Incoherent chaotic communication method and device based on deep learning

Abstract

The present invention discloses a non-coherent chaotic communication method and device based on deep learning, which relates to the field of data processing, including: obtaining a chaotic receiving signal to be demodulated and performing power spectrum density estimation, obtaining a PSD sequence and inputting it into a trained chaotic receiving signal demodulation model, receiving the PSD sequence through an input layer, the PSD sequence passes through a first BiLSTM layer, extracting a first time domain feature of the PSD sequence, and then passing through a self-attention layer to obtain a first intermediate feature, the first intermediate feature passes through a first Dropout layer to obtain a second intermediate feature, the second intermediate feature passes through a second BiLSTM layer, extracting a second time domain feature of the PSD sequence, and then passing through a second Dropout layer to obtain a third intermediate feature; the third intermediate feature is input into a fully connected layer and a Softmax layer for feature integration to obtain a probability distribution vector, the probability distribution vector is subjected to maximum likelihood judgment through a classification layer, and estimated binary information bits are obtained. The present invention solves the problem of low spectrum efficiency of existing chaotic communication systems.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to an incoherent chaotic communication method and device based on deep learning. Background Art

[0002] With the rapid development of wireless communication technology, the requirements for the security, anti-interference ability and spectrum utilization of communication systems are constantly increasing. Chaotic communication, as a new type of communication technology, has attracted widespread attention due to its good confidentiality, anti-interference ability and spectrum utilization. Traditional chaotic communication systems are mainly divided into two categories: coherent detection and incoherent detection. The coherent detection method requires the reconstruction of the chaotic signal at the transmitter at the receiver, which is complex to implement and has high requirements on channel conditions, such as chaotic synchronized keying (CSK). Although the incoherent detection method does not require chaotic synchronization, it needs to send additional reference signals, which reduces the spectrum efficiency and security of the system, such as differential chaotic shift keying (DCSK).

[0003] In recent years, the application of deep learning technology in the field of communications has attracted widespread attention. Some studies have tried to apply deep learning to chaotic communication systems to improve system performance. For example, a study proposed an OFDM-DCSK receiver based on deep learning, which uses a neural network to achieve signal detection. However, it still needs to send a reference signal, and does not fundamentally solve the spectrum efficiency problem of the DCSK system. Summary of the invention

[0004] The purpose of this application is to propose an incoherent chaotic communication method and device based on deep learning to address the above-mentioned technical problems.

[0005] In a first aspect, the present invention provides an incoherent chaotic communication method based on deep learning, comprising the following steps:

[0006] Obtain the chaotic received signal to be demodulated and perform power spectrum density estimation to obtain a PSD sequence;

[0007] A chaotic reception signal demodulation model based on a deep neural network is constructed and trained to obtain a trained chaotic reception signal demodulation model, wherein the chaotic reception signal demodulation model includes an input layer, a hidden layer, and an output layer, wherein the hidden layer includes a first BiLSTM layer, a self-attention layer, a first Dropout layer, a second BiLSTM layer, a second Dropout layer, a fully connected layer, and a Softmax layer connected in sequence, and the output layer includes a classification layer;

[0008] The PSD sequence is input into the trained chaotic receiving signal demodulation model, the PSD sequence is received through the input layer, the PSD sequence is input into the hidden layer, the PSD sequence passes through the first BiLSTM layer, the first time domain feature of the PSD sequence is extracted, the first time domain feature of the PSD sequence passes through the self-attention layer, the first intermediate feature is obtained, the first intermediate feature is randomly deactivated through the first Dropout layer to obtain the second intermediate feature, the second intermediate feature is passed through the second BiLSTM layer, the second time domain feature of the PSD sequence is extracted, the second time domain feature of the PSD sequence is randomly deactivated through the second Dropout layer to obtain the third intermediate feature; the third intermediate feature is input into the fully connected layer and the Softmax layer for feature integration to obtain a probability distribution vector, the probability distribution vector is input into the output layer, the maximum likelihood decision is made through the classification layer, and the estimated binary information bit is obtained.

