Uni-vector-sensor ESPRIT for multi-source azimuth-elevation angle-estimation

Abstract

This paper proposes a novel eigenstructure ESPRIT-based azimuth-elevation direction-finding (DF) algorithm that exploits source polarization diversity (in addition to spatial diversity), requires no a priori knowledge of signal frequencies, suffers no frequency-DOA ambiguity, pairs automatically the x-axis direction-cosines with the y-axis direction-cosines, reduces hardware and computational costs greatly, eliminates array inter-element calibration, can resolve up to five narrowband non-coherent sources, but uses only a single vector-sensor and two time-delayed data sets. Uni-vector-sensor ESPRIT is based on a matrix-pencil pair of temporally-displaced data sets collected from one single vector-sensor consisting of six spatially co-located non-isotropic polarization-sensitive antennas, measuring all six electromagnetic-field components of the incident wave-field. It impressively outperforms customary unpolarized scalar-sensor phased-arrays of comparable array-manifold size and computational load. In one scenario, uni-vector-sensor ESPRIT offers 33 dB SNR gain or a 7-fold reduction in hardware and software complexity.