Electromagnetic signal transmission method, system and computer readable storage medium

Abstract

The present application relates to an electromagnetic signal transmission method, system, and computer-readable storage medium, wherein the method includes: inputting an acquired electromagnetic signal into a first noise generator, performing noise addition processing on the electromagnetic signal based on a first anti-learning interference noise of the first noise generator, and generating a noise interference electromagnetic signal; the first noise generator inputs acquired first training data into an initial neural network model, and injects a first preset interference noise into each network layer in the initial neural network model, and performs multi-level fusion processing on the first preset interference noise and the first training data through each network layer, and iteratively trains the generated signal; the first anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model; and transmitting the noise interference electromagnetic signal to a signal receiving device. Through this application, the problem of low security of electromagnetic signal transmission is solved.

Description

Technical Field

[0001] The present application relates to the field of signal transmission, and in particular to electromagnetic signal transmission methods, systems, and computer-readable storage media. Background Art

[0002] With the advent of the information age, the technological level and business scale of electromagnetic signal transmission have achieved leapfrog growth. Electromagnetic signals have achieved significant improvements in speed, stability, distance, and efficiency, and the deep integration of artificial intelligence technologies has further promoted the intelligent and convenient processing of electromagnetic signals. In particular, deep neural networks have demonstrated superior performance to traditional algorithms in tasks such as signal recognition, modulation and demodulation, and interference suppression, significantly meeting people's actual needs for communication services. However, the inherent "black box characteristics" and "vulnerability to adversarial attacks" of deep learning models make them vulnerable to targeted attacks, making it difficult to meet the requirements for electromagnetic signal data protection in related technologies and making them unsuitable for confidential transmission of communication signals, thus affecting the security of electromagnetic signal transmission.

[0003] Currently, no effective solution has been proposed to address the problem of low security of electromagnetic signal transmission in related technologies. Summary of the Invention

[0004] The embodiments of the present application provide an electromagnetic signal transmission method, system, and computer-readable storage medium to at least solve the problem of low security of electromagnetic signal transmission in related technologies.

[0005] In a first aspect, an embodiment of the present application provides an electromagnetic signal transmission method, applied to a signal sending device, the method comprising:

[0006] obtaining an electromagnetic signal to be transmitted;

[0007] Inputting the electromagnetic signal into a trained first noise generator, performing noise addition processing on the electromagnetic signal based on a first anti-learning interference noise carried by the first noise generator, to generate a noise interference electromagnetic signal;

[0008] The first noise generator is generated by inputting the acquired first training data into an initial neural network model, injecting the first preset interference noise into each network layer of the initial neural network model, performing multi-level fusion processing on the first preset interference noise and the first training data through each network layer, and iteratively training; the first anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model;

[0009] The noise interference electromagnetic signal is transmitted to a signal receiving device.

[0010] In some embodiments, the signal receiving device is deployed with a first interference remover for performing noise removal processing on the noise interference electromagnetic signal, and the training process of the first noise generator and the first interference remover includes:

[0011] Inputting the first training data into a first autoencoder model; wherein the initial neural network model is set to the first autoencoder model;

[0012] performing multi-layer feature extraction processing on the first training data via a first encoder network of the first autoencoder model to obtain hierarchical feature data corresponding to each network level in the first encoder network, and fusing the injected first preset interference noise with the hierarchical feature data at each network level in the first encoder network to output a first fused training feature;

[0013] Inputting the first fused training features into a first decoder network of the first autoencoder model for decoding processing to obtain first reconstructed feature data;

[0014] Based on the first training data and the first reconstruction feature data, a first reconstruction loss function result is calculated, and the gradient of the first reconstruction loss function result is back-transferred to the first autoencoder model for iterative training to obtain a first target encoder model; the first target encoder model includes the first noise generator and the first interference remover.

[0015] In some embodiments, the calculating a first reconstruction loss function result based on the first training data and the first reconstruction feature data, and back-transferring the gradient of the first reconstruction loss function result to the first autoencoder model for iterative training to obtain a first target encoder model includes:

[0016] In a first training stage, calculating the first reconstruction loss function result;

[0017] In the second training stage, a preset stage disturbance is input into the first autoencoder model, and the first preset interference noise, the stage disturbance and the hierarchical feature data are fused using the first autoencoder model to output a second fused training feature, and the second fused training feature is input into the first decoder network for decoding to obtain second reconstructed feature data;

[0018] Calculating a second reconstruction loss function result based on the first training data and the second reconstruction feature data;

[0019] The gradients of the first reconstruction loss function result and the second reconstruction loss function result are back-transferred to the first autoencoder model for iterative training to obtain the first target encoder model.

[0020] In some embodiments, the first training data carries a classification label; and using the first autoencoder model to fuse the first preset interference noise, the stage disturbance, and the hierarchical feature data to output a second fused training feature includes:

[0021] Using the first autoencoder model, fusing the first preset interference noise, the stage disturbance, and the hierarchical feature data to output a second fused training feature, performing classification processing based on the second fused training feature, and outputting a classification result;

[0022] The step of back-transferring the gradients of the first reconstruction loss function result and the second reconstruction loss function result to the first autoencoder model for iterative training to obtain the first target encoder model includes:

[0023] Calculating a classification loss function result based on the classification result and the classification label, and calculating a joint loss function result based on the classification loss function result and the second reconstruction loss function result;

[0024] The gradient of the first reconstruction loss function result and the joint loss function result is back-transferred to the first autoencoder model for iterative training to obtain the first target encoder model.

