An adaptive jamming method based on target characteristics

Abstract

An adaptive interference method based on target characteristics, according to the real-time and changeable working mode characteristics of the cognitive system, establishes a dynamic knowledge base based on individual target characteristics, and establishes a feasible interference effect evaluation method based on individual characteristics, and combines the radiation source characteristics of the target to form a targeted interference strategy and generate a corresponding interference signal, thereby forming an adaptive interference closed loop from reconnaissance to interference and then to evaluation, and continuously optimizes the interference strategy according to the interference effect evaluation, thereby forming an efficient interference suppression method; at the same time, based on the radiation source characteristics of the target individual, accurate tracking interference can be accurately implemented to effectively cope with the changeable working mode of the cognitive system, with the characteristics of strong operability and strong target targeting.

Description

Technical field:

[0001] The invention belongs to the technical field of radio communication interference, and mainly relates to an adaptive interference method based on target characteristics. Background technology:

[0002] Adaptive jamming method refers to the jamming suppression method for systems with cognitive radio functions (hereinafter referred to as cognitive systems). It detects the electromagnetic environment, understands and learns the characteristics of the electromagnetic environment, and adjusts the jamming configuration in real time to achieve effective jamming of the cognitive system.

[0003] There are three main adaptive jamming methods commonly used at home and abroad:

[0004] The first is an interference method based on a reinforcement learning decision model. By building a reinforcement learning decision model that integrates the effective variance confidence upper bound, the interference parameters are determined and the optimal interference action is output.

[0005] The second is a cognitive jamming method based on Markov process decision-making, which collects radio frequency energy from wireless communication signals including wireless communication stations, mobile phones and dedicated data links to assess wireless communication threats and then jam the threatening energy level;

[0006] The third is a jamming method based on signal features. It extracts signal features and matches them with signals in the signal library, assesses their threat levels, and sorts the threat levels to select the most threatening signal for jamming.

[0007] The shortcomings of the above technologies can be summarized as the lack of recognition of individual characteristics or radiation sources of the cognitive system and the lack of effective interference effect evaluation methods based on such characteristics. There is no targeted learning of the working characteristics and target characteristics of the cognitive system itself, so it is difficult to implement accurate and real-time effective interference. Summary of the invention:

[0008] In order to overcome the above-mentioned shortcomings, the present invention provides an adaptive interference method based on target characteristics.

[0009] The technical solution adopted by the present invention to solve its technical problem is:

[0010] An adaptive jamming method based on target features comprises the following steps:

[0011] S1. Establish a signal library, an evaluation criteria library and an interference strategy library;

[0012] S2. Start the reconnaissance mode, receive the target signal that needs to be interfered, and analyze the target characteristics according to the received target signal. The target characteristic analysis includes the following steps:

[0013] S21. Extracting signal features by detecting the target signal;

[0014] S22. Signal pattern recognition

[0015] The signal features extracted in step S21 are identified by modulation pattern using GA-ELM genetic algorithm and extreme learning machine, and a steady-state signal is output;

[0016] S23. Target individual feature recognition

[0017] The steady-state signal output in step S22 is identified using a classification and recognition method based on the fingerprint characteristics of the radiation source signal to obtain the target individual characteristics;

[0018] S3. If the target signal is an existing signal in the signal library, search in the interference strategy library, match the corresponding interference strategy, and then generate an interference waveform;

[0019] If the target signal is a newly discovered signal, first generate an interference signal based on a similarity matching method. At the same time, through case reuse and case modification, combined with the interference effect, the evaluation criteria library is used for online evaluation to generate the optimal interference strategy, directly generate interference waveforms for unknown radiation sources, and update the signal library through machine learning;

[0020] S4. Transmit the interference waveform described in step S3 to start interference.

[0021] The specific steps of extracting signal features by detecting the target signal in step S21 are:

[0022] Perform data preprocessing on the collected signals;

[0023] For each frame of the acquired signal, firstly, a coarse frequency sweep is performed based on the energy detection technology to obtain the key frequency;

[0024] Then, based on the cyclostationary detection technology, the key frequency signal is precisely scanned to complete the preliminary analysis of the signal;

[0025] Finally, the signal is sent to the modulation mode identification module for identification.

[0026] The data preprocessing of the collected signals comprises the following steps:

[0027] a) remove the front end and the tail end of the received signal sequence, and then take the middle segment of the received signal sequence;

[0028] b) remove the points close to zero in the signal sequence;

[0029] c) Normalize the data to the range of 0 to 1.

