Performance evaluation of time-delay estimation in non-Gaussian conditions

Abstract

Time delay estimation (TDE) processors for non-Gaussian conditions have previously been developed and it has also been shown that exploiting the non-Gaussianity of the processes can lead to improved performance, relative to the Gaussian case. However, while for the Gaussian case standard tools (such as the Cramer-Rao bound) can easily be applied to predict the achievable performance, for the general non-Gaussian case using such tools rarely leads to analytical solutions. In this paper we present a close-form expression for a figure of merit (FOM) for any TDE processor under any statistical model. It is then evaluated for different TDE processors and it is confirmed (by simulations) that the proposed FOM accurately predicts the performance of these processors for reasonable values of the time-bandwidth product.