Local estimation of delay parameter following robust detection

A procedure is outlined for finding the variance of a local estimate of delay of a known signal in Gaussian noise following a detection process which is robust against contamination of the underlying probability distributions. The detection procedure is a likelihood ratio processor followed by a soft limiter. Maximum likelihood estimation of the delay parameter is used in the linear region of the soft limiter and the instants of saturation and desaturation are employed when the upper level of the soft limiter is crossed. The paper presents results for this estimator and compares them with the Cramer-Rao lower bound. Applications to Radar and Sonar are briefly discussed.