Multi-modal interference confrontation method and device based on mutual information

Abstract

The invention discloses a multi-modal interference confrontation method and device based on mutual information. The method comprises the following steps: acquiring a slow time complex envelope sequence of an echo signal and a track sequence of a target; respectively preprocessing the slow time complex envelope sequence and the track sequence to obtain a preprocessed complex envelope sequence and a preprocessed track sequence; inputting the preprocessed complex envelope sequence and the preprocessed track sequence into a trained multi-modal fusion network based on mutual information to obtain an identification result of the echo signal, the identification result being used for indicating whether the echo signal is an interference signal; wherein the total loss function used for constraining multi-modal fusion network training is composed of a mutual information loss item and a binary cross entropy loss item, and the mutual information loss item is calculated according to mutual information between a sample slow time complex envelope sequence and a sample track sequence. According to the method, mutual information among different modal data can be fully utilized to train the network, so that the recognition accuracy of the method is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of signal recognition, and in particular relates to a method and device for countering multimodal interference based on mutual information. Background Art

[0002] As the electromagnetic environment becomes increasingly complex, jamming techniques, particularly electronic jamming (ELJ), pose significant challenges and threats to radar system detection performance. Compared to suppressive jamming, deceptive jammers generate false target information through delay, modulation, and forwarding. This information is then applied to the radar signal receiving system, causing the radar to receive multiple deceptive target signals, impacting detection and tracking. To achieve target detection, radar systems require the use of appropriate electronic countermeasures (ECMs). Strong jamming countermeasures are essential for radar operation. Existing research has shown that as the number of quantization bits in digital radio frequency memory (DRFM) continues to increase, the effectiveness of deceptive jamming countermeasures using quantization error has significantly decreased, making countermeasures increasingly difficult for single-station radars. Single-station radars face drawbacks such as a single perspective and limited information during deceptive jamming countermeasures, limiting their jamming countermeasure capabilities. Multi-station radar systems have achieved significant progress in improving jamming countermeasure accuracy. In particular, intelligent jamming countermeasures, based on traditional manual feature extraction methods, have developed rapidly in recent years.

[0003] However, traditional jamming countermeasures rely on manual feature extraction, utilizing only intuitive features such as echo amplitude correlation and spatial location. This can lead to problems such as insufficient signal resource utilization and overly ideal feature extraction conditions. Existing intelligent methods have improved these shortcomings to some extent, but they often struggle to effectively process both signal-level and data-level information simultaneously. Single-dimensional data used for jamming countermeasures has a low efficiency in utilizing echo information.

[0004] Therefore, the current interference countermeasure method has the problem of insufficient recognition accuracy due to low information utilization. Summary of the Invention

[0005] The embodiments of the present invention provide a method and apparatus for countering multimodal interference based on mutual information, which can solve the above problems.

[0006] In a first aspect, an embodiment of the present invention provides a multimodal interference countermeasure method based on mutual information, the method comprising:

[0007] Acquire a slow-time complex envelope sequence of the echo signal and a target track sequence, wherein the slow-time complex envelope sequence is signal-level data and the track sequence is data-level data;

[0008] Preprocessing the slow-time complex envelope sequence and the track sequence respectively to obtain a preprocessed complex envelope sequence and a preprocessed track sequence;

[0009] Inputting the preprocessed complex envelope sequence and the preprocessed track sequence into a trained multimodal fusion network based on mutual information to obtain a recognition result of the echo signal, wherein the recognition result is used to indicate whether the echo signal is an interference signal;

[0010] Among them, the total loss function used to constrain the training of the multimodal fusion network is composed of a mutual information loss term and a binary cross entropy loss term, and the mutual information loss term is calculated based on the mutual information between the sample slow-time complex envelope sequence and the sample track sequence.

[0011] In a second aspect, an embodiment of the present invention provides a multimodal interference countermeasure device based on mutual information, comprising:

[0012] an acquisition unit, configured to acquire a slow-time complex envelope sequence of an echo signal and a target track sequence, wherein the slow-time complex envelope sequence is signal-level data and the track sequence is data-level data;

[0013] a preprocessing unit, configured to preprocess the slow-time complex envelope sequence and the track sequence respectively to obtain a preprocessed complex envelope sequence and a preprocessed track sequence;

[0014] an identification unit, configured to input the preprocessed complex envelope sequence and the preprocessed track sequence into a trained multimodal fusion network based on mutual information to obtain an identification result of the echo signal, wherein the identification result is used to indicate whether the echo signal is an interference signal;

[0015] Among them, the total loss function used to constrain the training of the multimodal fusion network is composed of a mutual information loss term and a binary cross entropy loss term, and the mutual information loss term is calculated based on the mutual information between the sample slow-time complex envelope sequence and the sample track sequence.

