摘要
目前常用的零陷展宽算法都可以归纳到协方差矩阵锥化(CMT)的范畴,按常规计算方式,CMT法的运算量为O(M3)。文章首先通过对CMT法中的锥化矩阵进行特征值分解,提出了一种递推实现的零陷展宽算法,将运算量降为O(JM2)(J为锥化矩阵的秩);将该递推方法与对角加载算法结合,大大提高了算法的鲁棒性;最后针对几种常用的锥化矩阵进行了分析,确定了Mailloux算法中虚拟干扰源的选取原则,对MZ算法的锥化矩阵进行了降秩近似,进一步降低了运算量。计算机仿真分析表明,递推CMT算法在与原算法性能相当的情况下运算速度大大提高,与对角加载算法结合后,可以以较低的运算量实现较好的稳健性。
Most of common used null broadening algorithms belong to the covariance matrix tapered（CMT） approach.Calculated in the conventional way,the CMT approach has a computational complexity of O（M3）.Through the eigen-decomposition of the tapered matrix,a recursive null broadening approach is proposed to reduce the computational complexity to O（JM2）,where J is the rank of the tapered matrix.Then the proposed approach is combined with the diagonal-loading algorithm to improve robustness of the method.Finally,some common used taper matrixes are analyzed.The number of fictitious sources in Mailloux algorithm is chosen,and low-rank approximation of the taper matrix in MZ approach is studied to reduce the computational cost further.Computer simulations demonstrate that the recursive CMT algorithm provides similar performance and much faster calculating speed compared with the original one,and when combining it with the diagonal-loading algorithm,more robustness can be achieved with lower computational cost.
出处
《宇航学报》
EI
CAS
CSCD
北大核心
2011年第4期911-916,共6页
Journal of Astronautics
关键词
零陷展宽
协方差矩阵锥化
低秩近似
对角加载
Null broadening
Covariance matrix tapered
Low-rank approximation
Diagonal loading
