A method for intelligent identification of communication radiation sources

Abstract

The present invention provides a method for intelligently identifying a communication radiation source, and relates to the technical field of communication radiation source identification. The present invention calculates the signal-to-noise ratio of a communication signal of a radiation source to be identified, decomposes the communication signal through wavelet transform, and respectively obtains a noise signal and a data signal in the signal; when the signal-to-noise ratio is less than a lower limit of a signal quality threshold, a feature of the noise signal is obtained by performing short-time Fourier transform on the noise signal; when the signal-to-noise ratio is greater than an upper limit of a signal quality threshold, an empirical mode decomposition is performed on the data signal to obtain a data signal feature state distribution model; when the signal-to-noise ratio is within the range of an upper and lower limit of the signal quality threshold, features are extracted from the noise signal and the data signal, and comprehensive features of the communication signal are output according to the signal-to-noise ratio; and a comprehensive identification model based on Bayes is established to identify the communication radiation source according to features of the communication signal obtained under different signal-to-noise ratios.

Description

Technical Field

[0001] The present invention relates to the technical field of communication radiation source identification, and in particular to a communication radiation source intelligent identification method. Background Art

[0002] With the rapid development of wireless communication technology, more and more wireless communication radiation source equipment has been put into use, providing great convenience for people's lives. However, with the increasing number of ways to obtain wireless communication radiation sources and the popularization of smart Internet of Things technology, there are often fake devices and communication radiation sources with identity fraud that endanger social security. How to accurately identify communication radiation sources has become a hot research field at present. By accurately identifying communication radiation sources, it provides strong support for communication security and strengthens the security of communication networks. At the same time, through targeted monitoring and management, the management of radio spectrum is more efficient and accurate. Whether in the military or civilian fields, the identification technology of subtle features of radiation sources is becoming more and more important. How to accurately adjust communication signals according to complex application scenarios has become an urgent problem to be solved.

[0003] In the prior art, publication number CN115169415A discloses an efficient communication radiation source identification method, which uses the same receiver to receive the radiation source, collects and marks the signal emitted by the radiation source device to be identified, and pre-processes the signal data; inputs the normalized data into the convolutional neural network model, and uses the objective function combining the cross entropy loss function and the center loss function as the loss function of the convolutional neural network to train the convolutional neural network model; uses the trained convolutional neural network to realize the identification of the communication radiation source, but the scheme has a high dependence on training data, and the communication signal is constantly changing. When receiving the communication signal, changes in the external environment and the internal environment will cause the quality of the communication signal to change, lack of dynamic recognition of dynamic signals, and poor adaptability.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a method for intelligently identifying a communication radiation source to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for intelligently identifying a communication radiation source, the specific steps comprising:

[0008] Step 1: Obtain the communication signal of the radiation source to be identified and the sample signal of the radiation source, calculate the signal-to-noise ratio of the communication signal to be identified by measuring the average power of the communication signal to be identified and the average power of the noise signal, set the signal quality threshold, and judge the communication signal quality according to the signal-to-noise ratio;

[0009] Step 2: Decompose the communication signal through wavelet transform, and obtain the noise signal and data signal in the signal by setting the noise threshold. When the signal-to-noise ratio is less than the lower limit of the signal quality threshold, obtain the noise signal feature by performing short-time Fourier transform on the noise signal;

[0010] Step 3: When the signal-to-noise ratio is greater than the upper limit of the signal quality threshold, multiple intrinsic mode functions are obtained by performing empirical mode decomposition on the data signal, and each intrinsic mode function is subjected to Hilbert transform to obtain an analytical signal. The feature distribution is reconstructed according to the analytical signal to obtain a data signal feature state distribution model;

[0011] Step 4: When the signal-to-noise ratio is within the upper and lower limits of the signal quality threshold, extract features from the noise signal and the data signal respectively, and calculate the weight ratio of the data signal feature to the noise signal feature according to the signal-to-noise ratio, and output the comprehensive features of the communication signal according to the weight ratio;

[0012] Step 5: Establish a Bayesian-based comprehensive recognition model, collect statistics based on the signal type and signal characteristics of the communication signal in the sample signal of the radiation source, update the model parameters, calculate the communication signal characteristics based on the acquired signal-to-noise ratio of the radiation source communication signal, and use it as the input of the updated comprehensive recognition model to identify the radiation source.

