基于SVD的二维子空间拟合DOA估计

改进传统子空间拟合波达方向(DOA)估计方法，以快拍数据矩阵的奇异值分解代替接收数据协方差矩阵的特征值分解，用奇异值和奇异值矢量进行信源数估计，避免协方差矩阵估计，减少运算量和矩阵估计误差。根据已有子空间拟合的一维修正变化投影(MVP)算法原理，推导出二维MVP算法实现步骤，对基于均匀圆阵的接收信号进行二维DOA估计。

DOA Estimation with 2D Subspace Fitting Based on SVD

This paper improves the traditional Direction Of Arrival(DOA) estimation method with subspace fitting. The Singular Value Decomposition(SVD) of the data matrix is employed to replace the Eigenvalue Decomposition(ED) of the covariance matrix which is estimated by the received data snapshots. The singular values and singular value vectors accomplish the estimation for the number of the impinging signals, and the covariance matrix estimation is avoided to mitigate the computation load and estimation error. The 2D Modified Varying Projection(MVP) algorithm is illustrated according to the principle of the 1D MVP. The estimation of the 2D DOA is adopted on the received signal based on the uniform circle array.