基于细胞神经网络保密通信研究

【作者】 王曦；

【导师】 张小红；

【作者基本信息】 江西理工大学 ， 计算机应用技术， 2008， 硕士

【摘要】 细胞神经网络(Cellular Neural Networks,简称CNN)动力学是神经网络与动力学系统交叉结合的一门新型学科,它的理论和应用研究成了当前的研究热点。CNN是一个高速非线性动力学系统,不同的模板参数选择表现出复杂的动力学行为,在满足一定条件下,系统会出现分岔和混沌现象。由于混沌系统具有对初始条件和系统模板参数的高度敏感性,是伪随机系列的天然候选者,这为CNN在保密通信中的应用提供了可靠的理论基础。本文在研究CNN的稳定性前提下,分析系统的动力学行为,设计出合适的CNN混沌与超混沌模板参数,并将其应用于保密通信中,主要研究成果体现在:(1)运用Lipschittz连续条件、不动点定理、Lyapunov函数稳定理及不等式方法,分析了CNN的全局渐近稳定性,设计推导的稳定性判断标准为不等式表达式,简单实用,对CNN邻域内细胞间相互作用为线性的稳定性分析具有通用性。(2)根据Lyapunov第一近似理论分析了CNN邻域内细胞间相互作用为非线性的稳定性情况,所采用的判断方法对全互连CNN模型的构造与稳定性分析具有通用性。(3)结合CNN稳定性判断方法构造了CNN混沌与高维超混沌系统,应用于音频保密通信中,实现了收发端的同步,使原始信号在接收端得到了较好的恢复。(4)设计了基于五维CNN全维和降维状态观测器的超混沌同步系统,通过仿真与数值测试得出:如果CNN超混沌系统的状态变量差值保持在一定范围内,可以较精确实现保密通信。更多还原

【Abstract】 Cellular Neural Networks (CNN) dynamics is a new kind of interdisciplinary sciences combined with neural networks and dynamic system, the theory and application of CNN have become a new focus recently. It is known that the CNN is a nonlinear dynamic system with high speed, different parameters selection for system template will lead to different dynamic behaviors such as bifurcation phenomenon and chaotic character, which provide reliable theory basis for secure communication because chaotic systems have a high sensitivity for changes of system initial conditions and template parameters. In this thesis, we gave attentions to the researches of CNN’s stability and nonlinear dynamic behaviors. At the same time, we designed some good template parameters for CNN application in secure communication. Main research works are shown as follows.Firstly, a novel criterion for CNN’s global asymptotic stability analysis is derived based on Lipschitz condition, fixed point theory and Lyapunov functional stability analysis. It is expressed by inequality equations and is universal for CNN’s stability judgment when the interaction between cells is linear. Some numerical examples are given to show the effectiveness.Secondly, we have applied the first method of Lyapunov to analyze CNN’s stability when the interaction between cells is nonlinear. Another novel criterion for the global asymptotic stability of cellular neural networks has been adopted with eigen-values analysis for Lyapunov’s linear-matrix, it is feasible by experiment test.Thirdly, a 5-order CNN hyper-chaotic system is designed according to the criterions put forward above, and it is applied to secure communication.Fourthly, we have constructed CNN chaotic and hyper-chaotic secure communication system based on the theories for chaos synchronization and state observer, a new technical requirement has been proposed: if the state values of CNN can be arbitrarily controlled within a certain control-range, the secure communication can be realized precisely.更多还原