摘要
针对直接序列扩频通信信号伪噪声(PN)码的干扰侦察问题,在已知信号伪码参数的前提下对直扩信号进行特征分析,用基于变步长的PCA神经网络方法来实现伪码序列的盲估计。该方法利用基于Hebb学习规则的无监督多主分量神经网络,结合自适应变步长学习算法,在估计在线特征值的基础上来控制步长的变化,以使神经网络最终达到较好的稳态收敛。理论分析和仿真结果表明,本方法能在较低信噪比的情况下对较长伪码进行准确的估计。
In order to resolve the interference caused by pseudo-noise (PN) sequence from low SNR DS/SS signal in communication reconnaissance, the characteristics of DS signal are analyzed on the basis of the provided parameters of PN. The PCA neural network based on the variable step-size is used to esti- mate PN sequence. The unsupervised multiple PCNN based on Hebb rule and self-adaptive variable stepsize learning algorithm is applied in this method to control the variation of step size in on the basis of estimating on-line characteristic value to make a better steady convergence of neural network. Theoretical analysis and computer simulation results show that the method can work well on lower SNR DS/SS for longer PN sequence.
出处
《现代防御技术》
北大核心
2010年第6期85-91,共7页
Modern Defence Technology
基金
国家自然科学基金项目(61071196)
国家自然科学基金——中物院NSAF联合基金项目(10776040);教育部新世纪优秀人才支持计划(NCET-10-0927);信号与信息处理重庆市市级重点实验室建设项目(CSTC,2009CA2003);重庆市自然科学基金项目(CSTC,2009BB2287)的资助.
关键词
通信侦察
无监督神经网络
主分量分析
自适应变步长学习算法
直接序列扩频信号
伪码序列
signal reconnaissance
unsupervised neural network
principal components analysis (PCA)
adaptive variable step-size learning algorithm
direct sequence spread spectrum signal
pseudo noise(PN) sequence