[0013] Furthermore, the specific steps of calculating the signal-to-noise ratio of the communication signal to be identified by detecting the power of the communication signal to be identified and the noise signal in the communication signal are:

[0014] By detecting the power of the communication signal to be identified and the noise signal in the communication signal at the same time interval, the average power of the communication signal to be identified and the average power of the noise are calculated. The specific calculation formula is:

[0015] , in, are the average power of the communication signal and the average power of the noise signal, respectively. are the power detection times of communication signal and noise respectively, For the The power of the communication signal detected at all times, For the The power of the noise signal detected at each moment;

[0016] The calculation formula of the signal-to-noise ratio of the communication signal to be identified is:

[0017] in, is the signal-to-noise ratio of the communication signal to be identified;

[0018] when When , the signal quality is judged to be poor;

[0019] when When , the signal quality is judged to be average;

[0020] when When the signal quality is good;

[0021] in, are the lower and upper limits of the signal quality threshold range, respectively. .

[0022] Furthermore, the specific steps of decomposing the communication signal by wavelet transform and obtaining the noise signal and the data signal in the signal by setting the noise threshold are as follows:

[0023] Firstly, the communication signal is subjected to wavelet transform and decomposed into a series of approximate coefficients and detail coefficients of different frequencies;

[0024] in, is the communication signal in the sample data set, is the approximate coefficient, is the detail coefficient, is the scale function, is the wavelet function, is the acquisition time point of the communication signal, are scale parameter and displacement parameter respectively;

[0025] By setting the noise threshold to process the wavelet coefficients, the detail coefficients affected by the retained noise are removed respectively;

[0026] , in, is the detail coefficient after denoising, is the detail coefficient of the retained noise, is the noise threshold to be set;

[0027] Use the processed wavelet coefficients to perform inverse wavelet transform to reconstruct the denoised data signal and the noise signal;

[0028] , in, is the denoised data signal, is the reserved noise signal.

[0029] Furthermore, the noise signal characteristics are obtained by performing short-time Fourier transform on the noise signal, and the specific operation steps are as follows:

[0030] By sliding the window function on the noise signal through short-time Fourier transform, the noise signal is divided into shorter segments, and then Fourier transform is performed on each segment, so that the frequency domain characteristics of the noise signal can be obtained at the same time. The specific calculation formula is:

[0031] in, For The frequency domain characteristics of the noise signal at the moment, is the retained noise signal, is the window function, The center position of the time window at time , is the frequency variable, is the number of samples in the window, Is an imaginary unit.

[0032] Furthermore, the specific steps of obtaining the data signal characteristic state distribution model are:

[0033] Decomposing the data signal x(t) into multiple intrinsic mode functions and a residual term through empirical mode decomposition;

[0034] in, is the data signal, For the The intrinsic mode functions, is the remaining item;

[0035] For each intrinsic mode function Hilbert transform, calculate its analytical signal;

[0036] in, is the analytical signal of the intrinsic mode function, is an imaginary unit, for The Hilbert transform of

[0037] Acquire characteristic distribution of data signal according to analytical signal;

[0038] in, is the characteristic distribution of the data signal, is the Dirac function, For time, is the frequency.

[0039] Furthermore, the Hilbert transform equation is:

[0040] in, is the result of Albert transformation, is the intrinsic mode function, is the Cauchy principal value, is the current time point, is the integration variable, is the standard normalization factor of the Hilbert transform.

[0041] Furthermore, the calculation formula for calculating the weight ratio of the signal feature to the noise feature according to the signal-to-noise ratio is:

[0042] in, is the weight ratio of signal feature to noise feature, is the signal-to-noise ratio, is the scaling factor of the signal-to-noise ratio, are the lower and upper limits of the signal quality threshold range, respectively. ;

[0043] The specific formula for outputting the comprehensive features of the communication signal according to the weight ratio to form the comprehensive feature data set is:

[0044] in, For comprehensive characteristics, is the weight ratio of data signal features to noise signal features, is the data signal characteristic, is the noise signal characteristic.

