Secret communication method and system based on storage pool chaotic prediction and interleaving coding

Abstract

A chaotic encryption module is characterized in that a light source output by a first continuous wave laser (CW) is modulated by a parallel quadrature modulator (IQM), so that a signal output by an arbitrary waveform generator (AWG) is converted and transmitted, and pseudo-random interleaving coding of the signal is realized; a transmitted signal is loaded to a chaos carrier through a first optical coupler FC to realize chaos hiding, then the chaos optical carrier predicted in advance by a storage pool is utilized in a chaos decryption module through an information transmission module, and the chaos optical carrier is modulated to a continuous optical carrier output by a second continuous wave laser CW through a Mach-Zehnder modulator MZM. And the 90-degree frequency mixer carries out decryption so as to preliminarily recover an original transmission signal. And finally, the obtained signal is processed by a local oscillator (LO) and a coherent receiver, and digital information processing (DSP) is carried out to realize compensation, recovery and decryption of the digital signal, so that chaotic secrecy coherent optical communication is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of chaotic secure communication in the field of optical fiber communication, and in particular relates to a secure communication method and system based on reserve pool chaotic prediction and interleaving coding. Background Art

[0002] Confidential communication based on chaotic synchronization was first proposed in the early 1990s. Since then, due to the potential application value of chaotic optical communication in high-speed confidential communication, many new schemes for realizing chaotic optical communication have been proposed. Among them, the chaotic confidential communication scheme based on semiconductor lasers generates optical chaotic signals by a nonlinear dynamic system based on lasers. However, due to the performance and technical bottlenecks of various electro-optical components, the capacity and rate of chaotic optical confidential communication systems are seriously affected. Therefore, the bandwidth and delay of chaotic sources in high-speed and long-distance confidential communication need to be further improved.

[0003] For the chaotic optical secure communication system based on semiconductor lasers, one of its core technologies is to achieve stable and lasting high-quality chaotic synchronization at the transceiver. In recent years, a machine learning method that can be used to predict the evolutionary behavior of nonlinear dynamic systems - reservoir computing (RC) has attracted the interest of many researchers. For example, the existing patent CN112822003A discloses a laser chaotic synchronization secure communication method and system based on reservoir computing. Its basic chaos prediction and chaos synchronization principle is to train and learn the behavior of the system under the target nonlinear dynamic system, and return to the reservoir through feedback. The trained reservoir can approach the target nonlinear dynamic system and can perform the next prediction, and finally achieve chaotic synchronization. However, the nonlinear expression of the commonly used reservoir system is limited, and the reservoir calculation method has the problem of low prediction accuracy. Its prediction accuracy is difficult to achieve good synchronization for the dynamic system under long-distance transmission, resulting in inevitable loss and error after information decryption. At the same time, high-speed and long-distance signal transmission between the transmitter and the receiver will cause great attenuation and distortion of the signal, resulting in information distortion after transmission. Therefore, how to solve the signal processing technology involved in the process of high-speed, long-distance and confidential transmission of signals and the efficient chaotic synchronization between the transmitter and receiver are issues that need to be urgently addressed in the current field of chaotic confidential communications.

[0004] At present, with the maturity of coherent optical communication technology, the distortion and loss in the signal transmission process are solved by digital signal processing technology, which greatly improves the transmission distance of information and provides the possibility for high-speed and long-distance chaotic secure communication. In addition, the continuous development of digital encryption algorithms also provides further guarantee for the secure transmission of information. By combining digital encryption algorithms with chaotic secure communication technology, double encryption of information transmission can be realized, which can further improve the security of data transmission. Based on this, the present invention proposes a secure communication method based on reserve pool chaos prediction and interleaved coding to solve some problems existing in the prior art, in order to promote the application and development of chaotic secure communication technology in related fields. Summary of the invention

[0005] The purpose of the present invention is to solve the problems of signal processing technology involved in the process of high-speed long-distance confidential transmission of signals and efficient chaotic synchronization of the transmitting and receiving ends.

[0006] To achieve the above object, the present invention provides a secure communication system based on reserve pool chaos prediction and interleaving coding, including a chaos encryption module, a reserve pool calculation neural network prediction module, a chaos decryption module, an information transmission module and an information receiving module;

[0007] The chaotic encryption module includes a plane reflector M, the output end of which is sequentially connected to a driving laser SL, a phase modulator PM, a first dispersion compensation optical fiber DCF and a second optical coupler FC; the output end of the second optical coupler FC is connected to the first optical coupler FC and a photodetector PD;

[0008] The input end of the first optical coupler FC is connected to a parallel quadrature modulator IQM, and the parallel quadrature modulator IQM is connected to a first continuous wave laser CW and an arbitrary waveform generator AWG; the output end of the first optical coupler FC is connected to the input end of the erbium-doped fiber amplifier EDFA;

