EF-ESPRIT: A simplified ESPRIT algorithm without generalized eigenvalue calculation

ESPRIT is an algorithm for DOA (Direction-of-Arrival) estimation. It provides a high angular resolution compared to the MUSIC algorithm, yet it has to calculate generalized eigenvalues of matrix pencils. In this paper, by introducing the sub-space decomposition into the original ESPRIT algorithm and exploiting Hermite symmetry of the correlation matrix of the received signals of an antenna array, we present an improved version of ESPRIT: EF-ESPRIT (Eigenvalue Free-ESPRIT), to reduce the computational burden involved in the ESPRIT algorithm. Through the theoretical derivation of the EF-ESPRIT algorithm, we have also identified a spectral scheme which manages to visualize the DOA estimation result, making the signal arrival angle easier to recognize. Simulation results show that the proposed algorithm estimates DOA more accurately and efficiently.