[0045] Furthermore, the specific steps of establishing the Bayesian-based comprehensive recognition model are:

[0046] Calculate the prior probability of the communication signal type according to the communication signal in the sample signal;

[0047] in, Communication signal type The prior probability of is the number of all signal types of the sample signal, is the indicative function, For the data set Communication signal type, is the signal type of the sample signal, is the smoothing coefficient;

[0048] The formula of the indicative function is:

[0049] ;

[0050] Calculating the likelihood probability of the feature set of each communication signal;

[0051] in, For a given signal type The signal feature set is observed The probability of For a given signal type The signal characteristics were observed under The probability of is the number of signal feature types;

[0052] Calculate the posterior probability of each signal type according to Bayes' theorem;

[0053] in, Signal type The posterior probability of is a normalizing constant;

[0054] Establish a classification decision model;

[0055] in, The signal type finally identified.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention calculates the signal-to-noise ratio of the communication signal of the acquired radiation source to be identified, decomposes the communication signal through wavelet transform, and respectively obtains the noise signal and the data signal in the signal; when the signal-to-noise ratio is less than the lower limit of the signal quality threshold, the noise signal feature is obtained by performing short-time Fourier transform on the noise signal; when the signal-to-noise ratio is greater than the upper limit of the signal quality threshold, the data signal feature state distribution model is obtained by performing empirical mode decomposition on the data signal; when the signal-to-noise ratio is within the range of the upper and lower limits of the signal quality threshold, the noise signal and the data signal are respectively extracted, and the comprehensive feature of the communication signal is output according to the signal-to-noise ratio; by establishing a comprehensive recognition model based on Bayes, the communication radiation source is identified according to the features of the communication signal obtained under different signal-to-noise ratios.

[0058] The present invention adaptively selects different processing methods according to different signal-to-noise ratios, effectively processes signals of different qualities, can adapt to the identification of radiation sources in various complex scenes, resist interference of different properties and intensities, ensure accurate extraction of signal features and identification of radiation sources, and centrally processes noise signals and obtains their features under low signal-to-noise ratio conditions, which helps to effectively suppress and identify environmental noise, thereby improving the accuracy of signal processing. Under medium signal-to-noise ratio conditions, a comprehensive signal feature view is formed by simultaneously analyzing signals and noise and fusing information under different signal-to-noise ratios. This hybrid analysis enables the recognition model to utilize all effective information of low and high signal-to-noise ratios in the environment, uses the empirical mode decomposition method to decompose signals under high signal-to-noise ratio conditions, can finely capture the intrinsic pattern distribution characteristics of the signal, achieve more in-depth signal analysis and more accurate feature extraction, and adopts the Bayesian method to construct the model, which can effectively fuse multiple signal features, improve the accuracy and reliability of recognition, and quantify the uncertainty of recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION

[0060] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0061] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0062] Example

[0063] See also Figure 1 , the present invention provides a technical solution:

[0064] A method for intelligently identifying a communication radiation source, the specific steps comprising:

[0065] Step 1: Obtain the communication signal of the radiation source to be identified and the sample signal of the radiation source, calculate the signal-to-noise ratio of the communication signal to be identified by measuring the average power of the communication signal to be identified and the average power of the noise signal, set the signal quality threshold, and judge the communication signal quality according to the signal-to-noise ratio.

[0066] Communication radiation source refers to the equipment or system used to send electromagnetic waves during the communication process. Due to the individual differences of radiation source equipment, these differences are reflected in the signal, and it is inevitable to produce subtle differences in characteristics. These subtle differences generally do not affect the transmission of information in the signal, and are detectable and reproducible. They are called individual subtle characteristics of the radiation source. Due to the tolerance of electronic components inside the transmitter, there will be small differences in the radiation source signal, which are manifested in the transmitted signal in the form of amplitude, phase, frequency, etc., which can more accurately identify the type and characteristics of the communication source. At the same time, the noise itself can also carry information about the type, location and operating status of the communication source equipment. Electronic equipment will generate noise of a specific frequency when it is running. By analyzing these specific noise characteristics, different characteristic information that can fully characterize the characteristics of the radiation source is captured and extracted from the signal transmitted by the communication radiation source equipment. This process can accurately distinguish and identify different communication radiation source equipment, thereby achieving the purpose of accurate identification. The signal-to-noise ratio is an important parameter to describe the quality of communication signals. It represents the ratio of signal strength to background noise intensity. A high signal-to-noise ratio means that the signal strength is much greater than the noise intensity and the communication quality is better; conversely, it means that the noise level is high, which may affect the reliability and clarity of communication.