[0009] The reserve pool calculation neural network prediction module includes a photodetector PD, the output end of the photodetector PD is connected to an ADC module and a reserve pool neural network RC in sequence; the output end of the reserve pool neural network RC is connected to a Mach-Zehnder modulator MZM;

[0010] The chaos decryption module includes a Mach-Zehnder modulator MZM, and the Mach-Zehnder modulator MZM is connected to a second continuous wave laser CW and a 90° mixer;

[0011] The information transmission module includes N optical fiber link modules, each of which is composed of an erbium-doped optical fiber amplifier EDFA, a single-mode optical fiber SSFM and a second dispersion-compensating optical fiber DCF; the output end of the erbium-doped optical fiber amplifier EDFA is sequentially connected to the single-mode optical fiber SSFM and the second dispersion-compensating optical fiber DCF, the second dispersion-compensating optical fiber DCF is connected to a 90° mixer; the 90° mixer is connected to a coherent receiver;

[0012] The information receiving module comprises a local oscillator laser LO and a coherent receiver; the coherent receiver is connected to the output end of the 90° mixer.

[0013] Furthermore, the reservoir neural network RC includes an input layer, a reservoir layer and an output layer; each layer includes a number of neurons.

[0014] Furthermore, the reservoir layer is composed of two reservoirs with the same parameters through linear coupling. The number of neurons in the reservoir can be set according to the complexity of the problem; the reservoir size N = 500, the spectral radius rho = 1.2, and the regularization parameter reg = 1e -6 ; When the coupling coefficient α = 0, the results of the two reserve pool calculations do not interact and are output in sequence; when α = 1, the calculations of the two reserve pools are in a unidirectional coupling output relationship.

[0015] A secure communication method based on reserve pool chaos prediction and interleaving coding comprises the following steps:

[0016] S1: The optical feedback structure formed by the driving laser SL and the plane reflector M is used to generate chaotic laser. After being optimized by the phase modulator PM and the first dispersion compensation fiber DCF, the chaotic laser outputs two identical optical signals through the second coupler FC with a splitting ratio of 50:50. One optical signal is used as a driving signal and is converted into a chaotic electrical signal through the photoelectric conversion of the photodetector PD and the ADC module and input into the reserve pool calculation neural network prediction module; the other optical signal is input into the first optical coupler FC as a driving signal;

[0017] S2: The first continuous wave laser CW generates an optical signal. The constellation signal to be transmitted is encoded offline by an arbitrary waveform generator AWG, and then modulated onto the optical signal generated by the first continuous wave laser CW by a parallel quadrature modulator IQM. The constellation signal needs to be modulated onto the optical signal after an encryption algorithm, thereby generating the original information m(t).

[0018] S3, the optical signal input into the first optical coupler FC in step S1 is superimposed with the original information m(t) in step S2 to generate a chaotic encryption signal c(t)+m(t), and the chaotic encryption signal is sequentially transmitted to the chaotic decryption module via the information transmission module composed of the erbium-doped fiber amplifier EDFA, the single-mode fiber SSFM, and the second dispersion compensation fiber DCF;

[0019] S4: The chaotic electrical signal input to the photodetector PD and the ADC module in S1 is used as the input of the reserve pool calculation neural network prediction module, and the weight of the reserve pool neural network RC is determined by optimizing the reserve pool neural network RC training phase, so that the output signal of the output layer of the reserve pool neural network RC is synchronized with the input signal of the input layer of the reserve pool neural network RC, so as to realize the chaotic prediction of the input signal, which is the chaotic prediction signal;

[0020] S5: In the RC test phase of the reserve pool neural network, the output layer outputs a chaotic prediction signal, and outputs a chaotic optical carrier c'(t) synchronized with the output of the chaotic encryption module through the Mach-Zehnder modulator MZM under the synchronous control of the second continuous wave laser CW; the chaotic encryption signal c(t)+m(t) output by the information transmission module and the chaotic optical carrier c'(t) are decrypted through a 90° mixer to obtain the original transmission signal m'(t);

[0021] S6: The original transmission signal m'(t) is input to the information receiving module, and after coherent reception by the local oscillator laser LO and the coherent receiver, offline digital information processing DSP is performed to optimize and compensate the original transmission signal m'(t).

[0022] Furthermore, the generation of the constellation signal in step S2 is to map the binary bit data to different symbol points of the constellation according to the phase amplitude, thereby generating a QAM signal to be transmitted.