[0030] The step S22 of signal pattern recognition comprises the following steps:

[0031] A total of five features including information entropy features and cyclic spectrum features are extracted to form feature vectors, which are used as the input of the neural network. After the trained network, the feature set extracted from the signal is divided into a labeled training set and an unlabeled test set, which will be used to train the GA-ELM network and test the recognition performance respectively;

[0032] After designing the GA-ELM network structure, it is trained using labeled training data to obtain a trained GA-ELM network;

[0033] Then the unlabeled data to be identified is input into the trained network to obtain the identification result, and finally the modulation identification of the communication signal is completed.

[0034] The information entropy features include singular spectrum Shannon entropy, singular spectrum exponential entropy, power spectrum Shannon entropy and power spectrum exponential entropy.

[0035] The step S23, identifying target individual features, comprises the following steps:

[0036] First, the steady-state signal is acquired;

[0037] Then, the signal fingerprint features are extracted from three aspects: nonlinearity, non-stationarity, and non-Gaussianity;

[0038] Finally, the fingerprint feature set of the radiation source signal is comprehensively extracted and the ITD is decomposed by time scale;

[0039] Time scale decomposition (ITD) decomposes the received signal into a series of rotating components with gradually decreasing instantaneous frequencies, and then extracts the combined features of the original signal and each rotating component, characterizing the signal fingerprint through multiple levels, dimensions and angles.

[0040] The signal fingerprint features include fractal features, statistical features and energy distribution features.

[0041] Due to the adoption of the above-mentioned technical solution, the present invention has the following advantages:

[0042] The present invention provides an adaptive interference method based on target characteristics. According to the real-time and changeable working mode characteristics of the cognitive system, a dynamic knowledge base based on individual target characteristics is established, and a feasible interference effect evaluation method is established based on the individual characteristics. In combination with the radiation source characteristics of the target, a targeted interference strategy is formed, and a corresponding interference signal is generated, thereby forming an adaptive interference closed loop from reconnaissance to interference and then to evaluation. According to the interference effect evaluation, the interference strategy is continuously optimized to form an efficient interference suppression method. At the same time, based on the radiation source characteristics of the target individual, accurate tracking interference can be accurately implemented to effectively cope with the changeable working mode of the cognitive system. The method has the characteristics of strong operability and strong target targeting. Description of the drawings:

[0043] Figure 1 It is a schematic diagram of the process of the present invention;

[0044] Figure 2 Signal detection flow chart in step S21 of the present invention;

[0045] Figure 3 The signal pattern recognition flow chart of step S22 of the present invention;

[0046] Figure 4 Flow chart of target individual feature recognition in step S23 of the present invention;

[0047] Figure 5 A schematic diagram of an embodiment of ITD for signal decomposition; Specific implementation method:

[0048] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0049] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] The method for adaptive interference based on target features described in conjunction with the accompanying drawings comprises the following steps:

[0051] S1. Establish a signal library, an evaluation criteria library and an interference strategy library;

[0052] S2. Start the reconnaissance mode, receive the target signal that needs to be interfered, and analyze the target characteristics according to the received target signal. The target characteristic analysis includes the following steps:

[0053] S21. Extracting signal features by detecting the target signal;

[0054] S22. Signal pattern recognition

[0055] The signal features extracted in step S21 are identified by modulation pattern using GA-ELM genetic algorithm and extreme learning machine, and a steady-state signal is output;

[0056] S23. Target individual feature recognition

[0057] The steady-state signal output in step S22 is identified using a classification and recognition method based on the fingerprint characteristics of the radiation source signal to obtain the target individual characteristics;

[0058] S3. If the target signal is already in the signal library, stop reconnaissance, turn on the interference mode, search in the interference strategy library, match the corresponding interference strategy, and then generate an interference waveform;

[0059] If the target signal is a newly discovered signal, first generate an interference signal based on a similarity matching method. At the same time, through case reuse and case modification, combined with the interference effect, the evaluation criteria library is used for online evaluation to generate the optimal interference strategy, directly generate interference waveforms for unknown radiation sources, and update the signal library through machine learning;

[0060] S4. Transmit the interference waveform described in step S3 to start interference.

[0061] The adaptive jamming method proposed in this invention can directly implement tracking jamming for the individual characteristics of the cognitive system, and has the characteristics of strong operability and strong target targeting. This method can provide an effective and feasible implementation strategy for cognitive electronic warfare, provide construction ideas for the construction of cognitive electronic warfare systems, and provide a reference plan for the development of electronic countermeasures towards self-adaptation and intelligence.

[0062] Specifically, Figure 2 As shown, the specific steps of extracting signal features by detecting the target signal in step S21 are:

[0063] Perform data preprocessing on the collected signals;

[0064] For each frame of the acquired signal, firstly, a coarse frequency sweep is performed based on the energy detection technology to obtain the key frequency;

[0065] Then, based on the cyclostationary detection technology, the key frequency signal is precisely scanned to complete the preliminary analysis of the signal;

[0066] Finally, the signal is sent to the modulation mode identification module for identification.

