A performance comparison of three localization algorithms

Abstract

In the practical application of localizing micro seismic events in potash mines, a total signal energy (TSE) algorithm described in this paper can yield satisfying source location estimates directly, without having to do any parameter estimation first, as most traditional two-step localization algorithms requiring TDOA (time difference of arrival) estimation do. Using Matlab simulations, the TSE algorithm is compared with the approximated maximum likelihood (AML) algorithm, which is also a one-step algorithm and is equivalent to the maximum cross-correlation criterion in single source cases. Then a performance comparison is done between the two one-step algorithms just mentioned and a two-step direct search (DS) algorithm utilizing the Nelder-Mead simplex method to find the source location with the aid of estimated TDOA's. In our simulation of the DS algorithm, the TDOA's are estimated using the generalized cross correlation (GCC) technique. Simulation results showed that the DS algorithm has the worst performance and the TSE algorithm has an overall best performance in the case of locating a near-field, single acoustic source with a short time duration