Vector Approximate Message Passing for 3D MIMO Radar

Compressive sensing has been widely used in MIMO radar to reduce the number of measurement antennas while preserving high spatial resolution. In this paper, we deal with the reconstruction of a 3D target scene in the array near-field. By modeling the depth dimension of the target as a property of the reflection coefficients, we can transform the problem of 3D target recovery into an equivalent 2D recovery task. In turn, we have to convert the measurement model to a distributed compressive sensing model to preserve the linearity of the measurement model. For this alternative problem, we develop a vector approximate message passing algorithm. We validate our proposed algorithm through numerical investigations.