[0067] Since the actual data may have problems such as unstable signals caused by power on and off and inconsistent signal amplitudes, the data must be preprocessed first. The preprocessing process can be divided into three steps:

[0068] a) Since the transmitter may be turned on and off during the transmission of the signal, the signals in these two processes are unstable. Therefore, the front end and the tail end of the received signal sequence are removed, and then the middle segment of the received signal sequence is taken.

[0069] b) During the actual signal reception process, the transmitter may not transmit a signal for a certain period of time, so the points close to zero in the signal sequence are removed.

[0070] c) Normalize the data to the range of 0 to 1.

[0071] Specifically, Figure 3 As shown, the step S22 signal pattern recognition includes the following steps:

[0072] A total of five features including information entropy features and cyclic spectrum features are extracted to form feature vectors, which are used as the input of the neural network. After the trained network, the feature set extracted from the signal is divided into a labeled training set and an unlabeled test set, which will be used to train the GA-ELM network and test the recognition performance respectively;

[0073] After designing the GA-ELM network structure, it is trained using labeled training data to obtain a trained GA-ELM network;

[0074] Then the unlabeled data to be identified is input into the trained network to obtain the identification result, and finally the modulation identification of the communication signal is completed.

[0075] The information entropy features include singular spectrum Shannon entropy, singular spectrum exponential entropy, power spectrum Shannon entropy and power spectrum exponential entropy.

[0076] In information theory, information entropy is used to measure the amount of information. The greater the uncertainty of the information, the greater the entropy value. Common types of entropy are Shannon entropy and exponential entropy. Shannon entropy is defined as:

[0077]

[0078] The exponential entropy is defined as:

[0079]

[0080] In terms of feature extraction, since information entropy has the characteristic of characterizing overall characteristics, information entropy can be combined with various signal processing methods to extract information entropy features from the time domain and frequency domain of the signal respectively.

[0081] Singular spectrum analysis is a time series dimension estimation method that combines phase space reconstruction and singular value decomposition. Let a discrete signal sequence be

[0082] X=[x1,x2…,x N ] (3)

[0083] Assuming the embedding dimension and delay time are m and n respectively, the reconstructed phase space matrix is:

[0084]

[0085] Perform singular value decomposition on the above formula:

[0086]

[0087] U and V are orthogonal matrices, and ∑ is a K×J matrix

[0088]

[0089] The values ​​outside the diagonal elements are all zero, and the non-zero elements on the diagonal constitute the singular value spectrum:

[0090] σ={σ1,σ2,…σ i ,…,σ j |j≤K} (7)

[0091] Define the non-zero singular values ​​σ i The proportion of the sum of all non-zero singular values ​​is the probability P i , we get:

[0092]

[0093] Substituting (1) into (2) we can obtain the singular spectrum Shannon entropy and the singular spectrum exponential entropy.

[0094] The power spectrum represents the signal power per unit frequency band, reflecting the relationship between signal power and frequency. Performing Fourier transform on the discrete time series yields:

[0095]

[0096] Where k is an integer and k = 0, 1, ..., N-1. The power spectrum value is further calculated:

[0097]

[0098] All power spectrum values ​​constitute the power spectrum sequence S,

[0099] S={S(k),k=0,1,…,N-1} (11)

[0100] Denoted as:

[0101]

[0102] Substituting (12) into (1) and (2) respectively, we can obtain the power spectrum Shannon entropy and power spectrum exponential entropy.

[0103] Specifically, Figure 4 As shown, the step S23 target individual feature recognition includes the following steps:

[0104] First, the steady-state signal is acquired;

[0105] Then, the signal fingerprint features are extracted from three aspects: nonlinearity, non-stationarity, and non-Gaussianity;

[0106] Finally, the fingerprint feature set of the radiation source signal is comprehensively extracted and the ITD is decomposed by time scale;

[0107] Time scale decomposition (ITD) decomposes the received signal into a series of rotating components with gradually decreasing instantaneous frequencies, and then extracts the combined features of the original signal and each rotating component, characterizing the signal fingerprint through multiple levels, dimensions and angles.

[0108] The signal fingerprint features include fractal features, statistical features and energy distribution features.

[0109] The time-scale decomposition model decomposes the signal into a series of effective intrinsic rotational components, which well define the instantaneous frequency, instantaneous amplitude and instantaneous phase. Additional features and morphological characteristics can be obtained by using the instantaneous amplitude and frequency / phase information obtained and through single wave analysis. Feature extraction and feature-based filtering can be performed in real time, and the feature signals of interest are extracted at the time scale where they occur, preserving their morphology and related phases.

