A chaotic encryption method, application, computer device and storage medium based on variational autoencoder

Abstract

The present invention discloses a chaotic encryption method, application, computer equipment and storage medium based on a variational autoencoder. The chaotic encryption method includes obtaining an original chaotic sequence; inputting the original chaotic sequence data into a pre-trained variational autoencoder model for reconstruction processing, and outputting a reconstructed chaotic sequence; and performing phase masking encryption and frequency masking 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 alternating residual connection blocks and self-attention layers, and the decoder network block includes a convolution layer and a loss layer arranged after each convolution layer. By introducing a variational autoencoder model, the generation rules of chaotic sequences are learned, replacing the inefficient chaotic sequence generation of linear solution of differential equations. By using a large amount of chaotic sequence training data, the performance and generalization capabilities of the training model are improved, the generation efficiency of the chaotic sequence can be effectively improved, and the transmission performance can be guaranteed.

Description

Technical Field

[0001] The present invention relates to a chaotic encryption method, application, computer equipment and storage medium based on a variational autoencoder, and belongs to the technical field of communication transmission. Background Art

[0002] With the development of technologies such as 5G networks, big data, and the Internet of Things, the requirements for transmission network capacity and efficiency continue to increase. Passive Optical Access Networks (PONs) are a critical component of data transmission at the physical layer of communications. Downlink signal transmission operates in a broadcast mode, making the transmitted data vulnerable to eavesdropping and attacks by unauthorized users. To improve the security of PON data transmission, data security technologies based on the physical layer have become a research hotspot.

[0003] Chaotic secure communication uses a chaotic model to generate chaotic sequences, which are then used to mask and encrypt signals in communication systems. In optical communication systems, the pseudo-randomness and initial value sensitivity of chaotic models are key safeguards for their security. However, chaotic sequence models are calculated using linear iterative differential equations, which typically require significant computational resources for serial computation. As the number of chaotic sequences required increases, the efficiency of chaotic communication encryption schemes is affected. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a chaotic encryption method, application, computer device and storage medium based on a variational autoencoder. The chaotic encryption method based on the variational autoencoder can effectively improve the generation efficiency of chaotic sequences and ensure transmission performance. Its application in an orthogonal frequency division multiplexing passive optical access (OFDM-PON) system can effectively improve the security of the OFDM-PON system.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a chaotic encryption method based on a variational autoencoder, comprising:

[0007] The variational autoencoder model includes an encoder network and a decoder network, wherein the encoder network includes alternating residual connection blocks and self-attention layers, and the decoder network block includes convolutional layers and a loss layer after each convolutional layer;

[0008] The chaotic encryption method comprises:

[0009] Get the original chaotic sequence;

[0010] The original chaotic sequence data is input into the pre-trained variational autoencoder model for reconstruction, and the reconstructed chaotic sequence is output;

[0011] Performing phase masking encryption and frequency masking encryption on the communication data according to the obtained reconstructed chaotic sequence;

[0012] The reconstructed chaotic sequence includes a first chaotic sequence capable of performing phase masking on communication data and a second chaotic sequence capable of performing frequency masking on communication data.

[0013] Furthermore, the training method of the variational autoencoder model includes:

[0014] Get the original sequence for training;

[0015] Input the obtained original sequence into the encoder network for data encoding processing, and output an intermediate sequence that deviates from the original sequence, wherein the deviation of the intermediate sequence includes a standard deviation and an average value vector;

[0016] The intermediate sequence is input into the decoder network for data decoding processing. By setting an adjustable training cycle for iterative training, the model loss function converges, and the reconstructed sequence is output to complete the training of the variational autoencoder model.

[0017] Furthermore, the encoder network includes 1 convolutional layer, 3 residual connection blocks and 2 self-attention layers;

[0018] The original input sequence is first compressed by the convolutional layer, and the ReLu function is used as the activation function. Then it is reduced in dimension by three alternating residual connection blocks and two self-attention layers, and finally a low-dimensional intermediate sequence is output.

[0019] Furthermore, the convolution kernel size is 5 and the step size is 2.

[0020] Furthermore, the decoder network includes two 3×3 convolutional layers, two batch normalization layers, and one fully connected layer, and each of the 3×3 convolutional layers is connected to a dropout layer;

[0021] The intermediate sequence output by the encoder network first passes through a 3×3 convolution layer, a dropout layer, and a normalization layer, and is processed by the ReLu activation function. It then passes through a 3×3 convolution layer, a dropout layer, and a normalization layer, and is processed by a fully connected layer. The output is reconstructed for training. If the loss function of the model has reached the preset conditions, the intermediate sequence data is directly used as the output of the decoder network.