[0016] In a third aspect, an embodiment of the present invention provides an electronic device comprising a processor and a memory, wherein the memory is used to store computer programs; the processor can be used to execute the computer program (instructions) stored in the memory to implement the method of the first aspect above.

[0017] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed, the method of the first aspect described above can be implemented.

[0018] It can be understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0019] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: since there may be a high degree of correlation between features in high-dimensional data, mutual information helps to identify and remove redundant features, reduce model complexity, prevent overfitting, and improve the universality of the model; and the total loss function used in the network training provided by the present invention includes a mutual information loss term, which is calculated by the maximum or minimum mutual information between data-level data and signal-level data. Therefore, the network provided by the present invention can fully utilize the mutual information between different types of data during training, thereby improving the accuracy and robustness of network processing. Therefore, the method provided by the present invention is based on this network for identification, which can improve the detection effect of interference signals and enhance anti-interference capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of the structure of a multimodal fusion network based on mutual information provided by an embodiment of the present invention;

[0021] Figure 2 A schematic diagram of a scenario in which a feature fusion module performs a fusion operation according to an embodiment of the present invention;

[0022] Figure 3 A flowchart of a multi-modal interference countermeasure method based on mutual information provided by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of a scenario in which a multi-station radar system acquires echo signals according to an embodiment of the present invention;

[0024] Figure 5 A schematic diagram of a scenario for preprocessing data of different modalities provided by an embodiment of the present invention;

[0025] Figure 6 A schematic structural diagram of a multimodal interference countermeasure device based on mutual information provided by an embodiment of the present invention;

[0026] Figure 7 A schematic diagram comparing the identification performance of networks under different PRIs provided in an embodiment of the present invention;

[0027] Figure 8 A schematic diagram comparing network authentication performance at different deception distances provided by an embodiment of the present invention;

[0028] Figure 9 A schematic diagram comparing network identification performance under different signal-to-noise ratios provided by an embodiment of the present invention;

[0029] Figure 10 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0031] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0032] It will also be understood that the term "and / or" used in the present description and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0033] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0034] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0035] References to "one embodiment" or "some embodiments" in the present specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present invention. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in yet other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0036] The present invention will be further described in detail below with reference to specific examples, but the embodiments of the present invention are not limited thereto.

[0037] Figure 1 The diagram shows a schematic diagram of the structure of a multimodal fusion network based on mutual information provided by an embodiment of the present invention. As an example and not a limitation, the network 100 may include a first CNN network 111, a second CNN network 112, a first Transformer network 121, a second Transformer network 122, a feature fusion module 130, and a fully connected layer 140.

[0038] In one possible implementation, see Figure 1 The first CNN network 111 and the second CNN network 112 can both be used to extract the basic features of the slow-time complex envelope sequence of the echo signal to obtain the first signal feature and the second signal feature. The first Transformer network 121 and the second Transformer network 122 can both be used to extract the attention features of the target track sequence to obtain the first data feature and the second data feature. The fusion module 130 can fuse these four features to obtain a fused feature, and the fully connected layer 140 can obtain the recognition result of the echo signal based on the fused feature.

[0039] Exemplarily, the track sequence belongs to data-level data, and the slow-time complex envelope sequence belongs to signal-level data.

[0040] In one example, see Figure 1 Both the first CNN network 111 and the second CNN network 112 can be 2D-CNN structures, where each convolutional layer of the CNN is followed by a batch normalization layer, a ReLU activation function, and a maximum pooling layer. The slow-time complex envelope of the preprocessed echo signal undergoes four layers of convolution and pooling operations, and the output is passed to a two-layer fully connected neural network, which ultimately outputs the signal features.

[0041] In one example, see Figure 1 , the first Transformer network 121 and the second Transformer network 122 can both be traditional Transformer networks.