[0067] In this embodiment, the specific steps of calculating the signal-to-noise ratio of the communication signal to be identified by detecting the power of the communication signal to be identified and the noise signal in the communication signal are:

[0068] By detecting the power of the communication signal to be identified and the noise signal in the communication signal at the same time interval, the average power of the communication signal to be identified and the average power of the noise are calculated. The specific calculation formula is:

[0069] , in, are the average power of the communication signal and the average power of the noise signal, respectively. are the power detection times of communication signal and noise respectively, For the The power of the communication signal detected at all times, For the The power of the noise signal detected at each moment;

[0070] The calculation formula of the signal-to-noise ratio of the communication signal to be identified is:

[0071] in, is the signal-to-noise ratio of the communication signal to be identified;

[0072] When the signal quality is poor and the noise accounts for too much, the noise signal is identified;

[0073] When the signal quality is average, the signal and noise ratio is balanced. The noise is stripped off and the communication signal after noise stripping and the stripped noise are comprehensively identified.

[0074] When the signal quality is good and the signal accounts for a large proportion, the communication signal after denoising is identified;

[0075] in, are the lower and upper limits of the signal quality threshold range, respectively. .

[0076] The signal quality threshold is set based on historical data or industry standards. By analyzing the relationship between signal quality and communication effect in historical data, a reasonable threshold can be set.

[0077] The system can adapt to different signal conditions by evaluating the signal-to-noise ratio to select the feature extraction method that best suits the current environment. This flexible approach improves the effectiveness and reliability of the system in changing environments. When the signal quality is good, extracting features directly from the signal is usually more direct and effective, and fast recognition can be achieved using fewer computing resources. When the signal quality is poor, focusing on noise features can avoid ineffective signal processing attempts, save resources, and improve recognition accuracy. In the case of poor signal quality, traditional signal-based feature extraction methods may not be effective enough. At this time, relying on noise features can serve as a supplement to help more accurately identify the communication source.

[0078] Step 2: Decompose the communication signal through wavelet transform, and obtain the noise signal and data signal in the signal by setting the noise threshold. When the signal-to-noise ratio is less than the lower limit of the signal quality threshold, obtain the noise signal characteristics by performing short-time Fourier transform on the noise signal.

[0079] Wavelet transform is a powerful mathematical tool for analyzing and processing communication signals, especially in signal decomposition and denoising. Wavelet transform can separate signals from noise because they usually show different characteristics in the wavelet domain. The main information of the signal is often concentrated in certain specific areas of the wavelet coefficients, while the noise is more dispersed. Through threshold processing or soft and hard threshold methods, the noise component can be removed or separated while retaining the main characteristics of the signal.

[0080] In this embodiment, the specific steps of decomposing the communication signal by wavelet transform and obtaining the noise signal and the data signal in the signal by setting the noise threshold are as follows:

[0081] Firstly, the communication signal is subjected to wavelet transform and decomposed into a series of approximate coefficients and detail coefficients of different frequencies;

[0082] in, is the communication signal in the sample data set, is the approximate coefficient, is the detail coefficient, is the scale function, is the wavelet function, is the acquisition time point of the communication signal, are the scale parameter and displacement parameter respectively.

[0083] The approximation coefficients represent the low-frequency components of the signal, that is, the basic trend or smooth part of the signal. At each level of the multi-layer wavelet decomposition, the approximation coefficients are calculated by applying the wavelet's low-pass filter and downsampling, usually at half the sampling rate. The detail coefficients represent the high-frequency components of the signal, that is, the details or fast-changing parts of the signal, such as edges or noise. They are calculated by applying the wavelet's high-pass filter and downsampling.

