OFDM encrypted signal PAPR suppression method and system based on auto-encoder

Abstract

The invention provides an OFDM (Orthogonal Frequency Division Multiplexing) encrypted signal PAPR (Peak to Average Power Ratio) suppression method and system based on an auto-encoder, and the method comprises the steps: collecting a data sequence of an OFDM system, splitting a complex signal of the data sequence into a real part and an imaginary part, and obtaining an input sequence; inputting a chaotic neural network; enabling the chaotic neural network to perform chaotic encryption on the input sequence to obtain encrypted data; inputting the encrypted data into an encoder, enabling the encoder to perform feature extraction on the encrypted data, and compressing the encrypted data to low latitude to obtain a low latitude result; inputting the low-latitude result into a decoder to obtain a reconstruction result; performing reverse fast Fourier transform on the reconstruction result to obtain a time domain signal; converting into a frequency domain form for reconstruction, and recovering an original signal; and restoring the data sequence according to the original signal. According to the method, the performance of the encrypted OFDM signal is remarkably improved, compared with the performance of the OFDM signal without adopting the suppression algorithm, the performance is improved by 10.7 dB, and compared with an existing PRNet scheme, the performance is improved by 2.1 dB.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication and signal processing technology, and specifically, relates to an OFDM encrypted signal PAPR suppression method and system based on an autoencoder. More specifically, the method combines the advantages of a chaotic neural network and an autoencoder, effectively reduces the peak-to-average power ratio of an OFDM signal, ensures the system bit error rate performance, and improves the security performance of an OFDM encryption system. Background Art

[0002] OFDM technology is a multi-carrier transmission technology widely used in wireless communication systems. It has the advantages of combating frequency selective fading and effectively utilizing spectrum resources.

[0003] However, the signal of OFDM system has a high PAPR, which leads to low power amplifier efficiency and easily causes nonlinear distortion of the signal, thus seriously affecting the transmission performance of the system. Therefore, many methods have been proposed to reduce the PAPR of OFDM system, including traditional clipping method, partial transmission sequence method and PAPR suppression method based on deep learning.

[0004] Specifically, chaotic systems have randomness, initial value sensitivity and nonlinear characteristics, which make them show great potential in the fields of communication encryption and signal processing. Chaotic encryption can effectively cover up signals in OFDM systems and improve system security by utilizing the high sensitivity and complex dynamic characteristics of chaotic systems. Chaotic neural networks combine the characteristics of chaotic systems and artificial neural networks, introduce chaotic dynamics into neural networks, and bring stronger nonlinear information processing capabilities, which helps to achieve better results in encryption and feature extraction.

[0005] The existing deep learning-based PRNet scheme has achieved certain results in PAPR suppression of OFDM systems, but it does not consider the new physical layer encryption system, has limited control over the signal bit error rate, and has certain deficiencies in the neural network architecture and loss function design, making it difficult to balance PAPR suppression with OFDM system security performance. Therefore, a new PAPR suppression method based on chaotic encryption and autoencoder is proposed to improve the existing PRNet scheme, so as to better balance PAPR suppression performance, system bit error rate performance and communication security.

[0006] Patent document CN114915398A discloses a chaotic encryption method, application, computer equipment and storage medium based on variational autoencoder. The scheme includes obtaining the original chaotic sequence; inputting the original chaotic sequence data into the pre-trained variational autoencoder model for reconstruction processing, and outputting the reconstructed chaotic sequence; and performing phase mask encryption and frequency mask encryption on the communication data according to the obtained reconstructed chaotic sequence. The variational autoencoder model includes an encoder network and a decoder network, the encoder network includes an alternating residual connection block and a self-attention layer, and the decoder network block includes a convolution layer and a loss layer arranged after each convolution layer. By introducing the variational autoencoder model, learning the generation rules of the chaotic sequence, replacing the inefficient chaotic sequence generation of linear solution differential equations, and through a large amount of chaotic sequence training data, the model performance ability and generalization ability are trained, which can effectively improve the generation efficiency of the chaotic sequence. The scheme has limited control over the signal bit error rate, and there are certain deficiencies in the architecture of the neural network and the design of the loss function, making it difficult to balance the PAPR suppression and the safety performance of the OFDM system at the same time. This problem needs to be solved urgently. Summary of the invention

[0007] In view of the defects in the prior art, the object of the present invention is to provide a method and system for suppressing the PAPR of an OFDM encrypted signal based on an autoencoder.