[0042] In one example, see Figure 2 , the fusion module 130 can be a multi-layer long short-term memory network (ML-LSTM).

[0043] Specifically, see Figure 2 , the fusion module 130 can be in the first signal feature On the basis of Second signal characteristic Second data feature Fusion is performed to obtain the fusion feature F sd .

[0044] In one example, see Figure 1 After the fusion features are input into the fully connected layer, the Sigmoid activation function can be used to map the output of the fully connected layer, and the mapped result is output as the recognition result.

[0045] In one possible implementation, the model 100 may be trained based on a back-propagation algorithm under the constraints of a total loss function.

[0046] Exemplarily, the total loss function can be composed of a mutual information loss term and a binary cross entropy loss term.

[0047] For example, see Figure 1 , the mutual information loss term can be calculated based on the mutual information between the sample slow-time complex envelope sequence and the sample track sequence.

[0048] In an example, the total loss function can be expressed as:

[0049] LOSS = loss - α (I1 - I2)

[0050] Among them, LOSS is the total loss function, I1 and I2 are used to represent the maximum mutual information and minimum mutual information between the sample slow-time complex envelope sequence and the sample track sequence, respectively, (I1-I2) is the mutual information loss term, loss is the binary cross entropy loss term, and α is the weight parameter.

[0051] For example, the binary cross entropy loss term can be expressed as:

[0052]

[0053] Among them, p i To represent the probability that the multimodal fusion network predicts that the i-th sample echo signal is true, y i is the true label of the i-th sample echo signal, N is the total number of sample echo signals; the sample slow-time complex envelope sequence is extracted from the sample echo signal.

[0054] Specifically, y i The value of can be 0 or 1. If it is 0, it means that the sample echo signal is a false target signal (ie, an interference signal). If it is 1, it means that the sample echo signal is a true target signal.

[0055] For example, the maximum mutual information and the minimum mutual information may satisfy the following formulas:

[0056]

[0057] in, Indicates that the maximum mutual information is based on the first signal feature and the first data feature Sure, for and The joint distribution of and They are and The marginal distribution of . Indicates that the minimum mutual information is based on the second signal feature and the second data feature Sure, for and The joint distribution of and They are and The marginal distribution of .

[0058] Because features in high-dimensional data may be highly correlated, mutual information helps identify and remove redundant features, reducing model complexity, preventing overfitting, and improving model universality. The total loss function used in network training provided by the present invention includes a mutual information loss term, which is calculated by the maximum or minimum mutual information between data-level and signal-level data. Therefore, the network provided by the present invention can fully utilize the mutual information between different types of data during training, improving the accuracy and robustness of network processing.

[0059] The mutual information-based multimodal interference countermeasure method provided in the embodiment of the present invention can be applied to electronic devices such as mobile terminals, personal laptops, supercomputers, etc. The embodiment of the present invention does not impose any restrictions on the specific type of electronic devices.

[0060] Figure 3 The flowchart shown is a multi-modal interference countermeasure method based on mutual information according to an embodiment of the present invention. As an example and not a limitation, the method may include steps S301-S303, each of which is described below.

[0061] S301, acquiring a slow-time complex envelope sequence of an echo signal and a target track sequence.

[0062] For example, the return signal may be a multi-station radar system (see Figure 4 ) The detection signal received is reflected by the target, or it can be an interference signal emitted by the jammer.

[0063] In one example, the echo data of different radar stations in a multi-station radar system can be first time-synchronized and phase-synchronized to capture target echo signals from different radar stations in the same spatial resolution unit; then, pulse compression is performed on the multiple echo signals from different radar stations to obtain a slow-time complex envelope sequence of the multiple echo signals.

[0064] For example, when performing time synchronization on echo data, time alignment may be performed based on a unified time, and when performing phase synchronization, grid search may be performed as a basis.

[0065] For example, considering that a single sampling can only obtain one continuous pulse repetition interval (PRI), the sample data volume is too low to accurately describe the target characteristics. Therefore, multiple samplings of multiple consecutive PRIs in the slow time domain can be used as the slow time complex amplitude envelope sequence of the nth radar station.

[0066] In one example, the target's track within the observation time range can be tracked and filtered to obtain the target's track sequence in D k a It represents the radial distance R of the receiving station a relative to the target after Kalman filtering on the track at the kth moment. k , azimuth angle θ k and pitch angle

[0067] S302 , preprocessing the slow-time complex envelope sequence and the track sequence respectively to obtain a preprocessed complex envelope sequence and a preprocessed track sequence.

