Method and device for detecting interference signal

Abstract

The embodiment of the invention provides a method and device for detecting an interference signal, and the method comprises the steps: processing the one-dimensional signal energy data of a to-be-detected signal, and obtaining the two-dimensional image data of the to-be-detected signal; reconstructing the two-dimensional image data of the to-be-detected signal to obtain reconstructed image data of the to-be-detected signal; and determining whether the to-be-detected signal is an interference signal according to the two-dimensional image data of the to-be-detected signal and the reconstructed image data of the to-be-detected signal. According to the invention, the problem that the accuracy of interference signal detection is low is solved, and the effect of improving the accuracy of interference signal detection is achieved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of communications, and more particularly, to a method and apparatus for detecting interference signals. Background Art

[0002] With the rapid development of satellite network technology, a large-scale deployment of a giant constellation satellite network with a wide coverage range and high communication rate is underway. Currently, tens of thousands of satellites are competing for limited orbital resources and spectrum resources. However, wireless communication spectrum resources are limited and scarce, and a large number of Non-Geostationary-Orbit (NGSO) satellite systems and Geostationary-Orbit (GSO) satellite systems use the same frequency band for communication. With the continuous increase in the number of NGSO satellites, the problem of downlink communication interference between them and GSO systems has become increasingly prominent.

[0003] Currently, detecting interference signals between NGSO and GSO relies on manual experience, which is highly subjective and heavily dependent on experience, resulting in a low accuracy rate for detecting interference signals.

[0004] In view of the above problems, there is currently no effective solution. Summary of the Invention

[0005] The embodiments of the present invention provide a method and apparatus for detecting interference signals to at least solve the problem of low accuracy rate of interference signals in the related art.

[0006] According to an embodiment of the present invention, there is provided a method for detecting interference signals, which is applied to detect interference signals in a satellite network, including: processing one-dimensional signal energy data of a signal to be detected to obtain two-dimensional image data of the signal to be detected; reconstructing the two-dimensional image data of the signal to be detected to obtain reconstructed image data of the signal to be detected; and determining whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected.

[0007] In an exemplary embodiment, determining whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected includes: determining a reconstruction error of the signal to be detected according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected; and determining whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and a target reconstruction error threshold.

[0008] In an exemplary embodiment, before determining whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and the target reconstruction error threshold, the method further includes: determining the reconstruction error between the two-dimensional image data of N first normal signals and the reconstructed image data of the first normal signals, obtaining N first reconstruction errors, where the first normal signals are normal communication signals of the NGSO or normal communication signals of the GSO, and N is an integer; determining the target reconstruction error threshold according to the N first reconstruction errors.

[0009] In an exemplary embodiment, determining the target reconstruction error threshold according to the N first reconstruction errors includes: determining the target reconstruction error threshold according to the mean of the N first reconstruction errors and the variance of the N first reconstruction errors.

[0010] In an exemplary embodiment, determining the reconstruction error between the two-dimensional image data of N first normal signals and the reconstructed image data of the first normal signals includes: inputting the two-dimensional image data of a first target normal signal into a target autoencoder, encoding the two-dimensional image data of the first target normal signal through an encoding layer of the target autoencoder to obtain an encoded feature map of the first target normal signal, where the N first normal signals include the first target normal signal; decoding the encoded feature map of the first target normal signal through a decoding layer of the target autoencoder to obtain the reconstructed image data of the first target normal signal; determining the reconstruction error of the first target normal signal according to the two-dimensional image data of the first target normal signal and the reconstructed image data of the first target normal signal.

[0011] In an exemplary embodiment, reconstructing the two-dimensional image data of the signal to be detected to obtain the reconstructed image data of the signal to be detected includes: inputting the two-dimensional image data of the signal to be detected into a target autoencoder, encoding the two-dimensional image data of the signal to be detected through an encoding layer of the target autoencoder to obtain an encoded feature map of the signal to be detected; decoding the encoded feature map of the signal to be detected through a decoding layer of the target autoencoder to obtain the reconstructed image data of the signal to be detected.

[0012] In an exemplary embodiment, the method further includes: training an autoencoder using a training sample data set, where the training sample data set includes a training sample data set and a test sample data set, the training sample data set includes a plurality of training sample data, the training sample data includes two-dimensional image data of a normal signal training sample, the test sample data set includes a plurality of test sample data, and the test sample data includes two-dimensional image data of an interference signal test sample in which a first system and a second system are at different angles; ending the training to obtain the target autoencoder when the test result of the test sample data set by the autoencoder meets a preset condition.

[0013] In an exemplary embodiment, the method further includes: inputting target test sample data into the autoencoder obtained by the i-th round of training to obtain reconstructed image data of the target test sample data, where the test sample data set includes the target test sample data, and i is an integer; determining a reconstruction error between the target test sample data and the reconstructed image data of the target test sample data to obtain the reconstruction error of the target test sample; and determining the test result of the target test sample data according to the relationship between the reconstruction error of the target test sample and the reconstruction error threshold obtained by the i-th round of training.