[0008] According to the present invention, a method for suppressing the PAPR of an OFDM encrypted signal based on an autoencoder is provided, comprising:

[0009] Step S1: collecting a data sequence of an OFDM system, splitting a complex signal of the data sequence into a real part and an imaginary part, and obtaining an input sequence according to the real part and the imaginary part;

[0010] Step S2: inputting the input sequence into the chaotic neural network; allowing the chaotic neural network to perform chaotic encryption on the input sequence to obtain encrypted data;

[0011] Step S3: inputting the encrypted data into an encoder, allowing the encoder to extract features from the encrypted data and compress it to a low latitude to obtain a low-latitude result; inputting the low-latitude result into a decoder to obtain a reconstruction result;

[0012] Step S4: performing an inverse fast Fourier transform on the reconstruction result to obtain a time domain signal;

[0013] Step S5: Convert the time domain signal into a frequency domain form for reconstruction to restore the original signal; restore the data sequence in step S1 according to the original signal.

[0014] Preferably, in step S1, the real part and the imaginary part are arranged in the order of splitting;

[0015] In step S2, the initial values ​​of the chaotic neural network, i.e., x1, x2, x3, x4, x5, x6, are set as keys, the keys are input into the ode45 function, the solution is looped, the sum of the first group of chaotic sequences is calculated and the remainder thereof and 6 are taken to obtain the encrypted data.

[0016] Preferably, in the step S3, the encrypted data is mapped through linear changes and activation functions to obtain a low-latitude result;

[0017] The low-dimensional result is the encrypted data after the linear change and activation function mapping to remove signal redundancy; the activation function includes: Sigmoid function or ReLU function;

[0018] In the step S4, the time domain signal is converted into digital-to-analog format and then wirelessly transmitted.

[0019] Preferably, in step S3, the encoder and the decoder determine the loss function of the chaotic neural network; the mathematical expression of the loss function is:

[0020]

[0021] in, represents the loss function of the neural network, which is determined by the encoder and the decoder; λ represents a preset parameter; X represents the original input data; represents the output of the decoder;

[0022]

[0023] Wherein, b represents another preset parameter; represents the damage of the signal after passing through the channel, and the loss function is defined according to the difference value; the L1, that is, The mathematical expression is:

[0024]

[0025] Wherein, X represents the signal point transmitted by the transmitting end; represents the output signal point of the decoder, FFF represents fast Fourier transform; H represents the received data after the signal is affected by the channel; f represents the output function of the encoder, IFFT represents inverse fast Fourier transform; g represents the output function of the decoder; ||||2 represents the two-norm.

[0026] According to the present invention, an OFDM encrypted signal PAPR suppression system based on an autoencoder is provided, comprising:

[0027] Module M1: collects a data sequence of an OFDM system, splits a complex signal of the data sequence into a real part and an imaginary part, and obtains an input sequence according to the real part and the imaginary part;

[0028] Module M2: input the input sequence into the chaotic neural network; and let the chaotic neural network perform chaotic encryption on the input sequence to obtain encrypted data;

[0029] Module M3: input the encrypted data into an encoder, and let the encoder extract features from the encrypted data and compress it to a low latitude to obtain a low-latitude result; input the low-latitude result into a decoder to obtain a reconstruction result;

[0030] Module M4: performing inverse fast Fourier transform on the reconstruction result to obtain a time domain signal;

[0031] Module M5: convert the time domain signal into a frequency domain form for reconstruction to restore the original signal; restore the data sequence in module M1 according to the original signal.

