Fast near field source localization algorithm based on Cumulant

In this paper, a new fast parameter estimation method based on cumulant is proposed for near-field sources. Firstly, a cumulant matrix is constructed and its singular value decomposition (SVD) is performed. Secondly, the closed form solutions of direction-of-arrival (DOA) and distance are obtained through the Root-MUSIC method from its left and right singular vectors, respectively. By exploiting the high-order cumulant matrix, we reduce the array aperture loss and improves the maximum number of sources that can be distinguished. Compared with other methods based on the high-order cumulants, our proposed method constructs only one cumulant matrix and performs singular value decomposition (SVD) only once, obtaining the closed form solution which completely avoids the peak search. As a result, the computation burden is greatly reduced, and the resolution probability and accuracy of the estimation are improved. Furthermore, our method can be applied in the mixed field sources without much additional computational burden. The simulation results show the better resolution and accuracy of our method.