Fast and robust super-resolution DOA estimation for UAV swarms

Abstract

Unmanned aerial vehicle (UAV) swarms have shown great potentials in civilian and military applications. Consequently, there is a high demand for accurate UAV swarms detection. In response to resolve the closely spaced UAVs, we propose three super-resolution direction of arrival (DOA) estimation algorithms, i.e., frequency-selective reweighted atomic-norm minimization (FSRAM), fast Fourier transform (FFT)-reweighted atomic-norm minimization (FFT-RAM) and FFT-FSRAM. These proposed three algorithms take full account of advantages of prior knowledge, effective information extraction and gridless sparse technique, i.e., i) the use of prior knowledge can improve the accuracy of DOA estimation; ii) the effective information extraction can improve the signal-to-noise ratio to enhance the robustness and reduce the computational complexity; iii) the gridless sparse technique is insensitive to signal correlations. Complexity analysis and numerical simulations are performed to demonstrate that, compared with the Beamforming method, multiple signal classification (MUSIC) and reweighted atomic-norm minimization (RAM), the proposed three algorithms are insensitive to signal correlations and the FFT-RAM and FFT-FSRAM are more robust and faster for super-resolution DOA estimation of UAV swarms under the noisy environment. Additionally, the real experiment with C-band radar is also conducted to verify the effectiveness of the proposed super-resolution DOA estimation algorithms.