Interpolation-Based Direction-of-Arrival Estimation for Coprime Arrays via Covariance Matrix Fitting

• Main Authors: Zhen Chen, Chongyi Fan, Xiaotao Huang
• Format: Article
• Language: English
• Published: IEEE 2020-01-01
• Series: IEEE Access
• Subjects: Coprime arrays; DOA estimation; virtual array interpolation; covariance matrix fitting; hyperparameter-free
• Online Access: https://ieeexplore.ieee.org/document/9146166/

Coprime arrays can highly increase degree-of-freedom (DOF) by exploiting the equivalent virtual signal. However, since the corresponding virtual array constructed by the coprime array is always a non-uniform linear array (non-ULA), most existing direction-of-arrival (DOA) estimation algorithms fail to utilize all received information and result in performance degradation. To address this issue, we propose a novel interpolation approach for coprime arrays to convert the virtual array into a ULA with which all received information can be efficiently utilized. In this paper, we consider a weighted covariance matrix fitting criterion to formulate a semi-definite programming (SDP) problem with respect to the interpolated virtual signal. After that, we can reconstruct a Hermitian Toeplitz covariance matrix corresponding to the interpolated ULA in a gridless manner, and the number of detectable targets is ulteriorly increased with the reconstructed covariance matrix. The proposed approach is hyperparameter-free so that the tedious process of selecting regularization parameters is avoided. Numerical experiments validate the superiority of the proposed interpolation-based DOA estimation algorithm in terms of DOF characteristic, resolution ability and estimation accuracy compared with several existing techniques.