Coherent MIMO radar range imaging with block sparse recovery

We consider range imaging for coherent MIMO radar, that is, the joint estimation of the channel matrices for all range bins of the radar scene. The fundamental problem is the lack of ideal ambiguity properties of the available waveforms. This is why matched filters are generally suboptimal estimators. Approaches like the instrumental variable filter overcome this problem to a certain extent, but need prior knowledge of the interfering range bins. We take a different approach and show that range imaging can be formulated as a block sparse recovery problem. The block structure arises as the coefficients of the channel matrix of a range bin are either all zero or nonzero. In a second step, high-resolution methods for azimuth estimation can be used. This is in contrast to other sparse recovery approaches in coherent MIMO radar imaging where range, Doppler, and azimuth estimation is performed simultaneously and resolution is limited by coarse grids. We make a first step towards the analysis of sparse recovery based range imaging for coherent MIMO radar by presenting numerical recovery results using iterative algorithms. Our simulation results demonstrate that the channel matrices for all range bins can be estimated reliably, even for a large number of targets.