[0009] Preferably, the power spectrum density estimation process adopts the Welch method, and the calculation formula of the data at the mth position in the PSD sequence is as follows:

[0010] ;

[0011] in, represents the chaotic received signal to be demodulated, Indicates the first chaotic received signal to be demodulated data of locations, is the length of the data segment, An index indicating the length of the data segment. is the number of segments, k' represents the index of the segment number, is the overlap length between adjacent segments, is the number of fast Fourier sampling points, is the Hamming window function.

[0012] Preferably, the first BiLSTM layer and the second BiLSTM layer both adopt a BiLSTM structure.

[0013] Preferably, the calculation process of the self-attention layer is as follows:

[0014] The query matrix, key matrix and value matrix are calculated from the first time domain feature of the PSD sequence as shown in the following formula:

[0015] ;

[0016] ;

[0017] ;

[0018] in, , and denote the query matrix, key matrix and value matrix respectively, , and denote the weight matrices corresponding to the query matrix, key matrix, and value matrix respectively, and P denotes the first time domain feature of the PSD sequence;

[0019] The attention matrix is ​​calculated based on the query matrix, key matrix and value matrix, as shown in the following formula:

[0020]

[0021] Where T represents the transposed matrix, d represents the dimension of the key matrix, and A represents the attention matrix;

[0022] The first intermediate feature is calculated using the following formula:

[0023] ;

[0024] Wherein, Z represents the first intermediate feature.

[0025] As a preference, the probability distribution vector is expressed as ,in, and They represent the probability of the chaotic received signal corresponding to the binary bit "0" and "1", and the calculation process of the probability distribution vector is as follows:

[0026] ;

[0027] in, represents the parameter matrix, represents the bias parameter, Indicates the output unit, corresponding to binary bits "0" and "1", Indicates the third intermediate feature;

[0028] The estimated binary information bits are calculated as follows:

[0029] like > ,but ;

[0030] like < ,but ;

[0031] in, represents the estimated binary information bits.

[0032] Preferably, the acquisition process of the chaotic received signal to be demodulated is as follows:

[0033] Obtaining binary information bits to be sent;

[0034] According to the binary information bits to be sent, a chaotic mapping function is selected and modulated to obtain a chaotic transmission signal. The binary information bits to be sent are expressed as , where n represents the nth binary information to be sent. If , then select the Logistic mapping function to map the binary information bits to be sent. If , then select Cubic mapping to map the binary information bits to be sent, as shown in the following formula:

[0035] ;

[0036] in, represents the chaotic sequence generated by the Logistic map, represents the chaotic sequence generated by the Cubic mapping, represents the value of the kth position in the chaotic sequence, is the expansion factor, represents the chaotic emission signal;

[0037] The chaotic transmission signal is transmitted through the channel to obtain the chaotic reception signal to be demodulated.

[0038] In a second aspect, the present invention provides an incoherent chaotic communication device based on deep learning, comprising:

[0039] The signal processing module is configured to obtain the chaotic received signal to be demodulated and perform power spectrum density estimation to obtain a PSD sequence;

[0040] A model building module is configured to build and train a chaotic reception signal demodulation model based on a deep neural network to obtain a trained chaotic reception signal demodulation model, wherein the chaotic reception signal demodulation model includes an input layer, a hidden layer, and an output layer, wherein the hidden layer includes a first BiLSTM layer, a self-attention layer, a first Dropout layer, a second BiLSTM layer, a second Dropout layer, a fully connected layer, and a Softmax layer connected in sequence, and the output layer includes a classification layer;

[0041] The demodulation module is configured to input the PSD sequence into a trained chaotic receiving signal demodulation model, receive the PSD sequence through the input layer, input the PSD sequence into the hidden layer, pass the PSD sequence through the first BiLSTM layer, extract the first time domain feature of the PSD sequence, pass the first time domain feature of the PSD sequence through the self-attention layer, obtain the first intermediate feature, pass the first intermediate feature through the first Dropout layer for random inactivation processing, obtain the second intermediate feature, pass the second intermediate feature through the second BiLSTM layer, extract the second time domain feature of the PSD sequence, pass the second time domain feature of the PSD sequence through the second Dropout layer for random inactivation processing, obtain the third intermediate feature; input the third intermediate feature into the fully connected layer and the Softmax layer for feature integration to obtain a probability distribution vector, input the probability distribution vector into the output layer, perform maximum likelihood judgment through the classification layer, and obtain estimated binary information bits.