[0025] In some embodiments, after obtaining the electromagnetic signal to be transmitted, the method further includes:

[0026] Performing semantic feature extraction processing on the electromagnetic signal to obtain a semantic feature vector;

[0027] Inputting the semantic feature vector into a trained second noise generator, performing noise addition processing on the electromagnetic signal based on a second anti-learning interference noise carried by the second noise generator, and generating a noise interference semantic vector;

[0028] The second noise generator is generated by inputting the acquired second training data into an initial neural network model, injecting a second preset interference noise into each network layer of the initial neural network model, performing multi-level fusion processing on the second preset interference noise and the second training data through each network layer, and iteratively training; the second anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model;

[0029] The noise interference semantic vector is transmitted to the signal receiving device.

[0030] In some embodiments, performing semantic feature extraction processing on the electromagnetic signal to obtain a semantic feature vector includes:

[0031] The electromagnetic signal is input into a trained signal encoder for semantic feature extraction to obtain the semantic feature vector.

[0032] In some embodiments, the signal receiving device is deployed with a signal decoder for restoring features of the noise interference semantic vector; the training process of the signal encoder and the signal decoder includes:

[0033] Obtaining third training data;

[0034] Inputting the third training data into a second encoder network in a second autoencoder model, and encoding the third training data to obtain training semantic feature data;

[0035] Inputting the training semantic feature data into a second decoder network in the second autoencoder model, and reconstructing the training semantic feature data to obtain training reconstructed data;

[0036] Calculating a semantic feature loss function result based on the third training data and the training reconstruction data;

[0037] The gradient of the semantic feature loss function result is back-transferred to the second autoencoder model for iterative training, and a second target encoder model is generated; the second target encoder model includes the signal encoder and the signal decoder.

[0038] In some embodiments, transmitting the noise interference electromagnetic signal to a signal receiving device includes:

[0039] Get the preset random encryption seed;

[0040] Based on the random encryption seed, a channel dynamic sorting operation is performed on the noise interference electromagnetic signal to obtain a random encryption signal, and the random encryption signal is transmitted to the signal receiving device.

[0041] In a second aspect, an embodiment of the present application provides an electromagnetic signal transmission system, including a signal sending device and a signal receiving device;

[0042] The signal sending device is communicatively connected to the signal receiving device, and is used to execute the electromagnetic signal transmission method as described in the first aspect above.

[0043] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the electromagnetic signal transmission method as described in the first aspect above.

[0044] Compared with the related art, the electromagnetic signal transmission method, system and computer-readable storage medium provided in the embodiments of the present application obtain an electromagnetic signal to be transmitted; input the electromagnetic signal into a trained first noise generator, and perform noise addition processing on the electromagnetic signal based on the first anti-learning interference noise carried by the first noise generator to generate a noise interference electromagnetic signal; wherein the first noise generator is generated by inputting the acquired first training data into an initial neural network model, and injecting the first preset interference noise into each network layer in the initial neural network model, performing multi-level fusion processing on the first preset interference noise and the first training data through each network layer, and iterative training; the first anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model; and transmitting the noise interference electromagnetic signal to a signal receiving device.

[0045] Based on this, the first anti-learning interference noise is added to the electromagnetic signal through the first noise generator, and the noise will be input through multiple levels to achieve multi-scale fusion with the electromagnetic signal, so as to retain information as much as possible and realize dynamic injection of multiple layers of anti-learning interference noise. The generated anti-learning interference noise is combined with the original features to effectively prevent data from being attacked and stolen to generate attack samples, and can effectively enhance the anti-interference and anti-eavesdropping capabilities of the original electromagnetic signal, and realize confidential transmission of electromagnetic signals, thereby solving the problem of low security of electromagnetic signal transmission.

[0046] The details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0048] Figure 1 This is a hardware structure block diagram of a terminal according to an electromagnetic signal transmission method according to an embodiment of the present application;

[0049] Figure 2 is a flow chart of a method for transmitting an electromagnetic signal according to an embodiment of the present application;

[0050] Figure 3 is a structural diagram of a first autoencoder model according to an embodiment of the present application;

[0051] Figure 4 is a flowchart of a training method for a first target encoder model according to an embodiment of the present application;

[0052] Figure 5 is a structural diagram of a second autoencoder model according to an embodiment of the present application;

[0053] Figure 6 is a schematic diagram of an electromagnetic signal transmission system architecture according to an embodiment of the present application;

[0054] Figure 7 This is a flowchart of a method for constructing an electromagnetic signal transmission system architecture according to an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is described and illustrated below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely used to explain this application and are not intended to limit this application. Based on the embodiments provided in this application, all other embodiments obtained by those of ordinary skill in the art without making any creative efforts are within the scope of protection of this application.

[0056] Obviously, the drawings described below are merely examples or embodiments of the present application. Those skilled in the art can, without inventive effort, apply the present application to other similar scenarios based on these drawings. Furthermore, it is also understood that, although the effort involved in such a development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, changes in design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as an insufficiency of the content disclosed in this application.

[0057] References to "embodiments" in this application mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it refer to independent or alternative embodiments that are mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described in this application may be combined with other embodiments unless there is a conflict.