[0068] In one example, see Figure 5 In (a), the signal modes of the echo signal can be spliced, that is, the slow time complex envelope ζ of radar 1 1 、Slow time complex envelope of radar 2 2 , ..., the slow-time complex envelope ζ of radar N N , perform horizontal linking, and obtain the two-dimensional pre-processed complex envelope sequence signal = [ζ 1 ζ 2 ;...;ζ N ].

[0069] In one example, see Figure 5 In (b), the track sequences of different targets obtained by the same receiving station can be spliced into a one-dimensional data block T 1 To T N, and then these one-dimensional data blocks are spliced to obtain the two-dimensional pre-processed track sequence data = [T 1 ;T 2 ;...;T N ].

[0070] S303: Input the pre-processed complex envelope sequence and the pre-processed track sequence into a trained multimodal fusion network based on mutual information to obtain a recognition result of the echo signal.

[0071] Illustratively, the multimodal fusion network based on mutual information may be the aforementioned network 100 .

[0072] Exemplarily, the recognition result of the echo signal is used to indicate whether the echo signal is an interference signal.

[0073] Since there may be a high degree of correlation between features in high-dimensional data, mutual information helps to identify and remove redundant features, reduce model complexity, prevent overfitting, and improve the universality of the model; and the total loss function used in the network training provided by the present invention includes a mutual information loss term, which is calculated by the maximum or minimum mutual information between data-level data and signal-level data. Therefore, the network provided by the present invention can make full use of the mutual information between different types of data during training, improving the accuracy and robustness of network processing. Therefore, the method provided by the present invention is based on this network for identification, which can improve the detection effect of interference signals and enhance anti-interference capabilities.

[0074] Figure 6 The diagram shows a schematic diagram of the structure of a multi-modal interference countermeasure device based on mutual information provided by an embodiment of the present invention. As an example and not a limitation, the device may include an acquisition unit 610, a pre-processing unit 620, and an identification unit 630.

[0075] Exemplarily, the acquisition unit 610 is used to acquire the slow-time complex envelope sequence of the echo signal and the track sequence of the target, wherein the slow-time complex envelope sequence is signal-level data and the track sequence is data-level data; the preprocessing unit 620 is used to preprocess the slow-time complex envelope sequence and the track sequence respectively to obtain a preprocessed complex envelope sequence and a preprocessed track sequence; the identification unit 630 is used to input the preprocessed complex envelope sequence and the preprocessed track sequence into a trained multimodal fusion network based on mutual information to obtain an identification result of the echo signal, and the identification result is used to indicate whether the echo signal is an interference signal; wherein, the total loss function used to constrain the training of the multimodal fusion network is composed of a mutual information loss term and a binary cross entropy loss term, and the mutual information loss term is calculated based on the mutual information between the sample slow-time complex envelope sequence and the sample track sequence.

[0076] In order to better illustrate the beneficial effects of the present invention, the following simulation experiments were conducted:

[0077] For example, the parameters of network 100 in the simulation experiment can be shown in Table 1 below. Here, Number is the number of convolution kernels in a convolutional layer or the output dimension size of other layers; Filter is the size of the convolution kernel, which in the Transformer model represents the number of encoder and decoder layers; Step is the step size, which in the Transformer model represents the number of attention heads; and Dropout is the proportion of neurons temporarily discarded from the network. To ensure classification accuracy, sigmoid is used as the network activation function.

[0078] Simulation Experiment 1

[0079] For example, in simulation experiment 1, the method provided by the present invention can be compared with a method that processes signal-level data using only 2D-CNN, a method that processes data-level data using only LSTM and Transformer, and a method that processes multimodal data using only 2D-CNN after removing the mutual information loss term. Furthermore, each method in the experiment used the same dataset to ensure consistency and eliminate any interference other than the different methods themselves. The simulation results obtained after 20 rounds of training at PRI = 36, SNR = 0dB, and spoofing distance dis = 25m are shown in Table 2 below.

