Interference identification method and device based on soft decision multi-node cooperative sensing

Abstract

The application relates to an interference identification method and device based on soft decision multi-node cooperative sensing. The method comprises the following steps: setting and identifying interference of a communication signal according to a classifier in a sensing node to obtain an interference identification result; inputting the interference identification result of each sensing node into a fusion center, processing all the interference identification results according to soft decision cooperative sensing to obtain a final interference identification result; quantitatively processing the interference identification result of each sensing node according to the fusion center, giving a label to a to-be-labeled interference sample according to the obtained quantitative result, inputting the final interference identification result and the label into the classifier in the sensing node through the fusion center for self-training to obtain a new trained classifier. The method can identify interference based on soft decision multi-node cooperative sensing, improves the accuracy of interference identification, and realizes high-quality self-training of the classifier through the fusion center, so that a new classifier with higher interference identification precision is obtained.

Description

Technical Field

[0001] This application relates to the field of wireless communication technology, and in particular to an interference identification method and apparatus based on soft decision-making multi-node cooperative sensing. Background Technology

[0002] Due to their inherent openness and broadcast nature, wireless communications are increasingly vulnerable to malicious interference attacks. In response, various techniques have been proposed to defend against these attacks, such as frequency hopping (FH), direct sequence spread spectrum (DSSS), and more recent advanced methods like game theory-based schemes, interference alignment, and optimal power control. Notably, a key objective behind these effective anti-interference methods is the accurate identification of interference attacks. Specifically, interference identification can effectively pinpoint interference attacks, thereby guiding the implementation of existing anti-interference methods. Furthermore, successful interference identification can even be used to predict the behavior of interferers, as real-time, continuously changing interference patterns are extremely costly and have low implementation rates. Therefore, an effective and efficient interference identification method is urgently needed.

[0003] However, interference identification is far more difficult than identification of legitimate communication signals because interference signals do not require coordination issues such as synchronization design, leading to the emergence of various interference patterns. In particular, interferers can adaptively adjust their interference strategies to maximize the destructive effect on legitimate nodes at the lowest cost, making it difficult to obtain interference detection probabilities and accurate interference signal sets. Furthermore, building a large set of labeled interference signals belonging to various patterns for classifier training requires extensive manual labeling, resulting in very high implementation costs. Even more seriously, due to the potentially time-varying characteristics of interference signals, there is a real-world need for a fast interference identification method that is low-complexity, high-accuracy, and does not require extensive labeling of training samples. Summary of the Invention

[0004] Therefore, it is necessary to provide an interference identification method and apparatus based on soft decision-making multi-node cooperative perception that can reduce the computational complexity of interference identification and improve the accuracy of interference identification, in order to address the above-mentioned technical problems.

[0005] An interference identification method based on soft decision-making multi-node cooperative sensing, the method comprising:

[0006] A sensing node is set up to receive communication signals. The classifier in the sensing node performs interference identification on the communication signals to obtain interference identification results. The interference identification results include the interference identification probability of each interference pattern in the communication signal.

[0007] The interference identification results of each sensing node are input into the fusion center, and all interference identification results are processed according to soft decision cooperative sensing to obtain the final interference identification result.

[0008] The fusion center quantifies the interference identification results of each sensing node to obtain the quantification results. Based on the quantification results, the interference samples to be labeled are assigned labels. The fusion center then inputs the final interference identification results and labels into the sensing nodes to perform self-training of the classifier, resulting in a well-trained new classifier.

[0009] In one embodiment, a sensing node is configured to receive communication signals, and an interference identification result is obtained by classifying the communication signals using a classifier in the sensing node, including:

[0010] A sensing node is set up to receive communication signals. The communication signals are normalized to obtain a normalized signal. The normalized signal is then subjected to a fast Fourier transform to obtain a single-sided spectrum signal.

[0011] The single-sided spectrum signal is converted to obtain interference features, which include carrier factor coefficient, the ratio of the variance of the communication signal to the square of the average value of the communication signal, flatness spectrum coefficient and third-order bias coefficient.

[0012] The interference features are input into the classifier in the sensing node for interference identification, and the interference identification result is obtained.