[0058] Unless otherwise defined, technical or scientific terms used herein shall have the ordinary meaning as understood by persons of ordinary skill in the art to which this application belongs. The terms "a," "an," "an," "the," and similar expressions used herein do not denote quantitative limitations and may refer to either the singular or the plural. The terms "comprise," "include," "have," and any variations thereof, used herein, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or modules (units) is not limited to the listed steps or units but may also include steps or units not listed, or may include other steps or units inherent to the process, method, product, or apparatus. The terms "connected," "connected," "coupled," and similar expressions used herein are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. As used herein, "plurality" means two or more. "And / or" describes an association between associated objects, indicating that three possible relationships exist. For example, "A and / or B" may mean: A exists alone; A and B exist simultaneously; or B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0059] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. Taking running on a terminal as an example, Figure 1 FIG is a hardware structure diagram of a terminal according to an electromagnetic signal transmission method of an embodiment of the present application. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown) a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0060] Memory 104 can be used to store computer programs, such as software programs and modules for application software, such as the computer program corresponding to the electromagnetic signal transmission method in the embodiments of the present application. Processor 102 executes the computer programs stored in memory 104 to perform various functional applications and data processing, thereby implementing the aforementioned methods. Memory 104 can include high-speed random access memory (RAM) and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some embodiments, memory 104 can further include memory remotely located from processor 102, which can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0061] Transmission device 106 is used to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by the terminal's communications provider. In one embodiment, transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0062] This embodiment also provides an electromagnetic signal transmission method. Figure 2 is a flow chart of an electromagnetic signal transmission method according to an embodiment of the present application, such as Figure 2 As shown, the process includes the following steps:

[0063] Step S210: Acquire the electromagnetic signal to be transmitted.

[0064] The acquisition of electromagnetic signals is usually completed through sensors or antenna arrays. Its essence is to modulate information (such as voice, images, data, etc.) onto electromagnetic waves in a specific way, and complete signal acquisition, preprocessing and encoding.

[0065] It's also worth noting that after acquiring the electromagnetic signal emitted by the signal source, it can also undergo a series of signal preprocessing, such as signal modulation. Specifically, let's take an electromagnetic signal that has undergone IQ modulation as an example. IQ modulation involves splitting the data into two channels, each modulated with a carrier wave, with the two carrier waves being orthogonal to each other. Here, I stands for in-phase, and Q stands for quadrature. The IQ signal to be transmitted is then preprocessed, and the I and Q component matrices of the signal are concatenated to produce the electromagnetic signal to be transmitted.

[0066] Step S220: Input the electromagnetic signal to the trained first noise generator, perform noise addition processing on the electromagnetic signal based on the first anti-learning interference noise carried by the first noise generator, and generate a noise interference electromagnetic signal.

[0067] Among them, the first noise generator is generated by inputting the acquired first training data into the initial neural network model, and injecting the first preset interference noise into each network layer in the initial neural network model, performing multi-level fusion processing on the first preset interference noise and the first training data through each network layer, and iterative training; the first anti-learning interference noise is obtained by minimizing the learning of the first preset interference noise during the iterative training process of the initial neural network model.

[0068] In this step, the trained neural network model (the first noise generator) is used to generate specific interference noise and inject it into the electromagnetic signal. The following is a detailed description of the training process of the first noise generator:

[0069] The initial neural network model typically employs a deep learning model, such as a U-shaped network (UNet), convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN), or autoencoder, consisting of multiple hidden layers (convolutional layers, fully connected layers, etc.). Its core design involves injecting a pre-set interference noise (such as Gaussian white noise or impulse noise) as additional input into multiple layers of the neural network during training, rather than a single layer. This multi-layer fusion achieves dynamic adaptation of noise and signal. More specifically, the initial training data (typical electromagnetic signal samples, such as speech, images, or modulated signals) is fed into the initial neural network model. During the forward propagation of each network layer, the pre-set interference noise is fused with the feature map of the current layer (e.g., pixel-by-pixel addition or channel-by-channel concatenation). Noise and signal features repeatedly interact at multiple levels of the neural network, forming a deep coupling. For example, noise in convolutional layers affects local feature extraction (such as edges and textures), while noise in fully connected layers interferes with global feature mapping (such as classification decision boundaries). Backpropagation is then used to adjust network weights so that the generated noise (i.e., first-class anti-learning interference noise) meets the following objectives: minimize learnability, meaning the noise pattern must be difficult for third-party neural networks to model or remove; and preserve signal usability, meaning that the signal, after adding noise, must still be restored to its original state through inverse processing (such as a noise separation network) at the receiver.

[0070] After the first noise generator is trained and deployed on the signal transmitting device, in the actual application phase, the electromagnetic signal (analog or digital) obtained in step S210 is input into the first noise generator. The trained first noise generator dynamically generates matching anti-learning interference noise based on the statistical characteristics of the input signal (such as spectral distribution and time domain waveform). The noise and signal are then fused at multiple network levels to form a noise-interference electromagnetic signal. The resulting noise-interference electromagnetic signal masks the characteristics of the original signal, reducing the risk of eavesdropping or active interference. Even if an attacker knows the noise injection mechanism, it is difficult to eliminate the noise effect through reverse training due to the deep binding of the noise to the signal, thus achieving resistance to deep learning attacks.

[0071] Step S230: transmitting the noise interference electromagnetic signal to a signal receiving device.

[0072] The noise interference electromagnetic signal generated by the above step S220 is transmitted to the signal receiving device through a physical channel. The signal receiving device uses a neural network paired with the noise generator to reversely separate the dynamic noise and restore the original signal. At the same time, the transmission security is ensured through dynamic key negotiation and physical layer authentication, and finally high-reliability and low-probability-of-interception electromagnetic signal transmission is achieved in a complex electromagnetic environment.

[0073] In related technologies, the transmission system is vulnerable to attacks, and signal data can be easily used by unauthorized model training to generate attack samples, reduce model performance, and leave attack loopholes, thus threatening the safe and reliable transmission of signals.

[0074] In contrast, the embodiment of the present application uses the above-mentioned electromagnetic signal transmission method to add the first anti-learning interference noise to the electromagnetic signal via the first noise generator, and the noise will be input through multiple levels to achieve multi-scale fusion with the electromagnetic signal, thereby retaining information as much as possible and realizing dynamic injection of multiple layers of anti-learning interference noise. The generated anti-learning interference noise combined with the original features can effectively prevent data from being attacked and stolen to generate attack samples, and can effectively enhance the anti-interference and anti-eavesdropping capabilities of the original electromagnetic signal, realize confidential transmission of electromagnetic signals, and solve the problem of low security of electromagnetic signal transmission.

