Dimensionality Reduction In High Resolution Direction Of Arrival Estimation

Abstract

Reducing data dimension prior to application of direction of arrival estimation algorithms is shown to lower computational requirements and improve certain aspects of performance. In this paper we utilize linear transformations for mapping full dimension data into a lower dimensional space. The transformation is designed to maximize the average signal to noise ratio over a set of likely signal scenarios. Reduced dimension versions of the MUSIC and minimum norm algorithms are presented and discussed. An example illustrates the effectiveness of the method.