MIMO radar is a promising concept and many theoretical studies have demonstrated its interest. The colored transmission makes it possible to extract each waveform and form the transmission beam by digital processing on receive. One strong limitation of this technique is that waveforms cannot be perfectly separated in practice, because of the intrinsic lack of orthogonality of the waveforms family or hardware defaults. In case of conventional radar processing, this badly impacts the performance on target detection and localization. In this paper, we explain in which way the usual approach of adapted filter is not adapted for MIMO radar signals. Then, we introduce different approaches to estimate the target amplitude based on adaptive processing, and we deal with their performance and limitations when applied in the context of non perfectly orthogonal waveforms. We especially focus on Orthogonal Matching Pursuit (OMP) procedure which aims at detecting and cleaning each target, successively. We point out the problematic effects due to the granularity of the target amplitude grid and neighbor targets influence. To solve the problem, we propose, at each step of the OMP procedure, not only to clean the estimated position of the target, but also the close neighbor positions. We demonstrate that this “extended rejection” increases the robustness of OMP on realistic MIMO waveforms. Eventually, we apply and compare the classical, IAA (Iterative Adaptive Approach) and OMP approaches on experimental MIMO signals.
展开▼