[0075] In some embodiments, the signal receiving device is equipped with a first interference remover for removing noise from the electromagnetic signal. The training process of the first noise generator and the first interference remover may further include the following steps:

[0076] The first training data is input into the first autoencoder model; wherein the initial neural network model is set to the first autoencoder model. The specific structures of the first encoder and the first decoder in the first autoencoder model may not be limited; the first encoder is responsible for abstracting the input data (such as electromagnetic signals) layer by layer into high-level features (potential representations), and the first decoder is responsible for reconstructing the original data from the high-level features. In addition, the above-mentioned first training data can be a data set constructed based on the collected signal samples. All signal samples are collected from ultra-short wave radio stations, and the acquisition conditions are set to direct and diffraction; in this embodiment, the IQ signals in the samples are preprocessed. First, for the IQ signal with a length of L, the signal's Quantity and The component matrices are spliced ​​to obtain the input signal :

[0077] X=concat(I,Q);

[0078] in and The component size is (1, L), after splicing The size is (2, L), then Split into K segments, each of size (2, N), satisfying the relationship L = K × N, and flattening the signal segments into size (1, 2 N); Reconstruct the divided fragments into the input matrix X K,2N , and use it as the first training data. This operation can save computing resources while mining the correlation between different time series segments.

[0079] The first encoder network of the first autoencoder model is used to perform multi-layer feature extraction processing on the first training data to obtain hierarchical feature data corresponding to each network level in the first encoder network, and at each network level in the first encoder network, the injected first preset interference noise is fused with the hierarchical feature data to output a first fused training feature.

[0080] Specifically, see Figure 3 The input data (first training data) passes through the multi-layer convolutional network of the first encoder, extracting features layer by layer. At each encoder level, the preset interference noise is fused with the feature map of the current layer. Fusion methods can include: element-by-element addition, where the noise is directly superimposed on the feature map to simulate additive interference in the channel; channel splicing, where the noise is spliced ​​with the feature map as an additional channel to increase the feature dimension; or, an attention mechanism can be introduced to dynamically adjust the weight of the noise's influence on the feature through learnable parameters. The fused features (first fused training features) contain both the original signal features and the dynamic noise, and serve as the input to the decoder.

[0081] Next, the first fused training features are input into the first decoder network of the first autoencoder model for decoding, yielding first reconstructed feature data. Based on the first training data and the first reconstructed feature data, a first reconstruction loss function is calculated, and the gradient of the first reconstruction loss function is backpropagated to the first autoencoder model for iterative training, yielding a first target encoder model. The first decoder receives the fused features, attempts to restore the original input data, and outputs the first reconstructed feature data. This process forces the decoder to learn the function of the first interference remover, namely, extracting effective information from noise-contaminated features. The reconstruction loss (such as mean squared error (MSE) or cross-entropy loss) is calculated by comparing the reconstructed data (first reconstructed feature data) with the original data (first training data). The gradient of the reconstruction loss function is backpropagated to the entire autoencoder, and the weights of the first encoder and decoder are updated using an optimizer (such as the adaptive moment estimator Adam). Iterative training continues until a stopping condition (e.g., reaching the number of training iterations, the training duration, or model convergence) is met. At this point, iterations are terminated, and a fully trained first target encoder model is generated.

[0082] The first target encoder model includes a first noise generator (i.e., a trained first encoder) and a first interference remover (i.e., a trained first decoder). The trained first noise generator and first interference remover are deployed on the signal transmitting device and the signal receiving device, respectively.

[0083] Through the above embodiment, by iteratively training the first autoencoder model, the noise generator and the interference remover are jointly trained, thereby reducing the model complexity while ensuring performance.

[0084] In some embodiments, the above-mentioned steps of calculating the first reconstruction loss function result based on the first training data and the first reconstruction feature data, and back-transferring the gradient of the first reconstruction loss function result to the first autoencoder model for iterative training to obtain the first target encoder model may further include the following steps:

[0085] In the first training stage, the first reconstruction loss function result is calculated; in the second training stage, the preset stage disturbance is input into the first autoencoder model, and the first preset interference noise, stage disturbance and hierarchical feature data are fused using the first autoencoder model to output the second fused training feature, and the second fused training feature is input into the first decoder network for decoding to obtain the second reconstructed feature data; based on the first training data and the second reconstructed feature data, the second reconstruction loss function result is calculated; the gradient of the first reconstruction loss function result and the second reconstruction loss function result are back-transferred to the first autoencoder model for iterative training to obtain the first target encoder model.

[0086] Specifically, for the above-mentioned first autoencoder model, training adopts a two-stage iterative training framework; training data is obtained from the training data set, signal encoding is performed in the network after adding anti-learning interference noise, a reconstructed signal is obtained through reconstruction, the reconstruction loss is calculated based on the reconstructed signal and the input signal, the classification result is obtained through classification, and the classification loss is calculated based on the classification result and the classification label for calculating the gradient, and the projected gradient descent method is used as the optimization algorithm.

[0087] In the two-stage iterative training, each round of training samples a batch of data from the training dataset and initializes the perturbation starting point ; In each round of training, the first stage of training is repeated iteratively times, and then the second phase of training is repeated times, and then update the model parameters. Each round of iterative training is divided into:

[0088] In the first training phase, each iteration includes adding anti-learning interference noise, calculating gradients, updating directions, and projecting back to the constraint space. The goal is to minimize the amount of information that can be learned by the model after adding anti-learning interference noise, as shown in the following formula:

[0089] ;

[0090] In the above formula, This is used to represent the calculation formula for the first reconstruction loss function. In this step, it can be MSE loss or other loss calculation methods. X represents the first input training data. represents the first preset interference noise added, Represents noise The maximum allowed amplitude. Represents the first autoencoder model Input to the attack The predicted output.

