摘要
多数DOA估计算法(如MUSIC算法和ESPRIT算法)都需要对复数协方差矩阵进行特征值分解,存在计算量大的问题,本文提出一种DOA估计的实数特征值分解算法,将协方差矩阵实部和虚部分离,利用Toeplitz矩阵和Hermitian矩阵的性质,构造新的实数矩阵,将复数协方差矩阵的特征值分解问题转化为对实数矩阵的特征值分解,减小了运算量,仿真结果表明,所提实数特征值分解算法与复数特征值分解算法精度相当。
To most of DOA estimation algorithms (such as MUSIC and ESPRIT), eigenvalue decomposition to complex covariance matrix is needed, which would lead to high computation load. A real eigenvalue decomposition algorithm for DOA estimation is proposed by dividing the covariance matrix to real part and imaginary part and then composing a new real matrix by utilizing characteristics of Toeplitz matrix and Hermitian matrix; therefore problem of complex matrix eigenvalue decomposition is converted to real matrix eigenvalue decomposition, computation load got reduced. Simulation results show that precision of the real eigenvalue decomposition algorithm is equivalent to that of the complex eigenvalue decomposition algorithm.
出处
《火控雷达技术》
2013年第1期6-8,共3页
Fire Control Radar Technology