[0110] like Figure 5 The figure shows the waveform of the rotation component and monotonic trend of a radio station signal after ITD decomposition. It can be seen from the figure that the signal is decomposed into a series of rotation components and residual trend components by ITD, and the decomposition sequentially obtains rotation components with high to low frequencies, and the frequency components contained in each rotation component change with the signal itself, which contains the deep information of the signal. By extracting the features of the rotation components, it is helpful to grasp the subtle scale transformation information of the signal.

[0111] According to the real-time and changeable working mode characteristics of the cognitive system, the present invention establishes a dynamic knowledge base based on individual target characteristics, establishes a feasible interference effect evaluation method based on individual characteristics, and forms a targeted interference strategy in combination with the target characteristics to generate corresponding interference signals, thereby forming an adaptive interference closed loop from reconnaissance to interference and then to evaluation, and continuously optimizes the interference strategy according to the interference effect evaluation, thereby forming an efficient interference suppression method; at the same time, based on the individual characteristics of the target, accurate tracking interference can be accurately implemented to effectively cope with the changeable working mode of the cognitive system, and has the characteristics of strong operability and strong target targeting.

[0112] The parts not described in detail in the above content are prior art, so they are not described in detail.

Claims

1. An adaptive jamming method based on target features, characterized in that: The steps include: S1. Establish a signal library, an evaluation criteria library and an interference strategy library; S2. Start the reconnaissance mode, receive the target signal to be interfered, and analyze the target characteristics according to the received target signal. The target characteristic analysis includes the following steps: S21. Extracting signal features by detecting the target signal; S22. Signal pattern recognition The signal features extracted in step S21 are identified by modulation pattern using GA-ELM genetic algorithm and extreme learning machine, and a steady-state signal is output; S23. Target individual feature recognition The steady-state signal output in step S22 is identified using a classification and recognition method based on the fingerprint characteristics of the radiation source signal to obtain the target individual characteristics; S3. If the target signal is a signal already in the signal library, search in the interference strategy library, match the corresponding interference strategy, and then generate an interference waveform; If the target signal is a newly discovered signal, first generate an interference signal based on a similarity matching method. At the same time, through case reuse and case modification, combined with the interference effect, the evaluation criteria library is used for online evaluation to generate the optimal interference strategy, directly generate interference waveforms for unknown radiation sources, and update the signal library through machine learning. S4. Transmit the interference waveform described in step S3 to start interference.

2. The method of adaptive jamming based on target features according to claim 1, characterized in that: The specific steps of extracting signal features by detecting the target signal in step S21 are: Perform data preprocessing on the collected signals; For each frame of the acquired signal, firstly, a coarse frequency sweep is performed based on the energy detection technology to obtain the key frequency; Then, based on the cyclostationary detection technology, the key frequency signal is precisely scanned to complete the preliminary analysis of the signal; Finally, the signal is sent to the modulation mode identification module for identification.

3. The method of adaptive interference based on target features according to claim 2, characterized in that: The data preprocessing of the collected signal comprises the following steps: a) removing the front end and the tail end of the received signal sequence, and then taking the middle segment of the received signal sequence; b) remove the points close to zero in the signal sequence; c) Normalize the data to the range of 0 to 1.

4. The method of adaptive jamming based on target features according to claim 1, characterized in that: The step S22 of signal pattern recognition comprises the following steps: A total of five features including information entropy features and cyclic spectrum features are extracted to form feature vectors, which are used as the input of the neural network. After the trained network, the feature set extracted from the signal is divided into a labeled training set and an unlabeled test set, which will be used to train the GA-ELM network and test the recognition performance respectively; After designing the GA-ELM network structure, it is trained using labeled training data to obtain a trained GA-ELM network; Then the unlabeled data to be identified is input into the trained network to obtain the identification result, and finally the modulation identification of the communication signal is completed.

5. The method of adaptive jamming based on target features according to claim 4, characterized in that: The information entropy features include singular spectrum Shannon entropy, singular spectrum exponential entropy, power spectrum Shannon entropy and power spectrum exponential entropy.

6. The method of adaptive jamming based on target characteristics according to claim 1, characterized in that: The step S23 targets individual feature identification, The following steps are involved: First, the steady-state signal is acquired; Then, the signal fingerprint features are extracted from three aspects: nonlinearity, non-stationarity, and non-Gaussianity; Finally, the fingerprint feature set of the radiation source signal is comprehensively extracted and the ITD is decomposed by time scale; Time scale decomposition (ITD) decomposes the received signal into a series of rotating components with gradually decreasing instantaneous frequencies, and then extracts the combined features of the original signal and each rotating component, characterizing the signal fingerprint through multiple levels, dimensions and angles.

7. The method of adaptive jamming based on target characteristics according to claim 6, characterized in that: The signal fingerprint features include fractal features, statistical features and energy distribution features.

Images