Generalized Signed Rank Estimator for Two-Dimensional Sinusoidal Signal Model

Abstract

Estimation of the unknown amplitudes and frequencies parameters of a two-dimensional (2-D) sinusoidal signal model observed in the additive random field is one of an important and difficult problem. In this paper, we can propose a robust sequential method for estimating the unknown parameters of the two-dimensional sinusoidal components when the signal is affected by different types of heavy-tailed noise. The proposed method is sequential generalized signed rank estimator method (SGSR), where each of the sequential least square estimator method (SLS) and sequential least absolute deviation estimator method (SLAD) are particular cases of SGSR estimator Method. Some numerical experiments have been performed for different sample size, and different heavy-tailed noise with various level of variance to observe the performance of the robust estimation method (SGSR) compared with the two other methods (SLS),(SLAD). A simulation study shows that Sequential generalized signed rank (SGSR) estimator performs better in estimation parameter than others (SLS, SLAD) when error distribution is heavy-tailed.