[0032] Preferably, in the module M1, the real part and the imaginary part are arranged in the order of splitting;

[0033] In the module M2, the initial values ​​of the chaotic neural network, i.e. x1, x2, x3, x4, x5, x6, are set as keys, the keys are input into the ode45 function, the solution is looped, the sum of the first group of chaotic sequences is calculated and the remainder thereof and 6 are taken to obtain the encrypted data.

[0034] Preferably, in the module M3, the encrypted data is mapped by linear change and activation function to obtain a low-latitude result;

[0035] The low-dimensional result is the encrypted data after the linear change and activation function mapping to remove signal redundancy; the activation function includes: Sigmoid function or ReLU function;

[0036] In the module M4, the time domain signal is converted into digital-to-analog format and then wirelessly transmitted.

[0037] Preferably, in the module M3, the encoder and the decoder determine the loss function of the chaotic neural network; the mathematical expression of the loss function is:

[0038]

[0039] in, represents the loss function of the neural network, which is determined by the encoder and the decoder; λ represents a preset parameter; X represents the original input data; represents the output of the decoder;

[0040]

[0041] Wherein, b represents another preset parameter; represents the damage of the signal after passing through the channel, and the loss function is defined according to the difference value; the L1, that is, The mathematical expression is:

[0042]

[0043] Wherein, X represents the signal point transmitted by the transmitting end; represents the output signal point of the decoder, FFT represents fast Fourier transform; H represents the received data after the signal is affected by the channel; f represents the output function of the encoder, IFFT represents inverse fast Fourier transform; g represents the output function of the decoder; ||||2 represents the two-norm.

[0044] According to a computer-readable storage medium storing a computer program provided by the present invention, when the computer program is executed by a processor, the steps of the OFDM encrypted signal PAPR suppression method based on an autoencoder are implemented.

[0045] An electronic device provided according to the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the steps of the OFDM encrypted signal PAPR suppression method based on the autoencoder are implemented.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1. The present invention discloses an encrypted orthogonal frequency division multiplexing based on an autoencoder, namely, Orthogonal frequency-division multiplexing, referred to as an OFDM signal peak to average power ratio suppression method, namely, Peak to Average Power Ratio, referred to as PAPR suppression method; the present invention utilizes chaotic encryption to effectively protect the signal, and extracts and reconstructs the encrypted signal through an autoencoder to reduce the PAPR of the signal in the encrypted OFDM system.

[0048] 2. The present invention does not require the design of additional complex accessories, and can achieve low PAPR and high-security signal transmission through simple encryption and self-encoding processes.

[0049] 3. The loss function adopted by the present invention includes multiple optimization items for controlling bit error rate, PAPR and encryption security, ensuring that the system maintains a low bit error rate and high encryption strength while achieving PAPR suppression.

[0050] 4. The present invention has a simple structure and high computational efficiency. It can significantly improve security performance in complex wireless communication environments and improve the transmission performance of the system. It has a significant improvement in reducing the PAPR performance of encrypted OFDM signals, which is 10.7 dB higher than that without the suppression algorithm and 2.1 dB higher than the existing PRNet solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0052] Figure 1 A schematic diagram of the process provided by the present invention;

[0053] Figure 2 A schematic diagram of the phase space of a chaotic neural network provided by the present invention;

[0054] Figure 3 A schematic diagram of an encoder and a decoder provided by the present invention;

[0055] Figure 4 This is a block diagram of the internal structure of the encoder provided by the present invention. DETAILED DESCRIPTION

[0056] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0057] Since the loss function and network structure of the chaotic neural network and the autoencoder are combined, the encrypted system will cause the PAPR to tend to an unstable state. The present invention aims to solve the problem that the PAPR suppression method of the OFDM system in the prior art has a large impact on the system bit error rate and insufficient performance optimization, and proposes a PAPR suppression method based on chaotic encryption and autoencoder. The method combines the chaotic neural network with the autoencoder network in the OFDM system to intelligently learn the intrinsic characteristics of the input signal, thereby applying the encoder at the transmitting end for encryption and PAPR suppression, and applying the decoder at the receiving end to restore the original signal.