[0014] In an exemplary embodiment, before determining the test result of the target test sample data according to the relationship between the reconstruction error of the target test sample and the reconstruction error threshold obtained by the i-th round of training, the method further includes: inputting two-dimensional image data of M second normal signals into the autoencoder obtained by the i-th round of training to obtain reconstructed image data of the M second normal signals, where the second normal signal is a normal communication signal and M is an integer; determining a reconstruction error between the two-dimensional image data of the M second normal signals and the reconstructed image data of the second normal signals to obtain M second reconstruction errors; and determining the reconstruction error threshold obtained by the i-th round of training according to the M second reconstruction errors, where if the i-th round of training is the last round of training of the autoencoder, the reconstruction error threshold obtained by the i-th round of training is the target reconstruction error threshold.

[0015] In an exemplary embodiment, the first system is a non-geostationary orbit satellite (NGSO) system, the second system is a geostationary orbit satellite (GSO) system, and the interference signal is an interference signal between the non-geostationary orbit satellite (NGSO) system and the geostationary orbit satellite (GSO) system.

[0016] According to another embodiment of the present invention, there is provided a device for detecting interference signals, which is applied to detect interference signals in a satellite network, including: a processing module for processing one-dimensional signal energy data of a signal to be detected to obtain two-dimensional image data of the signal to be detected; a reconstruction module for reconstructing the two-dimensional image data of the signal to be detected to obtain reconstructed image data of the signal to be detected; and a determination module for determining whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected.

[0017] According to still another embodiment of the present invention, there is also provided a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0018] According to still another embodiment of the present invention, there is also provided an electronic device including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0019] According to still another embodiment of the present invention, there is also provided a computer program product including a computer program, wherein when the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0020] Through the present invention, since the one-dimensional signal energy data of the signal to be detected is processed to obtain the two-dimensional image data of the signal to be detected; the two-dimensional image data of the signal to be detected is reconstructed to obtain the reconstructed image data of the signal to be detected; and it is determined whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected. Therefore, the problem of low accuracy in detecting interference signals in the related art can be solved, and the effect of improving the accuracy of detecting interference signals can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a hardware structure block diagram of a mobile terminal of a method for detecting interference signals according to an embodiment of the present invention;

[0022] Figure 2 is a flowchart of a method for detecting interference signals according to an embodiment of the present invention;

[0023] Figure 3 is a schematic diagram of visualizing time series data according to an embodiment of the present invention;

[0024] Figure 4 is a schematic diagram of a core module according to an embodiment of the present invention;

[0025] Figure 5Schematic diagram of a satellite constellation interference scenario according to an embodiment of the present invention;

[0026] Figure 6 Experimental results at different included angles between the GSO and the ground station antenna according to an embodiment of the present invention;

[0027] Figure 7 Structural block diagram of a device for detecting interference signals according to an embodiment of the present invention. Specific embodiments

[0028] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0029] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence.

[0030] The method embodiments provided in the embodiments of the present application can be executed on a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 Hardware structural block diagram of a mobile terminal of a method for detecting interference signals according to an embodiment of the present invention. As Figure 1 shown, the mobile terminal may include one or more ( Figure 1 only one is shown in the figure) processors 102 (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Among them, the above-mentioned mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those of ordinary skill in the art can understand that Figure 1 the structure shown is only schematic and does not limit the structure of the above-mentioned mobile terminal. For example, the mobile terminal may further include more or fewer components than Figure 1 shown in the figure, or have a different configuration from Figure 1 shown in the figure.

[0031] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for detecting interference signals in the embodiments of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implements the above-mentioned method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some instances, the memory 104 may further include a memory remotely disposed relative to the processor 102, and these remote memories can be connected to the mobile terminal through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0032] The transmission device 106 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by a communication provider of the mobile terminal. In one instance, the transmission device 106 includes a network adapter (Network Interface Controller, abbreviated as NIC), which can be connected to other network devices through a base station and thus communicate with the Internet. In one instance, the transmission device 106 can be a radio frequency (Radio Frequency, abbreviated as RF) module, which is used to communicate with the Internet wirelessly.

[0033] In this embodiment, a method for detecting interference signals running on the above-mentioned mobile terminal is provided, which is applied to detecting interference signals in a satellite network. Figure 2 It is a flowchart of the method for detecting interference signals according to the embodiments of the present invention, as Figure 2 shown, and the process includes the following steps:

[0034] Step S202, process the one-dimensional signal energy data of the signal to be detected to obtain two-dimensional image data of the signal to be detected;

[0035] Among them, the above-mentioned signal to be detected can be a normal signal of NGSO, or a normal signal of GSO, or an interference signal between NGSO and GSO.

[0036] The signal preprocessing module converts the I / Q data of the signal to be detected received through the following formula into one-dimensional energy data:

[0037] ;

[0038] Among them, is the I / Q data of the signal to be detected, is the one-dimensional signal energy data of the signal to be detected.

[0039] The data imaging module converts the one-dimensional energy data of the signal to be detected into an image in the manner of Figure 3 . By converting the time-series one-dimensional energy data into two-dimensional image data, the spatio-temporal distribution characteristics of the signal energy can be increased. The processed two-dimensional image has one channel, similar to a grayscale image.

