Enhanced Reactance-Domain ESPRIT Algorithm Employing Multiple Beams and Translational-Invariance Soft Selection for Direction-of-Arrival Estimation in the Full Azimuth

An enhanced
reactance-domain
estimation of signal parameters via rotational invariance techniques (RD-ESPRIT) algorithm is proposed for direction-of-arrival (DoA) estimation in the full-azimuth plane. Unlike the classical RD-ESPRIT algorithm, the proposed method employs a generalized RD correlation matrix combined with soft selection. The proposed algorithm is expected to allow estimating more than one signal DoA with improved estimation precision. The generalized RD correlation matrix is formed by utilizing a number of beam patterns greater than or equal to the number of array elements. The soft selection employed in the
enhanced
RD-ESPRIT consists of two steps. First, the estimates obtained by applying the generalized RD-ESPRIT to three translational-invariance configurations are gathered to form a set of DoA estimate candidates. Second, selection functions based on a modified MUSIC function are used to select the DoA estimates from among the estimate candidate set. The proposed algorithm is investigated through computer simulations with a seven-element electronically steerable parasitic array radiator (ESPAR) antenna. The simulation showed that the proposed algorithm can perform effective DoA estimation of up to four incoming signals in the full-azimuth plane with slightly improved DoA-estimation precision. So far, the classical RD-ESPRIT is not effective because it can estimate up to one signal DoA. Moreover, the proposed method showed robust estimation capability for up to two incoming signals. A robust estimation refers to the fact that a number of DoA estimates equal to the number of incoming signals can always be selected from trial to trial. A one-signal estimation experiment was conducted in an anechoic chamber with a fabricated seven-element ESPAR antenna. In general, the proposed algorithm showed as accurate estimation precision as the classical algorithm. In particular, using soft selection instead of hard selection increases the estimation precision, whereas augmenting the number of beam patterns brings an increase in the estimation precision only when used with hard selection.