[0058] The improvements of the present invention include the optimization of the loss function and model structure of the chaotic neural network and the autoencoder. In the design of the loss function, the present invention proposes a new form of loss function, which includes three parts: one part is a term for controlling the bit error rate performance, another part is a term for controlling the PAPR performance, and the third part is a term for encryption security. During the training process, when the system bit error rate reaches the expected target value, the loss function will further tend to reduce the PAPR while ensuring the security of chaotic encryption. In addition, the present invention has made improvements in the design of the autoencoder structure, replacing the activation function of the last layer of the autoencoder from ReLU to eLU to improve the reconstruction effect of the signal and ensure that the decoder can completely restore the input data, thereby further optimizing the PAPR suppression performance.

[0059] According to the present invention, a method for suppressing the PAPR of an OFDM encrypted signal based on an autoencoder is provided, comprising:

[0060] S1: Signal preprocessing, modulate the original data sequence of the OFDM system, split the complex signal into real and imaginary parts, and combine them in sequence to form the input sequence of the autoencoder. This step ensures that the input signal is suitable for subsequent autoencoder processing.

[0061] S2: Chaotic neural network encryption, input data sequence into the chaotic neural network, the chaotic neural network uses an activation function based on the combination of multiple frequency cosine signal, that is, Multiple Frequency Cosine Signal, referred to as MFCS and piecewise linear function. Through this network, the input signal is chaotically encrypted and the encrypted data is output.

[0062] Specifically, by calling the ode45 function in MATLAB to solve the differential equation, the chaotic system will continuously iterate x1, x2, x 3, The values ​​of x4, x5, and x6 are used to change the length of the chaotic sequence by setting the calculation step of the ode45 algorithm without changing its time approximation to select the number of loops. That is, there are 6 groups of independent chaotic sequences. Figure 2 is the phase space diagram of the generated chaotic sequence, the initial value of the chaotic system, that is, x1, x2, x 3, x4, x5, x6 are keys. Calculate the sum of the first group of chaotic sequences and take the remainder of 6. The number obtained is the key. XOR encrypt the signal to obtain the encrypted signal. Since the encrypted signal affects the distribution of the OFDM signal, the PAPR value will be affected to a certain extent.

[0063] S3: Encoder processing, inputting the chaotically encrypted data into the autoencoder model, such as Figure 3As shown in the figure, the autoencoder consists of two parts: the encoder and the decoder. First, the encoder extracts features from the data and compresses the high-dimensional input into a low-dimensional representation, that is, the original input is a high-dimensional complex and redundant signal, which will bring high computational complexity. It is mapped into a low-dimensional vector through linear transformation and activation function, and then the compressed representation is input into the decoder for reconstruction. The loss function is defined as a representation related to the bit error rate. At the decoder end, an improved eLU activation function is used to improve the accuracy of signal restoration. The activation function includes: Sigmoid function or ReLU function.

[0064] In the autoencoder, the loss function of the neural network is defined as

[0065]

[0066] in, represents the loss function of the neural network, which is determined by the encoder and the decoder; λ represents a preset parameter; X represents the signal point transmitted by the transmitter; Represents the output signal point of the decoder;

[0067] Wherein, L2(X) represents the loss value of the PAPR of the signal; Represents the loss value before and after encoding, and the expression is:

[0068]

[0069] Among them, b represents another preset parameter, Represents the damage of the signal after passing through the channel, and the loss function is defined according to the difference value, where The definition is as follows:

[0070]

[0071] In other words, That is, the mathematical expression of L1 is:

[0072]

[0073] Among them, X represents the signal point transmitted by the transmitter, represents the output of the decoder, FFT represents fast Fourier transform; H represents the received data after the signal is affected by the channel; f represents the output function of the encoder, IFFT represents the inverse fast Fourier transform; MSE represents the mean square error; g represents the generating function of the decoder; ||||2 represents the two-norm, which is the square root of the sum of the squares of a given vector;

[0074] L2= APR(x(n))= APR(IfFT(f(X))) (4)

[0075] Among them, L2 represents the PAPR value of the signal after transmission; PAPR represents the peak-to-average power ratio of the input signal; x(n) represents the original input signal of n points, where n represents n signal points; IFFT represents inverse fast Fourier transform.

