Exploiting Compressive Sensing Basis Selection to Improve 2 × 2 MIMO Radar Image

This paper presents a novel technique suitable to build a basis matrix for image recovery in Compressive Sensing Multiple-Input Multiple-Output (CS-MIMO) radar. The proposed technique selects the best sparsifying basis matrix through the use of Gaussian noise, achieving the $\mathrm{R}^{N}$ orthonormal space base with the sparsest structure. A comparison is made between the performance of this optimized basis matrix with both the Fast Fourier Transformation (FFT) and the Haar wavelet. Improvement with respect to optimum Nyquist criterion is quantitatively evaluated by using the effective Target peak to Secondary peak Ratio (TSR). Experimental data on a MIMO radar shows that this basic matrix maintains the Field of View (FOV), while improving the angular resolution with respect to the prior sparsity matrix.