Abstract: Traditional improved MUSIC algorithm decomposes the signal covariance matrix to make the signals subspace be orthogonal to the noise subspace, so decreases the effect of the noise. But when the signals interval are very small, traditional improved MUSIC algorithm has been unable to distinguish the signals as the SNR decreases. To this problem, a modified MUSIC is proposed which makes full use of signals subspace and its eigenvalue, on basis of the signals subspace obtained be orthogonal to the noise subspace, construct spectrum search function and then estimate the signals. Through computer simulation, it has been approved that when the signals interval are very small under the condition of the lower SNR and exist the correlated signals, modified MUSIC can estimate the signals accurately that traditional improved MUSIC algorithm can’t estimate.

Key words: Direction of Arrival（DOA） estimation, Multiple Signal Classification（MUSIC） algorithm, improved MUSIC algorithm, modified MUSIC algorithm

摘要： 传统改进 MUSIC 算法通过对接收信号协方差矩阵作预处理，使信号协方差矩阵分解得到信号子空间与噪声子空间正交，从而降低噪声的影响。但当信号间隔很小时，随着信噪比的降低，传统改进MUSIC算法已无法分辨出信号。基于此问题提出的修正MUSIC算法在使信号子空间与噪声子空间正交的基础上，充分利用了噪声子空间及其特征值对噪声子空间的修正，进而构造谱峰搜索函数估计出信号。通过仿真实验，证实了在信噪比很低的情况下，信号间隔很小且存在相关信号时，修正MUSIC算法能准确地估计出传统改进MUSIC算法不能估计的信号。

关键词: 波达方向估计, MUSIC 算法, 改进MUSIC算法, 修正MUSIC算法