[0076] Therefore, the loss of the neural network is related to the PAPR value, and the direction of the gradient descent of the neural network is consistent with the direction of the decrease of the PAPR value.

[0077] In the formula, λ and b parameters are artificially set and can be changed according to specific application scenarios. When L1 is greater than b, the system bit error rate does not reach the expected effect. According to formula (2), L3 is not 0, and L is still related to the system bit error rate. When changing network parameters after training, the bit error rate performance needs to be considered; when L1 is less than b, it means that the system bit error rate has been optimized to below b. According to formula (2), it can be calculated that L3 is equal to 0. At this time, it is only related to the system λL2(X) value, and L2 is related to PAPR((())), that is, it is related to PARP. The training phase will be more inclined to reduce PAPR and maximize the effect of PAPR suppression.

[0078] Specifically, the OFDM system has N subcarriers, and the signal on the i-th subcarrier is x' i and is a complex signal, then the sequence to be input to the encoder is X' = [x'1, x'2, ..., x' N ].like Figure 4 As shown, the autoencoder does not accept complex input, so the input sequence needs to be processed before entering the encoder to split each signal in X' into the real part x' Re,i and the imaginary part x' Im,i And arrange them in a one-to-one correspondence. Similarly, recombining the real and imaginary parts into a complex signal requires the sequence y' i is the frequency domain signal on the i-th subcarrier. f(X) then passes through the N-point IFFT module and DAC module before entering the channel for transmission. At the receiver end, after the received signal is processed by the ADC module and the FFT module, the signal input to the decoder is FFT(H*IFFT(f(X))).

[0079] The trained decoder will reconstruct the input sequence based on the input Contrary to the operation on the receiver, you need to The sequence in the middle is reassembled into complex numbers in order, and then the output sequence is obtained by parallel-to-serial conversion. The original data can then be obtained through demodulation.

[0080] S4: Inverse Fast Fourier Transform, or IFFT, in S3, the input signal is encoded using the trained encoder to obtain a low PAPR signal, and the signal is subjected to an Inverse Fast Fourier Transform, or IFFT, to convert it into a time domain signal, which is then transmitted through a digital-to-analog converter, or DAC, into a channel. The purpose of this step is to convert the signal into a form suitable for wireless transmission.

[0081] S5: Signal processing at the receiving end. At the receiving end, the received time domain signal is first processed by analog-to-digital conversion, i.e., ADC processing, and then fast Fourier transform, i.e., FFT, to convert the signal back to frequency domain form. This step ensures that the signal can be correctly processed by the decoder.

[0082] S6: Decoding and decryption, input the frequency domain signal after FFT processing into the decoder obtained by S3, reconstruct the signal and restore the original signal. Then decrypt it through the inverse process of chaotic neural network to restore the original data signal.

[0083] S7: Signal restoration, recombining the decoded signal into a complex form, demodulating it, and obtaining the final original data sequence. Through this process, the original data of the transmitter is successfully restored to ensure the communication quality of the system.

[0084] Specifically, in step S1, the real part and the imaginary part of the data after splitting are arranged in order to ensure that each component of the complex signal is correctly input into the autoencoder and encrypted one by one according to the subsequent chaotic sequence.

[0085] Specifically, in step S2, the activation function of the chaotic neural network is formed by superimposing multiple single-frequency cosine signals to form a complex nonlinear mapping relationship, thereby ensuring the chaos and security of encryption.

[0086] Specifically, in step S3, the activation function of the last layer of the autoencoder adopts eLU to ensure that the negative part of the signal can be correctly reconstructed and the accuracy of signal restoration is enhanced, and the loss function is reconstructed to ensure that the PAPR value can be maximally suppressed.

[0087] Specifically, in step S4, a standard IFFT algorithm is used to convert the frequency domain signal into a time domain signal, so as to ensure that the signal can adapt to the transmission requirements of the OFDM system after being processed by chaotic encryption and autoencoder.