[0040] Step S204: Reconstruct the two-dimensional image data of the signal to be detected to obtain the reconstructed image data of the signal to be detected;

[0041] Specifically, input the two-dimensional image data of the signal to be detected into the target autoencoder. Encode the two-dimensional image data of the signal to be detected through the encoding layer of the target autoencoder to obtain the encoded feature map of the signal to be detected; decode the encoded feature map of the signal to be detected through the decoding layer of the target autoencoder to obtain the reconstructed image data of the signal to be detected.

[0042] Among them, the above-mentioned target autoencoder is a two-dimensional convolutional autoencoder, and the target autoencoder is a trained two-dimensional convolutional autoencoder that has been trained.

[0043] Step S206: Determine whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected.

[0044] The above-mentioned interference signal is an interference signal between a non-geostationary orbit satellite (NGSO) and a geostationary orbit satellite (GSO).

[0045] Specifically, determine the reconstruction error of the signal to be detected according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected; determine whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and the target reconstruction error threshold.

[0046] The calculation formula of the reconstruction error is as follows:

[0047] ;

[0048] Among them, is the two-dimensional image data of the signal to be detected, is the reconstruction error of the signal to be detected.

[0049] When the reconstruction error of the signal to be detected is greater than or equal to the target reconstruction error threshold, the signal to be detected is an interference signal; when the reconstruction error of the signal to be detected is less than the target reconstruction error threshold, the signal to be detected is a normal signal.

[0050] Optionally, the execution entity of the above steps may be a background processor, or other devices with similar processing capabilities, or may also be a machine integrated with at least an image acquisition device and a data processing device. Among them, the image acquisition device may include a graphic acquisition module such as a camera, and the data processing device may include terminals such as a computer and a mobile phone, but is not limited thereto.

[0051] Through the above steps, the signal to be detected is first converted into one-dimensional signal energy data by the signal preprocessing module, and then through the data imaging module, the one-dimensional signal energy data is converted into two-dimensional image data. Then, the two-dimensional image data passes through the encoding layer and the decoding layer of the two-dimensional convolutional autoencoder to obtain the reconstructed image data of the signal to be detected, and the reconstruction error between the two-dimensional image data of the signal to be detected and the reconstructed image restored by the two-dimensional convolutional autoencoder is calculated. The reconstruction error is compared with the target reconstruction error threshold. When the reconstruction error is greater than or equal to the target reconstruction error threshold, the signal to be detected is an interference signal. When the reconstruction error is less than the target reconstruction error threshold, the signal to be detected is a normal signal. Since the two-dimensional convolutional autoencoder can autonomously learn the difference between normal signals and interference signals, automatically learn the reconstruction error threshold, and better conform to the variation law of normal signals and signals, the accuracy of detecting interference signals is higher.

[0052] In an exemplary embodiment, before determining whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and the target reconstruction error threshold, the method further includes:

[0053] Determine the reconstruction error between the two-dimensional image data of N first normal signals and the reconstructed image data of the first normal signals to obtain N first reconstruction errors, where the first normal signals are normal communication signals, and N is an integer; determine the target reconstruction error threshold according to the N first reconstruction errors.

[0054] The first normal signal is the normal communication signal of the NGSO or the normal communication signal of the GSO.

[0055] Taking the above first normal signal as the normal signal of the NGSO as an example, the specific value of N can be determined according to the actual situation.

[0056] The reconstruction error of any one of the N first normal signals (the first target normal signal is any one of the N first normal signals) can be obtained in the following manner:

[0057] Input the two-dimensional image data of the first target normal signal into the target autoencoder, and encode the two-dimensional image data of the first target normal signal through the encoding layer of the target autoencoder to obtain the encoded feature map of the first target normal signal, where the N first normal signals include the first target normal signal; decode the encoded feature map of the first target normal signal through the decoding layer of the target autoencoder to obtain the reconstructed image data of the first target normal signal; determine the reconstruction error of the first target normal signal according to the two-dimensional image data of the first target normal signal and the reconstructed image data of the first target normal signal.

[0058] In an exemplary embodiment, determining the target reconstruction error threshold according to the N first reconstruction errors includes: determining the target reconstruction error threshold according to the mean of the N first reconstruction errors and the variance of the N first reconstruction errors.

[0059] ;

[0060] where is the j-th first reconstruction error, k is a preset weight value, and the value of k can be adjusted according to actual situations, for example, adjusted according to the distribution of interference signals.

[0061] The above target autoencoder is obtained by training an autoencoder. Specifically, the autoencoder is trained using a training sample dataset, where the training sample dataset includes: a training sample dataset and a test sample dataset. The training sample dataset includes multiple training sample data, and the training sample data includes: two-dimensional image data of normal signal training samples. The test sample dataset includes multiple test sample data, and the test sample data includes: two-dimensional image data of interference signal test samples when the first system and the second system are at different angles; end the training to obtain the target autoencoder when the test results of the autoencoder for the test sample dataset meet the preset conditions.

[0062] The above first system is a non-geostationary orbit satellite (NGSO) system, and the second system is a geostationary orbit satellite (GSO) system.

[0063] The above preset conditions include but are not limited to that in the test results of the test sample dataset, the test results of more than a preset proportion (which can be determined according to actual situations, such as 50%) of the test sample data are accurate. For example, assume that the test sample dataset includes 200 two-dimensional image data of interference signal test samples. If the results output by the trained autoencoder indicate that the results of more than 100 interference signal test samples are accurate, then end the training to obtain the target autoencoder.