[0023] Furthermore, the specific process of the encryption algorithm in step S2 is:

[0024] The pseudo-random interleaving encoding method of the signal is to perform pseudo-random interleaving encoding on the QAM binary sequence through the Lorenz three-dimensional chaotic system and the specific random numbers generated by the function to achieve encryption; the Lorenz three-dimensional chaotic equation is as follows:

[0025]

[0026] Among them, σ, ρ, β are dimensionless constants, x, y, z are state components of the chaotic system, and t is time. After solving the chaotic equation and digitizing the chaotic sequence, the obtained x, y, z are integerized, as shown in the following formula:

[0027] P x，i =mod(Extract(x i , m, n, p), M)

[0028] P y，i =mod(Extract(y i , m, n, p), M)

[0029] P z，i =mod(Extract(z i , m, n, p), M)

[0030] where x i ,y i 、z i is a chaotic sequence, m, n, and p are specific bit numbers extracted from the decimal part, and the Extract function is implemented to extract specific bit numbers from the decimal part. The resulting random sequence {P x，i}、{P y，i} and {P z，i}, using {P x，i}Generate a chaotic synchronization sequence and insert it into the front of the QAM signal frame; use the digitized sequence {P y，i} and {P z，i} to construct a scrambling sequence, and the QAM signal and the scrambling sequence can be pseudo-randomly interleaved to achieve signal encryption.

[0031] Furthermore, the decryption in step S5 uses the same chaotic equation, the same initial value and the same number of iterations as those used in the chaotic encryption module to construct the same scrambling sequence and obtain the same digital encryption information, thereby restoring the original transmission signal m'(t) after scrambling coding to the original constellation diagram.

[0032] Beneficial effects:

[0033] (1) The chaotic light source based on optical feedback is optimized by combining a phase modulator with a dispersion-compensating fiber, which increases the bandwidth of the chaotic source and effectively suppresses the delay characteristics.

[0034] (2) The reservoir computing neural network prediction module acts as a receiver of the chaotic synchronization communication system. Compared with the traditional reservoir computing prediction model, the output of the first reservoir is used as a new feedback input to the second reservoir during linear coupled reservoir computing training, that is, the input of the second reservoir is replaced by the output of the first reservoir. The two reservoirs can store information through the connection of each unit in the loop, where the previous input affects the next response. After training, the reservoir becomes an autonomous power system that imitates the system from the input end. This model solves the problem of insufficient prediction accuracy of traditional methods, and achieves the effect of smaller prediction error and longer prediction time.

[0035] (3) Since the computational neural network prediction module of the reserve pool system uses a third-party laser as its input, its synchronization quality is independent of the effective information coverage and is easy to control. It can still obtain good demodulation results at extremely low coverage and ensure safety. The third-party driving laser does not need to be synchronized with the main laser and the reserve pool computational neural network prediction model, which can effectively prevent eavesdropping and greatly enhance the security of the system.

[0036] (4) The transmitted constellation diagram is pre-coded by pseudo-random interleaving during the information encoding process, and the binary sequence converted by the QAM signal is scrambled and encoded by a specific random sequence generated by a digital chaotic system and a function to achieve encryption. At the receiving end, the recovered information is processed by a coherent receiver through DSP, which can optimize, compensate and deinterleave the signal loss of the original signal after high-speed and long-distance transmission. The security of the system is increased through secondary digital encryption, and the system transmission performance is also significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the device structure of the present invention;

[0038] Figure 2 It is an internal schematic diagram of the reserve pool calculation neural network prediction module of the present invention.

[0039] Figure numerals: 1-first continuous wave laser CW, 2-parallel orthogonal modulator IQM, 3-arbitrary waveform generator AWG, 4-first optical coupler FC, 5-plane reflector M, 6-driving laser SL, 7-phase modulator PM, 8-first dispersion compensation fiber DCF, 9-second optical coupler FC, 10-photodetector PD, 11-ADC module, 12-reserve pool neural network RC, 13-erbium-doped fiber amplifier EDFA, 14-single mode fiber SSFM, 15-second dispersion compensation fiber DCF, 16-second continuous wave laser CW, 17-Mach-Zehnder modulator MZM, 18-90° mixer, 19-coherent receiver Coherent Receiver, 20-local oscillator laser LO, 21-chaotic encryption module, 22-reserve pool calculation neural network prediction module, 23-chaotic decryption module, 24-information transmission module, 25-information receiving module. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] The application principle of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0042] Embodiment 1:

[0043] according to Figure 1 As shown, this embodiment proposes a secure communication method based on reserve pool chaos prediction and interleaving coding, and the specific steps are as follows:

[0044] S1: A driving laser SL6 and a plane reflector M5 are used to form an optical feedback structure to generate a chaotic laser, and the generated chaotic laser is further optimized by a phase modulator PM7 and a first dispersion compensation fiber DCF8. The nonlinear effect of the phase modulator PM7 and the first dispersion compensation fiber DCF8 can make the driving laser SL6 output a chaotic carrier signal with high bandwidth and suppressed delay characteristics to enhance system safety; the laser chaotic carrier then passes through a second optical coupler FC9 with a splitting ratio of 50:50 to output two identical optical signals, one of which is used as a driving signal and is converted into a chaotic electrical signal through the photoelectric conversion of the photodetector PD10 and the ADC11 module, wherein the chaotic electrical signal is offline input into the reserve pool calculation neural network prediction module 22 for chaotic prediction; the other optical signal (i.e., the chaotic optical signal c(t)) is input into the first optical coupler FC4 as a driving signal.