[0088] Specifically, in step S6, the chaotic neural network decryption process adopts parameter settings opposite to those of the encryption process, ensuring that the legitimate recipient can correctly restore the original data, while the illegal recipient cannot decipher the signal.

[0089] In addition, the method of the present invention adopts a dynamic learning rate and an Adam optimizer for training, ensuring that the model can converge quickly and effectively during the training process, thereby achieving the optimal suppression effect on PAPR and maintaining signal security without degrading the system bit error rate.

[0090] The present invention also provides an OFDM encrypted signal PAPR suppression system based on an autoencoder. The OFDM encrypted signal PAPR suppression system based on an autoencoder can be implemented by executing the process steps of the OFDM encrypted signal PAPR suppression method based on an autoencoder, that is, a person skilled in the art can understand the OFDM encrypted signal PAPR suppression method based on an autoencoder as a preferred implementation of the OFDM encrypted signal PAPR suppression system based on an autoencoder.

[0091] According to the present invention, an OFDM encrypted signal PAPR suppression system based on an autoencoder is provided, comprising:

[0092] Module M1: collects a data sequence of an OFDM system, splits a complex signal of the data sequence into a real part and an imaginary part, and obtains an input sequence according to the real part and the imaginary part;

[0093] Module M2: input the input sequence into the chaotic neural network; and let the chaotic neural network perform chaotic encryption on the input sequence to obtain encrypted data;

[0094] Module M3: input the encrypted data into an encoder, and let the encoder extract features from the encrypted data and compress it to a low latitude to obtain a low-latitude result; input the low-latitude result into a decoder to obtain a reconstruction result;

[0095] Module M4: performing inverse fast Fourier transform on the reconstruction result to obtain a time domain signal;

[0096] Module M5: convert the time domain signal into a frequency domain form for reconstruction to restore the original signal; restore the data sequence in module M1 according to the original signal.

[0097] Those skilled in the art know that, in addition to realizing the system and its various devices, modules, and units provided by the present invention in a purely computer-readable program code, it is entirely possible to realize the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for realizing various functions can also be regarded as structures within the hardware component; the devices, modules, and units for realizing various functions can also be regarded as both software modules for realizing the method and structures within the hardware component.

[0098] In the description of the present application, it should be understood that the terms "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0099] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for suppressing PAPR of OFDM encrypted signals based on autoencoder, characterized in that: include: Step S1: collecting a data sequence of an OFDM system, splitting a complex signal of the data sequence into a real part and an imaginary part, and obtaining an input sequence according to the real part and the imaginary part; Step S2: inputting the input sequence into the chaotic neural network; The chaotic neural network is used to perform chaotic encryption on the input sequence to obtain encrypted data; Step S3: inputting the encrypted data into an encoder, allowing the encoder to extract features from the encrypted data and compress it to a low latitude to obtain a low-latitude result; Inputting the low-latitude result into a decoder to obtain a reconstruction result; Step S4: performing an inverse fast Fourier transform on the reconstruction result to obtain a time domain signal; Step S5: Convert the time domain signal into a frequency domain form for reconstruction to restore the original signal; restore the data sequence in step S1 according to the original signal.

2. The method for suppressing the PAPR of an OFDM encrypted signal based on an autoencoder according to claim 1, characterized in that: In the step S1, the real part and the imaginary part are arranged in the order of splitting; In step S2, the initial values ​​of the chaotic neural network, i.e., x1, x2, x3, x4, x5, x6, are set as keys, the keys are input into the ode45 function, the solution is looped, the sum of the first group of chaotic sequences is calculated and the remainder thereof and 6 are taken to obtain the encrypted data.

3. The method for suppressing the PAPR of an OFDM encrypted signal based on an autoencoder according to claim 1, characterized in that: In the step S3, the encrypted data is mapped through linear changes and activation functions to obtain a low-latitude result; The low-dimensional result is the encrypted data after the linear transformation and activation function mapping to remove signal redundancy; The activation function includes: a Sigmoid function or a ReLU function; In the step S4, the time domain signal is converted into digital-to-analog format and then wirelessly transmitted.

