On the subsample time delay estimation of narrowband ultrasonicechoes

摘要

Correlation based time delay estimators, optimal under certainnconditions, exhibit the well-known threshold effect of poor performancenat low signal-to-noise ratio (SNR). This sudden reduction in performancenof the correlation based time delay estimators at low SNR arises fromnthe misidentification of one unique “extremum” in very noisynconditions and from the peak fitting procedure in the case of thensubsample time delay estimation. In this paper, two new estimators-thenMSX and MXS-for the estimation of subsample time delays in narrow-bandnsignals are proposed. In these estimators, cross-correlations andnautocorrelations are matched at a number of different lags to yield annumber of time delay estimates which are subsequently combined to obtainnone robust time delay estimate. They seem to perform adequately over thenSNR range used in simulations of -5 to 20 dB. Their performances arencompared to those of two cross-correlation based estimators. Usingnsimulated data, it is demonstrated that all four estimators perform wellnat high SNR, but at low SNR the proposed MSX and MXS estimators offernsignificant improvements in the bias and variance of the estimates.nAdditionally, these findings are verified using ultrasonic experimentalndata at three different SNR