相干信号源DOA估计的一种改进SVD算法

摘要

多重信号(MUSIC)算法是波达方向(DOA)估计中的一种标志性算法,在理想条件下具有良好的性能,但是当信号源相干时,算法的性能就会变得很差.为了使其在低信噪比、小角度条件下对相干信号源有着更好的分辨能力和稳定性,通过时解相干重要算法--矢量奇异值(SVD)算法的研究,并针对SVD算法在低信噪比、小角度条件下分辨能力的不足,提出了一种改进的矢量奇异值算法(NSVD),即利用信号协方差矩阵的最大特征矢量,按一定规则构造出新矩阵,然后对矩阵进行修正,再利用奇异值分解算法估计出信号相关信息.最后通过大量的计算机仿真证明了算法的良好性能.%MUSIC algorithm is one of the significant algorithms for spatial spectrum estimation.It works well with fine performance in the ideal condition, but if signal source is coherent, the result of algorithm would be bad.Basing on the research of SVD algorithm, a new algorithm named NSVD algorithm (that is using the maximum feature vector of the signal covariance matrix to construct a new matrix according to certain rules, then correct the matrix, use singular value decomposition algorithm to estimate the signal information) is proposed, which could work better when under the low-SNR and low-angle condition.At last, a good performance ia verified through lots of computer emulation.