[0091] In the second training phase, each iteration involves adding perturbations, calculating gradients, updating directions, and projecting back into the constraint space. The goal is to improve model capabilities through adversarial learning, learn as much information as possible from the data, and eliminate the effects of noise. In other words, a preset stage perturbation and a first preset interference noise are input into the first autoencoder model. These perturbations and noise can be random or specifically designed to simulate various interferences that may be encountered in real applications. In the first autoencoder model, the stage perturbation, the first preset interference noise, and the hierarchical feature data extracted in the first training phase are fused to generate a second fused training feature. This step aims to enable the model to learn how to extract effective features in the presence of perturbations and noise. The second fused training feature is then input into the first decoder network for decoding, generating second reconstructed feature data. A second reconstruction loss function is calculated based on the first training data and the second reconstructed feature data. Compared to the first training phase, the loss function in this second training phase also measures the difference between the reconstructed data and the original data, but this time it is calculated in the presence of perturbations and noise.

[0092] During the two training phases, the gradients of the first and second reconstruction loss functions with respect to the model parameters are calculated. The gradients from these two phases are backpropagated into the first autoencoder model. This means that the model parameters are updated based on the gradients of these two loss functions. By iterating this process multiple times (i.e., repeating the operations of the first and second training phases and calculating the losses and gradients), the parameters of the first autoencoder model are continuously optimized. Ultimately, when the model meets the preset stopping criteria (e.g., the loss function value no longer decreases significantly, the number of iterations reaches the upper limit, etc.), the first target encoder model is obtained.

[0093] Through the above embodiment, iterative training is divided into two stages. In the first stage, basic reconstruction capabilities are established. In the second stage, perturbations and noise are introduced to enhance the robustness of the model. Combined with the gradient back propagation of the dual loss function, the model can significantly improve its adaptability and generalization performance to complex interference while maintaining the feature extraction accuracy.

[0094] It should also be noted that in the second training phase, the first training data and the second reconstruction feature data can be used to calculate the second reconstruction loss function, and the gradient of the calculated second reconstruction loss function result is back-transferred to the model for iterative training. Alternatively, in another embodiment, stage training can also be performed in conjunction with the classification results. Specifically, the first training data carries a classification label. In this case, the first training data is another batch of data with classification labels sampled from the above-mentioned signal sample data set. The process includes the following steps:

[0095] Using the first autoencoder model, the first preset interference noise, stage disturbance and hierarchical feature data are fused and processed to output the second fused training feature, and classification processing is performed based on the second fused training feature to output the classification result. Figure 3 In the second training phase, the input features (i.e., the first training data with classification labels) are fed into each convolutional network layer of the first encoder of the first autoencoder model. After multi-scale fusion with the first preset interference noise injected by each convolutional network layer, the fused features are fed into a batch normalization layer for normalization to obtain the processed second fused training features. The second fused training features are then fed into a fully connected layer serving as the classification head for classification, or fed into an auxiliary classification network for classification to obtain a classification result for the electromagnetic signal.

[0096] Next, the classification loss function result is calculated based on the classification result and the classification label, and the joint loss function result is calculated based on the classification loss function result and the second reconstruction loss function result; the gradients of the first reconstruction loss function result and the joint loss function result are back-transferred to the first autoencoder model for iterative training to obtain the first target encoder model.

[0097] More specifically, this is as shown in the following formula:

[0098] ;

[0099] in, Calculation formula for the second reconstruction loss function. represents the added stage disturbance, Represents disturbance The maximum allowed amplitude. Represents the first autoencoder model Input to the attack The predicted output.

[0100] Through the above embodiment, the first autoencoder model is iteratively trained by combining reconstruction loss with classification loss. Compared to training constrained by reconstruction loss alone, the use of multi-task training can further constrain model training and improve model performance. Furthermore, by introducing classification loss only in the second training phase, the model weights are first trained to a certain extent using unlabeled data, and then the classification loss is added using labeled data to further constrain training. This can effectively reduce label dependence and address the issue of limited labeled data.

[0101] For ease of understanding, the following describes the training process of the first target encoder model with reference to the accompanying figures. Figure 4 , the process includes the following steps:

[0102] Step S401: input first training data.

[0103] Step S402: Initialize model parameters, including: learning rate , PGD parameters of the first training stage (perturbation range , step length , number of iterations ), PGD parameters of the second training stage (perturbation range , step length , number of iterations ).

[0104] Step S403: Determine whether m rounds of training have been completed. If so, terminate the training and generate a first target encoder model.

[0105] Step S404: If the judgment result of the above step S403 is no, data sampling is performed and disturbance is initialized.

[0106] Step S405: executing the first training phase.

[0107] Step S406: executing the second training phase.

[0108] Step S407: calculate the gradient, update the parameters, and return to the above step S403 to determine whether to stop training.

[0109] In some embodiments, after obtaining the electromagnetic signal to be transmitted, the electromagnetic signal transmission method further includes the following steps:

[0110] Semantic feature extraction and processing are performed on electromagnetic signals to obtain semantic feature vectors. Through signal preprocessing, multi-domain feature fusion, and deep learning coding, high-level semantic information reflecting the essential attributes of the original electromagnetic signal is parsed and converted into a structured vector representation to facilitate subsequent tasks such as signal classification, recognition, or decision-making.