[0042] In a third aspect, the present invention provides an electronic device comprising one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0043] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any implementation manner in the first aspect.

[0044] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, which, when executed by a processor, implements the method described in any implementation manner in the first aspect.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) The incoherent chaotic communication method based on deep learning proposed in the present invention utilizes a deep neural network (DNN) to learn and demodulate the power spectral density characteristics of the chaotic received signal. It can achieve efficient and reliable incoherent demodulation without sending a reference signal, effectively improving the spectrum efficiency and security of the system.

[0047] (2) The incoherent chaotic communication method based on deep learning proposed in the present invention uses a deep neural network to process the PSD sequence, which can effectively extract the time domain characteristics and frequency domain characteristics of the chaotic signal and realize efficient and reliable incoherent demodulation.

[0048] (3) The incoherent chaotic communication method based on deep learning proposed in the present invention adopts two different chaotic mapping functions to generate chaotic sequences, which is beneficial to increase the chaos of the signal and improve the security of the communication system. The multipath Rayleigh fading channel is used as the channel model, which makes the communication environment closer to the real communication scenario and effectively verifies the performance of the proposed system in a complex channel environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 A flowchart of an incoherent chaotic communication method based on deep learning according to an embodiment of the present application;

[0051] Figure 2 The overall framework diagram of the incoherent chaotic communication method based on deep learning in an embodiment of the present application;

[0052] Figure 3 A structural schematic diagram of a chaotic reception signal demodulation model of an incoherent chaotic communication method based on deep learning according to an embodiment of the present application;

[0053] Figure 4 A schematic diagram of the power spectral density characteristics of a chaotic transmission signal of an incoherent chaotic communication method based on deep learning according to an embodiment of the present application;

[0054] Figure 5 A schematic diagram of the power spectral density characteristics of a chaotic received signal after passing through a multipath Rayleigh fading channel in an incoherent chaotic communication method based on deep learning according to an embodiment of the present application;

[0055] Figure 6 A comparison chart of the bit error rate (BER) performance of the incoherent chaotic communication method based on deep learning in the embodiment of the present application, the traditional DCSK method and the ideal antipodal CSK method in a multipath Rayleigh fading channel environment is given;

[0056] Figure 7 A comparison chart of the bit error rate (BER) performance of the incoherent chaotic communication method based on deep learning in the embodiment of the present application, the DLCSK method, and the OFDM-DCSK method based on deep learning in a multipath Rayleigh fading channel environment is given;

[0057] Figure 8A schematic diagram of an incoherent chaotic communication device based on deep learning according to an embodiment of the present application;

[0058] Fig. 9 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Figure 1 An incoherent chaotic communication method based on deep learning provided by an embodiment of the present application is shown, comprising the following steps:

[0061] S1, obtain the chaotic received signal to be demodulated and perform power spectrum density estimation to obtain a PSD sequence.

[0062] In a specific embodiment, the acquisition process of the chaotic received signal to be demodulated is as follows:

[0063] Obtaining binary information bits to be sent;

[0064] According to the binary information bits to be sent, a chaotic mapping function is selected and modulated to obtain a chaotic transmission signal. The binary information bits to be sent are expressed as , where n represents the nth binary information to be sent. If , then select the Logistic mapping function to map the binary information bits to be sent. If , then select Cubic mapping to map the binary information bits to be sent, as shown in the following formula:

[0065] ;

[0066] in, represents the chaotic sequence generated by the Logistic map, represents the chaotic sequence generated by the Cubic mapping, represents the value of the kth position in the chaotic sequence, is the expansion factor, represents the chaotic emission signal;

[0067] The chaotic transmission signal is transmitted through the channel to obtain the chaotic reception signal to be demodulated.