[0045] S2, the first continuous wave laser CW1 generates an optical signal, and the constellation signal to be transmitted is encoded offline through the arbitrary waveform generator AWG3, wherein the generation of the constellation signal is to map the binary bit data to different symbol points of the constellation according to the phase amplitude, thereby generating the QAM signal (constellation signal) to be transmitted. The constellation signal is modulated onto the optical signal generated by the first continuous wave laser CW1 through the parallel orthogonal modulator IQM2. Among them, the constellation signal needs to be modulated onto the optical signal after the encryption algorithm, and its encryption method is based on the digital chaotic system, that is, the constellation signal is encrypted by "secondary modulation", thereby generating the pre-processed original information m(t) to be transmitted.

[0046] S3, the chaotic optical signal c(t) input to the first optical coupler FC4 in step S1 is superimposed with the original information m(t) in step S2 to generate a chaotic encrypted signal c(t)+m(t), and the chaotic encrypted signal is sequentially transmitted to the chaotic decryption module 23 via the erbium-doped fiber amplifier EDFA13, the single-mode fiber SSFM14, and the second dispersion compensation fiber DCF15 to form an information transmission module 24;

[0047] S4: The chaotic electrical signal input to the photodetector PD10 and the ADC module 11 in S1 is used as the input of the reserve pool calculation neural network prediction module 22. The weight of the reserve pool neural network RC12 is determined by optimizing the training phase of the reserve pool neural network RC12, so that the output signal of the output layer of the reserve pool neural network RC12 is synchronized with the input signal of the input layer of the reserve pool neural network RC12, that is, the chaotic prediction of the input signal is realized, which is a chaotic prediction signal.

[0048] The steps of the training phase of the reserve pool neural network RC12 are as follows: the output of the driving laser SL6 is used as the input of the reserve pool calculation neural network prediction module 22 after passing through the photodetector PD10, and is used to iteratively generate the internal node states of the reserve pool;

[0049] During the training phase, the chaotic electrical signal of the input layer is transmitted to the reservoir layer in a back-to-back manner and used as the target output data calculated by the reservoir layer to calculate the connection weight between the reservoir layer and the output layer.

[0050] S5: During the test phase of the reserve pool neural network RC12, the output layer outputs a chaotic prediction signal, and outputs a chaotic optical carrier c'(t) synchronized with the output of the chaotic encryption module 21 through the Mach-Zehnder modulator MZM17 under the synchronous control of the second continuous wave laser CW16. The chaotic encryption signal c(t)+m(t) output by the information transmission module 24 and the synchronized chaotic optical carrier c'(t) are decrypted by the 90° mixer 18 to obtain a useful original transmission signal m'(t). The method for decrypting the constellation digital signal (i.e., decrypting the original transmission signal m'(t)) is to reuse the same chaotic equation, the same initial value and the same number of iterations used by the transmitting end, construct the same scrambling sequence, obtain the same digital encryption information, and then restore the original transmission signal m'(t) after scrambling coding to the original constellation.

[0051] S6: The original transmission information m'(t) output by the chaotic decryption module 23 will have loss and error after long-distance transmission. It is input into the information receiving module 25, and after coherent reception by the local oscillator laser LO20 and the coherent receiver CoherentReceiver19, offline digital information processing DSP is performed to optimize and compensate for the loss of the original transmission signal m'(t) after high-speed and long-distance transmission.

[0052] The specific process of the encryption algorithm in step S2 is: the signal is encrypted by pseudo-random interleaving coding, and the specific method is to perform pseudo-random interleaving coding on the binary sequence converted by the QAM signal through the Lorenz three-dimensional chaotic system and the specific random number generated by the function function to achieve encryption. Lorenz three-dimensional chaotic equation:

[0053]

[0054] Among them, σ, ρ, β are dimensionless constants, x, y, z are state components of the chaotic system, and t is time. After solving the chaotic equation and digitizing the chaotic sequence using computing software, the obtained x, y, z are integerized, as shown in the following formula:

[0055] P x，i =mod(Extract(x i ,m,n,p),M) (4)

[0056] P y,i =mod(Extract(y i , m, n, p), M) ( 5)

[0057] P z，i =mod(Extract(z i,m,n,p),M) (6)

[0058] where x i ,y i 、z i is a chaotic sequence, m, n, and p are specific bit numbers extracted from the decimal part, and their values ​​here are 2, 4, and 6 respectively. M is the decimal number of the bit. Since this system adopts 16QAM constellation diagram to transmit signals, its value is 256.