[0111] The semantic feature vector is input into the trained second noise generator, and noise is added to the electromagnetic signal based on the second anti-learning interference noise carried by the second noise generator to generate a noise interference semantic vector; wherein, the second noise generator is generated by inputting the acquired second training data into the initial neural network model, and injecting the second preset interference noise into each network layer in the initial neural network model, performing multi-level fusion processing on the second preset interference noise and the second training data through each network layer, and iterative training; the second anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model; and the noise interference semantic vector is transmitted to the signal receiving device.

[0112] It should be understood that the training process of the second noise generator is similar to that of the first noise generator, the only difference being the different training data input during the training process. The first training data used to train the first noise generator is electromagnetic signal data, while the second training data used to train the second noise generator is training data that converts the electromagnetic signal data into a semantic feature vector Feat; that is, the training data set containing the second training data input into the training framework , represents the i-th training data extracted as semantic feature vector, and N represents the number of samples in the dataset.

[0113] Through the above embodiment, the electromagnetic signal is extracted as a high-dimensional semantic feature, which is transmitted instead of the original signal. In this way, only the local features of the signal related to the task need to be transmitted each time, avoiding information redundancy and excessive resource occupation, and greatly improving the transmission efficiency of the communication system; at the same time, the data information is projected from the low-dimensional space to the high-dimensional space through the transmission of semantic features, which is an effective data encryption method, protecting sensitive information from being stolen, thereby further improving the security of signal transmission.

[0114] In some embodiments, the semantic feature extraction process performed on the electromagnetic signal to obtain the semantic feature vector includes:

[0115] The electromagnetic signal is input into a trained signal encoder for semantic feature extraction, generating a semantic feature vector. Specifically, the original electromagnetic signal is first framed, windowed, and denoised (e.g., using a wavelet packet transform or adaptive filtering) to suppress spectral leakage and enhance information in key frequency bands. The signal is then converted into a time-frequency spectrum (e.g., a two-dimensional time-frequency matrix generated by a short-time Fourier transform) as the input representation for the deep learning model. Finally, a trained signal encoder (e.g., a convolutional neural network or autoencoder) automatically extracts abstract features through multiple layers of nonlinear transformations.

[0116] Through the above embodiments, through end-to-end training of the deep learning model, traditional time-frequency analysis and semantic modeling are integrated into a unified framework, which significantly improves the automation level and generalization ability of feature extraction.

[0117] In some embodiments, the signal receiving device is equipped with a signal decoder for restoring features of a noise interference semantic vector; the training process of the signal encoder and the signal decoder includes:

[0118] Acquire third training data; wherein, the third training data can be electromagnetic signal data from the same batch as the first training data, or can be training data from different batches.

[0119] Next, the third training data is input into the second encoder network in the second autoencoder model, and the third training data is encoded to obtain training semantic feature data; the training semantic feature data is input into the second decoder network in the second autoencoder model, and the training semantic feature data is reconstructed to obtain training reconstructed data; based on the third training data and the training reconstructed data, the semantic feature loss function result is calculated; the gradient of the semantic feature loss function result is back-transferred to the second autoencoder model for iterative training, and a second target encoder model is generated; the second target encoder model includes a signal encoder and a signal decoder.

[0120] Specifically, the third training data is input into the second encoder network of the second autoencoder model, and high-order semantic features are extracted through nonlinear mapping to generate training semantic feature data; the feature data is then input into the second decoder network, and the signal is reconstructed using deconvolution or transposed convolution operations, and the training reconstructed data is output; based on the difference between the original third training data and the reconstructed data (such as mean square error or cross entropy loss), the semantic feature loss function value is calculated; through the gradient backpropagation algorithm, the loss gradient is propagated back to the encoder and decoder networks, and the parameters of both are optimized simultaneously; after multiple rounds of iterative training, the model converges to the second target encoder model, the signal encoder it contains can accurately capture the essential semantic information of the electromagnetic signal, and the signal decoder has high-fidelity reconstruction capabilities, forming an end-to-end semantic feature extraction and reconstruction system as a whole.

[0121] See also Figure 5 The input data (third training data) passes through the second encoder's multi-layer convolutional network, extracting features layer by layer, from low-level features (such as edges and textures) to high-level features (such as semantic information), and outputting training semantic feature data. This training semantic feature data is then input into the second decoder's multi-layer deconvolutional network, which undergoes layer-by-layer deconvolution operations and outputs training reconstructed data. A loss function is calculated based on the input third training data and the output training reconstructed data to minimize the difference between the input signal and the reconstructed result, achieving iterative training.

[0122] Through the above embodiment, by iteratively training the second autoencoder model, the signal encoder and signal decoder are jointly trained, so that the encoder and decoder can promote each other and optimize collaboratively during the training process. Specifically, the encoder can learn more accurate and discriminative semantic feature representations under the feedback of the decoder. These features can better capture the essential properties of electromagnetic signals; at the same time, the decoder can continuously improve the accuracy and fidelity of signal reconstruction under the feature input of the encoder, so as to minimize the difference between the reconstructed signal and the original signal. This joint training method not only helps to improve the performance of the entire autoencoder model, but also enables the model to show stronger robustness and generalization ability in electromagnetic signal processing tasks.

[0123] In some embodiments, the above-mentioned step of transmitting the noise interference electromagnetic signal to the signal receiving device may further include the following steps:

[0124] Obtain a preset random encryption seed; based on the random encryption seed, perform a channel dynamic sorting operation on the noise interference electromagnetic signal to obtain a random encryption signal, and transmit the random encryption signal to a signal receiving device.

