SNR-dependent filtering for Time Of Arrival estimation in high noise

Time of Arrival (ToA) estimation is a cornerstone of many of the remote sensing applications including radar, sonar, and reflective seismology. The conventional Matched Filter Maximum Likelihood (MFML) ToA estimator suffers from rapid deterioration in the accuracy as Signal to Noise Ratio (SNR) falls below certain threshold value. In this paper we suggest an alternative method for ToA estimation based on the fusion of measurements from biased estimators which are obtained using a pair of unmatched filters. Suboptimal but not perfectly correlated estimators are combined together to produce a robust estimator for ToA estimation in high noise. The unmatched filters pair is parameterized by a single parameter (phase shift) which is selected based on estimated SNR level.