[0059] The function is implemented as Extract function to extract specific bit numbers from the decimal part, and the resulting random sequence {P x，i}、{P y，i} and {P z，i}, using {P x，i}Generate a chaotic synchronization sequence and insert it into the QAM signal frame. This synchronization sequence is actually a pilot sequence that can be used to complete signal estimation and symbol synchronization. Using the other two digitized sequences {P y，i} and {P z，i} to construct a scrambling sequence, and the QAM signal and the scrambling sequence can be pseudo-randomly interleaved to achieve signal encryption. In the information receiving module 25, as long as the same chaotic equation, the same initial value and the same number of iterations are used to construct the same scrambling sequence, the same digital encrypted information can be obtained, and then the deinterleaving can be correctly achieved to restore the original data order.

[0060] A secure communication system based on reserve pool chaos prediction and interleaving coding comprises a chaos encryption module 21, a reserve pool calculation neural network prediction module 22, a chaos decryption module 23, an information transmission module 24 and an information receiving module 25.

[0061] The chaotic encryption module 21 includes a plane mirror M5, a driving laser SL6, a phase modulator PM7, a first dispersion compensating fiber DCF8, a second optical coupler FC9, a first optical coupler FC4, a first continuous laser CW1, an arbitrary waveform generator AWG3, and a parallel orthogonal modulator IQM2. The output end of the plane mirror M5 is sequentially connected to the driving laser SL6, the phase modulator PM7, the first dispersion compensating fiber DCF8, and the second optical coupler FC9; the output end of the second optical coupler FC9 is connected to the first optical coupler FC4 and the photodetector PD10. The plane mirror M5 and the driving laser SL6 generate chaotic lasers through optical feedback, and the chaotic lasers are optimized after passing through the phase modulator PM7 and the first dispersion compensating fiber DCF8. The input end of the first optical coupler FC4 is connected to a parallel quadrature modulator IQM2, and the parallel quadrature modulator IQM2 is connected to a first continuous wave laser CW1 and an arbitrary waveform generator AWG3; the output end of the first optical coupler FC4 is connected to the input end of the erbium-doped fiber amplifier EDFA13. In the chaotic encryption module 21, the binary information to be transmitted is mapped into a quadrature amplitude modulation QAM signal, and the timing of the transmission signal is changed in a random manner, and the continuous data is rearranged according to a certain rule to achieve pseudo-random interleaving coding. The random method here is a random sequence extracted by a digital chaotic system and a specific function, and then a digital-to-analog conversion is performed. After being modulated by the parallel quadrature modulator IQM2, it is modulated onto the optical signal of the first continuous wave laser CW1 and transmitted through the optical fiber. The chaotic optical signal and the light source loaded with information after modulation are subjected to chaotic masking through the first optical coupler FC4;

[0062] The reserve pool calculation neural network prediction module 22 includes a photodetector PD10, an ADC module 11 and a reserve pool neural network RC12; the output end of the photodetector PD10 is connected to the ADC module 11 and the reserve pool neural network RC12 in sequence; the output end of the reserve pool neural network RC12 is connected to the Mach-Zehnder modulator MZM17; the optical signal output by the chaotic encryption module 21 of the reserve pool calculation neural network prediction module 22 is converted into an electrical signal by the photodetector PD10 and then input into the ADC module 11, and then input into the reserve pool neural network RC12 for offline prediction of chaotic electrical signals. The reserve pool neural network RC12 is composed of two reserve pools with consistent parameters through linear coupling.

[0063] The information transmission module 24 includes N optical fiber link modules, each of which is composed of an erbium-doped fiber amplifier EDFA13, a single-mode fiber SSFM14, and a second dispersion-compensating fiber DCF15; the output end of the erbium-doped fiber amplifier EDFA13 is connected in sequence to the single-mode fiber SSFM14 and the second dispersion-compensating fiber DCF15, the second dispersion-compensating fiber DCF15 is connected to a 90° mixer 18; the 90° mixer 18 is connected to a coherent receiver 19. The second dispersion-compensating fiber DCF15 is used to compensate for the dispersion effect of optical signals caused by long-distance transmission, and the erbium-doped fiber amplifier EDFA13 compensates for the attenuation of optical signals.

[0064] The chaos decryption module 23 includes a second continuous wave laser CW16 , a Mach-Zehnder modulator 17 and a 90° mixer 18 , and the Mach-Zehnder modulator MZM17 is connected to the second continuous wave laser CW16 and the 90° mixer 18 .

[0065] The information receiving module 25 includes a local oscillator laser LO20 and a coherent receiver 19. The coherent receiver 19 is connected to the output end of the 90° mixer 18.

