The asymptotic risk in a signal parameter estimation problem

In estimating the unknown location of a rectangular signal observed with white noise, the asymptotic risks of three important estimators are compared under L/sub 1//L/sub 2/ losses. A different numerical scheme is used to improve the accuracy of I.A. Ibragimov and R.Z. Hasminskii's (Statistical Estimation-Asymptotic Theory, 2nd. ed., New York: Springer-Verlag, 1985) result, which also leads to further information and numerical comparisons about the problem.<
>