Robust Modified Newton Algorithms Using Tikhonov Regularization for TDOA Source Localization

This paper addresses a modified Newton (MNT) algorithm for the source localization problem utilizing time difference of arrival. As the improvement of the Newton (NT) method, the proposed algorithm can guarantee convergence stability in the case of bad initial values by using the Tikhonov (TI) regularization theory. Moreover, a two-stage MNT algorithm is proposed for the source localization with sensor position errors. Theoretical analysis is provided to illustrate that the two-stage MNT algorithm has less computational load and faster convergence speed compared with the MNT algorithm. Simulation results show the superior location accuracy and better convergence performance of the proposed MNT and two-stage MNT algorithms in comparison with relative methods.