[0068] Specifically, refer to Figure 2The transmitter includes two chaotic sequence generators and a modulator; the chaotic sequence generator contains two chaotic mapping functions, namely Logistic mapping and Cubic mapping; the modulator selects the corresponding chaotic sequence for modulation according to the binary information bits to be sent to obtain the chaotic transmission signal of the transmitter. The receiver includes a power spectral density (PSD) estimator and a demodulator; the PSD estimator is used to estimate the PSD of the chaotic receiving signal to obtain the PSD sequence, and the demodulator is used to demodulate the PSD sequence to obtain the estimated binary information bits.

[0069] In the transmitter, different chaotic signal generators are used to map the binary information bits to be transmitted. That is, two different chaotic sequences are generated using Logistic mapping and Cubic mapping. The expression of chaotic mapping is:

[0070] Logistic Mapping: ;

[0071] Cubic Mapping: ;

[0072] in, represents the chaotic sequence generated by the Logistic map data of locations, represents the chaotic sequence generated by the Logistic map data of locations, Indicates the chaotic sequence generated by the Cubic mapping. data of locations, Indicates the chaotic sequence generated by the Cubic mapping. data in each position, expansion factor is the length of the chaotic sequence.

[0073] The chaotic emission signal is composed of two different chaotic sequences. , chaotic emission signal Transmit through the channel and get the received signal In the embodiment of the present application, a multipath Rayleigh fading channel is used as the channel model, and the expression of the data at the kth position in the received signal is as follows:

[0074] ;

[0075] in, is an additive Gaussian white noise vector with a mean of 0 and a variance of ; is the number of Rayleigh channel paths; For the The Rayleigh channel fading coefficient of the path; is the delay between adjacent paths.

[0076] In a specific embodiment, the power spectrum density estimation process adopts the Welch method, and the calculation formula of the data at the mth position in the PSD sequence is as follows:

[0077] ;

[0078] in, represents the chaotic received signal to be demodulated, Indicates the first chaotic received signal to be demodulated Data for each location, is the length of the data segment, An index indicating the length of the data segment, is the number of segments, k' represents the index of the segment number, is the overlap length between adjacent segments, is the number of fast Fourier sampling points, is the Hamming window function.

[0079] Specifically, the power spectral density (PSD) of the received signal is estimated, and the estimation process adopts the Welch method.

[0080] S2, construct and train a chaotic reception signal demodulation model based on a deep neural network to obtain a trained chaotic reception signal demodulation model, wherein the chaotic reception signal demodulation model includes an input layer, a hidden layer, and an output layer, wherein the hidden layer includes a first BiLSTM layer, a self-attention layer, a first Dropout layer, a second BiLSTM layer, a second Dropout layer, a fully connected layer, and a Softmax layer connected in sequence, and the output layer includes a classification layer.

[0081] In a specific embodiment, both the first BiLSTM layer and the second BiLSTM layer adopt a BiLSTM structure.

[0082] Specifically, refer to Figure 3 ,The chaotic received signal demodulation model adopts a deep neural network consisting of an input layer, a first bidirectional long short-term memory (BiLSTM) layer, a self-attention layer, a first Dropout layer, a second bidirectional long short-term memory (BiLSTM) layer, a second Dropout layer, a fully connected layer, a Softmax layer, and a classification layer.

[0083] S3, input the PSD sequence into the trained chaotic receiving signal demodulation model, receive the PSD sequence through the input layer, input the PSD sequence into the hidden layer, pass the PSD sequence through the first BiLSTM layer, extract the first time domain feature of the PSD sequence, pass the first time domain feature of the PSD sequence through the self-attention layer, obtain the first intermediate feature, pass the first intermediate feature through the first Dropout layer for random inactivation to obtain the second intermediate feature, pass the second intermediate feature through the second BiLSTM layer, extract the second time domain feature of the PSD sequence, pass the second time domain feature of the PSD sequence through the second Dropout layer for random inactivation to obtain the third intermediate feature; input the third intermediate feature into the fully connected layer and the Softmax layer for feature integration to obtain a probability distribution vector, input the probability distribution vector into the output layer, perform maximum likelihood judgment through the classification layer, and obtain estimated binary information bits.