[0080] As shown in Table 2 below, the method provided by the present invention outperforms other methods in all evaluation metrics, with each metric exceeding 0.9888. Comparisons with methods for processing unimodal and multimodal data demonstrate that the network provided by the present invention is more advantageous for multimodal data input. Compared with the second method for processing multimodal data, single-variable controlled experiments demonstrate that both the training method and network structure of the present invention are optimal.

[0081] Table 1

[0082]

[0083] Table 2

[0084]

[0085] Simulation Experiment 2

[0086] For example, in simulation experiment 2, the influence of track length, PRI, signal-to-noise ratio (SNR), and deception distance on the identification capability of network 100 can be studied in combination with the actual environment. Each time, one parameter among track length, PRI, deception distance, and SNR is adjusted. For example, keep Δr k= ±30m, SNR is 4dB, the track length is increased from 1s to 15s, and the corresponding PRI is increased from 3 to 45. The network identification effect at each track length and PRI is obtained, and the simulation is obtained. Figure 7 Similarly, keep the track length at 4s, PRI at 12, SNR at 3dB, and set Δr k = ±5~±75m simulation Figure 8 ; Keep the track length at 4s, PRI at 12, Δr k = ±50m, set SNR = -4 ~ 10dB, simulation results Figure 9 .

[0087] from Figure 7 As can be seen, the network's identification performance gradually improves as the track length increases from 1s to 15s and the PRI increases from 3 to 45. After a track length of 6s and a PRI of 18, the network's identification performance stabilizes, with accuracy values exceeding 0.95. This is because the network has acquired sufficient target identification information, meeting its identification requirements.

[0088] from Figure 8 It can be seen that Δr k = ±5~±75m, the identification effect of the network gradually increases. This is because as the deception distance increases, the characteristics of the real and fake targets become more obvious, and the network identification effect becomes better. k When the distance is ≥±40m, the identification effect of the network is stable and the accuracy can be guaranteed to be greater than 0.95.

[0089] from Figure 9 As can be seen, the network's identification performance gradually improves as the signal-to-noise ratio (SNR) increases from -4dB to 10dB. This is because the increase in SNR reduces the impact of noise on the echo signal and makes it easier for the network to detect the essential characteristics of the real target or active interference, thereby improving network identification performance. When the SNR exceeds 4dB, the network's identification performance stabilizes and maintains an accuracy rate greater than 0.95.

[0090] Therefore, the network provided by the present invention can fully utilize the mutual information between different types of data during training, improving the accuracy and robustness of network processing. The method provided by the present invention is based on this network for recognition, which can improve the detection effect of interference signals and enhance the anti-interference ability.

[0091] Figure 10 FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 10 The electronic device 1000 shown may include: at least one processor 1010 ( Figure 10Only one processor is shown in the figure), a memory 1020, and a computer program 1030 stored in the memory 1020 and executable on the at least one processor 1010, wherein the processor 1010 implements the steps of any of the above-mentioned method embodiments when executing the computer program 1030.

[0092] The electronic device 1000 may be a processing device such as a robot that can implement the above method. The embodiment of the present invention does not impose any limitation on the specific type of the electronic device.

[0093] Those skilled in the art will understand that Figure 10 The electronic device 1000 is merely an example and does not constitute a limitation on the electronic device. The electronic device 1000 may include more or fewer components than shown in the figure, or may combine certain components or different components. For example, the electronic device 1000 may also include an input and output interface.

[0094] The processor 1010 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASTC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gates, or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor.

[0095] In some embodiments, the memory 1020 may be an internal storage unit, such as a hard disk or a memory. In other embodiments, the memory 1020 may also be an external storage device, such as a plug-in hard disk, a smart memory card (Smart Memory Card, SMC), a secure digital (Secure Digital, SD) card, a flash card, etc. Furthermore, the memory 1020 may also include both an internal storage unit and an external storage device. The memory 1020 is used to store an operating system, an application program, a boot loader (Boot Loader), data, and other programs, such as the program code of the computer program. The memory 1020 may also be used to temporarily store data that has been output or is about to be output.

[0096] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0097] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0098] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments can be implemented.

[0099] An embodiment of the present invention provides a computer program product. When the computer program product is run on an electronic device, the electronic device can implement the steps in the above-mentioned various method embodiments when executing the computer program product.

[0100] If the integrated unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the process steps in the above-mentioned method embodiments by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a camera / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. Examples include USB flash drives, removable hard drives, magnetic disks, or optical disks. In some jurisdictions, based on legislation and patent practice, computer-readable media cannot be electric carrier signals or telecommunication signals.