[0022] Furthermore, the convolution kernel size of the 3×3 convolution layer is 1, and the dropout rate of the dropout layer is 0.2.

[0023] Furthermore, the loss function is a minimization loss function, and the calculation formula includes:

[0024]

[0025] Where σ is the standard deviation of the encoder network output; m is the mean vector of the encoder network output; n is the data dimension of the original sequence of the encoder network input; i represents the i-th value in the standard deviation vector; p E (x|y) is the probability of outputting y when the encoder network input is x; p D (y|x) is the probability of outputting x when the decoder network input is y;

[0026] A second object of the present invention is to provide an application of a chaotic encryption method based on a variational autoencoder, wherein the chaotic encryption method based on the variational autoencoder is used in an orthogonal frequency division multiplexing passive optical access communication system for data encryption transmission. The communication system comprises:

[0027] The transmitting end is used to first perform serial-to-parallel switching on the original data sequence input, and then map the serial-to-parallel switched original data sequence to a first chaotic sequence through orthogonal amplitude modulation for phase masking; map the orthogonal amplitude modulated original data sequence to a second chaotic sequence through subcarrier modulation for frequency masking; and convert the original data sequence after subcarrier modulation from the frequency domain to the time domain through fast Fourier transform, and after parallel-to-serial conversion, input the signal into the optical fiber;

[0028] The receiving end is used to receive the signal outputted into the optical fiber by the transmitting end and decrypt the signal. The decryption process is the inverse transformation of the transmitting end.

[0029] A third object of the present invention is to provide a computer device comprising a processor and a storage medium;

[0030] The storage medium is used to store instructions;

[0031] The processor is configured to operate according to the instructions to execute the steps of any of the above methods.

[0032] A fourth object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the program implements the steps of any of the above methods when executed by a processor.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] (1) The present invention introduces a variational autoencoder model to learn the generation rules of chaotic sequences, and takes advantage of the parallel computing of deep networks to replace the inefficient chaotic sequence generation of linear differential equations. By training a large amount of chaotic sequence training data, the model performance and generalization ability of the network model are improved, which can effectively improve the generation efficiency of chaotic sequences and ensure transmission performance.

[0035] (2) By applying the chaotic encryption method based on the variational autoencoder model in the orthogonal frequency division multiplexing passive optical access network (OFDM-PON) and using training materials of different chaotic security schemes, the encoder can learn the complex structure of data distribution in various chaotic models and ultimately has the ability to generate a large space key group, effectively improving the security of the OFDM-PON system. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of the chaotic encryption method in Example 1;

[0037] Figure 2 is a diagram of the variational autoencoder network structure in Example 1;

[0038] Figure 3 This is a block diagram of the encoder network structure in Example 1;

[0039] Figure 4 This is a block diagram of the decoder network structure in Example 1;

[0040] Figure 5 This is a block diagram of the chaotic encryption experimental model in Example 1;

[0041] Figure 6 This is a graph showing the bit error rate of the signal of the probability shaping (PS)-16 quadrature amplitude modulation (QAM) VS received optical power in the chaotic encryption experiment in Example 1;

[0042] Figure 7 is the calculation time of the chaotic encryption experiment in Example 1 VS the number of chaotic sequence iterations;

[0043] Figure 8 This is a flow chart of the method for applying the chaotic encryption method in the OFDM-PON system in Example 2. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0045] The purpose of this patent is to disclose a chaotic encryption method based on a variational autoencoder (VAE), suitable for use in chaotic secure communications. This chaotic sequence encryption scheme for elastic optical networks based on secure communications introduces a deep generative model, namely a variational autoencoder (VAE) encoder model. The VE is trained using the original chaotic sequence, and the sequence generated by the VE encoder model is used for encryption. This chaotic encryption method is then applied to the encrypted transmission of quadrature amplitude modulation (16QAM) signals. The sequence generated by the VE model is used to encrypt and mask the constellation phase distribution of uniform signal points and the subcarrier positions corresponding to each signal at the transmitter. At the receiver, the same initial key is used to generate a decryption sequence. Due to the robustness of the VE, the same decryption sequence as that used at the transmitter is obtained. By performing inverse operations on the subcarrier positions and constellation phase distribution, the original data is retrieved. This encryption method generates a large key space through deep learning and encrypts the transmitted signal twice, achieving a "one-time, one-pad" encryption method where the key length is equal to the transmitted data. This ensures the confidentiality of important files during transmission, greatly enhancing the complexity and security of the system. At the same time, this scheme can learn the characteristics of different chaotic models and obtain chaotic sequences with good confidentiality performance; it can make full use of the advantages of parallel computing of graphics processors in deep learning, effectively improve the efficiency of chaotic encryption systems, and ensure good transmission performance.

