includes an encoder network for extracting the latent variable mean vector and standard deviation vector of the input feature vector; a reparameterization sampling module for calculating the latent variable embedding vector according to the latent variable mean and standard deviation; And use the embedding vector as the input feature for subsequent drone signal discrimination, which is used to improve the classification and recognition accuracy of camouflaged, frequency-hopping or abnormal RF communication signals.

[0048] In the present invention, the embedding vector is input into a signal classification model to determine whether the signal belongs to drone communication. The signal classification model is selected from a graph neural network, specifically including: Use the embedding vector representing the output of the variational autoencoder as the node feature; each target signal segment corresponds to a node, and the embedding vector represents the attribute feature of the node, which is used to construct a graph structure.

[0049] Construct a graph \(G=(V, E)\), where: \(V\) represents the set of nodes, and each node corresponds to a segment of RF signal; \(E\) represents the set of edges, indicating the correlation between nodes (such as temporal proximity, frequency proximity, feature similarity); Optional ways to construct edges: Temporal edge: Connect signal segments that are consecutive in time; Similarity edge: Based on metrics such as cosine similarity and Euclidean distance to measure proximity in the feature space; Construct a graph neural network model (such as GCN, GraphSAGE, GAT): Each layer aggregates and updates the information of neighboring nodes, and the expression is: ; where: represents the feature of the \(i\)-th node at the \(l\)-th layer; represents the set of neighbors of the \(i\)-th node; represents the learnable weight matrix; \(\sigma\) represents the activation function (such as ReLU); The output of the last layer is the node label probability, indicating whether the signal is a UAV communication.

[0050] Output a binary classification label for each node (i.e., each segment of embedding vector): Label 1: Indicates that the signal segment belongs to UAV communication (control signal or video transmission); Label 0: Indicates that the signal segment is non-UAV communication or background interference; At the same time, output the classification confidence (softmax or sigmoid) for reference by the interference decision module.

[0051] If the recognition result indicates that the signal is an illegal UAV communication signal, then generate an interference strategy according to its frequency, modulation characteristics and temporal law, and perform dynamic interference on the target signal segment, specifically including: Interference frequency selection (frequency alignment) includes: ; represents the center frequency of the interference signal, represents the center frequency of the target signal (provided by the recognition module) \(\Delta f\) represents the frequency offset compensation coefficient for dynamic frequency tracking, and the initial value can be 0 and can be adjusted to cope with frequency hopping behavior.

[0052] Interference bandwidth matching includes: ; In the formula, represents the bandwidth of the interference signal, represents the effective bandwidth of the target communication signal, and \(\alpha\) represents the safety frequency multiplication factor (the recommended value range is 1.1~1.5), which is used to ensure coverage of the target frequency spectrum range and can effectively suppress even in the case of frequency offset or frequency hopping.

[0053] Interference modulation mode adaptation includes: ; In the formula, represents the interference modulation mode (such as CW, Noise, Chirp, OFDM, Sweep, etc.), Indicates the modulation method of the target signal (such as QPSK, FSK, DSSS, etc.); Match is the modulation adaptation function, and the rule can be set as: If is frequency hopping (FHSS), then broadband noise is adopted; if is fixed modulation video transmission (such as OFDM), then frequency point suppression or anti-modulation deception is adopted.

[0054] Dynamic power control (based on signal strength) includes: ; where represents the interference transmission power, represents the target signal strength (which can be obtained through RSSI or energy estimation); β represents the power margin coefficient (recommended 1.5 - 3.0); Z represents the minimum interference threshold of the device (considering the environmental noise floor).

[0055] Interference timing control (synchronous time window) includes: ; where represents the set of time windows for applying interference, represents the detected target signal activity moment; represents the target communication cycle (the time interval of the target video transmission or control signal can be estimated); the interference is executed at the key alignment points of the target communication cycle to improve the interference efficiency and reduce the energy consumption.

[0056] Through the dynamic interference strategy generation method, precise suppression or deception of complex UAV signals such as frequency hopping, time division, and multi-modulation can be achieved; at the same time, the power output and false kill risk are reduced on the basis of ensuring the interference effectiveness; it is beneficial to construct a composite interference system architecture driven by rules + optimized by learning.