[0125] A high-entropy random encryption seed (e.g., a pseudo-random sequence generated using a linear feedback shift register) is pre-generated. This seed determines the initial state and dynamic behavior of the subsequent channel sorting algorithm. The seed can be distributed to both communicating parties via an encryption protocol (e.g., a key exchange algorithm) to ensure synchronization and security. In electromagnetic environments subject to noise interference, the channel priority of signal transmission is dynamically adjusted based on the random encryption seed. For example, in a dual-channel architecture, the pseudo-random sequence generated by the seed controls the switching timing of the primary and backup channels, randomly distributing noise interference signals across the channels. Specifically, by real-time monitoring of channel quality (e.g., signal-to-noise ratio and bit error rate) and combining seed-defined sorting rules (e.g., round-robin or random weighting), channels with severe interference are temporarily blocked or weighted down, forming a dynamically adaptive, anti-interference channel matrix. After channel sorting, the transmitted signal is embedded in a randomized transmission path determined by the seed, forming a randomized encrypted signal encrypted at the physical layer. The signal receiving device uses the synchronization seed to restore the channel sorting rules and accurately extract valid signals. Unauthorized eavesdroppers, lacking the seed information, struggle to parse the signal path, thus preventing data leakage or tampering.

[0126] More specifically, a trained random encryptor and inverse random decryptor are deployed on the signal sending device and the signal receiving device respectively. The input of the random encryptor F is the electromagnetic signal with noise added output by the first noise generator, or the semantic feature with noise added output by the second signal noise generator, and the output is spliced ​​with the random encryption seed and sent through the signal sending unit. In this embodiment, the random transmitter uses a random channel shuffling method to further encrypt the signal. For the semantic feature signal with anti-learning interference noise added, the channel dimension is randomly shuffled, and the random encryption seed is The final signal to be sent is:

[0127] Signal=[shuffle(x;r),r];

[0128] In the above formula, x represents the input signal to the random encryptor, and shuffle(·) represents the random shuffling of the channel dimension of the original semantic feature signal. The shuffling rule is controlled by the random encryption seed r.

[0129] Through the above embodiment, by obtaining a preset random encryption seed to generate a dynamic sorting rule, the channel of the electromagnetic signal carrying noise interference is rearranged and encrypted, so that the signal presents a random spectrum distribution feature during the transmission process, and finally the random encrypted signal that is difficult to be parsed by the eavesdropper is transmitted to the signal receiving device, thereby realizing the concealment of the real transmission path of the signal through the channel switching mechanism driven by the random encryption seed, and effectively improving the anti-interception and anti-interference capabilities of the electromagnetic signal.

[0130] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0131] This embodiment also provides an electromagnetic signal transmission system, including a signal sending device and a signal receiving device; the signal sending device is communicatively connected to the signal receiving device and is used to execute the electromagnetic signal transmission method described in any of the above embodiments.

[0132] The present application is described in detail below with reference to specific embodiments. Figure 6 is a schematic diagram of an electromagnetic signal transmission system architecture according to an embodiment of the present application, such as Figure 6 As shown, the system includes a signal sending device and a signal receiving device.

[0133] The signal transmission device includes a data preprocessor, a signal encoder, a second noise generator, a random encryptor, and a signal transmitter. The data preprocessor is used to preprocess electromagnetic signal data to meet the input requirements of the transmission system; the signal encoder is used to extract the characteristic representation of the electromagnetic signal and obtain the signal's high-dimensional semantic feature vector; the second noise generator is used to generate a noise vector that is resistant to learning interference and perform multi-scale fusion with the semantic features; the random encryptor is used to further randomly encrypt the output of the noise generator based on a random seed, and then concatenate the encrypted result with the encryption seed to obtain the complete signal to be transmitted; and the signal transmitter is used to transmit the signal. In this way, the electromagnetic signal and the random encryption seed are input into the signal transmission device, and the various modules in the signal transmission device perform processing and signal transmission operations to achieve signal transmission.

[0134] The signal receiving device includes a signal receiver, a derandom decryptor, a second interference remover, and a signal decoder. The signal receiver is used to receive the signal; the derandom decryptor is used to decode the random encryption seed to implement the reverse random encryption process; the second interference remover is used to eliminate interference information in the received signal; and the signal decoder is used to recover the information and ultimately obtain the output signal.

[0135] It is understandable that the signal transmitter and signal receiver mentioned above can be transmitters and receivers commonly used in communication networks.

[0136] Furthermore, for the above electromagnetic signal transmission system architecture, its construction process can be found in Figure 7 , the process includes the following steps:

[0137] Step S701: collect samples and construct a signal data set.

[0138] Step S702: Building an electromagnetic signal secure transmission system.

[0139] Step S703: training the signal encoder and the signal decoder.

[0140] Step S704: training the second noise generator and the second interference remover.

[0141] Step S705: Complete the training and perform signal transmission.

[0142] This embodiment further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0143] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0144] Optionally, in this embodiment, the processor may be configured to execute the following steps through a computer program:

[0145] S1, obtaining an electromagnetic signal to be transmitted.

[0146] S2, inputting the electromagnetic signal into the trained first noise generator, performing noise addition processing on the electromagnetic signal based on the first anti-learning interference noise carried by the first noise generator, and generating a noise interference electromagnetic signal; wherein, the first noise generator is generated by inputting the acquired first training data into the initial neural network model, and injecting the first preset interference noise into each network layer in the initial neural network model, performing multi-level fusion processing on the first preset interference noise and the first training data through each network layer, and iterative training; the first anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model.

[0147] S3, transmitting the noise interference electromagnetic signal to a signal receiving device.

[0148] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be repeated here.

[0149] In addition, in conjunction with the electromagnetic signal transmission method in the above embodiments, the present application can provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, any one of the electromagnetic signal transmission methods in the above embodiments is implemented.