[0046] Example 1:

[0047] like Figure 1 As shown, this embodiment provides a chaotic encryption method based on a variational autoencoder, including:

[0048] S01: Get the original chaotic sequence;

[0049] S02: Input the original chaotic sequence data into the pre-trained variational autoencoder model for reconstruction, and output the reconstructed chaotic sequence;

[0050] S03: Perform phase masking encryption and frequency masking encryption on the communication data according to the obtained reconstructed chaotic sequence.

[0051] The reconstructed chaotic sequence includes a first chaotic sequence capable of performing phase masking on the communication data and a second chaotic sequence capable of performing frequency masking on the communication data.

[0052] The above-mentioned chaotic encryption method based on variational autoencoder is implemented through the following stages.

[0053] 1. Building a Variational Autoencoder Model

[0054] like Figure 3 and 4As shown, the variational autoencoder model described in this embodiment includes an encoder network and a decoder network. The encoder network includes 1 convolutional layer, 3 residual connection blocks and 2 self-attention layers. The input original sequence is first subjected to data compression processing by the convolutional layer, and the ReLu function is used as the activation function, the pooling layer is cancelled, and the dimension reduction of the data is achieved by Stride; then, the dimensionality reduction processing is performed by 3 residual connection blocks and 2 self-attention layers set alternately, and finally a low-dimensional intermediate sequence is output. The convolution kernel size of the convolution layer in the encoder network is 5, and the stride is 2. The decoder network includes 2 3×3 convolutional layers, 2 batch normalization layers and 1 fully connected layer, and each 3×3 convolution layer is connected to a loss layer. The intermediate sequence output by the encoder network first passes through a 3×3 convolutional layer, a dropout layer, and a normalization layer, and is processed by the ReLu activation function. It then passes through a 3×3 convolutional layer, a dropout layer, and a normalization layer, and then through a fully connected layer. The output is reconstructed for training. If the model's loss function meets the preset conditions, the intermediate sequence data is directly used as the output of the decoder network. The kernel size of the 3×3 convolutional layer in the decoder network is 1, and the dropout layer has a dropout rate of 0.2.

[0055] 2. Training the Variational Autoencoder Model

[0056] like Figure 2 As shown in Figure 2, the training method of the variational autoencoder model includes:

[0057] S11: obtaining an original sequence for training, and using a plurality of chaotic sequences output by a chaotic model in the prior art as training sequences;

[0058] S12: Input the obtained original sequence into the encoder network for data encoding processing, and output an intermediate sequence that deviates from the original sequence, where the deviation of the intermediate sequence includes a standard deviation and a mean vector;

[0059] S13: Input the intermediate sequence into the decoder network for data decoding processing, perform iterative training by setting an adjustable training cycle to converge the model loss function, and output the reconstructed sequence to complete the training of the variational autoencoder model.

[0060] In order to minimize the reconstruction error, this embodiment adopts the minimization loss function. By minimizing the loss function, the variational autoencoder model can learn the data distribution of various original sequences and reconstruct new data samples. The expression is:

[0061]

[0062] Where σ is the standard deviation of the encoder network output; m is the mean vector of the encoder network output; n is the data dimension of the original sequence of the encoder network input; i represents the i-th value in the standard deviation vector; pE (x|y) is the probability of outputting y when the encoder network input is x; p D (y|x) is the probability of outputting x when the decoder network input is y;

[0063] 3. Reconstructing the original chaotic sequence

[0064] The original chaotic sequence is input into the trained variational autoencoder model to output a reconstructed chaotic sequence, which includes a first chaotic sequence that can mask the phase of the communication data and a second chaotic sequence that can mask the frequency of the communication data.

