Space/Time Random Processes and Optimum Array Processing,

摘要

Representations of space/time signals as stationary, homogeneous random processes are used to formulate the theory of optimum array processing. Generalized Fourier expansions for one, two, and three dimensions are discussed. These representations include directional, isotropic signal fields, as well as all previously published ambient noise models. Receiving arrays are modeled in terms of how they observe signal fields with the roles of their geometry and the observation noise analyzed. The design of optimum beamformers is formulated and coupled to the representation of the signal field and the receiving array. The structure of optimum beamforms are analyzed and related to beam displacement, null placement, and superdirective methods. Processing performance obtained by exact methods and approximate methods are compared for various signal fields and array geometries. The closeness of the results indicate the usefulness of the representations and array analysis in predicting the performance of optimum systems. Finally, the structure of preformed beams and clustered receiving arrays are analyzed with these methods. (Author)