[0150] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0151] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0152] Those skilled in the art should understand that the various technical features of the above-described embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0153] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for transmitting electromagnetic signals, characterized in that: Applied to a signal sending device, the method includes: obtaining an electromagnetic signal to be transmitted; Inputting the electromagnetic signal into a trained first noise generator, performing noise addition processing on the electromagnetic signal based on a first anti-learning interference noise carried by the first noise generator, to generate a noise interference electromagnetic signal; The first noise generator is generated by inputting the acquired first training data into an initial neural network model, injecting the first preset interference noise into each network layer of the initial neural network model, performing multi-level fusion processing on the first preset interference noise and the first training data through each network layer, and iteratively training; the first anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model; The noise interference electromagnetic signal is transmitted to a signal receiving device; the signal receiving device is equipped with a first interference remover for performing noise removal processing on the noise interference electromagnetic signal, and the training process of the first noise generator and the first interference remover includes: Inputting the first training data into a first autoencoder model; wherein the initial neural network model is set to the first autoencoder model; performing multi-layer feature extraction processing on the first training data via a first encoder network of the first autoencoder model to obtain hierarchical feature data corresponding to each network level in the first encoder network, and fusing the injected first preset interference noise with the hierarchical feature data at each network level in the first encoder network to output a first fused training feature; Inputting the first fused training features into a first decoder network of the first autoencoder model for decoding processing to obtain first reconstructed feature data; Based on the first training data and the first reconstruction feature data, a first reconstruction loss function result is calculated, and the gradient of the first reconstruction loss function result is back-transferred to the first autoencoder model for iterative training to obtain a first target encoder model; the first target encoder model includes the first noise generator and the first interference remover.

2. The electromagnetic signal transmission method according to claim 1, wherein: The method of calculating a first reconstruction loss function result based on the first training data and the first reconstruction feature data, and back-transferring the gradient of the first reconstruction loss function result to the first autoencoder model for iterative training to obtain a first target encoder model includes: In a first training stage, calculating the first reconstruction loss function result; In the second training stage, a preset stage disturbance is input into the first autoencoder model, and the first preset interference noise, the stage disturbance and the hierarchical feature data are fused using the first autoencoder model to output a second fused training feature, and the second fused training feature is input into the first decoder network for decoding to obtain second reconstructed feature data; Calculating a second reconstruction loss function result based on the first training data and the second reconstruction feature data; The gradients of the first reconstruction loss function result and the second reconstruction loss function result are back-transferred to the first autoencoder model for iterative training to obtain the first target encoder model.

3. The electromagnetic signal transmission method according to claim 2, wherein: The first training data carries a classification label; and the first preset interference noise, the stage disturbance, and the hierarchical feature data are fused using the first autoencoder model to output a second fused training feature, including: Using the first autoencoder model, fusing the first preset interference noise, the stage disturbance, and the hierarchical feature data to output a second fused training feature, performing classification processing based on the second fused training feature, and outputting a classification result; The step of back-transferring the gradients of the first reconstruction loss function result and the second reconstruction loss function result to the first autoencoder model for iterative training to obtain the first target encoder model includes: Calculating a classification loss function result based on the classification result and the classification label, and calculating a joint loss function result based on the classification loss function result and the second reconstruction loss function result; The gradient of the first reconstruction loss function result and the joint loss function result is back-transferred to the first autoencoder model for iterative training to obtain the first target encoder model.

4. The electromagnetic signal transmission method according to claim 1, wherein: After acquiring the electromagnetic signal to be transmitted, the method further includes: Performing semantic feature extraction processing on the electromagnetic signal to obtain a semantic feature vector; Inputting the semantic feature vector into a trained second noise generator, performing noise addition processing on the electromagnetic signal based on a second anti-learning interference noise carried by the second noise generator, and generating a noise interference semantic vector; The second noise generator is generated by inputting the acquired second training data into an initial neural network model, injecting a second preset interference noise into each network layer of the initial neural network model, performing multi-level fusion processing on the second preset interference noise and the second training data through each network layer, and iteratively training; the second anti-learning interference noise is obtained by minimizing the first preset interference noise during the iterative training process of the initial neural network model; The noise interference semantic vector is transmitted to the signal receiving device.

5. The electromagnetic signal transmission method according to claim 4, characterized in that: The performing semantic feature extraction processing on the electromagnetic signal to obtain a semantic feature vector includes: The electromagnetic signal is input into a trained signal encoder for semantic feature extraction to obtain the semantic feature vector.

6. The electromagnetic signal transmission method according to claim 5, characterized in that: The signal receiving device is equipped with a signal decoder for restoring features of the noise interference semantic vector; the training process of the signal encoder and the signal decoder includes: Obtaining third training data; Inputting the third training data into a second encoder network in a second autoencoder model, and encoding the third training data to obtain training semantic feature data; Inputting the training semantic feature data into a second decoder network in the second autoencoder model, and reconstructing the training semantic feature data to obtain training reconstructed data; Calculating a semantic feature loss function result based on the third training data and the training reconstruction data; The gradient of the semantic feature loss function result is back-transferred to the second autoencoder model for iterative training, and a second target encoder model is generated; the second target encoder model includes the signal encoder and the signal decoder.

7. The electromagnetic signal transmission method according to any one of claims 1 to 6, characterized in that: The transmitting the noise interference electromagnetic signal to a signal receiving device includes: Get the preset random encryption seed; Based on the random encryption seed, a channel dynamic sorting operation is performed on the noise interference electromagnetic signal to obtain a random encryption signal, and the random encryption signal is transmitted to the signal receiving device.

8. An electromagnetic signal transmission system, characterized in that: including signal sending equipment and signal receiving equipment; The signal sending device is communicatively connected to the signal receiving device, and is used to perform the electromagnetic signal transmission method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the electromagnetic signal transmission method according to any one of claims 1 to 7 is implemented.

Images