[0084] In a specific embodiment, the calculation process of the self-attention layer is as follows:

[0085] The query matrix, key matrix and value matrix are calculated from the first time domain feature of the PSD sequence as shown in the following formula:

[0086] ;

[0087] ;

[0088] ;

[0089] in, , and denote the query matrix, key matrix and value matrix respectively, , and denote the weight matrices corresponding to the query matrix, key matrix, and value matrix respectively, and P denotes the first time domain feature of the PSD sequence;

[0090] The attention matrix is ​​calculated based on the query matrix, key matrix and value matrix, as shown in the following formula:

[0091]

[0092] Where T represents the transposed matrix, d represents the dimension of the key matrix, and A represents the attention matrix;

[0093] The first intermediate feature is calculated using the following formula:

[0094] ;

[0095] Wherein, Z represents the first intermediate feature.

[0096] In a specific embodiment, the probability distribution vector is expressed as ,in, and They represent the probability of the chaotic received signal corresponding to the binary bit "0" and "1", and the calculation process of the probability distribution vector is as follows:

[0097] ;

[0098] in, represents the parameter matrix, represents the bias parameter, Indicates the output unit, corresponding to binary bits "0" and "1", Indicates the third intermediate feature;

[0099] The estimated binary information bits are calculated as follows:

[0100] like > ,but ;

[0101] like < ,but ;

[0102] in, represents the estimated binary information bits.

[0103] Specifically, the estimated PSD sequence Input the trained chaotic receiving signal demodulation model, first receive the PSD sequence through the input layer ; PSD sequence Input to the first BiLSTM layer for processing, by PSD sequence Perform BiLSTM calculations at each time step to obtain the first time-frequency features of the PSD sequence at each time step, which ultimately constitute the first time-frequency features of the PSD sequence. , It represents the first time domain feature of the PSD sequence at time step t. In the BiLSTM structure, the input sequence is considered to have time correlation, so the definition is that each position is a time step, so the time step t is consistent with the sequence length.

[0104] Furthermore, the self-attention layer is used to perform weighted processing on the first time-frequency feature of the PSD sequence to obtain the first intermediate feature Z; the first intermediate feature Z output by the self-attention layer is randomly inactivated through the first Dropout layer to prevent overfitting; the processed second intermediate feature is input into the second BiLSTM layer for deep feature extraction to obtain the second time domain feature of the PSD sequence, and the structure of the second BiLSTM layer is the same as that of the first BiLSTM layer; the second time domain feature of the PSD sequence is randomly inactivated through the second Dropout layer to obtain the third intermediate feature Z'. After the above series of linear and nonlinear operations, the third intermediate feature Z' is input into the fully connected layer and the Softmax layer for feature integration to obtain the probability distribution vector , according to the probability distribution vector Perform maximum likelihood decision to obtain estimated binary information bits .

[0105] The following embodiments of the present application analyze the time domain characteristics and frequency domain characteristics of the chaotic signal in the incoherent chaotic communication method (β=100) based on deep learning proposed in the embodiments of the present application.

[0106] like Figure 4 The power spectral density characteristics of the chaotic emission signals generated by the two mappings are shown, and they each present unique frequency domain distribution characteristics; Figure 5 The power spectral density characteristics of the chaotic received signal are shown. Despite the channel transmission, the difference in frequency domain characteristics of the two mappings remains stable, which proves the feasibility of using power spectral density characteristics for demodulation.

