Electromagnetic coupling matrix modeling and ESPRIT-based direction finding: a case study using a uniform linear array of identical dipoles

Abstract

For the estimation of an incident source's direction-of-arrival (DOA) based on data collected from a uniform linear array (ULA) of identical antennas that are electromagnetically coupled mutually, several references have modifications to the well known ESPRIT algorithm, in order to mitigate the mutual coupling effects. These modified ESPRIT-based algorithms discard the data collected by antennas at the two ends of the linear array. Those to-be-overlooked antennas are labeled “auxiliary” antennas or “dummy” antennas. However, those methods presume the antenna-array's mutual coupling matrix mathematically to be both Toeplitz and banded. These exacting presumptions are tested here by a case study where the antennas are half-wavelength dipoles. This paper discovers that the Toeplitz-and-banded coupling-matrix assumption is actually invalid, based on antenna-electromagnetics computations via the “method of moments”. This paper then points out the serious consequences of this Toeplitz-and-banded mis-modeling on the estimation of the incident source's direction-of-arrival.