[0013] In one embodiment, a sensing node is configured to receive a communication signal, the communication signal is normalized to obtain a normalized signal, and a fast Fourier transform is performed on the normalized signal to obtain a single-sided spectrum signal, including:

[0014] Set N c Each sensing node receives a communication signal, represented as

[0015] r i =H i s J +ω i

[0016] Where, r i This represents the communication signal received by the i-th sensing node, where i = 1, 2, ..., N c H i =diag(h i ) represents the channel matrix, h i Represents the channel coefficient, s J Indicates interference signal, ω i Indicates white Gaussian noise;

[0017] The communication signal is normalized to obtain a normalized signal, and the normalized signal is then subjected to a Fast Fourier Transform to obtain a single-sided spectrum signal y. i .

[0018] In one embodiment, interference features are input into a classifier in the sensing node for interference identification, resulting in interference identification results, including:

[0019] The feature vector composed of interference features is represented as

[0020]

[0021] Among them, t s Represents a time slot, C represents the carrier factor coefficient, and R... f The ratio of the variance to the square of the mean of the communication signal is represented by Fse, the flatness spectral coefficient is represented by b3, the third-order bias coefficient is represented by T, and the eigenvector is transposed.

[0022] The feature vector is input into the classifier in the sensing node for interference identification, and the interference identification result is obtained, represented as follows:

[0023]

[0024]

[0025] Where, p i This represents the interference identification result of the i-th sensing node. Let represent the interference identification probability of the c-th type of interference pattern in the communication signal received by the i-th sensing node, n represents that the feature vector follows an n-dimensional normal distribution, μ represents the mean vector of the labeled interference features, Σ represents the covariance matrix of the labeled interference features, and α represents the total number of interference patterns.

[0026] In one embodiment, the interference identification results of each sensing node are input into the fusion center, and all interference identification results are processed according to soft decision cooperative sensing to obtain the final interference identification result, including:

[0027] The interference identification results from each sensing node are input into the fusion center. Based on soft-decision cooperative sensing, all interference identification results are aggregated to obtain the total interference identification probability, denoted as:

[0028]

[0029] The total interference identification probability is calculated to obtain the final interference identification result, which is expressed as follows:

[0030]

[0031] Where, p FC (c) represents the interference recognition probability of the c-th type of interference pattern.

[0032] In one embodiment, the interference identification results of each sensing node by the fusion center are quantized to obtain the quantization result, including:

[0033] Based on the interference identification probabilities of each interference pattern in the interference identification results of the i-th sensing node obtained by the fusion center, the interference identification probabilities are sorted by magnitude, and the highest interference identification probability is selected for quantization to obtain the interference identification index of the i-th sensing node, denoted as .

[0034]

[0035] N c The interference identification index of each sensing node is summarized to obtain the quantization result, which is represented as:

[0036] In one embodiment, labels are assigned to the interference samples to be labeled based on the quantization results. The final interference identification results and labels are then input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a newly trained classifier, including:

[0037] Decision conditions are set based on the quantification results, and the decision conditions are expressed as follows:

[0038]

[0039] in, Indicates that the quantization result η is equal to Quantity, N represents the value that appears most frequently in the quantization result η. th Represents the threshold, and N th ≤N c , Indicates a label, Indicates the interference sample to be labeled;

[0040] When the result of the judgment condition is When a positive label is assigned to the interference sample to be labeled, it is represented as... Otherwise, assign a negative label to the interference sample to be labeled, denoted as

[0041] The final interference identification result and label are input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a well-trained new classifier.

[0042] In one embodiment, the final interference identification result and label are input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a trained new classifier, including:

[0043] The final interference identification result and label are input into the sensing node through the fusion center. The interference features of the interference samples to be labeled corresponding to the positive label are stored to obtain the interference feature set. The classifier is trained based on the interference feature set to obtain the trained new classifier.

[0044] In one embodiment, before configuring the sensing node to receive communication signals, the method further includes:

[0045] The sensing nodes are initialized based on the covariance matrix and average value of each interference pattern in the communication signal. The initialized sensing nodes are then set up around the fusion center to receive communication signals.

[0046] An interference identification device based on soft-decision multi-node cooperative sensing, the device comprising:

[0047] The interference identification module is used to set up the sensing node to receive communication signals, and to identify interference in the communication signals according to the classifier in the sensing node to obtain the interference identification result; wherein, the interference identification result includes the interference identification probability of each interference pattern in the communication signal;

[0048] The cooperative perception module is used to input the interference identification results of each perception node into the fusion center, and process all interference identification results according to the soft decision cooperative perception to obtain the final interference identification result.

