Efficient 2-D MUSIC algorithm for super-resolution moving target tracking based on an FMCW radar

Frequency modulated continuous wave(FMCW) radar is an advantageous sensor scheme for target estimation and environmental perception. However, existing algorithms based on discrete Fourier transform(DFT), multiple signal classification(MUSIC) and compressed sensing, etc., cannot achieve both low complexity and high resolution simultaneously. This paper proposes an efficient 2-D MUSIC algorithm for super-resolution target estimation/tracking based on FMCW radar. Firstly, we enhance the efficiency of 2-D MUSIC azimuth-range spectrum estimation by incorporating 2-D DFT and multi-level resolution searching strategy. Secondly, we apply the gradient descent method to tightly integrate the spatial continuity of object motion into spectrum estimation when processing multi-epoch radar data,which improves the efficiency of continuous target tracking. These two approaches have improved the algorithm efficiency by nearly 2-4 orders of magnitude without losing accuracy and resolution. Simulation experiments are conducted to validate the effectiveness of the algorithm in both single-epoch estimation and multi-epoch tracking scenarios.