[0066] The reservoir layer calculation of the reservoir calculation neural network prediction module 22 of the present invention is divided into two stages: a training stage and a testing stage.

[0067] The calculation principle of the reserve pool layer is: in the training phase, the weight matrices Win1 and Win2 between the input layer and the reserve pool, and the internal weight matrices W1 and W2 of the nodes inside the reserve pool are randomly generated and will not change once generated. Only the connection weight Wout between the reserve pool and the output layer needs to be determined through the training process, and the training process first generates the state vector X(t) of the internal node of the reserve pool according to the input data u(t) iteration, and then calculates Wout according to the target output data Y of the training phase and the state vector X(t) of the node of the reserve pool.

[0068] During the testing phase, the reservoir system needs to be driven by input data to update the state of the system. After the state update, the reservoir no longer relies on new external input signals, but recursively updates the future state through its own internal state r(t). At each time step tt, the output of the reservoir system is the output connection weight Wout multiplied by the current state vector r(t) and the input data u(t). Once the reservoir system enters the prediction phase, the model will enter the self-driving mode, that is, the next prediction depends on the prediction result of the previous step.

[0069] The chaotic encryption module 21 uses a driving laser SL6 with optical feedback to generate a chaotic light source, which can output a high-quality chaotic light source with a wide bandwidth and suppressed delay characteristics through a phase modulator PM7 and a first dispersion compensation optical fiber DCF8. This chaotic light source enters the reserve pool neural network RC12 prediction module through the photodetector PD10 and the ADC module 11. After training, it can accurately predict the dynamic trajectory of the chaotic encryption system and achieve complete chaos synchronization in the information receiving module 25; the light source output by the first continuous wave laser CW1 is modulated by the parallel orthogonal modulator IQM2 so that the signal output by the arbitrary waveform generator AWG3 is modulated from the electrical domain to the optical domain for transmission. The signal is encoded in The code time is to change the timing of the transmission signal in a random manner, rearrange the continuous data according to a certain rule, so as to realize pseudo-random interleaving coding, and load the transmitted signal to the chaotic carrier through the first optical coupler FC4 to realize chaos hiding, and then transmit through the erbium-doped fiber amplifier EDFA13, the standard single-mode fiber SSFM14, and the second dispersion compensation fiber DCF15, and then modulate the chaotic prediction signal through the Mach-Zehnder modulator MZM17 to the continuous optical carrier output by the second continuous wave laser CW16 in the chaotic decryption module 23, and then the chaotic encryption signal output by the information transmission module 24 and the 90° mixer 18 are subtracted and decrypted to preliminarily restore the original transmission signal. Finally, the obtained original transmission signal is passed through the information receiving module 25 composed of the local oscillator laser LO20 and the coherent receiver CoherentReceiver19, and the digital information processing DSP is carried out to realize the compensation, recovery and decryption of the digital signal, thereby realizing chaotic confidential coherent optical communication.

[0070] The chaotic encryption module 21 optimizes the chaotic source and outputs it to the reserve pool computing neural network prediction module 22 and the information transmission module 24. The chaotic encryption module 21 performs chaotic modulation on the signal modulated by the parallel orthogonal modulator IQM2 and the chaotic encrypted signal output by the chaotic encryption module 21, and transmits it to the chaotic decryption module 23 over a long distance. The chaotic decryption module 23 then decrypts the required information, and the information receiving module performs signal compensation and optimization processing.

[0071] Embodiment 2:

[0072] according to Figure 2 As shown, the present invention adopts Figure 2 The structure of the reserve pool neural network RC in the reserve pool calculation neural network prediction module 22 is as follows: it includes an input layer, a reserve pool layer composed of two reserve pools with the same parameters; and an output layer. Each layer contains a number of neurons, and the number of neurons in the reserve pool can be set according to the complexity of the problem. The present invention takes one layer of input and output, each layer has N neurons, and the reserve pool layer has M neurons. The reserve pool calculation dynamics equation can be expressed as follows.

[0073]

[0074] W out =YX T (XX T +λη)(8)

[0075]

[0076] x(t+1)=αy 2,1 (t)+(1-α)y 1,2 (t)(10)

