Anti-Synchronization Of Time-delay Neural Networks And Its Application In Secure Communication

Neural network is a complex system composed of many neurons connected to each other,and has excellent associative memory and information storage capabilities.Owing to the limitation of information processing speed and amplifier switching rate,time delay often occurs in the practical application of neural networks.The existence of time delay leads to the abundant chaotic dynamic behavior of neural networks.Therefore,the synchronization research of time-delay neural network has wide application prospects in chaotic secure communication field.Anti-synchronization describes a synchronization behavior with the same amplitude but opposite signs.This dissertation explores the anti-synchronization of time-delay neural network and its application in the field of secure communication.The main works as follows:(1)The issue of anti-synchronization for a class of neural networks with constant delay is investigated.The network allows unknown parameters and stochastic noise.An adaptive control strategy is proposed to ensure the mixed and anti-synchronization of the drive-response systems.In the case of no stochastic noise,it is theoretically proved that the proposed control strategy is less conservative than an existing result.A numerical example illustrates the applicability of the proposed control strategy.(2)The issue of anti-synchronization for a class of neural networks with time-varying delay is investigated.The network allows semi-Markov jump parameters and reaction diffusion.Using the Lyapunov functional method and stochastic analysis techniques,an anti-synchronization criterion is obtained.Based on the analysis result,a non-fragile fault-tolerant control strategy is proposed with the aid of some decoupling techniques.Two numerical examples illustrate the superiority of the analysis method and the applicability of the control strategy.(3)The issue of anti-synchronization for a class of neural networks with both discrete and distributed time-delay is investigated.By using an enhanced two-sided looped functional and linear matrix inequality techniques,an anti-synchronization criterion is obtained.Based on the analysis result,a non-fragile sampled-data control strategy is proposed with the aid of some decoupling techniques.A numerical example illustrates the effectiveness and superiority of the proposed control strategy.(4)The application of anti-synchronization theory of time-delay neural network in secure communication is discussed.With the help of chaotic masking method,a chaotic secure communication scheme is designed based on anti-synchronization of time-delay neural network with unknown parameters and stochastic noise.A numerical example illustrates the effectiveness of the communication scheme.