CAJ | 학술논문

针对未知非平稳噪声背景下相干信源的波达方向估计问题,提出了一种T差分平滑快速(TDSF)算法,算法根据非平稳噪声对角占优的特性,通过构造一个T矩阵,对协方差矩阵进行T变差分平滑运算,可同时分辨相干和非相干信源.算法采用幂迭代法直接获得噪声子空间,无需特征值分解,运算量较小,可有效地消除非平稳复噪声的影响,且不存在伪峰问题,适用于广泛的未知非平稳噪声和低信噪比环境.计算机仿真实验证实了算法的有效性.
침대미지비평은조성배경하상간신원적파체방향고계문제,제출료일충T차분평활쾌속(TDSF)산법,산법근거비평은조성대각점우적특성,통과구조일개T구진,대협방차구진진행T변차분평활운산,가동시분변상간화비상간신원.산법채용멱질대법직접획득조성자공간,무수특정치분해,운산량교소,가유효지소제비평은복조성적영향,차불존재위봉문제,괄용우엄범적미지비평은조성화저신조비배경.계산궤방진실험증실료산법적유효성.
For the need of coherent signal source DOA estimation in the presence of unknown non-stationary noise, a T difference smoothing fast (TDSF) Algorithm is proposed. According to the diagonal-dominated characteristic of non-stationary noise, a T ma-trix was constructed in this algorithm, which does the difference smoothing operation on the data covariance matrix. This algorithm can both resolve the correlated and uncorrelated signal sources. The algorithm adopts power iteration method to get the noise sub-space, doesnt need eigen-value decomposition, has small computation burden, can avoid the non-stationary complex noise, and doesnt have the pseudo-peak which appears in the algorithm suggested in paper. This algorithm can be applied to broad unknown non-stationary noise and low SNR environment. The simulation experiment results show that this algorithm is good and effective.