基于非对角元素方差最小感知字典的DOA估计

Direction-of-arrival estimation based on the minimum variance of non-diagonal elements for sensing dictionary

针对传统正交匹配追踪(OMP)算法性能受到冗余字典中原子间相关性影响的问题,提出了一种基于格拉姆型矩阵非对角元素方差最小的感知字典设计准则,利用感知字典对OMP算法进行修正,提高了重构性能。针对存在噪声情况的DOA估计,将感知字典与多测量向量的OMP算法相结合,克服了原子间强相关性造成的不良影响,提高了抗噪性能,在极低信噪比下能准确地进行DOA估计。

As the correlation of adjacent atoms in a redundant dictionary has affected the performance of ordinary or- thogonal matching pursuit （OMP） algorithm,a design criteria based on the minimization variance of Gramm-type matrix non-diagonal elements is proposed for sensing dictionary, a OMP algorithm is modified by using sensing dictionary and the reconstruction performance of the modified OMP algorithm is improved. For the noise existed case, sensing dictionary is combined with OMP algorithm of multiple measurement vector, the bad influence caused by the strong correlation of adiacent atoms is overcome, the proposed method can improve anti-noise ability so that it can make accurate DOA estimation in very low signal-to-noise ratio case.