[0049] The self-training module is used to quantify the interference identification results of each sensing node based on the fusion center, obtain the quantization results, assign labels to the interference samples to be labeled based on the quantization results, and input the final interference identification results and labels into the sensing nodes through the fusion center to perform self-training of the classifier and obtain a trained new classifier.

[0050] The aforementioned interference identification method and apparatus based on soft-decision multi-node cooperative perception fully utilizes the identification information from different sensing nodes by inputting the interference identification results of all sensing nodes into a fusion center and processing them according to soft-decision cooperative perception to obtain the final interference identification result. This reduces the impact of identification errors from individual sensing nodes on the interference identification result. Furthermore, the interference identification results are not quantized during the processing according to soft-decision cooperative perception, avoiding quantization errors and further improving the accuracy of the final interference identification result. Quantifying the interference identification results of each sensing node by the fusion center allows for the assignment of high-confidence labels to the interference samples to be labeled, reducing the cost of manual signal labeling. By inputting the high-accuracy final interference identification result and high-confidence labels into the sensing nodes through the fusion center, the classifiers in the sensing nodes undergo high-quality self-training, resulting in new classifiers with higher identification accuracy, thereby further improving the accuracy of subsequent interference identification. Attached Figure Description

[0051] Figure 1 This is an application scenario diagram of an interference identification method based on soft decision-making multi-node cooperative perception in one embodiment;

[0052] Figure 2 This is a flowchart illustrating an interference identification method based on soft decision-making multi-node cooperative perception in one embodiment.

[0053] Figure 3 This is a schematic diagram comparing the average recognition accuracy of the proposed method under different labeled samples in one embodiment with a number of sensing nodes of 4.

[0054] Figure 4 This is a schematic diagram comparing the average recognition accuracy of different interference recognition methods when the number of training samples is 10, as shown in one embodiment. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] The interference identification method based on soft decision-making multi-node cooperative sensing proposed in this invention can be applied to, for example... Figure 1 In the scenario shown. For example... Figure 1 As shown, when legitimate communication parties are communicating, the communication quality will be affected by a high bit error rate (BER) after a specific jammer starts working. At this time, the legitimate receiver stops sending acknowledgment characters (ACK), and then the legitimate transmitter stops communicating with the legitimate receiver. When the legitimate receiver cannot obtain information from the legitimate transmitter, the interference identification process is performed through the set sensing nodes and fusion center (FC).

[0057] In one embodiment, such as Figure 2 As shown, an interference identification method based on soft decision-making multi-node cooperative perception is provided, which is then applied to... Figure 1 Taking the scenario in the example, the following steps are included:

[0058] Step 202: Set up a sensing node to receive communication signals, and use the classifier in the sensing node to perform interference identification on the communication signals to obtain interference identification results; wherein, the interference identification results include the interference identification probability of each interference pattern in the communication signal.

[0059] It is understandable that deploying multiple sensing nodes around a legitimate receiver and considering the cooperative sensing among multiple sensing nodes can help identify interference. The classifiers in the sensing nodes are generally Bayesian classifiers, including normal Bayesian classifiers and naive Bayesian classifiers. Bayesian classifiers use causal reasoning to calculate the interference identification probability of each interference pattern in the communication signal.

[0060] Step 204: Input the interference identification results of each sensing node into the fusion center, process all interference identification results according to soft decision cooperative sensing, and obtain the final interference identification result.

[0061] It is understandable that the fusion center can also be considered a legitimate receiver. The fusion center receives interference identification data from all sensing nodes, fuses it, and provides the final identification result. Soft decision cooperative sensing is a multi-sensing method. In soft decision, each sensing node sends the interference identification probability of each interference pattern in the interference identification result to the fusion center. After receiving the interference identification results from all sensing nodes, the fusion center provides the final interference identification result by making full use of the interference identification information from different sensing nodes.

[0062] Step 206: Quantify the interference identification results of each sensing node according to the fusion center to obtain the quantification results. Assign labels to the interference samples to be labeled according to the quantification results. Input the final interference identification results and labels into the sensing nodes through the fusion center to perform self-training of the classifier and obtain a trained new classifier.

[0063] It is understandable that before the classifier self-trains, the fusion center needs to quantify the interference identification results of each sensing node, assign high-confidence labels to the interference samples to be labeled based on the quantification results, and finally feed back the final interference identification results and labels as feedback information to the sensing nodes to guide the classifiers in the sensing nodes to self-train, thereby identifying new classifiers with higher accuracy.