[0084] The scale parameter is used to control the expansion and contraction of the wavelet function, affecting the frequency of the analysis. By changing the scale parameter, the frequency resolution can be transformed in the time-frequency analysis, which makes wavelet analysis particularly suitable for the characteristic analysis of communication signals. The displacement parameter is used to control the translation of the wavelet function along the time axis. By changing the displacement parameter, the wavelet function can be analyzed locally at different positions of the signal, so that the local characteristics of the signal at different time points can be captured.

[0085] Determining the best values ​​for the scale and displacement parameters requires experimentation and iteration. You can start with a wide range of parameters and then refine your selection based on the results of your initial analysis, usually starting with a larger scale (low frequency analysis) and gradually reducing the scale (increasing frequency resolution) to observe the different frequency components of the signal.

[0086] By setting the noise threshold to process the wavelet coefficients, the detail coefficients affected by the retained noise are removed respectively;

[0087] , in, is the detail coefficient after denoising, is the detail coefficient of the retained noise, is the noise threshold to be set;

[0088] The setting of the noise threshold affects the noise removal effect and the degree of signal retention. The noise threshold is set in advance by the fixed threshold method. It is set according to the general noise level of the signal and uniformly applied to all detail coefficients. This method is relatively simple to operate.

[0089] Use the processed wavelet coefficients to perform inverse wavelet transform to reconstruct the denoised data signal and the noise signal;

[0090] , in, is the denoised data signal, is the reserved noise signal.

[0091] Short-time Fourier transform can provide the frequency component information of the signal at a specific time point, which is very useful for analyzing and identifying instantaneous noise signals in the signal. It divides the signal into shorter time periods and performs Fourier transform on the noise signal in each time period, thereby obtaining the time and frequency information of the noise signal at the same time.

[0092] The noise signal characteristics are obtained by performing short-time Fourier transform on the noise signal, and the specific operation steps are as follows:

[0093] By sliding the window function on the noise signal through short-time Fourier transform, the noise signal is divided into shorter segments, and then Fourier transform is performed on each segment, so that the frequency domain characteristics of the noise signal can be obtained at the same time. The specific calculation formula is:

[0094] in, For The frequency domain characteristics of the noise signal at the moment, is the retained noise signal, is the window function, The center position of the time window at time , is the frequency variable, is the number of samples in the window, Is an imaginary unit.

[0095] Step 3: When the signal-to-noise ratio is greater than the upper limit of the signal quality threshold, obtain multiple intrinsic mode functions by performing empirical mode decomposition on the data signal, perform Hilbert transform on each intrinsic mode function to obtain an analytical signal, reconstruct the feature distribution according to the analytical signal, and obtain the data signal feature state distribution model.

[0096] Empirical mode decomposition has been used to analyze and classify various radio signals, including signal classification, interference detection, modulation identification and other fields. Through these applications, the unique characteristics of different signal sources can be identified and the performance of automatic monitoring systems can be improved. The empirical mode decomposition method can process signals adaptively and obtain the time, frequency and energy distribution characteristics of the signal. The empirical mode decomposition is proposed mainly to solve the problem of nonlinearity. It can make the signal decomposition unique and can well localize the signal in the time domain and frequency domain. The purpose of time-frequency analysis is to reflect the energy of the signal on the scale of time and frequency at the same time, so that the information hidden in the time domain waveform can be clearly reflected and the characteristic information contained in the signal can be obtained.

[0097] In this embodiment, the specific steps of obtaining the data signal characteristic state distribution model are:

[0098] Decomposing the data signal x(t) into multiple intrinsic mode functions and a residual term through empirical mode decomposition;

[0099] in, is the data signal, For the The intrinsic mode functions, For the remaining items.

[0100] Each intrinsic mode function represents a different frequency component in the signal. By performing time-frequency analysis on each intrinsic mode function, the dynamic characteristics of the signal, such as instantaneous frequency and instantaneous amplitude, can be obtained. The residual term is the remaining part of the signal after all intrinsic mode functions are extracted, which is usually the low-frequency trend or long-term trend in the original signal.

