Design And Hardware Implementation Of Chaotic Secure Communication Based On Reservoir Computing

In the modern society with the rapid development of digital information,the security of information is the focus of social concern.Since its appearance,chaotic system has been widely used in information encryption because of its noise-like property and unpredictability.For secure communication systems encrypted by chaotic systems,the decryption end usually needs a system synchronized with the chaotic system at the encryption end.The traditional chaotic system synchronization method usually requires the establishment of an identical chaotic system to synchronize the two systems by means of self-drive or coupling,which limits the flexibility of the decryption synchronization system to a certain extent and costs too much in hardware implementation.Therefore,this paper proposes a method based on reservoir computing to synchronize chaotic systems with low error for a long time,and designs a chaotic secure communication mechanism.Finally,common information,such as images and sounds,is used as ciphertext for functional verification experiments and hardware implementation is carried out with FPGA.The specific research is as follows:1.For the first time,the reservoir computing is used to propose a long time synchronization scheme for complex chaotic systems with infinite dimension and delay.Take time delayed Lorenz system as an example.Firstly,the equations are solved in time series,and the equally-spaced sampled data are divided into train sets and validation sets.Secondly,the train set is used to train the initial reservoir computer to get the output matrixW_out.Finally,a single variable drive is used to drive the trained reservoir computer to synchronize them for a long time.Take Mackey＿Glass and Kuramoto-Sivashinsky systems for example,their training methods are similar.In the prediction stage,a long time synchronization is carried out by means of intermittent true value correction.Finally,the effects of the size N and leakage rateαon the prediction accuracy are studied.The results show that a reservoir computing with suitable parameters synchronizes the original system with an error of 2-3 orders of magnitude lower than the true value.2.A chaotic secure communication mechanism is designed based on the research of chaotic synchronization in reservoir computing.The reliability of the scheme is verified by image secure communication system based on time delayed Lorenz system,voice secure communication system based on Mackey＿Glass and Kuramoto-Sivashinsky system.First of all,the chaotic system is sampled and divided into intensive and train sets.The train set is sent to the receiving end of the reservoir computer to be trained.The intensive information is encrypted in any way to obtain ciphertext and sent to the receiving end for decryption.Finally,the original information is decrypted by synchronizing the original system with the reservoir.Through some performance index parameter test scheme of confidentiality,the picture in the secrecy system to encrypt the image pixel distribution is uniform,speech security system of the correlation of the original voice and encrypted voice close to 0,and the encrypted voice than the original voice information entropy,when the SNR is higher than about 60 d B,the error can reach about 10^-2.3.After the design and software simulation of the security communication scheme,this paper uses FPGA to implement the security communication scheme of time delayed Lorenz system.RTL design and comprehensive analysis of each functional module,and then the whole system is connected to the module test to get the best state and give reasonable hardware memory resource allocation and low power design analysis.It is worth mentioning that in hardware implementation,TMDS communication protocol is used to further optimize its anti-noise,and its error is basically less than 10^-3when the SNR is about 40d B.