[0064] Taking the i-th round of training in the training process of the autoencoder as an example (the i-th round can be any round of training), the 0-th round is the untrained initial autoencoder, and the target autoencoder is obtained after the last round of training. The target test sample data is any test sample data in the test sample dataset.

[0065] Input the target test sample data into the autoencoder obtained from the i-th round of training to obtain the reconstructed image data of the target test sample data. The test sample dataset includes the target test sample data, where i is an integer; determine the reconstruction error between the target test sample data and the reconstructed image data of the target test sample data to obtain the reconstruction error of the target test sample; determine the test result of the target test sample data according to the relationship between the reconstruction error of the target test sample and the reconstruction error threshold obtained from the i-th round of training.

[0066] In an exemplary embodiment, the reconstruction error threshold obtained from the above-mentioned i-th round of training is obtained in the following manner:

[0067] Input the two-dimensional image data of M second normal signals into the autoencoder obtained from the i-th round of training to obtain the reconstructed image data of the M second normal signals. The second normal signal is a normal communication signal, and M is an integer; determine the reconstruction error between the two-dimensional image data of the M second normal signals and the reconstructed image data of the second normal signals to obtain M second reconstruction errors; determine the reconstruction error threshold obtained from the i-th round of training according to the M second reconstruction errors. If the i-th round of training is the last round of training of the autoencoder, the reconstruction error threshold obtained from the i-th round of training is the target reconstruction error threshold.

[0068] The second normal signal is the normal communication signal of the NGSO or the normal communication signal of the GSO.

[0069] Determining the reconstruction error threshold obtained from the i-th round of training according to the M second reconstruction errors includes: determining the reconstruction error threshold obtained from the i-th round of training according to the mean of the M second reconstruction errors and the variance of the M second reconstruction errors. The above-mentioned second normal signal can be the normal signal of the NGSO or the GSO. The above-mentioned first normal signal and the above-mentioned second normal signal can be the same normal signal or different normal signals.

[0070] In the above embodiments, by converting the temporal energy data of signals (including signals to be detected and training sample signals) into two-dimensional image data, the spatio-temporal distribution law of signal energy is highlighted, and a two-dimensional convolutional autoencoder is designed to learn the spatio-temporal correlation features of signal energy fluctuations. Based on the reconstruction error of the normal signal by the two-dimensional autoencoder, a reconstruction error threshold is defined to effectively detect interference signals in multiple scenarios.

[0071] As Figure 4 shown is a schematic diagram of the core module, including a signal preprocessing module, a data imaging module, a feature learning module, and an interference detection module.

[0072] First, the signal preprocessing module converts the I / Q data of the received signal (including the signal to be detected or the training sample signal, the signal during the training process is the training sample signal, and the signal during the detection process is the signal to be detected) into one-dimensional energy data through the following formula, which can highlight the energy change of the signal more prominently.

[0073] ;

[0074] The data imaging module converts the one-dimensional energy data into an image form in the way of Figure 3 so that it can be processed by the two-dimensional convolutional autoencoder. By converting the temporal one-dimensional energy data into two-dimensional image data, the spatio-temporal distribution characteristics of signal energy can be increased, and then the spatio-temporal high-dimensional features of signal energy data can be extracted by combining with the subsequent two-dimensional convolutional autoencoder. The processed two-dimensional image has one channel, similar to a grayscale image, and is input into the two-dimensional convolutional autoencoder for feature learning.

[0075] The feature learning module mainly includes a two-dimensional convolutional autoencoder, and its structure is shown in Table 1.

[0076] Table 1

[0077]

[0078] The feature learning of the two-dimensional convolutional autoencoder is the training stage, which mainly includes three steps to achieve feature learning, namely the input signal encoding process, the feature decoding process, and the reconstruction loss calculation process.

[0079] First, the two-dimensional image data processed by the data imaging module (the two-dimensional image data of the training sample signal, that is, the two-dimensional image data of the normal signal training sample) is input into the encoding layer of the two-dimensional convolutional autoencoder. The encoding layer consists of multiple convolutional layers, activation function layers, and max pooling layers. Through the processing of the encoding layer, the dimensionality reduction and spatio-temporal feature mapping of the two-dimensional image data can be achieved, and the features include the spatio-temporal fluctuation law of signal energy. The encoding layer of the two-dimensional convolutional autoencoder learns the non-linear function for the input two-dimensional image data Perform feature mapping as shown in the following formula:

[0080] ;

[0081] Wherein, is the feature vector learned by the encoding layer, performs non-linear transformation by the activation function, W1 is the weight of the convolutional neurons in the encoding layer, and b1 is the bias of the convolutional neurons in the encoding layer.

