A computationally efficient estimator is proposed which can provide instant
aneous delay measurement for many practical applications such as radar. son
ar and geolocation via satellites. It consists of an adaptive FIR filter to
model the time difference of arrival (TDOA) between the received signals f
rom two sensors and a variable gain to adapt to the changing signal-to-nois
e ratio environment. The TDOA is obtained directly on a sample-by-sample ba
sis and no interpolation is necessary. The proposed algorithm gives an unbi
ased delay estimate, and its performance for both stationary and nonstation
ary conditions is evaluated rigorously. It is also proved that the least-sq
uares realisation of the estimator attains the Cramer-Rao lower bound.