[0101] For each intrinsic mode function Hilbert transform, calculate its analytical signal;

[0102] in, is the analytical signal of the intrinsic mode function, is an imaginary unit, for The Hilbert transform of

[0103] The Hilbert transform is used to construct an analytical signal from a real signal, thereby obtaining the instantaneous frequency and instantaneous amplitude of the signal, which helps to analyze the instantaneous characteristics of the signal, especially in communications, where it can help identify frequency modulation and time-varying characteristics.

[0104] Acquire characteristic distribution of data signal according to analytical signal;

[0105] in, is the characteristic distribution of the data signal, is the Dirac function, For time, is the frequency.

[0106] In this embodiment, the Hilbert transform equation is:

[0107] in, is the result of Albert transformation, is the intrinsic mode function, is the Cauchy principal value, is the current time point, is the integration variable, is the standard normalization factor of the Hilbert transform.

[0108] Step 4: When the signal-to-noise ratio is within the upper and lower limits of the signal quality threshold, extract features from the noise signal and the data signal respectively, and calculate the weight ratio of the data signal feature to the noise signal feature according to the signal-to-noise ratio, and output the comprehensive features of the communication signal according to the weight ratio.

[0109] In this embodiment, the calculation formula for calculating the weight ratio of the signal feature and the noise feature according to the signal-to-noise ratio is:

[0110] in, is the weight ratio of signal feature to noise feature, is the signal-to-noise ratio, is the scaling factor of the signal-to-noise ratio, are the lower and upper limits of the signal quality threshold range, respectively. .

[0111] The weight ratio reflects the proportion of each feature output according to the proportion of noise signal to data signal under normal signal quality conditions. By extracting noise and data signal features respectively, the effective components and interference components in the signal can be more accurately identified and distinguished, thereby improving the accuracy of signal processing. The weight ratio calculated according to the signal-to-noise ratio can dynamically adjust the degree of attention paid to signal features and noise features, thereby optimizing the signal decoding process. For example, in a high noise environment, the attention paid to noise features can be increased to improve the robustness of decoding. By combining the weight ratio of data and noise features to output the comprehensive features of the communication signal, it can better adapt to different communication environments and improve the overall performance of the system, such as reducing the bit error rate and improving the transmission efficiency.

[0112] The specific formula for outputting the comprehensive features of the communication signal according to the weight ratio to form the comprehensive feature data set is:

[0113] in, For comprehensive characteristics, is the weight ratio of data signal features to noise signal features, is the data signal characteristic, is the noise signal characteristic.

[0114] Step 5: Establish a Bayesian-based comprehensive recognition model, collect statistics based on the signal type and signal characteristics of the communication signal in the sample signal of the radiation source, update the model parameters, calculate the communication signal characteristics based on the acquired signal-to-noise ratio of the radiation source communication signal, and use it as the input of the updated comprehensive recognition model to identify the radiation source.

[0115] The Bayesian-based comprehensive recognition model is based on the Bayesian theorem and the theory of mutual independence assumption between feature attributes. The core idea is to calculate the conditional probability corresponding to each feature under a given category label, and then use these conditional probabilities for classification, which greatly reduces the difficulty of implementing the algorithm, and the algorithm can be incrementally trained based on real-time added data. In addition, Naive Bayes also has stable classification efficiency. In radiation source identification, communication radiation sources usually have multiple signal characteristics, including frequency, amplitude, phase, waveform, etc. Naive Bayes can efficiently process multidimensional data and combine multiple features to improve recognition accuracy.

[0116] In this embodiment, the specific steps of establishing the Bayesian-based comprehensive recognition model are:

[0117] Calculate the prior probability of the communication signal type according to the communication signal in the sample signal;

[0118] in, Communication signal type The prior probability of is the number of all signal types of the sample signal, is the indicative function, For the data set Communication signal type, is the signal type of the sample signal, is the smoothing coefficient.

[0119] Prior probability refers to the initial estimated probability of different communication signal types before the current signal features are observed. Prior probability is used to reflect the system's original assumptions about the signal type before processing a specific signal. For example, if a certain signal type appears frequently in historical data, a higher prior probability may be set for this signal category. In this way, the system will be more inclined to match features to this common signal type during the recognition process.

