阵列存在未知有界扰动的鲁棒加权子空间拟合

由于阵列响应难于精确标定,因此,阵列信号处理算法的性能不仅受测量噪声的影响同时还受阵列响应扰动的影响。信号回波方向估计DOA受该系统扰动的影响很大。以往的研究均将模型的扰动假设为随机误差,且往往假设扰动是相关矩阵已知的高斯分布。然而,实际系统误差很难满足这一假设。模型误差的更合理的假设是未知有界(UBB)的假设,这里,研究模型扰动是未知有界时,信号的DOA估计问题,提出了鲁棒的信号子空间拟合原理,并用二次锥规划来求解。数值仿真表明该算法是有效的。

Robust Signal Subspace Fitting for Unknown But Bounded Array Perturbation

For the array cannot be perfectly calibrated, the limiting factor in the performance of array signal processing algorithms is most often not measurement noise but rather perturbations in the array response model. Depending on the size of such errors, estimates of the directions of arrival ( DOA' s) may be significantly degraded. A number of techniques have been considered for improving the robustness of array processing algorithm. These techniques have assumed the model errors are stochastic, and can be drawn from some known a priori distribution. But in certain case, they are not, the reasonable assume is the perturbations is unknown but bounded ( UBB) . An robust signal subspace fitting algorithm for estimation of the directions of arrival is proposed in the case of general array errors. Numerical experiments verifies the effectiveness of the algorithm.