Adaptive optimum array processing [acoustic]
A global approach to the processing of information from an array of sensors is used in which the unprocessed outputs of the individual elements are considered as the observables, and the structure is allowed to evolve freely with optimum detection as the criterion. Beam forming is not imposed from the beginning, but might be an integral part of the resulting structure. It is found that if a priori knowledge is incorrectly incorporated into the array process or design the detection performance on the receiver operating characteristic curve can suffer greatly. When implemented sequentially, the optimal array processor exhibits natural learning or adaptive features. The Bayesian updating which takes place as an integral part of the sequential structure is illustrated by computer simulation
