Joint sparse recovery for direction of arrival based on the generalized MUSIC criterion

The core of compressed sensing (CS) is to recover a set of sparse signals from incomplete measurements. In this paper, to solve the coherent direction-of-arrival (DOA) estimation problem, it is formulated as a joint-sparse recovery problem by exploiting the signal sparsity in the spatial domain. When the sparsity level of the signals satisfies certain conditions, the signals are recoverable even in the case of coherent sources. By further considering the structure of the sparse signals, a more general condition for the recoverability is theoretically derived. Then a method is proposed based on the CS theory and the generalized MUSIC criterion. This criterion improves on MUSIC so that the support is reliably recovered under the case of rank-defect. In our method, we modify the criterion to make it applicable for continuous dictionaries. When the partial DOAs has been estimated, the modified criterion can estimate the remaining DOAs with higher accuracy. By alternately using the modified criterion, the proposed method achieves super-resolution. Simulation results demonstrate that the proposed method has satisfactory performance even in the case of coherent sources. (C)& nbsp;2021 Elsevier Inc. All rights reserved.

Automated summary

The paper formulates the coherent DOA estimation problem as a joint-sparse recovery problem by exploiting signal sparsity in the spatial domain. A method based on CS theory and the generalized MUSIC criterion is proposed, which improves on MUSIC and achieves reliable support recovery under rank-defect conditions. The proposed method achieves super-resolution by modifying the criterion and is shown to have satisfactory performance even in the case of coherent sources.