An Experimental Study of the Angular Superresolution of Two Correlated Signals Using the Minimum-Polynomial Method

Abstract

We present experimental results on the superresolution of two closely located signal sources, which were obtained using an automotive millimeter-wave radar. The signal sources were mutually correlated and the input process consisted of only one sample. The minimum-polynomial method (the root variant) was compared with the Minimum Description Length (MDL) criterion when determining the number of sources and the root MUltiple Signal Classification (MUSIC) method when estimating their angular location. The minimum-polynomial method is shown to have a higher efficiency compared with the MDL criterion and ensures the source superresolution for the angular distance which is a factor of 4–5 smaller than the width of the antenna-array pattern. In terms of accuracy of estimating the coordinates of the signal sources, the efficiency of this method almost coincides with that of the root-MUSIC method if the number of sources is considered to be known in the latter method.