A weighted OMP algorithm for Doppler superresolution

Radar target speed resolution is related to the accumulation time. In order to obtain higher frequency resolution, we need longer coherent accumulation time. However, long accumulation will bring many problems. Since the targets are sparse in the Doppler domain, Doppler super-resolution can be implemented in shorter accumulation time. Doppler frequency of different targets may be correlative, as well as data obtained by single sample, which can be solved effectively by using compressed sensing theory. For the time-frequency transform dictionary generated from frequency resolution, a weighted OMP (Orthogonal matching pursuit) algorithm is proposed to reconstruct the sensing dictionary. This algorithm is superior to OMP etc. greedy algorithm. Especially in the case of the target Doppler frequency close, it shows the high resolution accuracy.