Smoothing Frequency Estimation Method for Automotive Radar

We present a smoothing frequency method to jointly estimate range, azimuth angle and radial velocity of multiple targets for automotive applications with MIMO-FMCW radar. In order to obtain super-resolution estimations of targets in dense environment, we construct a shift invariant structure to smooth data in the time-space domain and exploit subspace-based algorithms. The number of detectable targets is not limited by the number of MIMO virtual antennas and the method only requires two chirps' data to achieve the functionality. Simulated and experimental results are shown by applying the proposed method to a 77-GHz MIMO-FMCW radar.