Direction Finding for Non-circular Sources Based on Weighted Unitary Nuclear Norm Minimization

In this paper, we propose a weighted unitary nuclear norm minimization approach for direction-of-arrival (DOA) estimation in the strictly non-circular sources case. The proposed method firstly utilizes the non-circularity characteristics of the source to extend the data model for doubling the array aperture. Then the complex-valued extend date model is turned into a real-valued data model by unitary transformation, which can suppress the noise by the real-valued structure. And then a unitary nuclear norm minimization framework is formulated via a block sparse model which has no influence of the unknown noncircularity phase. In addition, the noncircular Capon (NC-Capon) spectrum is used to design a weight matrix for unitary nuclear norm minimization algorithm. Finally, the DOA estimation is accomplished by searching the non-zero blocks of recovered matrix. Owing to utilizing the noncircularity of source, the real-valued data model and the twofold sparsity structure, the proposed method can provide higher spatial angular resolution and better angle estimation performance. Simulation results are used to verity the effectiveness and advantages of the proposed method.