[0082] For example, during the encoding process, the two-dimensional image data is first input into the first convolutional layer to obtain intermediate features of size (8, 28, 28), where 8 is the number of channels, and 28 is the length and width of the intermediate feature map. The first convolutional layer performs preliminary feature extraction and data dimensionality reduction on the spatial information in the two-dimensional image. Then, this intermediate feature is further input into the ReLU activation function layer, which can introduce non-linear factors into the feature mapping, enabling the model to fit the non-linear problem of signal energy feature extraction. Next, the non-linear feature map is input into the max pooling layer for feature screening and dimensionality reduction, obtaining an intermediate feature map of size (8, 14, 14). Furthermore, the intermediate feature map of size (8, 14, 14) is input into the second convolutional layer to further extract the feature map, obtaining an intermediate feature map of size (16, 14, 14), and then input into the ReLU activation function layer and the max pooling layer, obtaining an intermediate feature map of size (16, 7, 7). This intermediate feature map is then input into the third convolutional layer and the ReLU activation function layer to further extract spatio-temporal high-dimensional features. Finally, the intermediate feature map of size (16, 7, 7) is input into the fourth convolutional layer, the ReLU activation function layer, and the max pooling layer, obtaining a high-dimensional feature map of size (16, 3, 3) containing the spatio-temporal attributes of the signal.

[0083] The high-dimensional feature map of size (16, 3, 3) is the signal feature map learned by the two-dimensional convolutional autoencoder. A larger feature map can obtain richer features of the signal, but it will also lead to redundancy of features. A smaller feature map can obtain more accurate signal features, but it will lead to a weakening of the independence of different features. During the interference detection process of the downlink of low-earth orbit communication satellites, the feature map is set to (16, 3, 3), which can effectively learn the energy features of the signal.

[0084] Furthermore, the feature map containing the high-dimensional spatio-temporal key attributes is input into the decoding layer of the two-dimensional convolutional autoencoder. The decoding layer contains multiple deconvolutional layers and activation function layers. The key role of the decoding layer is to accurately restore the signal high-dimensional feature map to two-dimensional image data consistent with the input data. Therefore, the structure of the decoding layer is almost exactly the opposite of that of the encoding layer. The decoding layer of the two-dimensional convolutional autoencoder passes through the non-linear function for the feature vector Data restoration is performed as shown in the following formula:

[0085] ;

[0086] Among them, is the two-dimensional reconstructed image data output by the decoding layer, Z is the feature vector learned by the encoding layer, the activation function of the decoding layer implements a non-linear transformation, W2 is the weight of the convolutional neurons in the decoding layer, and b2 is the bias of the convolutional neurons in the decoding layer.

[0087] During the decoding process, a high-dimensional feature map of size (16, 3, 3) is input into the first deconvolution layer. Through the deconvolution operation, the decoding layer restores the feature map of size (16, 3, 3) to an intermediate image of size (8, 6, 6) according to the spatio-temporal correlation in the high-dimensional feature map. Then, the intermediate image is also input into the ReLU activation function layer to introduce a non-linear transformation during the decoding process and restore the non-linear change between the data and the feature map. Then, through the second deconvolution layer and the third deconvolution layer, the spatio-temporal correlation between the image data is further restored to obtain image data of size (4, 24, 24). Finally, the image data of size (4, 24, 24) is input into the fourth deconvolution layer for the final data restoration, and at the same time, the number of channels of the image is restored to 1, and the length and width are restored to 28 to obtain a reconstructed image consistent with the input image. During the transformation process of the fourth deconvolution layer, there is no activation function layer, and the purpose is to complete the linear mapping from the feature vector to the input two-dimensional image data to approximate the input data as much as possible.

[0088] Finally, in the reconstruction error calculation process, the reconstruction error between the input two-dimensional image data and the two-dimensional reconstructed image restored by the two-dimensional convolutional autoencoder through the encoding layer and the decoding layer is used as the loss function of the two-dimensional convolutional autoencoder. Stochastic gradient descent and backpropagation are used to drive the training of the two-dimensional convolutional autoencoder, so that the two-dimensional convolutional autoencoder learns the mapping process from the input data to the feature map and the restoration process from the feature map to the two-dimensional image data, thereby obtaining the learning ability of the spatio-temporal characteristics of the signal energy in the two-dimensional image data. In the reconstruction error calculation process, the mean square error (MSE) is used for calculation. MSE measures the encoding and decoding ability of the two-dimensional convolutional autoencoder by calculating the mean of the sum of the squared differences between the input two-dimensional image data and the two-dimensional reconstructed image restored by the two-dimensional convolutional autoencoder. The calculation formula of the reconstruction error is as follows:

[0089] ;

[0090] The interference detection module detects interference data based on a trained two-dimensional convolutional autoencoder. Since only normal communication signals without interference are used for model training during the feature learning training process of the two-dimensional convolutional autoencoder, the two-dimensional convolutional autoencoder can better learn the energy spatio-temporal fluctuation characteristics of normal communication signals and accurately reconstruct the two-dimensional image data of normal communication signals. However, when an interference signal is input into the two-dimensional convolutional autoencoder, the signal feature vector extracted by the two-dimensional convolutional autoencoder is different from that of normal communication signals. Therefore, the two-dimensional image data of the interference signal cannot be accurately reconstructed during the decoding process, resulting in a large reconstruction error. Therefore, normal communication signals have a small reconstruction error, while interference signals have a large reconstruction error. Therefore, in the interference detection module, the decision threshold is calculated based on the reconstruction error of normal communication signals to determine whether there is an interference signal, as shown in the following formula:

[0091] ;

[0092] where is the j-th first reconstruction error, k is a preset weight value, and the value of k can be adjusted according to the actual situation, such as adjusted according to the distribution of interference signals.