[0120] The formula of the indicative function is:

[0121] .

[0122] The indicator function is used to determine whether a signal feature meets a specific condition. Mathematically, the value of the indicator function is 1 when the condition is true, otherwise it is 0. In Naive Bayes, the indicator function can be used in the estimation stage of model parameters. For example, when calculating the frequency of each feature in a given category, the indicator function can help count the number of times the feature appears.

[0123] Calculating the likelihood probability of the feature set of each communication signal;

[0124] in, For a given signal type The signal feature set is observed The probability of For a given signal type The signal characteristics were observed under The probability of is the number of signal feature types.

[0125] Likelihood probability refers to the probability of observing a certain signal feature under a certain signal type. Likelihood probability is used to evaluate the likelihood of different features appearing when the signal type is known. For example, the probability of a feature that is considered to be a specific type of signal, such as parameters such as frequency and amplitude, appearing in a certain communication signal category is the likelihood probability of the signal feature. It helps infer the likelihood of a specific category by combining signal features.

[0126] Calculate the posterior probability of each signal type according to Bayes' theorem;

[0127] in, Signal type The posterior probability of is a normalizing constant.

[0128] The posterior probability is the updated probability of a specific communication radiation source category given the signal characteristics. The posterior probability combines the prior probability and the likelihood probability to ultimately determine which signal type a specific signal characteristic belongs to. By maximizing the posterior probability, the Naive Bayes algorithm can identify and classify the signal type.

[0129] Establish a classification decision model;

[0130] in, The signal type finally identified.

[0131] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0132] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0133] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0134] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for intelligently identifying a communication radiation source, characterized in that: The specific steps include: Step 1: Obtain the communication signal of the radiation source to be identified and the sample signal of the radiation source, calculate the signal-to-noise ratio of the communication signal to be identified by detecting the power of the communication signal to be identified and the noise signal in the communication signal, set the signal quality threshold, and judge the quality of the communication signal according to the signal-to-noise ratio; Step 2: Decompose the communication signal through wavelet transform, and obtain the noise signal and data signal in the signal by setting the noise threshold. When the signal-to-noise ratio is less than the lower limit of the signal quality threshold, obtain the noise signal feature by performing short-time Fourier transform on the noise signal; Step 3: When the signal-to-noise ratio is greater than the upper limit of the signal quality threshold, multiple intrinsic mode functions are obtained by performing empirical mode decomposition on the data signal, and each intrinsic mode function is subjected to Hilbert transform to obtain an analytical signal. The feature distribution is reconstructed according to the analytical signal to obtain a data signal feature state distribution model; Step 4: When the signal-to-noise ratio is within the upper and lower limits of the signal quality threshold, extract features from the noise signal and the data signal respectively, and calculate the weight ratio of the data signal feature to the noise signal feature according to the signal-to-noise ratio, and output the comprehensive features of the communication signal according to the weight ratio; Step 5: Establish a Bayesian-based comprehensive recognition model, collect statistics on the signal type and signal characteristics of the communication signal in the sample signal of the radiation source, update the model parameters, and use the noise signal characteristics, data signal characteristics and comprehensive characteristics obtained from the different signal-to-noise ratios of the radiation source communication signal as the input of the updated comprehensive recognition model to identify the communication radiation source.

2. A method for intelligently identifying communication radiation sources according to claim 1, characterized in that: The specific steps of calculating the signal-to-noise ratio of the communication signal to be identified by detecting the power of the communication signal to be identified and the noise signal in the communication signal are: By detecting the power of the communication signal to be identified and the noise signal in the communication signal at the same time interval, the average power of the communication signal to be identified and the average power of the noise are calculated. The specific calculation formula is: , in, are the average power of the communication signal and the average power of the noise signal, respectively. are the power detection times of communication signal and noise respectively, For the The power of the communication signal detected at all times, For the The power of the noise signal detected at each moment; The calculation formula of the signal-to-noise ratio of the communication signal to be identified is: in, is the signal-to-noise ratio of the communication signal to be identified; when When , the signal quality is judged to be poor; when When , the signal quality is judged to be average; when When the signal quality is good; in, are the lower and upper limits of the signal quality threshold range, respectively. .