[0064] The aforementioned interference identification method based on soft-decision multi-node cooperative perception inputs the interference identification results of all sensing nodes into a fusion center and processes these results according to soft-decision cooperative perception to obtain the final interference identification result. This fully utilizes the identification information from different sensing nodes, reducing the impact of individual sensing node identification errors on the interference identification result. Furthermore, when processing all interference identification results according to soft-decision cooperative perception, no quantization processing is performed on the interference identification results, avoiding quantization errors and further improving the accuracy of the final interference identification result. Quantifying the interference identification results of each sensing node by the fusion center allows for the assignment of high-confidence labels to the interference samples to be labeled, reducing the cost of manual signal labeling. By inputting the high-accuracy final interference identification result and high-confidence labels into the sensing nodes through the fusion center, the classifiers in the sensing nodes undergo high-quality self-training, resulting in new classifiers with higher identification accuracy, thereby further improving the accuracy of subsequent interference identification.

[0065] In one embodiment, a sensing node is configured to receive communication signals, and an interference identification result is obtained by classifying the communication signals using a classifier in the sensing node, including:

[0066] The sensing nodes are configured to receive communication signals. For the signal r received by the i-th sensing node... i First, for signal r i Normalization is performed to obtain a normalized signal, represented as [-1, 1]. Then, a fast Fourier transform is performed on the normalized signal to obtain the single-sided spectrum signal y. i ,in l is the length of the Fast Fourier Transform (FFT) in the calculation; then the single-sided spectral signal y i The interference features are converted into interference features, which include carrier factor coefficients, the ratio of the variance of the communication signal to the square of the average value of the communication signal, flatness spectral coefficients, and third-order bias coefficients. Finally, each sensing node inputs the interference features into a classifier in the sensing node, such as a normal Bayesian classifier or a Naive Bayes classifier. The interference is identified based on the classifier, and the interference identification result is output, which includes the interference identification probability of each interference pattern in the received communication signal.

[0067] Specifically, there are six types of interference patterns in communication signals: single-tone interference, multi-tone interference, narrowband interference, wideband interference, narrowband comb interference, and frequency sweeping interference.

[0068] In one embodiment, a sensing node is configured to receive a communication signal, the communication signal is normalized to obtain a normalized signal, and a fast Fourier transform is performed on the normalized signal to obtain a single-sided spectrum signal, including:

[0069] Set N c Each sensing node receives a communication signal, represented as

[0070] r i =H i s J +ω i

[0071] Where, r i This represents the communication signal received by the i-th sensing node, where i = 1, 2, ..., N c H i =diag(h i ) represents the channel matrix, h i Represents the channel coefficient, s J Indicates interference signal, ω i Indicates white Gaussian noise;

[0072] The communication signal is normalized to obtain a normalized signal, and the normalized signal is then subjected to a Fast Fourier Transform to obtain a single-sided spectrum signal y. i .

[0073] In one embodiment, interference features are input into a classifier in the sensing node for interference identification, resulting in interference identification results, including:

[0074] The feature vector composed of interference features is represented as

[0075]

[0076] Among them, t s Represents a time slot, C represents the carrier factor coefficient, and R... f The ratio of the variance to the square of the mean of the communication signal is represented by Fse, the flatness spectral coefficient is represented by b3, the third-order bias coefficient is represented by T, and the eigenvector is transposed.

[0077] The feature vector is input into the classifier in the sensing node for interference identification, and the interference identification result is obtained, represented as follows:

[0078]

[0079]

[0080] Where, p i This represents the interference identification result of the i-th sensing node. Let represent the interference identification probability of the c-th type of interference pattern in the communication signal received by the i-th sensing node, n represents that the feature vector follows an n-dimensional normal distribution, μ represents the mean vector of the labeled interference features, Σ represents the covariance matrix of the labeled interference features, and α represents the total number of interference patterns.

[0081] Specifically, let n=4, and input the feature vector that follows a 4-dimensional normal distribution into the normal Bayes classifier for interference identification. The interference features follow a multi-dimensional normal distribution.

[0082] It is understandable that information about the interference signal can be extracted using the four interference features in the feature vector: the carrier factor coefficient, the ratio of the variance of the communication signal to the square of the average value of the communication signal, the flatness spectrum coefficient, and the third-order bias coefficient. These interference features can then be used to identify the interference signal. Furthermore, using interference features can reduce the input dimension of the classifier from 1 to 4, which helps to eliminate computational complexity.