[0093] In the interference detection module, the signal to be detected is first converted into one-dimensional signal energy data through the signal preprocessing module, and then through the data imaging module, the one-dimensional signal energy data is converted into two-dimensional image data. Then, the two-dimensional image is successively passed through the encoding layer and the decoding layer of the two-dimensional convolutional autoencoder in the feature learning module, and the MSE error between the input two-dimensional image data and the reconstructed image restored by the two-dimensional convolutional autoencoder is calculated. The MSE reconstruction error is compared with the threshold. When the MSE reconstruction error is greater than the threshold, the input data is an interference signal; when the MSE reconstruction error is less than the threshold, the input data is a normal signal.

[0094] In the embodiment of the present invention, a satellite constellation interference scenario is constructed using MATLAB, including one GSO satellite, 40 NGSO satellites, and one NGSO ground station, as Figure 5 shown, including NGSO satellite 501, GSO satellite 502, and NGSO ground station. Among them, the GSO and NGSO satellites enable satellite operation through Two-Line Element (TLE) files. The TLE file contains a set of orbital elements of earth-orbiting satellites and can be used to calculate the actual operating state of satellites. Both GSO and NGSO use the Ku band of 10.7 - 12.7 GHz for communication. Therefore, the interference situation between GSO and NGSO can be effectively simulated.

[0095] In the embodiments of the present invention, according to the above satellite constellation interference scenario, 1200 I / Q samples of NGSO normal communication signals are generated, and 200 signal samples in the case of NGSO and GSO interference are generated. Among them, 1000 NGSO normal communication signals are used as the training set based on time-series data visualization and two-dimensional convolutional autoencoder, 200 NGSO normal communication signals are used as the interference-free test set, and 200 NGSO and GSO interference signal samples are used as the interference test set.

[0096] First, the I / Q samples of 1000 NGSO normal communication signals are input into the signal preprocessing module to obtain 1000 one-dimensional signal energy data of NGSO, and the one-dimensional signal energy data is input into the data visualization module to obtain two-dimensional image data with a size of (1, 28, 28), and then input into the two-dimensional convolutional autoencoder for encoding in the encoding layer, feature extraction of the feature map, and restoration in the decoding layer, to obtain 1000 two-dimensional reconstructed image data of NGSO restored by the output of the two-dimensional convolutional autoencoder. Then, the reconstruction error between the 1000 original two-dimensional image data of NGSO and the 1000 restored two-dimensional reconstructed image data of NGSO is calculated. After 50 rounds of training, the trained two-dimensional convolutional autoencoder is obtained, and the interference decision threshold is calculated according to the reconstruction error. Finally, the interference-free test set of 200 NGSO normal communication signals and the interference test set of 200 NGSO and GSO are sequentially input into the signal preprocessing module, the data visualization module, and the feature learning module to obtain their reconstruction errors, and compare them with the interference decision threshold to obtain the test results. In addition, when the included angle between the GSO satellite and the NGSO ground station antenna pointing is different, the influence on the NGSO communication link is different. When the included angle is small, the interference of GSO to NGSO is greater. When the included angle is large, the GSO signal received by the NGSO ground station is weak, and at this time, the interference of GSO to NGSO is small. In the embodiments of the present invention, the data when GSO is used as an interference signal at different included angles is simulated as the interference test data set, and the range of the included angle between the GSO and the NGSO ground receiving station antenna is 0-20°. Figure 6 Showing the experimental results of the present invention and the prior art when the included angle of the NGSO ground station antenna pointing is different, it can be seen that the detection accuracy of the present invention is significantly higher than that of the interference detection method based on the fully connected autoencoder and the interference detection method based on the energy threshold when the interference signal is weak.

[0097] This application introduces the time-series data visualization method to transform the one-dimensional signal energy data into two-dimensional image data, highlighting the spatio-temporal characteristics of the energy fluctuation, which is convenient for the convolutional autoencoder to extract effective features and improve the interference detection accuracy. The two-dimensional convolutional autoencoder is introduced. By inputting the two-dimensional image data of the signal energy, it autonomously learns the spatio-temporal high-dimensional features of the energy, and uses the image reconstruction error as the interference decision threshold to improve the interference detection accuracy of the satellite downlink.

[0098] Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases, the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that makes a contribution to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to enable a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0099] In this embodiment, a device for detecting interference signals is also provided, which is applied to detect interference signals in a satellite network. The device is used to implement the above embodiments and preferred implementation methods, and those that have been described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0100] Figure 7 is a structural block diagram of the device for detecting interference signals according to an embodiment of the present invention. As Figure 5 shown, the device includes:

[0101] A processing module 72, configured to process the one-dimensional signal energy data of the signal to be detected to obtain two-dimensional image data of the signal to be detected;

[0102] A reconstruction module 74, configured to reconstruct the two-dimensional image data of the signal to be detected to obtain reconstructed image data of the signal to be detected;

[0103] A determination module 76, configured to determine whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected.

[0104] In an exemplary embodiment, the above device is further configured to determine a reconstruction error of the signal to be detected according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected; and determine whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and a target reconstruction error threshold.

