Systolic preprocessors for linearly constrained beamforming

Systolic preprocessors for linearly constrained beamforming are developed using factorizations of the generalized sidelobe canceller. These preprocessors transform linearly constrained beamformers into unconstrained form, which permits unconstrained adaptive processors to be used for computing the beamformer output. Two preprocessors are developed for small numbers of constraints relative to the data dimension, and two for partially adaptive beamforming where the number of effective constraints is often large relative to the data dimension. All four preprocessors are applicable to arbitrary constraint sets. Tradeoffs between the various preprocessors are discussed.>