[0057] Example 2, please refer to Figure 2 As shown, the UAV individual countermeasure system based on radio frequency identification and dynamic interference described in this embodiment includes a radio frequency signal acquisition module, a signal preprocessing module, a micro-feature extraction module, a representation learning coding module, and a signal recognition and discrimination module; Radio frequency signal acquisition module: Use a wideband radio frequency receiving device to collect communication signals in the airspace and obtain complex IQ data streams; Signal preprocessing module: Perform band-pass filtering, window framing, and synchronization processing on the IQ data stream to extract the target signal segment; Micro-feature extraction module: Extract radio frequency high-order micro-features based on the target signal segment, and the micro-features include instantaneous frequency deviation fluctuation features and envelope entropy anomaly features; Representation learning coding module: Input the instantaneous frequency deviation fluctuation features and envelope entropy anomaly features into a representation learning model for high-dimensional coding processing, and the representation learning model is a variational autoencoder, and output an embedding vector for recognition and discrimination; Signal recognition and discrimination module: Input the embedded vector into a signal classification model to determine whether the signal belongs to UAV communication. The signal classification model is selected from graph neural networks; Dynamic interference decision-making and execution module: If the recognition result indicates that the signal is an illegal UAV communication signal, generate an interference strategy based on its frequency, modulation characteristics, and timing law, and perform dynamic interference on the target signal segment.

[0058] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0059] It should be understood that the term "and / or" in this article is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship. Specifically, it can be understood by referring to the context before and after.

[0060] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0061] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application.

Claims

1. A method for individual countermeasure of unmanned aerial vehicles based on radio frequency identification and dynamic interference, characterized in that: Including: Collecting communication signals in the airspace using a wide - band radio frequency receiving device to obtain complex IQ data streams; Performing band - pass filtering, window framing, and synchronization processing on the IQ data streams to extract target signal segments; Extracting radio frequency high - order micro - features based on the target signal segments, where the micro - features include instantaneous frequency deviation fluctuation features and envelope entropy anomaly features; Inputting the instantaneous frequency deviation fluctuation features and envelope entropy anomaly features into a representation learning model for high - dimensional encoding processing, where the representation learning model is a variational auto - encoder, and outputting an embedding vector for identification and discrimination; Inputting the embedding vector into a signal classification model to discriminate whether the signal belongs to unmanned aerial vehicle (UAV) communication, and the signal classification model is selected from graph neural networks; If the recognition result indicates that the signal is an illegal UAV communication signal, then generate an interference strategy according to its frequency, modulation characteristics, and timing rules, and perform dynamic interference on the target signal segment.

2. The individual countermeasure method for unmanned aerial vehicles based on radio frequency identification and dynamic interference according to claim 1, wherein: The specific process of collecting airspace communication signals using a wide-band RF receiving device includes: using a software-defined radio device with a frequency coverage range of 300 MHz to 6 GHz to collect wireless signals in the target airspace, and saving the collected original signals in the form of complex IQ samples. The IQ signal is expressed as ; where I[t] and Q[t] are the in-phase and quadrature components respectively, and j is the imaginary unit.

3. The method for individual countermeasure of drones based on radio frequency identification and dynamic interference according to claim 1, characterized in that: The method for obtaining the instantaneous frequency deviation fluctuation characteristics is as follows: Calculate the instantaneous frequency , which represents the derivative of the signal phase with respect to time and is defined as: ; where is the instantaneous phase of the IQ signal, and the calculation formula is: ; By calculating the phase change amount for consecutive sampling points and dividing by the sampling period T, a frequency offset sequence is obtained: ; In the formula, T is the sampling period; represents the instantaneous frequency value of each sampling point; Further calculate the standard deviation of the frequency sequence to obtain the frequency deviation fluctuation intensity value, and its formula is: ; In the formula, SG is the frequency deviation fluctuation intensity value; where is the average frequency offset, and M is the number of samples.

4. The individual countermeasure method for unmanned aerial vehicles based on radio frequency identification and dynamic interference according to claim 3, wherein: The method for obtaining the abnormal characteristics of envelope entropy is as follows: Let be the modulus value of the IQ signal, that is, the envelope, and the expression is: ; Normalize the signal envelope into a probability distribution form, and then calculate the entropy value based on the definition of Shannon entropy: ; $H$ is the envelope entropy, which represents the probability density of the $i$-th envelope amplitude interval; Let \(F\) be the number of segments. To detect anomalies, the envelope entropy anomaly deviation value is further defined. , and the expression is: ; where is the standard envelope entropy of known legal drones, represents the envelope entropy of the current signal.

