Coherent MIMO radar with sparse recovery: Joint vs. separate range and azimuth estimation

An optimal joint estimation of the ranges, azimuth angles, and reflection coefficients for multiple targets in coherent multiple-input multiple-output radar is computationally intractable. We provide a numerical comparison of different estimators that all use sparse recovery for the discretized range parameter to reduce algorithmic complexity. The methods differ in whether estimation of angles and ranges is performed jointly or separately and in whether the azimuth angle is discretized, too. Our results show that joint estimators perform better than separate estimators and that introducing an angular grid adds robustness for low SNR values, but decreases accuracy for high SNR values.