Comparison of DOA estimation performance for various types of sparse antenna array geometries

Three subclasses of geometries for nonuniform linear antenna arrays with a fixed number of sensors are compared in the sense of maximum possible direction-of-arrival (DOA) estimation accuracy. Cramer-Rao bound analysis is applied to compare the optimal accuracy for each geometry under some fixed source environment. Actual DOA estimation simulations, obtained by recently-introduced algorithms, are used to demonstrate the applicability of Cramer-Rao bound analysis for DOA estimation in these cases. We show that previous attempts to maximise the number of contiguous correlation lags and to avoid missing lags in certain array geometries does not necessarily lead to an improvement in DOA estimation performance.