[0105] In an exemplary embodiment, before determining whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and the target reconstruction error threshold, the device is further configured to determine the reconstruction error between the two-dimensional image data of N first normal signals and the reconstructed image data of the first normal signals, so as to obtain N first reconstruction errors, where the first normal signals are normal communication signals and N is an integer; and determine the target reconstruction error threshold according to the N first reconstruction errors.

[0106] In an exemplary embodiment, the device is further configured to determine the target reconstruction error threshold according to the mean value of the N first reconstruction errors and the variance of the N first reconstruction errors.

[0107] In an exemplary embodiment, the device is further configured to input the two-dimensional image data of a first target normal signal into a target autoencoder, encode the two-dimensional image data of the first target normal signal through an encoding layer of the target autoencoder to obtain an encoded feature map of the first target normal signal, where the N first normal signals include the first target normal signal; decode the encoded feature map of the first target normal signal through a decoding layer of the target autoencoder to obtain the reconstructed image data of the first target normal signal; and determine the reconstruction error of the first target normal signal according to the two-dimensional image data of the first target normal signal and the reconstructed image data of the first target normal signal.

[0108] In an exemplary embodiment, the device is further configured to input the two-dimensional image data of the signal to be detected into the target autoencoder, encode the two-dimensional image data of the signal to be detected through the encoding layer of the target autoencoder to obtain an encoded feature map of the signal to be detected; and decode the encoded feature map of the signal to be detected through the decoding layer of the target autoencoder to obtain the reconstructed image data of the signal to be detected.

[0109] In an exemplary embodiment, the device is further configured to train the autoencoder using a training sample data set, where the training sample data set includes: a training sample data set and a test sample data set, the training sample data set includes a plurality of training sample data, the training sample data includes: two-dimensional image data of a normal signal training sample, the test sample data set includes a plurality of test sample data, the test sample data includes: two-dimensional image data of an interference signal test sample when a first system and a second system are at different angles; and end the training to obtain the target autoencoder when the test result of the autoencoder for the test sample data set meets a preset condition.

[0110] In an exemplary embodiment, the above-mentioned device is further configured to input the target test sample data into the autoencoder obtained through the i-th round of training to obtain the reconstructed image data of the target test sample data, where the test sample data set includes the target test sample data, and i is an integer; determine the reconstruction error between the target test sample data and the reconstructed image data of the target test sample data to obtain the reconstruction error of the target test sample; and determine the test result of the target test sample data according to the relationship between the reconstruction error of the target test sample and the reconstruction error threshold obtained through the i-th round of training.

[0111] In an exemplary embodiment, before determining the test result of the target test sample data according to the relationship between the reconstruction error of the target test sample and the reconstruction error threshold obtained through the i-th round of training, the above-mentioned device is further configured to input the two-dimensional image data of M second normal signals into the autoencoder obtained through the i-th round of training to obtain the reconstructed image data of the M second normal signals, where the second normal signals are normal communication signals and M is an integer; determine the reconstruction error between the two-dimensional image data of the M second normal signals and the reconstructed image data of the second normal signals to obtain M second reconstruction errors; and determine the reconstruction error threshold obtained through the i-th round of training according to the M second reconstruction errors, where if the i-th round of training is the last round of training for the autoencoder, the reconstruction error threshold obtained through the i-th round of training is the target reconstruction error threshold.

[0112] In an exemplary embodiment, the first system is a non-geostationary orbit satellite (NGSO) system, the second system is a geostationary orbit satellite (GSO) system, and the interference signal is an interference signal between the non-geostationary orbit satellite (NGSO) system and the geostationary orbit satellite (GSO) system.

[0113] It should be noted that the above-mentioned various modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited thereto: the above-mentioned modules are all located in the same processor; or, the above-mentioned various modules are respectively located in different processors in any combination form.

[0114] An embodiment of the present invention further provides a computer-readable storage medium, in which a computer program is stored, where when the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0115] In an exemplary embodiment, the above-mentioned computer-readable storage medium may include, but is not limited to: various media that can store computer programs such as USB flash drives, read-only memory (ROM for short), random access memory (RAM for short), mobile hard disks, magnetic disks, or optical discs.

[0116] An embodiment of the present invention further provides an electronic device, including a memory and a processor. A computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0117] In an exemplary embodiment, the above-mentioned electronic device may further include a transmission device and an input / output device. Among them, the transmission device is connected to the above-mentioned processor, and the input / output device is connected to the above-mentioned processor.

[0118] Specific examples in this embodiment may refer to the examples described in the above embodiments and exemplary embodiments, and will not be repeated here.

[0119] An embodiment of the present invention further provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the steps of the methods in various embodiments of the present application.

[0120] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. They can be implemented by program codes executable by the computing device, so that they can be stored in a storage device and executed by the computing device. And in some cases, the steps shown or described can be executed in a different order from here, or they can be separately made into individual integrated circuit modules, or multiple modules or steps among them can be made into a single integrated circuit module to implement. In this way, the present invention is not limited to any specific combination of hardware and software.

[0121] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting interference signals, characterized in that, Applied to detecting interference signals in a satellite network, including: Processing the one-dimensional signal energy data of the signal to be detected to obtain the two-dimensional image data of the signal to be detected; Reconstructing the two-dimensional image data of the signal to be detected to obtain the reconstructed image data of the signal to be detected; Determining whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected.

