摘要
本文基于时间域前、后向线性预测方程,利用整体最小二乘方法来分析单次快拍空域谱估计问题。这种方法同时考虑数据矩阵与观察矢量噪声扰动的影响,所以,在低信噪比情况下,性能优于MUSIC等大多数高分辨空域谱估计方法。另外,由于它利用了数据矩阵奇异值分解技术,所以,对相干信号源也能实现良好分辨。本文摸拟实验结果证明了这一点。
This paper, on the basis of time domain forward and backward linear prediction equation, utillizes the total least square method to analyse single snapshot spatial domain spectrum estimation. This method considers both the effects of the noise perturbation of the data matrixs and the observed vectors so that its performances are better than most high resolution methods analogous to MUSIC's for lower power ratio of signal and noise. In addtion, thanks to utilizing data matrixs singular value decomposition (SVD) technique, this method is able to excellently resolve coherent signal sources, which are certificated by simulation results in this paper.
出处
《系统工程与电子技术》
EI
CSCD
1991年第11期35-40,共6页
Systems Engineering and Electronics
基金
高校博士点基金
关键词
线性预测
信息处理
频谱分析
Linear prediction, Total least square, Singular value decomposition, Noise subspace.
