Adaptive subarrays beamforming

Rank reduction prior to adaptive beamforming is required for large arrays in order to reduce the required number of samples. It is often processed by a conventional beamforming applied on a number of subarrays partitioning the whole array. The subarrays beams are then adaptively recombined by a second stage of beamforming. On the opposite, subarrays may also be adaptively processed and their beams conventionally, or even adaptively recombined. Several partition schemes are finally available. In this article, a general framework is introduced for those two-steps beamforming. Two basic partitions schemes of a linear array are then considered by forming subarrays of contiguous or interlaced sensors. The pros and the cons of each solution are analyzed in terms of resolution, anti-jamming, noise rejection, and directivity index.