[0077] In the above formula: x is the input data, t is the time, Δt is the change time, W * is the internal weight matrix, W in is the input weight matrix, W out is the connection matrix, u(t) is the input data, Y is the output matrix, X is the state matrix, and λ is the ridge regression parameter, which is used to avoid overfitting. η is the unit matrix. a is the leakage rate, and its value range is between (0, 1). tanh is the activation function, which is used to introduce nonlinear factors into neurons so that the reserve pool can arbitrarily approximate any nonlinear function, so that the reserve pool can be applied to many nonlinear models. According to formula (10), in order to enable the reserve pool calculation to obtain better prediction results, inspired by the diffusion coupling of oscillators, the two reserve pool calculations can achieve an interactive effect. The present invention adopts formula (10) to couple the two reserve pool calculations, where y1 and y2 are the outputs of reserve pool 1 and reserve pool 2 respectively. By coupling the outputs of the two reserve pool calculations, that is, mixing their outputs as new inputs, the two reserve pool calculations are linked together to achieve interaction between the two reserve pool calculations. Wherein the reserve pool size N is 500, the spectral radius rho is 1.2, and the regularization parameter reg is 1e -6 , α is the coupling coefficient between the two reservoir calculations. When α = 0, the results of the two reservoir calculations do not interact and are output in sequence; when α = 1, the two reservoir calculations are in a one-way coupling output relationship. This also shows that α describes the coupling mode and coupling strength between the two reservoir calculations.

[0078] The reserve pool calculation principle of the present invention can be briefly described as follows: in the training stage, the driving laser SL6 outputs chaotic laser which is output through the photodetector PD10 and the ADC module 11 and is used as the input u(t) of the reserve pool calculation neural network prediction module 22. Through the randomly determined connection weights Win1 and Win2 between the input layer and the two reserve pools and the connection weights W1 and W2 between the neurons inside the reserve pool, the reserve pool iteratively generates a state vector X(t). At the same time, the chaotic carrier data Y from the chaotic encryption module 21 is used as the target output to be combined with X(t) to calculate the connection weight Wout between the reserve pool and the output layer. It will not change after the training is completed and is used in the testing stage.

[0079] During the testing phase, the driving laser SL6 continues to output chaotic laser, which passes through the photodetector PD10 and the ADC module 11 and then outputs chaotic data as the input u(t) of the reserve pool calculation neural network prediction module 22. Through the weight matrices Win1, Win2 and the internal weight matrices W1, W2, the reserve pool iteratively generates the state vector X(t). The state vector X(t) quickly generates the output Y(t)=X(t)*Wout through the Wout linear regression. This output is the predicted chaotic carrier. When the prediction accuracy reaches the set range, the chaotic synchronization of the reserve pool calculation neural network prediction module 22 and the driving laser SL6 can be achieved.

[0080] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A secure communication system based on reserve pool chaos prediction and interleaving coding, characterized in that: It includes a chaos encryption module, a reserve pool calculation neural network prediction module, a chaos decryption module, an information transmission module and an information receiving module; The chaotic encryption module includes a plane reflector M, the output end of which is sequentially connected to a driving laser SL, a phase modulator PM, a first dispersion compensation optical fiber DCF and a second optical coupler FC; the output end of the second optical coupler FC is connected to the first optical coupler FC and a photodetector PD; The input end of the first optical coupler FC is connected to a parallel quadrature modulator IQM, and the parallel quadrature modulator IQM is connected to a first continuous wave laser CW and an arbitrary waveform generator AWG; the output end of the first optical coupler FC is connected to the input end of the erbium-doped fiber amplifier EDFA; The reserve pool calculation neural network prediction module includes a photodetector PD, the output end of the photodetector PD is connected to an ADC module and a reserve pool neural network RC in sequence; the output end of the reserve pool neural network RC is connected to a Mach-Zehnder modulator MZM; The chaos decryption module includes a Mach-Zehnder modulator MZM, and the Mach-Zehnder modulator MZM is connected to a second continuous wave laser CW and a 90° mixer; The information transmission module includes N optical fiber link modules, each of which is composed of an erbium-doped optical fiber amplifier EDFA, a single-mode optical fiber SSFM and a second dispersion-compensating optical fiber DCF; the output end of the erbium-doped optical fiber amplifier EDFA is sequentially connected to the single-mode optical fiber SSFM and the second dispersion-compensating optical fiber DCF, the second dispersion-compensating optical fiber DCF is connected to a 90° mixer; the 90° mixer is connected to a coherent receiver; The information receiving module comprises a local oscillator laser LO and a coherent receiver; the coherent receiver is connected to the output end of the 90° mixer.

2. The secure communication system based on reserve pool chaos prediction and interleaving coding according to claim 1 is characterized in that: The reservoir neural network RC includes an input layer, a reservoir layer and an output layer; each layer contains a number of neurons.

3. The secure communication system based on reserve pool chaos prediction and interleaving coding according to claim 2 is characterized in that: The reservoir layer consists of two reservoirs with consistent parameters through linear coupling. The number of neurons in the reservoir is set according to the complexity of the problem; the reservoir size N = 500, the spectral radius rho = 1.2, and the regularization parameter reg = 1e -6 ; When the coupling coefficient α=0, the results of the two reservoir calculations do not interact and are output in sequence; when α=1, the calculations of the two reservoirs are in a unidirectional coupling output relationship.

