SATURN: A Double-Recursive Deconvolution Algorithm for Suppressing Sidelobe Effects in Non-Nyquist SAR and MIMO Imaging Radars

The angular resolution of an imaging radar system is limited by the aperture size and its associated cost and complexity. Relaxing the
$\lambda /2$
element spacing condition for large arrays introduces sidelobe and grating lobe effects that severely degrade the image quality. In certain imaging applications, such as automotive radars, the targets are sparsely located and the imaging domain can be approximated as a collection of point scatterers. In such cases, it is possible to “deconvolve” these sidelobe effects to recover a clean image. In this article, a double-recursive deconvolution algorithm titled Sparse Array Target-Segregation Using Recursive Nulling (SATURN) is proposed. It differs from existing CLEAN deconvolution algorithms in two steps: target response estimation and multiple target decorrelation. The target response is estimated using a Sweep and Extinguish step that removes the reliance on the complex image phase. Correlation between targets is suppressed using a Recursive Nulling step that prevents the breakup of point targets. The algorithm is applied to synthetic aperture radars (SARs) and multiple-input multiple-output (MIMO) radars in sparse 3-D imaging scenarios with canonical targets and real-world targets. Dynamic range improvement of 25 dB and thinning factors of over 200 are experimentally demonstrated using a 31-element circular array of X-band U-slot patch antennas with an array diameter of
$30\lambda $
.