[0083] In one embodiment, the interference identification results of each sensing node are input into the fusion center, and all interference identification results are processed according to soft decision cooperative sensing to obtain the final interference identification result, including:

[0084] The interference identification results from each sensing node are input into the fusion center. Based on soft-decision cooperative sensing, all interference identification results are aggregated to obtain the total interference identification probability, denoted as:

[0085]

[0086] The total interference identification probability is calculated to obtain the final interference identification result, which is expressed as follows:

[0087]

[0088] Where, p FC (c) represents the interference recognition probability of the c-th type of interference pattern.

[0089] It is understandable that the fusion center processes the interference identification results of all sensing nodes based on soft decision-making collaborative perception to obtain the final interference identification result. This fully utilizes the identification information from different sensing nodes, reduces the impact of identification errors of individual sensing nodes on the identification result, and avoids quantification errors by not quantifying the interference identification results when processing all interference identification results based on soft decision-making collaborative perception. This further improves the accuracy of the final interference identification result.

[0090] In one embodiment, the interference identification results of each sensing node by the fusion center are quantized to obtain the quantization result, including:

[0091] Based on the interference identification probabilities of each interference pattern in the interference identification results of the i-th sensing node obtained by the fusion center, the interference identification probabilities are sorted by magnitude, and the highest interference identification probability is selected for quantization to obtain the interference identification index of the i-th sensing node, denoted as .

[0092]

[0093] N c The interference identification index of each sensing node is summarized to obtain the quantization result, which is represented as:

[0094] It is understandable that the purpose of quantifying the interference identification results of each sensing node is to provide high-confidence labels for the self-training process of the classifier in the sensing node, and to feed back the final interference identification results with high accuracy and high-confidence labels to the sensing node through the fusion center to guide the classifier to perform high-quality self-training.

[0095] In one embodiment, labels are assigned to the interference samples to be labeled based on the quantization results. The final interference identification results and labels are then input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a newly trained classifier, including:

[0096] Decision conditions are set based on the quantification results, and the decision conditions are expressed as follows:

[0097]

[0098] in, Indicates that the quantization result η is equal to Quantity, N represents the value that appears most frequently in the quantization result η. th Represents the threshold, and N th ≤N c , Indicates a label, Indicates the interference sample to be labeled;

[0099] When the result of the judgment condition is When a positive label is assigned to the interference sample to be labeled, it is represented as... Otherwise, assign a negative label to the interference sample to be labeled, denoted as

[0100] The final interference identification result and label are input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a well-trained new classifier.

[0101] In one embodiment, the final interference identification result and label are input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a trained new classifier, including:

[0102] The final interference identification result and label are input into the perception node through the fusion center. The perception node stores the interference features of the interference samples to be labeled corresponding to the positive label, and obtains the interference feature set. The classifier is trained based on the interference feature set to obtain a new trained classifier.

[0103] In a specific embodiment, the process of self-training of the fusion center-assisted classifier is shown in Table 1. The brackets {} in Table 1 represent the detailed steps of the sample labeling process and the self-training process in the perception node.

[0104] Table 1. The self-training process of the fusion center auxiliary classifier.

[0105]

[0106] It is understandable that assigning positive and negative labels is to filter out interference samples that can be used for training. A positive label indicates that the sample can be used for training, while a negative label indicates that the sample cannot be used for training. By assigning positive and negative labels, a high-confidence label is provided for the self-training of the classifier, avoiding the need for manual labeling of a large number of training samples and reducing the cost of sample labeling.

[0107] In one embodiment, before configuring the sensing node to receive communication signals, the method further includes:

[0108] The sensing nodes are initialized based on the covariance matrix and average value of each interference pattern in the communication signal. The initialized sensing nodes are then set up around the fusion center to receive communication signals.

[0109] It should be understood that, although Figure 2 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 2 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0110] To further illustrate the beneficial effects of the interference identification method based on soft decision-making multi-node cooperative perception proposed in this invention, simulation experiments were conducted for verification. Specifically, in the simulation verification, it was assumed that the received communication signal had already undergone demodulation and filtering processing, and the communication signal was sampled. The sampling time was set to T = 2 seconds, and the sampling frequency f was... s =1000Hz, meaning there are M = 2000 samples of the received communication signal. Let K be the number of unlabeled samples corresponding to each interference pattern. s This means that the number of samples for each interference pattern is the same, and there are a total of six interference patterns, which is the total number of unlabeled samples T. s =6K s The parameters of the interference patterns in the communication signals are shown in Table 2.