2. The method according to claim 1, wherein Determining whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected, including: Determining the reconstruction error of the signal to be detected according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected; Determining whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and the target reconstruction error threshold.

3. The method according to claim 2, wherein Before determining whether the signal to be detected is the interference signal according to the relationship between the reconstruction error of the signal to be detected and the target reconstruction error threshold, the method further includes: Determining the reconstruction errors between the two-dimensional image data of N first normal signals and the reconstructed image data of the first normal signals to obtain N first reconstruction errors, where the first normal signals are normal communication signals and N is an integer; Determining the target reconstruction error threshold according to the N first reconstruction errors.

4. The method according to claim 3, wherein Determining the target reconstruction error threshold according to the N first reconstruction errors, including: Determining the target reconstruction error threshold according to the mean of the N first reconstruction errors and the variance of the N first reconstruction errors.

5. The method according to claim 3, characterized in that Determining the reconstruction errors between the two-dimensional image data of N first normal signals and the reconstructed image data of the first normal signals, including: Inputting the two-dimensional image data of a first target normal signal into a target autoencoder, encoding the two-dimensional image data of the first target normal signal through the encoding layer of the target autoencoder to obtain the encoded feature map of the first target normal signal, where the N first normal signals include the first target normal signal; Decoding the encoded feature map of the first target normal signal through the decoding layer of the target autoencoder to obtain the reconstructed image data of the first target normal signal; Determining the reconstruction error of the first target normal signal according to the two-dimensional image data of the first target normal signal and the reconstructed image data of the first target normal signal.

6. The method according to claim 1, wherein Reconstructing the two-dimensional image data of the signal to be detected to obtain the reconstructed image data of the signal to be detected, including: Inputting the two-dimensional image data of the signal to be detected into a target autoencoder, encoding the two-dimensional image data of the signal to be detected through the encoding layer of the target autoencoder to obtain the encoded feature map of the signal to be detected; Decoding the encoded feature map of the signal to be detected through the decoding layer of the target autoencoder to obtain the reconstructed image data of the signal to be detected.

7. The method according to claim 5 or 6, characterized in that, The method further includes: Train the autoencoder using a training sample data set, where the training sample data set includes: a training sample data set and a test sample data set, the training sample data set includes a plurality of training sample data, the training sample data includes: two-dimensional image data of a normal signal training sample, and the test sample data set includes a plurality of test sample data, and the test sample data includes: two-dimensional image data of an interference signal test sample with the first system and the second system at different angles; End the training to obtain the target autoencoder when the test result of the test sample data set by the autoencoder meets the preset conditions.

8. The method according to claim 7, characterized in that, The method further includes: Input the target test sample data into the autoencoder obtained by the i-th round of training to obtain the reconstructed image data of the target test sample data. The test sample data set includes the target test sample data, where i is an integer; Determine the reconstruction error between the target test sample data and the reconstructed image data of the target test sample data to obtain the reconstruction error of the target test sample. Determine the test result of the target test sample data according to the relationship between the reconstruction error of the target test sample and the reconstruction error threshold obtained by the i-th round of training.

9. The method according to claim 8, characterized in that, Before determining the test result of the target test sample data according to the relationship between the reconstruction error of the target test sample and the reconstruction error threshold obtained by the i-th round of training, the method further includes: Input the two-dimensional image data of M second normal signals into the autoencoder obtained by the i-th round of training to obtain the reconstructed image data of the M second normal signals, where the second normal signal is a normal communication signal and M is an integer; Determine the reconstruction error between the two-dimensional image data of the M second normal signals and the reconstructed image data of the second normal signals to obtain M second reconstruction errors. Determine the reconstruction error threshold obtained by the i-th round of training according to the M second reconstruction errors, where if the i-th round of training is the last round of training of the autoencoder, the reconstruction error threshold obtained by the i-th round of training is the target reconstruction error threshold.

10. The method according to claim 7, wherein The first system is a non-geostationary orbit satellite NGSO system, the second system is a geostationary orbit satellite GSO system, and the interference signal is an interference signal between the non-geostationary orbit satellite NGSO system and the geostationary orbit satellite GSO system.

11. A device for detecting interference signals, characterized in that Applied to detect interference signals in a satellite network, including: A processing module for processing the one-dimensional signal energy data of the signal to be detected to obtain the two-dimensional image data of the signal to be detected; A reconstruction module for reconstructing the two-dimensional image data of the signal to be detected to obtain the reconstructed image data of the signal to be detected; A determination module for determining whether the signal to be detected is an interference signal according to the two-dimensional image data of the signal to be detected and the reconstructed image data of the signal to be detected.

12. A computer-readable storage medium, characterized in that, A computer program is stored in the computer-readable storage medium, where the computer program, when executed by a processor, implements the steps of the method described in any one of claims 1 to 10.

13. A chip, which includes at least one of an editable logic circuit and executable instructions, and runs in an electronic device to implement the method according to any one of claims 1 to 10.

14. An electronic device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory, and the processor is configured to run the computer program to execute the method according to any one of claims 1 to 10.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 10 are implemented.

Images