A Generalized 2-D DOA Estimation Method Based on Low-Rank Matrix Reconstruction

Low-Complexity Gridless 2D Harmonic Retrieval via Decoupled-ANM Covariance Reconstruction

“…This section presents numerical results to evaluate the performance and the complexity of the RR-D-ANM solutions, while the LRSCR [13], the conventional MUSIC algorithm 2 In this paper, we consider sufficient separated frequencies to match the minimum frequenc separation requirement [18]. Moreover, the reweighted ANM technique inspired by [24] can be utilized to enhance sparsity and resolution when unknown frequencies are spaced closely.…”

“…Because of this decoupling, the PSD matrix in (20d) is of size 2(2M − 1) × 2(2N − 1). However, the extra PSD constraint in (20e) is of size M N × M N , which raises the complexity order to be comparable to V-ANM [14] and LRSCR [13]. Fortunately, when L is reasonable large, this covariance-based constraint is mostly satisfied, and hence can be removed leading to a truncated version termed RR-D-ANM-w/o3, to balance the computational cost and performance.…”

“…Recently, to overcome this problem, two super-resolution gridless CS approaches are proposed for the MMV case based on low-rank structured covariance reconstruction (LRSCR) [13] and vectorization-based atomic norm minimization (V-ANM) [14], [15], respectively. Unfortunately, the computational complexities of both techniques scale exponentially with the dimensionality, and will become unacceptable for This work was supported in part by the US National Science Foundation grants #1527396 and #1547364, and the National Science Foundation of China grants #61871218, #61801211 and #61471191.…”

Two-Dimensional Separable Gridless Direction-of-Arrival Estimation Based on Finite Rate of Innovation

“…In this simulation, the 2D-SGFRI algorithm is compared with three other two-dimensional DOA estimation algorithms, including 2D-MUSIC algorithm for planar array with arbitrary geometry [3], decoupled atomic norm minimization algorithm for uniform rectangular array (2D-DANM) [18] and 2D Toeplitz matrix complete algorithm for sparse planar array (2D-TMC) [26]. Among them, 2D-MUSIC is suitable for the bi-orthogonal sparse linear array used in this paper, but 2D-DANM algorithm is only suitable for uniform rectangular array and 2D-TMC algorithm is suitable for uniform rectangular array and sparse rectangular array.…”

“…The computational complexity of 2D-DANM algorithm and 2D-TMC algorithm have been given respectively in references [18] and [26]. Similarly, the complexity of the proposed 2D-SGFRI algorithm is not difficult to give based on the analysis of SDP process complexity in the above two references.…”