4D Automotive Radar Exploiting Sparse Array Optimization and Compressive Sensing

Automotive radar systems require high resolution in four dimensions: range, Doppler, elevation and azimuth. The angular resolution of an automotive radar are determined by the antenna array aperture. Two-dimensional (2D) antenna arrays are necessary for angle estimation in both elevation and azimuth for automotive radar systems to enable drive-over and drive-under functions. Sparse arrays offer advantages such as reduced mutual coupling and lower hardware costs. The sparse array configurations like coprime and nested arrays, which require a large number of array snapshots, may not be suitable for highly dynamic automotive scenarios. Multiple-input and multiple-output (MIMO) radars are widely adopted in automotive radar applications due to their ability to synthesize a large virtual array. In this paper, our objective is to optimize the geometry of 2D MIMO sparse arrays while considering fabrication constraints, i.e., minimal spacing between antennas. This optimization aims to minimize the peak sidelobe level and half-power beam width (HPBW), thus enabling high-resolution imaging with single snapshot. Angle finding is accomplished through a 2D compressive sensing approach. Through extensive numerical experiments, we demonstrate that the proposed workflow offers a practical solution for 2D MIMO sparse arrays, ensuring high angular resolution in automotive radar systems.