[0107] like Figure 6 As shown in the figure, in a Rayleigh fading channel environment (G=2, τ=1), the incoherent chaotic communication method based on deep learning proposed in the embodiment of the present application is compared with the traditional DCSK method and the ideal antipodal CSK method. For different expansion factors β (β=64, 256, 512), the incoherent chaotic communication method based on deep learning proposed in the embodiment of the present application is better than the DCSK method. In particular, when β=512, the signal-to-noise ratio =30, a gain of about 10dB is obtained, and the performance is close to the ideal antipodal CSK method.

[0108] like Figure 7As shown in the figure, under the same Rayleigh fading channel conditions, the incoherent chaotic communication method based on deep learning proposed in the embodiment of the present application is compared with the DLCSK method and the OFDM-DCSK method based on deep learning. The experimental results show that the incoherent chaotic communication method based on deep learning proposed in the embodiment of the present application exhibits superior performance. Under the same expansion factor β, when When it is greater than 14 dB, the performance of the incoherent chaotic communication method based on deep learning proposed in the embodiment of the present application is better than the DLCSK system and the OFDM-DCSK method based on deep learning.

[0109] Further references Figure 8 As an implementation of the methods shown in the above figures, the present application provides an embodiment of an incoherent chaotic communication device based on deep learning, and the device embodiment is similar to Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0110] The embodiment of the present application provides an incoherent chaotic communication device based on deep learning, comprising:

[0111] The signal processing module 1 is configured to obtain the chaotic received signal to be demodulated and perform power spectrum density estimation to obtain a PSD sequence;

[0112] The model construction module 2 is configured to construct and train a chaotic reception signal demodulation model based on a deep neural network to obtain a trained chaotic reception signal demodulation model, wherein the chaotic reception signal demodulation model includes an input layer, a hidden layer, and an output layer, wherein the hidden layer includes a first BiLSTM layer, a self-attention layer, a first Dropout layer, a second BiLSTM layer, a second Dropout layer, a fully connected layer, and a Softmax layer connected in sequence, and the output layer includes a classification layer;

[0113] The demodulation module 3 is configured to input the PSD sequence into a trained chaotic receiving signal demodulation model, receive the PSD sequence through the input layer, input the PSD sequence into the hidden layer, pass the PSD sequence through the first BiLSTM layer, extract the first time domain feature of the PSD sequence, pass the first time domain feature of the PSD sequence through the self-attention layer, obtain the first intermediate feature, pass the first intermediate feature through the first Dropout layer for random inactivation processing, obtain the second intermediate feature, pass the second intermediate feature through the second BiLSTM layer, extract the second time domain feature of the PSD sequence, pass the second time domain feature of the PSD sequence through the second Dropout layer for random inactivation processing, obtain the third intermediate feature; input the third intermediate feature into the fully connected layer and the Softmax layer for feature integration to obtain a probability distribution vector, input the probability distribution vector into the output layer, perform maximum likelihood judgment through the classification layer, and obtain estimated binary information bits.

[0114] Fig. 9 Schematic diagram of the hardware structure of the electronic device provided by the embodiment of the present invention. Figure 7 As shown, the electronic device of this embodiment includes: a processor 901 and a memory 902; wherein the memory 902 is used to store computer-executable instructions; the processor 901 is used to execute the computer-executable instructions stored in the memory to implement the various steps performed by the electronic device in the above embodiment. For details, please refer to the relevant description in the above method embodiment.

[0115] Optionally, the memory 902 may be independent or integrated with the processor 901 .

[0116] When the memory 902 is independently provided, the electronic device further includes a bus 903 for connecting the memory 902 and the processor 901 .

[0117] The embodiment of the present invention further provides a computer storage medium, in which computer execution instructions are stored. When the processor 901 executes the computer execution instructions, the above method is implemented.

[0118] The embodiment of the present invention further provides a computer program product, including a computer program. When the computer program is executed by the processor 901, the above method is implemented.

[0119] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0120] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to implement the solution of this embodiment.

[0121] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each module may exist physically separately, or two or more modules may be integrated into one unit. The unit formed by the above modules may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0122] The above-mentioned integrated module implemented in the form of a software function module can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium, including a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor 901 to perform some steps of the methods of various embodiments of the present application.