[0065] Test Example 1:

[0066] In order to test the effect of the chaotic encryption method based on the variational autoencoder model in Example 1, Figure 5 As shown in the figure, the experimental system for the chaotic encryption method based on a variational autoencoder described in this test example is a dual-polarization coherent optical system. At the transmitter, the encrypted information is first generated by an arbitrary waveform generator and sent to the in-phase / quadrature modulator. The light source is a continuous-wave laser operating at 1550nm with an optical power of 14.5dBm. The optical signal output by the light source is injected into the in-phase / quadrature modulator and amplified by an erbium-doped fiber amplifier to obtain the modulated signal. The modulated signal is then transmitted through a 25km standard single-mode optical fiber and input to the receiver. At the receiver, an optical tunable filter is used to suppress the amplifier's spontaneous emission noise, and a coherent receiver receives and preliminarily demodulates the received signal. A local oscillator and mixed-signal oscilloscope are used to display and output the received signal as a digital file.

[0067] like Figure 6 As shown in the curve of bit error rate versus received optical power based on PS-16QAM, the bit error rate of illegal reception is maintained at around 0.5, which means that all signals are misidentified. -3 When the receiver sensitivity of PS-16QAM is 1.05dB higher than that of uniform 16QAM, the probability shaping (PS) technology effectively improves the receiver sensitivity.

[0068] like Figure 7 The graph showing the computation time vs. the number of chaotic sequence iterations in the chaotic encryption experiment shows the relationship between computation time and the number of generated chaotic sequence iterations. It can be seen that as the number of sequence iterations increases, the time consumed by the traditional iterative equation increases rapidly, while the time required by the variational autoencoder (VAE) remains around 0.2ms. This shows that the chaotic encryption model based on the variational autoencoder can shorten computation time and effectively reduce the complexity of the chaotic encryption scheme compared to the chaotic model based on the traditional iterative equation.

[0069] Example 2:

[0070] like Figure 8 As shown, this embodiment provides an application of a chaotic encryption method based on a variational autoencoder, and the chaotic encryption method based on a variational autoencoder described in Example 1 is used in an orthogonal frequency division multiplexing passive optical access communication system (OFDM-PON) for data encryption transmission.

[0071] The communication system includes:

[0072] The transmitting end is used to first perform serial-to-parallel switching on the original data sequence input, and then map the serial-to-parallel switched original data sequence to a first chaotic sequence through orthogonal amplitude modulation for phase masking; map the orthogonal amplitude modulated original data sequence to a second chaotic sequence through subcarrier modulation for frequency masking; and convert the original data sequence after subcarrier modulation from the frequency domain to the time domain through fast Fourier transform, and after parallel-to-serial conversion, input the signal into the optical fiber;

[0073] The receiving end is used to receive the signal output by the transmitting end into the optical fiber and decrypt the signal. The decryption process is the inverse of the transmitting end's transformation. In the decryption process of orthogonal amplitude demodulation and subcarrier demodulation, the same initial key group as the encryption at the transmitting end is used to generate a chaotic decryption sequence to achieve accurate decryption.

[0074] This paper introduces a deep generative model called a variational autoencoder to generate chaotic sequences for OFDM symbol encryption. By using training materials from different chaotic security schemes, the encoder can learn the complex structure of data distribution in various chaotic models, ultimately enabling the generation of large-scale key groups, effectively improving the security of OFDM-PON systems. Furthermore, thanks to GPU parallel computing, this scheme requires only 1.38% of the methods used in traditional encryption schemes.

[0075] Example 3:

[0076] An embodiment of the present invention further provides a computer device, including a processor and a storage medium;

[0077] The storage medium is used to store instructions;

[0078] The processor is configured to operate according to the stored instructions to perform the steps of the following method:

[0079] Get the original chaotic sequence;

[0080] The original chaotic sequence data is input into the pre-trained variational autoencoder model for reconstruction, and the reconstructed chaotic sequence is output;

[0081] Performing phase masking encryption and frequency masking encryption on the communication data according to the obtained reconstructed chaotic sequence;

[0082] The computer device provided in this embodiment can execute the chaotic encryption method based on the variational autoencoder in Example 1. Its implementation principle and technical effects are similar and will not be repeated here.

[0083] Example 4:

[0084] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the following method:

[0085] The storage medium is used to store instructions;

[0086] The processor is configured to operate according to the stored instructions to perform the steps of the following method:

[0087] Get the original chaotic sequence;

[0088] The original chaotic sequence data is input into the pre-trained variational autoencoder model for reconstruction, and the reconstructed chaotic sequence is output;

[0089] Performing phase masking encryption and frequency masking encryption on the communication data according to the obtained reconstructed chaotic sequence;

[0090] The storage medium provided in this embodiment can execute the chaotic encryption method based on the variational autoencoder in Example 1. Its implementation principle and technical effects are similar and will not be repeated here.

