This research is a treatment to resolve the problem of 2D angle-of-arrival estimation for the signals received by uniform circular arrays using the UCA-ESPRIT algorithm. A number of programs and program revisions were written to test the performance of this method, and to point out its advantages, and limitations. The observed conclusion was that this method is suitable for uncorrelated or not highly correlated signals and for high number of sensors, inspite of the worst cases (low SNR (0 dB) and low number of snapshots (1 snapshot)). But degradation can be noticed on the performance in two cases; when coherent or highly correlated signals are present, or if the number of sensors is low (less than 2M). To overcome this problem and to get butter resolution and to reduce the RMS error, a solution was proposed through the rooting techniques and preprocessing schemes like the spatial smoothing techniques. The Forward / Backward (FB) spatial smoothing can be applied to coherent or highly correlated signals by exploiting the centro-Hermitian property of beamspace manifold. Also this technique increases the number of sensors virtually. When applying the forward / backward spatial smoothing to the covariance matrix upon the application of the UCA-ESPRIT algorithm, good results were obtained. The new proposed algorithm was named as UCA-FB-ESPRIT algorithm. Using the abovementioned algorithm, the RMS error of coherent and highly correlated signals was reduced. Furthermore the problem of low number of sensors was also solved. The conclusion is that the new proposed algorithm is better than the previous one and it can be used efficiently in the area of direction-of-arrival estimation.