[0123] It should be understood that the processor 901 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or the processor 901 may be any conventional processor 901, etc. The steps of the method disclosed in the invention may be directly embodied in the hardware processor 901 for execution, or may be executed by a combination of hardware and software modules in the processor 901.

[0124] The memory 902 may include a high-speed RAM memory, and may also include a non-volatile storage NVM, such as at least one disk memory, and may also be a USB flash drive, a mobile hard disk, a read-only memory, a magnetic disk, or an optical disk.

[0125] The bus 903 may be an Industry Standard Architecture (ISA), a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus 903 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus 903 in the drawings of the present application is not limited to only one bus 903 or one type of bus 903.

[0126] The above storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The storage medium can be any available medium that can be accessed by a general or special purpose computer.

[0127] An exemplary storage medium is coupled to the processor 901, so that the processor 901 can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor 901. The processor 901 and the storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor 901 and the storage medium can also exist as discrete components in an electronic device or a main control device.

[0128] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A non-coherent chaotic communication method based on deep learning, characterized in that: The following steps are involved: The chaotic received signal to be demodulated is obtained and the power spectrum density is estimated to obtain a PSD sequence. The acquisition process of the chaotic received signal to be demodulated is as follows: Obtaining binary information bits to be sent; A chaotic mapping function is selected according to the binary information bit to be transmitted, and a chaotic transmission signal is obtained by modulation. The binary information bit to be transmitted is represented by d n ∈{0,1}, where n represents the nth binary information to be sent. If d n =0, then select the Logistic mapping function to map the binary information bits to be sent. If d n =1, then select Cubic mapping to map the binary information bits to be sent, as shown in the following formula: Among them, f1(x k ) represents the chaotic sequence generated by the Logistic map, f2(x k ) represents the chaotic sequence generated by the Cubic mapping, x k represents the value of the kth position in the chaotic sequence, β is the expansion factor, s n represents the chaotic emission signal; The chaotic transmission signal is transmitted through a channel to obtain the chaotic reception signal to be demodulated; A chaotic reception signal demodulation model based on a deep neural network is constructed and trained to obtain a trained chaotic reception signal demodulation model, wherein the chaotic reception signal demodulation model comprises an input layer, a hidden layer and an output layer, wherein the hidden layer comprises a first BiLSTM layer, a self-attention layer, a first Dropout layer, a second BiLSTM layer, a second Dropout layer, a fully connected layer and a Softmax layer connected in sequence, and the output layer comprises a classification layer; The PSD sequence is input into the trained chaotic receiving signal demodulation model, the PSD sequence is received through the input layer, the PSD sequence is input into the hidden layer, the PSD sequence passes through the first BiLSTM layer, the first time domain feature of the PSD sequence is extracted, the first time domain feature of the PSD sequence passes through the self-attention layer to obtain a first intermediate feature, the first intermediate feature passes through the first Dropout layer for random inactivation to obtain a second intermediate feature, the second intermediate feature passes through the second BiLSTM layer to extract the second time domain feature of the PSD sequence, the second time domain feature of the PSD sequence passes through the second Dropout layer for random inactivation to obtain a third intermediate feature; the third intermediate feature is input into the fully connected layer and the Softmax layer for feature integration to obtain a probability distribution vector, the probability distribution vector is input into the output layer, the maximum likelihood decision is made through the classification layer to obtain an estimated binary information bit.

2. The incoherent chaotic communication method based on deep learning according to claim 1, characterized in that: The power spectrum density estimation process adopts the Welch method, and the calculation formula of the data at the mth position in the PSD sequence is as follows: Among them, r n represents the chaotic received signal to be demodulated, r n [l+(k′-1)D] represents the data at the l+(k′-1)Dth position in the chaotic received signal to be demodulated, L is the data segment length, l represents the index of the data segment length, K is the number of segments, k' represents the index of the number of segments, D is the overlapping length between adjacent segments, M is the number of fast Fourier sampling points, and W(l) is the Hamming window function.

