MUSIC算法是一种基于矩阵特征空间分解的方法。从几何角度讲，信号处理的观测空间可以分解为信号子空间和噪声子空间，显然这两个空间是正交的。信号子空间由阵列接收到的数据协方差矩阵中与信号对应的特征向量组成，噪声子空间则由协方差矩阵中所有最小特征值（噪声方差）对应的特征向量组成。MUSIC算法就是利用这两个互补空间之间的正交特性来估计空间信号的方位。噪声子空间的所有向量被用来构造谱，所有空间方位谱中的峰值位置对应信号的来波方位。MUSIC算法大大提高了测向分辨率，同时适应于任意形状的天线阵列，但是原型MUSIC算法要求来波信号是不相干的。-MUSIC algorithm is a matrix decomposition method based on feature space. From the geometric point of view, the observed spatial signal processing can be decomposed into signal subspace and noise subspace, it is clear that the two spaces are orthogonal. The signal received by the array to the subspace of the covariance matrix of the signal component corresponding eigenvectors, the noise subspace of the covariance matrix of all by the smallest eigenvalue (noise variance) eigenvectors corresponding to the composition. MUSIC algorithm is the use of orthogonal properties between these two complementary space to estimate the spatial orientation of the signal. Noise subspace of all vectors are used to construct the spectrum, all the spatial orientation of the spectrum corresponding to the peak position of the signal wave direction. MUSIC algorithm greatly improves the resolution measurements, while the antenna array adapted to any shape, but the prototype MUSIC algorithm requires to wave sign