[0111] Table 2 Parameters of Interference Patterns

[0112]

[0113] In simulation experiments, such as Figure 3 As shown, the average recognition accuracy of the proposed method under different labeled samples was first studied when the number of sensing nodes was 4. Figure 3It can be seen that by comparing the results under different labeled samples, i.e. K=10, 30 and 100, it can be seen that as the number of labeled samples used for training increases, the average accuracy of the interference recognition method based on soft decision multi-node cooperative perception proposed in this invention becomes very obvious. This can be attributed to the fact that the probability (soft information) calculated by the normal Bayes classifier becomes more reliable as the number of labeled samples increases.

[0114] like Figure 4 As shown, with a training sample size of 10, the average recognition accuracy of the proposed method (i.e., the self-learning method based on soft fusion FC assistance in the figure) was compared with that of hard fusion, soft fusion, self-learning based on hard fusion FC assistance, and node self-learning methods. First, it can be observed that both the self-training schemes based on soft fusion FC assistance and hard fusion FC assistance can significantly improve the average recognition accuracy. This is because self-training utilizes the latent information in unlabeled samples, and the proposed method has a higher average recognition accuracy due to reduced quantization errors. Second, the FC-assisted Bayesian classifier has a better average accuracy than the node-based self-training version because FC-assisted self-training can reduce false labels through multi-node collaborative perception. Third, compared to the version without self-training, the proposed method can improve the average recognition accuracy through self-training, which can be explained by the improved reliability of soft information through self-training.

[0115] In one embodiment, an interference identification device based on soft-decision multi-node cooperative perception is provided, the device comprising:

[0116] The interference identification module is used to set up the sensing node to receive communication signals, and to identify interference in the communication signals according to the classifier in the sensing node to obtain the interference identification result; wherein, the interference identification result includes the interference identification probability of each interference pattern in the communication signal.

[0117] It is understandable that deploying multiple sensing nodes around a legitimate receiver and considering the cooperative sensing among multiple sensing nodes can help identify interference. The classifiers in the sensing nodes are generally Bayesian classifiers, including normal Bayesian classifiers and naive Bayesian classifiers. Bayesian classifiers use causal reasoning to calculate the interference identification probability of each interference pattern in the communication signal.

[0118] The cooperative perception module is used to input the interference identification results of each perception node into the fusion center, and process all interference identification results according to the soft decision cooperative perception to obtain the final interference identification result.

[0119] It is understandable that the fusion center can also be considered a legitimate receiver. The fusion center receives interference identification data from all sensing nodes, fuses it, and provides the final identification result. Soft decision cooperative sensing is a multi-sensing method. In soft decision, each sensing node sends the interference identification probability of each interference pattern in the interference identification result to the fusion center. After receiving the interference identification results from all sensing nodes, the fusion center provides the final interference identification result by making full use of the interference identification information from different sensing nodes.

[0120] The self-training module is used to quantify the interference identification results of each sensing node based on the fusion center, obtain the quantization results, assign labels to the interference samples to be labeled based on the quantization results, and input the final interference identification results and labels into the sensing nodes through the fusion center to perform self-training of the classifier and obtain a trained new classifier.

[0121] It is understandable that before the classifier self-trains, the fusion center needs to quantify the interference identification results of each sensing node, assign high-confidence labels to the interference samples to be labeled based on the quantification results, and finally feed back the final interference identification results and labels as feedback information to the sensing nodes to guide the classifiers in the sensing nodes to self-train, thereby identifying new classifiers with higher accuracy.

[0122] Specific limitations regarding the interference identification device based on soft-decision multi-node cooperative perception can be found in the limitations of the interference identification method based on soft-decision multi-node cooperative perception described above, and will not be repeated here. Each module in the aforementioned interference identification device based on soft-decision multi-node cooperative perception can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0124] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for interference identification based on soft-decision multi-node cooperative sensing, characterized in that, The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: According to the fusion center, the interference identification probability of each interference pattern in the interference identification result of the first i perception node is sorted by size, the maximum interference identification probability is selected for quantization processing, and the interference identification index of the first i perception node is obtained, which is represented as: ； Will The interference identification index of each sensing node is summarized to obtain the quantization result, which is represented as follows: ; The method comprises: The method comprises: ； wherein, denotes the quantization result is equal to the number of denotes the value that occurs most frequently in the quantization result denotes a threshold value, and , denotes a label, denotes an interference sample to be labeled; When the result of the decision condition is , a positive label is assigned to the to-be-labeled interference sample, denoted as ; otherwise, a negative label is assigned to the to-be-labeled interference sample, denoted as . The method comprises:

2. The method of claim 1, wherein, The method comprises: The method comprises: The method comprises: The method comprises:

3. The method of claim 2, wherein, The method comprises: Setting A perception node receives a communication signal, denoted as: ； wherein, represents a communication signal received by the i th perception node, , represents a channel matrix, represents a channel coefficient, represents an interference signal, represents white Gaussian noise; The communication signal is normalized to obtain a normalized signal, and the normalized signal is subjected to fast Fourier transform to obtain a single-side spectrum signal .

4. The method of claim 2, wherein, The method comprises: The method comprises: ； wherein, denotes a time slot, denotes the carrier factor coefficient, denotes the ratio of the variance to the square of the mean of the communication signal, denotes the flatness spectrum coefficient, denotes the 3rd order offset coefficient, T denotes a transposition of the feature vector; The method comprises: ； ； in, Indicates the first i Interference identification results of individual sensing nodes Indicates the first i The interference identification probability of the c-th type of interference pattern in the communication signals received by the sensing nodes, where n represents that the feature vector follows an n-dimensional normal distribution. This represents the mean vector of the labeled interference features. This represents the covariance matrix of the labeled interference features. This represents the total number of interference patterns.

5. The method of claim 1, wherein, The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: The method comprises: ； The total interference identification probability is calculated to obtain the final interference identification result, which is expressed as follows: ； wherein, denotes the interference recognition probability of the cth interference pattern.

6. The method of claim 1, wherein, The final interference identification result and the label are input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a trained new classifier, including: The final interference identification result and the label are input into the sensing node through the fusion center. The interference features of the interference samples to be labeled corresponding to the positive label are stored to obtain the interference feature set. The classifier is trained according to the interference feature set to obtain a trained new classifier.

7. The method of claim 1, wherein, Before configuring the sensing node to receive communication signals, the following steps are also included: The sensing node is initialized based on the covariance matrix and average value of each interference pattern in the communication signal to obtain an initialized sensing node, which is then set up around the fusion center to receive the communication signal.

8. An interference identification device based on soft-decision multi-node cooperative sensing, characterized in that, The device includes: An interference identification module is configured to set up a sensing node to receive a communication signal, and to perform interference identification on the communication signal according to a classifier in the sensing node to obtain an interference identification result; wherein, the interference identification result includes the interference identification probability of each interference pattern in the communication signal; The cooperative perception module is used to input the interference identification results of each perception node into the fusion center, and process all the interference identification results according to the soft decision cooperative perception to obtain the final interference identification result. The self-training module is used to quantize the interference identification results of each sensing node according to the fusion center to obtain the quantization result, assign a label to the interference sample to be labeled according to the quantization result, and input the final interference identification result and the label into the sensing node through the fusion center to perform self-training of the classifier to obtain a trained new classifier. The quantization results, obtained by quantizing the interference identification results of each sensing node based on the fusion center, include: According to the fusion center, the interference identification probability of each interference pattern in the interference identification result of the first i perception node is sorted by size, the maximum interference identification probability is selected for quantization processing, and the interference identification index of the first i perception node is obtained, which is represented as: ； Will The interference identification index of each sensing node is summarized to obtain the quantization result, which is represented as follows: ; Specifically, the process involves assigning labels to the interference samples to be labeled based on the quantization results, inputting the final interference identification results and the labels into the sensing node through the fusion center, and performing self-training of the classifier to obtain a trained new classifier, including: Based on the quantification results, decision conditions are set, and the decision conditions are expressed as follows: ； wherein, denotes the quantization result is equal to the number of denotes the value that occurs most frequently in the quantization result denotes a threshold value, and , , denotes a label, denotes an interference sample to be labeled; When the result of the decision condition is , a positive label is assigned to the to-be-labeled interference sample, denoted as ; otherwise, a negative label is assigned to the to-be-labeled interference sample, denoted as . The final interference identification result and the label are input into the sensing node through the fusion center to perform self-training of the classifier, resulting in a trained new classifier.

Images