Spatiotemporal Virtual Transformation Based Super-Resolution Angle Estimation Under 2-D Nonuniform Sparse MIMO Array

In traditional multiple input multiple output (MIMO) mmWave radar, missing estimation is a common problem in coherent source scenarios due to low angle resolution in elevation and azimuth dimensions. To address this, we propose a novel two-dimensional (2-D) super-resolution angle estimation method based on compressed sensing. Our approach involves a spatial virtual transformation of the nonuniform sparse array of
$M$
emitters and
$N$
receivers, generating a spatiotemporal virtual array (MIMO-II-VA) with two times aperture of the spatial virtual array using multiple snapshots based on nested array rules. The received echo signal of MIMO-II-VA is then recovered using compressed sensing technology, enabling accurate and stable angle estimation for multiple targets in the coherent source scenario. We first introduce the optimal sub-pattern assignment (OSPA) metric criterion to the angle estimation domain, which effectively measures under-estimated or over-estimated target errors compared to the conventional root mean square error criterion (RMSE). Our simulation results demonstrate that our proposed method achieves an angle resolution approaching azimuth 0.5° and elevation 1° using ARS548 RDI mmWave radar with 12 emitters and 16 receivers under actual physical antenna array when the signal to noise ratio (SNR) is higher than 5dB. Furthermore, with more than 16 snapshots, our approach achieves a 100% success rate in 5 coherent sources.