Doppler Frequency Shift Based Source Localization in Presence of Sensor Location Errors

Abstract

In this paper, the Doppler frequency shift-based localization problem in the presence of sensor location errors is addressed. Based on the measurement model, we present two methods, an explicit solution and a semidefinite relaxation (SDR) technique, to estimate the position of the stationary source. In the two proposed algorithms, the non-accurate information on sensor position and velocity is considered. In the former algorithm, based on two best linear unbiased estimator, an explicit solution of the source position can be obtained, whereas in the latter one, the SDR technique is applied to relax the original maximum likelihood estimation into a convex semidefinite programming problem, which provides accurate estimate without postprocessing. Compared with the methods where the sensor location errors are not taken into consideration, the proposed algorithms are able to achieve a more accurate localization result. Simulations verify a better performance of these two proposed algorithms.