Robust direct position determination methods in the presence of array model errors

Abstract

Direct position determination (DPD) methods are known to have many advantages over the traditional two-step localization method, especially for low signal-to-noise ratios (SNR) and/or short data records. However, similar to conventional direction-of-arrival (DOA) estimation methods, the performance of DPD estimators can be dramatically degraded by inaccuracies in the array model. In this paper, we present robust DPD methods that can mitigate the effects of these uncertainties in the array manifold. The proposed technique is related to the classical auto-calibration procedure under the assumption that prior knowledge of the array response errors is available. Localization is considered for the cases of both unknown and a priori known transmitted signals. The corresponding maximum a posteriori (MAP) estimators for these two cases are formulated, and two alternating minimization algorithms are derived to determine the source location directly from the observed signals. The Cramér-Rao bounds (CRBs) for position estimation are derived for both unknown and known signal waveforms. Simulation results demonstrate that the proposed algorithms are asymptotically efficient and very robust to array model errors.