Coherent processing for randomized stepped-frequency radar in multi-target scenario

Randomized stepped-frequency (RSF) radar has the potential of realizing joint high-resolution range-Doppler estimation. However, estimating multiple targets' two-dimensional parameters at high efficiency based on RSF waveforms is still a difficult issue. In this study, a novel algorithm, which estimates multiple targets' velocities by using different frequencies pulses and then generates their high-range-resolution profiles (HRRPs), is proposed for RSF radar. In the velocities estimation, by setting a certain difference-frequency parameter, a set of signals with constant range-phase terms and changing velocity-phase terms is constructed and then used to generate a new generalized Doppler spectrum called the difference-frequency Doppler spectrum (DFDS). To suppress rising noise floors caused by sidelobes and cross-terms in the DFDS, DFDSs corresponding to different difference-frequency parameters are non-coherently integrated to obtain a so-called accumulated DFDS. From the accumulated DFDS, targets are detected and their velocities are estimated. In the HRRPs generation, a special dictionary matrix is structured by using the estimated velocities, and each target HRRP, without interference from other targets, is generated via the compressed sensing (CS) technique. The proposed algorithm avoids huge computation burden by separately performing Doppler and range process, and guarantees the performance in a multi-target scenario via the noise floors suppression and CS.