5. The method for individual countermeasure of unmanned aerial vehicle based on radio frequency identification and dynamic interference according to claim 4, characterized in that: Inputting the instantaneous frequency deviation fluctuation features and envelope entropy anomaly features into a representation learning model for high - dimensional encoding processing, specifically including: The micro - features include instantaneous frequency deviation fluctuation features and envelope entropy anomaly features. Construct a two - dimensional input vector from the frequency deviation fluctuation intensity value and the envelope entropy anomaly deviation value, and input it into the variational auto - encoder model for high - dimensional encoding processing; The variational auto - encoder model includes an encoder network for extracting the latent variable mean vector and standard deviation vector of the input feature vector, and calculating the latent variable embedding vector according to the latent variable mean and standard deviation; And using the embedding vector as the input feature for subsequent UAV signal discrimination.

6. The method for individual countermeasure of unmanned aerial vehicle based on radio frequency identification and dynamic interference according to claim 5, characterized in that: Discriminating whether the signal belongs to UAV communication, and the signal classification model is selected from graph neural networks, specifically including: Using the embedding vector output by the variational auto - encoder as the node feature; each target signal segment corresponds to a node, and the embedding vector represents the attribute feature of the node, which is used to construct a graph structure; Construct a graph G=(V, E), where: V represents the set of nodes, and each node corresponds to a radio frequency signal; E represents the set of edges, indicating the correlation between nodes; construct a graph neural network model: each layer aggregates and updates the information of neighboring nodes, and the output of the last layer is the node label probability, indicating whether the target signal is UAV communication; Output a binary classification label for each node: Label 1: indicating that the target signal segment belongs to UAV communication; Label 0: indicating that the target signal segment is non - UAV communication or background interference.

7. The method for individual countermeasure of an unmanned aerial vehicle based on radio frequency identification and dynamic interference according to claim 6, characterized in that: The interference strategy generation step includes: According to the center frequency of the target signal calculate the center frequency of the interference, and the expression is: ; represents the center frequency of the interference signal, represents the center frequency of the target signal; Δf represents the frequency offset compensation coefficient; According to the target bandwidth Calculate the interference bandwidth, and the expression is: ; In the formula, represents the interference signal bandwidth, and α represents the safety frequency multiplication factor; According to the target modulation method Determine the interference modulation type, and the expression is: ; In the formula, represents the interference modulation method, and Match is the modulation adaptation function; According to the target signal strength Calculate the interference power, and the expression is as follows: ; In the formula, represents the interference transmission power, represents the target signal strength, β represents the power margin coefficient; Z represents the minimum interference threshold of the device; According to the target communication period and signal timing generate a set of interference time windows, and the expression is: ; In the formula, represents the set of time windows for interference application, represents the detected target signal activity moments; represents the target communication period.

8. An individual UAV countermeasure system based on RFID and dynamic interference, which is used to implement the individual UAV countermeasure method based on RFID and dynamic interference according to any one of claims 1-7, characterized in that: Including a radio frequency signal acquisition module, a signal pre - processing module, a micro - feature extraction module, a representation learning encoding module, and a signal recognition and discrimination module; Radio frequency signal acquisition module: Collecting communication signals in the airspace using a wide - band radio frequency receiving device to obtain complex IQ data streams; Signal pre - processing module: Performing band - pass filtering, window framing, and synchronization processing on the IQ data streams to extract target signal segments; Micro - feature extraction module: Extracting radio frequency high - order micro - features based on the target signal segments, where the micro - features include instantaneous frequency deviation fluctuation features and envelope entropy anomaly features; Representation learning encoding module: Input the instantaneous frequency deviation fluctuation feature and the envelope entropy anomaly feature into a representation learning model for high-dimensional encoding processing. The representation learning model is a variational autoencoder, and an embedding vector for identification and discrimination is output; Signal identification and discrimination module: Input the embedding vector into a signal classification model to discriminate whether the signal belongs to UAV communication. The signal classification model is selected from graph neural networks; Dynamic interference decision and execution module: If the recognition result indicates that the signal is an illegal UAV communication signal, generate an interference strategy according to its frequency, modulation feature and timing law, and perform dynamic interference on the target signal segment.

Images