基于神经网络混沌扩频序列的研究  

Study of Chaotic Spread-Spectrum Sequences Based on Neural Networks

作　　者：刘联会[1] 石军[1] 李白萍[1] 王建新[1] 

机构地区：[1]西安科技大学通信与信息工程学院,西安710054

出　　处：《电子与信息学报》2004年第1期137-141,共5页Journal of Electronics & Information Technology

摘　　要：应用神经网络的强大学习能力和具有全局最优的BP改进算法,提出了通过训练学习建立的具 有混沌性态的优化神经网络模型;利用网络权值调整的灵活性来产生混沌序列,该模型序列更换容易并且数 量巨大。实验与分析结果表明该模型产生的混沌扩频序列具有良好的相关特性、平衡特性以及理想的线性复 杂度,是最优加密密钥及扩频码的优选码型之一。The chaos generation neural network based on the excellent learning ability and synaptic weight database are built to generate many chaotic spread-spectrum sequences trained by the modified back-propagation algorithm with various discrete chaotic time series. The chaotic sequences are very easily generated by changing weights of neural network model, and their number is large. The computer simulation results show that, the output chaotic sequences have good correlation property, balance property and linear complexity, therefore they are good candidates for the optimal encrypting code and the spread spectrum code.

关 键 词：神经网络 混沌扩频序列 相关函数 线性复杂度 网络权值 系统模型 

分 类 号：TN914.42[电子电信—通信与信息系统]