DOA Estimations With Limited Snapshots Based on Improved Rank-One Correlation Model in Unknown Nonuniform Noise

• Published in: IEEE transactions on vehicular technology 2021-10, Vol.70 (10), p.10308-10319
• Main Authors: Fang, Yunfei, Zhu, Shengqi, Zeng, Cao, Gao, Yongchan, Li, Shidong
• Format: Article
• Language: English
• Subjects: Algorithms; alternating grid optimization (AGO); Arrays; Computational modeling; Correlation; Covariance matrices; Covariance matrix; Direction of arrival; Direction of arrival (DOA); Direction-of-arrival estimation; Estimation; Mathematical analysis; matrix completion (MC); Noise; Noise reduction; nonuniform noise; Optimization; Robustness (mathematics); sparse reconstruction (SR)
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2021.3105673

It is well known that classic direction-of-arrival (DOA) estimation techniques yield unsatisfactory performance with limited snapshots in unknown nonuniform noise. In this article, a sparse reconstruction (SR) DOA estimation method combining vectorized and reduced signal covariance matrix (SCM) is proposed. The new approach is based on an improved rank-one correlation model for denoising. With an extended virtual aperture array, this technique is able to provide a high-resolution DOA estimation with limited snapshots and is robust to nonuniform noise. In addition, two effective algorithms are also derived to obtain noise-free SCMs. One is to obtain diagonal entries of the noise-free SCM from the off-diagonal entries based on the fact that diagonal entries of SCM across the array are identical. The other is based on matrix completion (MC) theory which uses the low rank property of the SCM. Numerical results demonstrate that the new approaches yield superior performance than existing techniques in DOA estimations for multiple and compactly placed sources.