[0101] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0102] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

Claims

1. A multimodal interference countermeasure method based on mutual information, characterized in that: include: Acquire a slow-time complex envelope sequence of the echo signal and a target track sequence, wherein the slow-time complex envelope sequence is signal-level data and the track sequence is data-level data; Preprocessing the slow-time complex envelope sequence and the track sequence respectively to obtain a preprocessed complex envelope sequence and a preprocessed track sequence; Inputting the preprocessed complex envelope sequence and the preprocessed track sequence into a trained multimodal fusion network based on mutual information to obtain a recognition result of the echo signal, wherein the recognition result is used to indicate whether the echo signal is an interference signal; Among them, the total loss function used to constrain the training of the multimodal fusion network is composed of a mutual information loss term and a binary cross entropy loss term, and the mutual information loss term is calculated based on the mutual information between the sample slow-time complex envelope sequence and the sample track sequence.

2. The method according to claim 1, characterized in that The preprocessing of the complex envelope sequence and the track sequence to obtain a preprocessed complex envelope sequence and a preprocessed track sequence respectively includes: horizontally linking the plurality of slow-time complex envelope sequences in sequence to obtain the two-dimensional pre-processed complex envelope sequence; The track sequences acquired by the same radar receiving station are respectively spliced into one-dimensional data blocks, and a plurality of the one-dimensional data blocks are spliced to obtain the two-dimensional preprocessed track sequence.

3. The method according to claim 1, characterized in that The total loss function is expressed as: LOSS = loss - α (I1 - I2) Among them, LOSS is the total loss function, I1 and I2 are used to represent the maximum mutual information and minimum mutual information between the sample slow-time complex envelope sequence and the sample track sequence, respectively, (I1-I2) is the mutual information loss term, loss is the binary cross entropy loss term, and α is the weight parameter.

4. The method according to claim 3, characterized in that The binary cross entropy loss term is expressed as: Among them, p i represents the probability that the multimodal fusion network predicts that the i-th sample echo signal is true, y i is the true label of the i-th sample echo signal, N is the total number of sample echo signals; the sample slow-time complex envelope sequence is extracted from the sample echo signal.

5. The method according to claim 3, characterized in that The multimodal fusion network includes a first CNN network, a second CNN network, a first Transformer network, a second Transformer network, a feature fusion module, and a fully connected layer; The first CNN network and the second CNN network are both used to extract basic features of the slow-time complex envelope sequence to obtain a first signal feature and a second signal feature; The first Transformer network and the second Transformer network are both used to extract attention features of the track sequence to obtain first data features and second data features; The fusion module is used to fuse the first data feature, the second data feature, the first signal feature and the second signal feature to obtain a fusion feature; The fully connected layer is used to obtain the recognition result according to the fusion feature.

6. The method according to claim 5, characterized in that The maximum mutual information satisfies the following formula: in, Indicates that the maximum mutual information is based on the first signal feature and the first data feature Sure, for and The joint distribution of and They are and The marginal distribution of .

7. The method according to claim 5, characterized in that The minimum mutual information satisfies the following formula: in, Indicates that the minimum mutual information is based on the second signal feature And the second data feature Sure, for and The joint distribution of and They are and The marginal distribution of .

8. A multimodal interference countermeasure device based on mutual information, characterized in that: include: an acquisition unit, configured to acquire a slow-time complex envelope sequence of an echo signal and a target track sequence, wherein the slow-time complex envelope sequence is signal-level data and the track sequence is data-level data; a preprocessing unit, configured to preprocess the slow-time complex envelope sequence and the track sequence respectively to obtain a preprocessed complex envelope sequence and a preprocessed track sequence; an identification unit, configured to input the preprocessed complex envelope sequence and the preprocessed track sequence into a trained multimodal fusion network based on mutual information to obtain an identification result of the echo signal, wherein the identification result is used to indicate whether the echo signal is an interference signal; Among them, the total loss function used to constrain the training of the multimodal fusion network is composed of a mutual information loss term and a binary cross entropy loss term, and the mutual information loss term is calculated based on the mutual information between the sample slow-time complex envelope sequence and the sample track sequence.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the electronic device, the method according to any one of claims 1 to 7 is implemented.

Images