[0091] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0092] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0093] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0095] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A chaotic encryption method based on variational autoencoder, characterized in that: include: The variational autoencoder model includes an encoder network and a decoder network, wherein the encoder network includes alternating residual connection blocks and self-attention layers, and the decoder network block includes convolutional layers and a loss layer after each convolutional layer; The chaotic encryption method comprises: Get the original chaotic sequence; The original chaotic sequence data is input into the pre-trained variational autoencoder model for reconstruction, and the reconstructed chaotic sequence is output; Performing phase masking encryption and frequency masking encryption on the communication data according to the obtained reconstructed chaotic sequence; The reconstructed chaotic sequence includes a first chaotic sequence capable of performing phase masking on communication data and a second chaotic sequence capable of performing frequency masking on communication data.

2. The chaotic encryption method based on variational autoencoder according to claim 1, characterized in that: The training method of the variational autoencoder model includes: Get the original sequence for training; Input the obtained original sequence into the encoder network for data encoding processing, and output an intermediate sequence that deviates from the original sequence, wherein the deviation of the intermediate sequence includes a standard deviation and an average value vector; The intermediate sequence is input into the decoder network for data decoding processing. By setting an adjustable training cycle for iterative training, the model loss function converges, and the reconstructed sequence is output to complete the training of the variational autoencoder model.

3. The chaotic encryption method based on variational autoencoder according to claim 2, characterized in that: The encoder network consists of 1 convolutional layer, 3 residual connection blocks and 2 self-attention layers; The original input sequence is first compressed by the convolutional layer, and the ReLu function is used as the activation function. Then it is reduced in dimension by three alternating residual connection blocks and two self-attention layers, and finally a low-dimensional intermediate sequence is output.

4. The chaotic encryption method based on variational autoencoder according to claim 3 is characterized in that: The convolution kernel size is 5 and the step size is 2.

5. The chaotic encryption method based on variational autoencoder according to claim 3 is characterized in that: The decoder network includes two 3×3 convolutional layers, two batch normalization layers, and one fully connected layer, and each 3×3 convolutional layer is connected to a dropout layer; The intermediate sequence output by the encoder network first passes through a 3×3 convolution layer, a dropout layer, and a normalization layer, and is processed by the ReLu activation function. It then passes through a 3×3 convolution layer, a dropout layer, and a normalization layer, and is processed by a fully connected layer. The output is reconstructed for training. If the loss function of the model has reached the preset conditions, the intermediate sequence data is directly used as the output of the decoder network.

6. The chaotic encryption method based on variational autoencoder according to claim 5, characterized in that: The convolution kernel size of the 3×3 convolutional layer is 1, and the dropout rate of the dropout layer is 0.

2.

7. The chaotic encryption method based on variational autoencoder according to claim 2, characterized in that: The loss function is a minimization loss function, and the calculation formula includes: ； Where, is the standard deviation of the encoder network output; m is the mean vector of the encoder network output; n is the data dimension of the original sequence of the encoder network input; i represents the i-th value in the standard deviation vector; The probability of outputting y when the encoder network input is x; The probability of outputting x when the decoder network input is y.

8. An application system of a chaotic encryption method based on a variational autoencoder, characterized in that: The chaotic encryption method based on a variational autoencoder according to any one of claims 1 to 7 is used in an orthogonal frequency division multiplexing passive optical access communication system for data encryption transmission, wherein the communication system comprises: The transmitting end is used to first perform serial-to-parallel switching on the original data sequence input, and then map the serial-to-parallel switched original data sequence to a first chaotic sequence through orthogonal amplitude modulation for phase masking; map the orthogonal amplitude modulated original data sequence to a second chaotic sequence through subcarrier modulation for frequency masking; and convert the original data sequence after subcarrier modulation from the frequency domain to the time domain through fast Fourier transform, and after parallel-to-serial conversion, input the signal into the optical fiber; The receiving end is used to receive the signal outputted into the optical fiber by the transmitting end and decrypt the signal. The decryption process is the inverse transformation of the transmitting end.

9. A computer device, characterized in that: including processors and storage media; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 7.

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

Images