3. The method for intelligently identifying a communication radiation source according to claim 1, characterized in that: The specific steps of decomposing the communication signal by wavelet transform and respectively obtaining the noise signal and the data signal in the signal by setting the noise threshold are as follows: Firstly, the communication signal is subjected to wavelet transform and decomposed into a series of approximate coefficients and detail coefficients of different frequencies; in, is the communication signal in the sample data set, is the approximate coefficient, is the detail coefficient, is the scale function, is the wavelet function, is the acquisition time point of the communication signal, are the scale parameter and displacement parameter respectively; By setting the noise threshold to process the wavelet coefficients, the detail coefficients affected by the retained noise are removed respectively; , in, is the detail coefficient after denoising, is the detail coefficient of the retained noise, is the noise threshold to be set; Use the processed wavelet coefficients to perform inverse wavelet transform to reconstruct the denoised data signal and the noise signal; , in, is the denoised data signal, is the reserved noise signal.

4. The method for intelligently identifying a communication radiation source according to claim 1, characterized in that: The noise signal characteristics are obtained by performing short-time Fourier transform on the noise signal, and the specific operation steps are as follows: By sliding the window function on the noise signal through short-time Fourier transform, the noise signal is divided into shorter segments, and then Fourier transform is performed on each segment, so that the frequency domain characteristics of the noise signal can be obtained at the same time. The specific calculation formula is: in, For The frequency domain characteristics of the noise signal at the moment, is the retained noise signal, is the window function, The center position of the time window at time , is the frequency variable, is the number of samples in the window, Is an imaginary unit.

5. The method for intelligently identifying a communication radiation source according to claim 1, characterized in that: The specific steps of obtaining the data signal characteristic state distribution model are: Decomposing the data signal x(t) into multiple intrinsic mode functions and a residual term through empirical mode decomposition; in, is the data signal, For the The intrinsic mode functions, is the remaining item; For each intrinsic mode function Perform Hilbert transform and calculate its analytical signal; in, is the analytical signal of the intrinsic mode function, is an imaginary unit, for The Hilbert transform of Acquire characteristic distribution of data signal according to analytical signal; in, is the characteristic distribution of the data signal, is the Dirac function, For time, is the frequency.

6. A method for intelligently identifying communication radiation sources according to claim 5, characterized in that: The Hilbert transform equation is: in, is the result of Albert transformation, is the intrinsic mode function, is the Cauchy principal value, is the current time point, is the integration variable, is the standard normalization factor of the Hilbert transform.

7. The method for intelligently identifying a communication radiation source according to claim 1, characterized in that: The calculation formula for calculating the weight ratio of the signal feature and the noise feature according to the signal-to-noise ratio is: in, is the weight ratio of signal feature to noise feature, is the signal-to-noise ratio, is the scaling factor of the signal-to-noise ratio, are the lower and upper limits of the signal quality threshold range, respectively. ; The specific formula for outputting the comprehensive features of the communication signal according to the weight ratio to form the comprehensive feature data set is: in, For comprehensive characteristics, is the weight ratio of data signal features to noise signal features, is the data signal characteristic, is the noise signal characteristic.

8. The method for intelligently identifying a communication radiation source according to claim 1, characterized in that: The specific steps of establishing the Bayesian-based comprehensive recognition model are: Calculate the prior probability of the communication signal type according to the communication signal in the sample signal; in, Communication signal type The prior probability of is the number of all signal types of the sample signal, is the indicative function, For the data set Communication signal type, is the signal type of the sample signal, is the smoothing coefficient; The formula of the indicative function is: ； Calculating the likelihood probability of the feature set of each communication signal; in, For a given signal type The signal feature set is observed The probability of For a given signal type The signal characteristics were observed under The probability of is the number of signal feature types; Calculate the posterior probability of each signal type according to Bayes' theorem; in, Signal type The posterior probability of is a normalizing constant; Establish a classification decision model; in, The signal type finally identified.

Images