4. The method for suppressing the PAPR of an OFDM encrypted signal based on an autoencoder according to claim 3, characterized in that: In step S3, the encoder and the decoder determine the loss function of the chaotic neural network; the mathematical expression of the loss function is: in, represents the loss function of the neural network, which is determined by the encoder and the decoder; λ represents a preset parameter; X represents the original input data; represents the output of the decoder; L2(X) represents the loss value of the PAPR of the signal; Indicates the loss value before and after encoding; The mathematical expression is: Wherein, b represents another preset parameter; It represents the damage of the signal after passing through the channel, and the loss function is defined according to the difference value; the mathematical expression of L1 is: L1=||Xg(FFT((H×IFFT(f(X))))||2 (3) Among them, X represents the signal point transmitted by the transmitter, represents the output of the decoder, FFT represents fast Fourier transform; H represents the received data after the signal is affected by the channel; f represents the output function of the encoder, IFFT represents inverse fast Fourier transform; g represents the generating function of the decoder; ||||2 represents the two-norm.

5. An OFDM encrypted signal PAPR suppression system based on autoencoder, characterized in that: include: Module M1: collects a data sequence of an OFDM system, splits a complex signal of the data sequence into a real part and an imaginary part, and obtains an input sequence according to the real part and the imaginary part; Module M2: inputting the input sequence into the chaotic neural network; The chaotic neural network is used to perform chaotic encryption on the input sequence to obtain encrypted data; Module M3: inputting the encrypted data into an encoder, allowing the encoder to extract features from the encrypted data and compress it to a low latitude to obtain a low-latitude result; Inputting the low-latitude result into a decoder to obtain a reconstruction result; Module M4: performing inverse fast Fourier transform on the reconstruction result to obtain a time domain signal; Module M5: convert the time domain signal into a frequency domain form for reconstruction to restore the original signal; restore the data sequence in module M1 according to the original signal.

6. The OFDM encrypted signal PAPR suppression system based on autoencoder according to claim 5, characterized in that: In the module M1, the real part and the imaginary part are arranged in the order of splitting; In the module M2, the initial values ​​of the chaotic neural network, i.e. x1, x2, x3, x4, x5, x6, are set as keys, the keys are input into the ode45 function, the solution is looped, the sum of the first group of chaotic sequences is calculated and the remainder thereof and 6 are taken to obtain the encrypted data.

7. The OFDM encrypted signal PAPR suppression system based on autoencoder according to claim 5, characterized in that: In the module M3, the encrypted data is mapped through linear changes and activation functions to obtain a low-dimensional result; The low-dimensional result is the encrypted data after the linear transformation and activation function mapping to remove signal redundancy; The activation function includes: a Sigmoid function or a ReLU function; In the module M4, the time domain signal is converted into digital-to-analog format and then wirelessly transmitted.

8. The OFDM encrypted signal PAPR suppression system based on autoencoder according to claim 7, characterized in that: In the module M3, the encoder and the decoder determine the loss function of the chaotic neural network; the mathematical expression of the loss function is: in, represents the loss function of the neural network, which is determined by the encoder and the decoder; λ represents a preset parameter; X represents the original input data; represents the output of the decoder; Wherein, b represents another preset parameter; It represents the damage of the signal after passing through the channel, and the loss function is defined according to the difference value; the mathematical expression of L1 is: L1=||Xg(FFT(H×IFFT(f(X))))||2 (3) Among them, X represents the signal point transmitted by the transmitter, represents the output of the decoder, FFT represents fast Fourier transform; H represents the received data after the signal is affected by the channel; f represents the output function of the encoder, IFFT represents inverse fast Fourier transform; g represents the generating function of the decoder; ||||2 represents the two-norm.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the OFDM encrypted signal PAPR suppression method based on an autoencoder according to any one of claims 1 to 4 are implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by a processor, the steps of the OFDM encrypted signal PAPR suppression method based on an autoencoder according to any one of claims 1 to 4 are implemented.

Images