Direct Position Determination in Asynchronous Sensor Networks

Source positioning using time difference of arrival (TDOA) has received much research interest in the field of vehicular technology during the past decades. In practice, sensor networks may be disturbed by clock biases, and extra active anchor sources with known positions could be exploited to calibrate these clock biases. Although there exist numerous two-step localization methods on this issue, they are inherently suboptimal. In this paper, we address the TDOA-based source direct position determination (DPD) problem in asynchronous sensor networks. First, the clock biases are demonstrated via theoretical analyses to cast significant influences on localization accuracy negatively. Then, a DPD method is proposed, exploiting anchor sources to jointly calibrate clock biases and determine the unknown source position, which integrates an expectation-maximization (EM) algorithm and a Gauss–Newton algorithm for coarse and refined parameter estimation, respectively. In this way, the computational load of the considered DPD problem is reduced considerably compared with exhaustive search-based DPD method. Moreover, the Cramer–Rao lower bound is derived for the considered DPD problem, and how anchor sources affect the localization performance is further analyzed. Simulation results demonstrate the necessity of clock bias calibration and the satisfactory performance of the proposed method.