4. A secure communication method based on reserve pool chaos prediction and interleaving coding, characterized in that: The following steps are involved: S1: The optical feedback structure formed by the driving laser SL and the plane reflector M is used to generate chaotic laser. After being optimized by the phase modulator PM and the first dispersion compensation fiber DCF, the chaotic laser outputs two identical optical signals through the second coupler FC with a splitting ratio of 50:

50. One optical signal is used as a driving signal and is converted into a chaotic electrical signal through the photoelectric conversion of the photodetector PD and the ADC module and input into the reserve pool calculation neural network prediction module; the other optical signal is input into the first optical coupler FC as a driving signal; S2: The first continuous wave laser CW generates an optical signal. The constellation signal to be transmitted is encoded offline by an arbitrary waveform generator AWG, and then modulated onto the optical signal generated by the first continuous wave laser CW by a parallel quadrature modulator IQM. The constellation signal needs to be modulated onto the optical signal after an encryption algorithm, thereby generating the original information m(t). S3, the optical signal input into the first optical coupler FC in step S1 is superimposed with the original information m(t) in step S2 to generate a chaotic encryption signal c(t)+m(t), and the chaotic encryption signal is sequentially transmitted to the chaotic decryption module via the information transmission module composed of the erbium-doped fiber amplifier EDFA, the single-mode fiber SSFM, and the second dispersion compensation fiber DCF; S4: The chaotic electrical signal in S1 is used as the input of the reserve pool calculation neural network prediction module, and the weight of the reserve pool neural network RC is determined by optimizing the reserve pool neural network RC training phase, so that the output signal of the output layer of the reserve pool neural network RC is synchronized with the input signal of the input layer of the reserve pool neural network RC, so as to realize the chaotic prediction of the input signal, which is the chaotic prediction signal; S5: In the RC test phase of the reserve pool neural network, the output layer outputs a chaotic prediction signal, and outputs a chaotic optical carrier c'(t) synchronized with the output of the chaotic encryption module through the Mach-Zehnder modulator MZM under the synchronous control of the second continuous wave laser CW; the chaotic encryption signal c(t)+m(t) output by the information transmission module and the chaotic optical carrier c'(t) are decrypted through a 90° mixer to obtain the original transmission signal m'(t); S6: The original transmission signal m'(t) is input to the information receiving module, and after being coherently received by the local oscillator laser LO and the coherent receiver, an offline digital information processing DSP is performed to optimize and compensate the original transmission signal m'(t).

5. The secure communication method based on reserve pool chaos prediction and interleaving coding according to claim 4 is characterized in that: The generation of the constellation signal in step S2 is to map the binary bit data to different symbol points of the constellation according to the phase amplitude, thereby generating a QAM signal to be transmitted.

6. The secure communication method based on reserve pool chaos prediction and interleaving coding according to claim 4 is characterized in that: The specific process of the encryption algorithm in step S2: The Lorenz three-dimensional chaos equation is as follows: Where σ, ρ, β are dimensionless constants, x, y, z are state components of the chaotic system, and t is time. After solving the chaotic equation and digitizing the chaotic sequence, the obtained x, y, z are integerized, as shown in the following formula: P x,i =mod(Extract(x i ,m,n,p),M) P y,i =mod(Extract(y i ,m,n,p),M) P z,i =mod(Extract(z i ,m,n,p),M) where x i ,y i 、z i is a chaotic sequence, m, n, p are specific bit numbers extracted from the decimal part, M is the decimal number of the bit, and the Extract function is implemented to extract specific bit numbers from the decimal part to obtain a random sequence {P x,i }、{P y,i } and {P z,i }, using {P x,i }Generate a chaotic synchronization sequence and insert it into the front of the QAM signal frame; use the digitized sequence {P y,i } and {P z,i } to construct a scrambling sequence, and pseudo-randomly interleave the QAM signal with the scrambling sequence to achieve signal encryption.

7. The secure communication method based on reserve pool chaos prediction and interleaving coding according to claim 4 is characterized in that: The training phase in step S4 is specifically as follows: The output of the driving laser SL is used as the input of the reserve pool calculation neural network prediction module after passing through the photodetector PD, and is used to iteratively generate the internal node states of the reserve pool; During the training phase, the chaotic electrical signal of the input layer is transmitted to the reservoir layer in a back-to-back manner and used as the target output data calculated by the reservoir layer to calculate the connection weight between the reservoir layer and the output layer.

8. The secure communication method based on reserve pool chaos prediction and interleaving coding according to claim 4 is characterized in that: The decryption in step S5 uses the same chaotic equation, the same initial value and the same number of iterations as those used in the chaotic encryption module to construct the same scrambling sequence and obtain the same digital encryption information, thereby restoring the original transmission signal m'(t) after scrambling coding to the original constellation diagram.

Images