Refining Sensor Positions Based on Localization of Multi Sources in Wireless Sensor Networks

This paper investigates the problem of refining the sensor positions in a wireless sensor network utilizing the time difference of arrival (TDOA) observations from multiple disjoint sources. Since the TDOA observations correspond to the same sensor position uncertainties, it is expected to further improve the sensor position accuracy based on the localization of the sources. Rather than the traditional two-step weighted least-squares approach that provides an approximate solution to source localization and sensor refinement problem, we firstly uses an iterative constrained weighted least-squares method to arrive at a more accurate localization of the multiple sources, which can be used to construct a set of range measurements with respect to each sensor. Finally, an iterative sensor position refinement algorithm is developed to fuse this range information and the prior sensor location information. Numerical Monte Carlo simulations are designed to verify the performance of the proposed method. The results illustrate that the corresponding average mean squared errors of both the source localization and sensor refinement are considerably smaller that those using the previous method.