An efficient TDOA and FDOA based source localization algorithm via importance sampling

Abstract

The location and velocity information gives crucial impacts and supports in many applications. In this paper, we address the the problem of source localization based on time-difference-of-arrival (TDOA) and frequency-difference-of-arrival (FDOA) measurements. The nonlinear and nonconvex maximum likelihood estimation can be approximated by Monte Carlo importance sampling technique. We derive the importance function analogous to Gaussian distribution based on Fisher information matrix (FIM) where the global convergence can be guaranteed with sufficient samples. Then we consider the source localization under sensor location uncertainties. The localization information of both sensors and source is jointly estimated by importance sampling. The simulation results illustrate that the proposed method is superior than other existing methods and it can attain the Cramér-Rao lower bound at moderate noise levels.