Unitary cyclic ESPRIT-like direction finding

A unitary formulation of cyclic ESPRIT algorithm, with uniform linear array (ULA) configuration, is proposed by constructing the forward backward cyclic autocorrelation matrix by Hermitian mapping. This approach not only substantially reduces the computational complexity via real-valued decomposition, but also outperforms the forward backward spatial smoothing algorithm for uncorrelated or correlated sources. In addition, this method allows to select desired signals and to ignore interferences because of its signal-selective property. The performance analysis shows that for uncorrelated or correlated sources, the Hermitian mapping approach has a lower RMSE and a higher output SNR than the FBSS method. Finally, simulation results show the effectiveness of the proposed methods.