Investigation of theoretical limits for unconditional AoA estimations in multi-element antenna arrays by simulations

In this paper, we thoroughly investigate the performance of unconditional angle-of-arrival AoA estimation in multi-element antennas for a finite number of samples. Two AoA estimation algorithms are applied: a direct iterative algorithm for searching the maximum likelihood (ML) estimate and a three-step algorithm using the sample covariance matrix eigenvalue decomposition at the first step and the least-square (LS) method to improve the estimate at the next steps. Two configurations of antenna systems are considered: uniform linear array (ULA) and uniform circular array (UCA). It is shown that for the proposed algorithms and considered antenna configurations, the accuracy of AoA estimation for large SNR reaches a theoretical limit that depends only on the sample size N. In low SNR region the iterative algorithm at 35 % -50 % outperforms the accuracy of the three-step algorithm. The joint detection-estimation schemes for the considered estimators and antenna arrays were investigated. It was shown that the detection characteristics of these schemes are very close for the given numbers of antenna elements and samples.