3. The incoherent chaotic communication method based on deep learning according to claim 1, characterized in that: The first BiLSTM layer and the second BiLSTM layer both adopt a BiLSTM structure.

4. The incoherent chaotic communication method based on deep learning according to claim 1, characterized in that: The calculation process of the self-attention layer is as follows: The query matrix, key matrix and value matrix are calculated by the first time domain feature of the PSD sequence, as shown in the following formula: Q=PW Q ; K=PW K ; V=PW V ; Where Q, K, and V represent the query matrix, key matrix, and value matrix, respectively, and W Q , W K and W v denote the weight matrices corresponding to the query matrix, key matrix, and value matrix respectively, and P denotes the first time domain feature of the PSD sequence; The attention matrix is ​​calculated according to the query matrix, key matrix and value matrix, as shown in the following formula: Where T represents the transposed matrix, d represents the dimension of the key matrix, and A represents the attention matrix; The first intermediate feature is calculated using the following formula: Z = VA; Wherein, Z represents the first intermediate feature.

5. The incoherent chaotic communication method based on deep learning according to claim 1, characterized in that: The probability distribution vector is represented by e d =[e0,e1], where e0 and e1 represent the probability of the chaotic received signal corresponding to the binary bit "0" and "1" respectively. The calculation process of the probability distribution vector is as follows: Where W represents the parameter matrix, b represents the bias parameter, u represents the output unit, corresponding to the binary bits "0" and "1", and Z' represents the third intermediate feature; The estimated binary information bits are calculated as follows: If e0>e1, then If e0 < e1, then in, represents the estimated binary information bits.

6. A non-coherent chaotic communication device based on deep learning, characterized in that: include: The signal processing module is configured to obtain the chaotic received signal to be demodulated and perform power spectrum density estimation to obtain a PSD sequence. The acquisition process of the chaotic received signal to be demodulated is as follows: Obtaining binary information bits to be sent; A chaotic mapping function is selected according to the binary information bit to be transmitted, and a chaotic transmission signal is obtained by modulation. The binary information bit to be transmitted is represented by d n ∈{0,1}, where n represents the nth binary information to be sent. If d n =0, then select the Logistic mapping function to map the binary information bits to be sent. If d n =1, then select Cubic mapping to map the binary information bits to be sent, as shown in the following formula: Among them, f1(x k ) represents the chaotic sequence generated by the Logistic map, f2(x k ) represents the chaotic sequence generated by the Cubic mapping, x k represents the value of the kth position in the chaotic sequence, β is the expansion factor, s n represents the chaotic emission signal; The chaotic transmission signal is transmitted through a channel to obtain the chaotic reception signal to be demodulated; A model building module is configured to build and train a chaotic received signal demodulation model based on a deep neural network to obtain a trained chaotic received signal demodulation model, wherein the chaotic received signal demodulation model includes an input layer, a hidden layer, and an output layer, wherein the hidden layer includes a first BiLSTM layer, a self-attention layer, a first Dropout layer, a second BiLSTM layer, a second Dropout layer, a fully connected layer, and a Softmax layer connected in sequence, and the output layer includes a classification layer; The demodulation module is configured to input the PSD sequence into the trained chaotic receiving signal demodulation model, receive the PSD sequence through the input layer, input the PSD sequence into the hidden layer, pass the PSD sequence through the first BiLSTM layer, extract the first time domain feature of the PSD sequence, pass the first time domain feature of the PSD sequence through the self-attention layer to obtain the first intermediate feature, pass the first intermediate feature through the first Dropout layer to perform random inactivation processing to obtain the second intermediate feature, pass the second intermediate feature through the second BiLSTM layer to extract the second time domain feature of the PSD sequence, pass the second time domain feature of the PSD sequence through the second Dropout layer to perform random inactivation processing to obtain the third intermediate feature; input the third intermediate feature into the fully connected layer and the Softmax layer for feature integration to obtain a probability distribution vector, input the probability distribution vector into the output layer, perform maximum likelihood judgment through the classification layer, and obtain estimated binary information bits.

7. An electronic device comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images