用于高分辨率测向的噪声子空间抽取算法

高分辨率测向是有广阔应用前景的技术，当前的一个主要问题是其计算量巨大，噪声子空间抽取算法（ＮＯＳＥ）首先利用最小二乘法求得噪声子空间中的一组基或这组基的一个子集，然后形成空间谱并估计可能的波达方法。最后，再利用一次最小二乘法检测信号数目并确定真实波达方向。在存在相干源时，ＮＯＳＥ可利用空间平滑技术去相干后进行检测和估计。与标准的高分辨率测向算法（多信号分类算法ＭＵＳＩＣ）相比，ＮＯＳＥ不但运算量大为减少，而且分辨率也有提高，文中的信号数目的检测方法基于对波达方向的估计，具有简单准确的特点，可以推广到多种算法中去。

High resolution direction finding algorithm using noise subspace extraction

High resolution direction finding is a promising technique. One major problem with it is its large amount of computation. Noise subspace extraction(NOSE) first finds out a basis of the noise subspace or one of its subsets by a least squares procedure.Then spatial spectrum is constructed and possible directions of arrival(DOA's) are estimated. At last, another least squares procedure is used to determine the number of sources and real DOA's. When coherent sources are present,spatial smoothing technique can be employed to decorrelate coherence before NOSE works. Theory and simulations have both indicated that, compared with multi signal classification(MUSIC), NOSE needs less computation for higher resolution. The detection algorithm proposed in this paper is based on DOA estimation and can be applied to other methods as well.