Adaptive Combining of Wideband Array Data for Optimal Reception

Abstract

A study of several closely related adaptive processors for array data has been completed. Processors were designed to converge approximately to a minimum-variance linear unbiased estimator of an unknown signal common to all elements of the array. Any spatial structure of the background noise is used by such a system to enhance the output signal-to-noise ratio. Possible areas of application include sonar, underwater communication, space communication, and seismology. The basic linear adaptive processor has variable coefficients adjusted by a rule similar to that for the minimization by the projection gradient method of a quadratic form which is subject to a linear constraint. Modifications of the basic adjustment procedure have been introduced to reduce system sensitivity to data anomalies, decrease computational requirements, and decrease memory requirements. Experimental evaluation of the adaptive array processors has been completed using data from an actual array of seismometers. Both transient recovery from initially poor processor coefficients and steady-state operation have been found quite satisfactory.