FAIR AND ANDREWS WEIGHTING-BASED IRWLS ALGORITHMS FOR TIME-DELAY ESTIMATION IN UNDERWATER TARGET TRACKING

Underwater target tracking relies on a model relating the target state s to time-delay and bearing measurements. This furnishes linearized me asurement models. Problems arise due to fitting models using the least squares procedure, whose success may depend on the assumption that th e data noise distribution is Gaussian. For many cases of non-Gaussian errors, performance of the least squares estimators is far from optima l. Robust regression procedures have been proposed to improve the perf ormance of the least squares procedures for non-Gaussian errors, and t o enhance their performance for the Gaussian errors. In this paper fil ters for time-delay estimation based on the Fair and Andrews's weighti ng functions of the iteratively reweighted least squares method are pr oposed. Computational results are given to illustrate and compare the performance of the two filters as well as that due to ordinary least s quares filters.