摘要
本文介绍信号相关矩阵的奇异值分解(SVD)与特征根结构分解(ED)之间的关系及用信号特征矢量表示平稳随机过程和信号子空间的方法。利用信号子空间对信号进行信息提取,可减少噪声对估计参数精度的影响。文中论述了提高前向预测定向精度的方法。SVD能把信号空间与噪声空间分开,以提出互相关矩阵中信号信息。
This paper introduces the relationship between singular val-ue decomposition and eigenstructure of the correlation matrix and themethods of describing stationary stochastic process and signal subspace us-ing signal eigenvectors. Based on signal subspace, we can get noise free pa-rameter estimation. The method of improving the precision of forwardprediction direction finding is given. SVD can decompose received signalinto signal subspace and noise subspace, abatracting the information in-cluded in cross-correlation matrix.
关键词
信号处理
ED
SVD
相关矩阵
correlation
matrices
characteristic value
structure
information processing
singular solution
