Maximizing Information Extraction of Extended Radar Targets Through MIMO Beamforming

We jointly design an information-theoretic transmit and receive radar beamformers for spatially near multiple extended targets. We maximize the mutual information (MI) between the received signals and the targets signatures that allows the extraction of the unknown features, which may include shape, dimensions, and material. However, high interference caused by spatially near targets might obstruct the information extraction, and directing the beamformers toward the steering vector as done in conventional beamformers does not solve this problem, especially for extended targets. In this letter, an iterative algorithm is presented to solve this problem using alternative minimization, dividing it into two blocks. The first block is solving for the transmit beamformers successively using block coordinate descent, and the second one is solving for the receiver beamformers using the minimum variance distortionless response. We also show the effect of using our beamformers on the waveform design problem. Numerical results indicate that this algorithm can achieve substantially higher MI than the existing conventional methods. Thus, except for some degenerate cases, having fixed beamformers instead of optimized ones lead to significant performance degradation.