Performance analysis of 2-D DOA estimation via L-shaped array

Signal direction-of-arrival (DOA) estimation using an L-shaped array of sensors configured by two uniform linear arrays (ULA) has been an active research topic in array signal processing. A simple but efficient DOA estimation method has recently been proposed by exploiting the L-shape array geometry and the cross-correlation information of sensor data. In this paper, the asymptotic variance of the estimated azimuth and elevation angles is discussed. Unlike the traditional asymptotic analysis of ESPRIT algorithm where the eigen-decomposition of auto-correlation is used, herein we will focus on the first order perturbation for the cross-correlation matrix where the noise is eliminated. The theoretical analysis is verified through numerical results.