Super-resolution method for coherent DOA estimation of multiple wideband sources

Abstract

We focus on coherent Direction Of Arrival (DOA) estimation of wideband sources based on spatial sparsity. By development of super-resolution algorithms, DOA estimation of narrowband sources has already been addressed based on gridless sparse algorithms and atomic norm minimization. In this paper, a super-resolution based method is proposed for coherent DOA estimation of multiple wideband sources. To do so, we first develop a new focusing method to map the subband with the largest center frequency to the other ones with no requirement for any initial estimates. Then, using the latter focusing method, we introduce an atomic norm problem by defining a new set of atoms and exploiting the signal joint sparsity of different frequency subbands in a continuous spatial domain. This problem is then cast as a semidefinite program, which leads to implementing a new coherent DOA estimation method with higher resolution and more robustness to noise. Our method only needs one single snapshot for each frequency subband, although it can also be used for multiple snapshots. Numerical simulations show the outperformance of the proposed method compared to the conventional ones.