TLS-ESPRIT in a time delay estimation

In this paper TLS-ESPRIT is applied to a time delay estimation. The correlation method and a frequency domain deconvolution are used to convert the time delay estimation problem to a estimation of frequencies of complex sinusoids in a white nonstationary noise. TLS-ESPRIT is then applied to estimate the unknown frequencies, i.e., the time delays. It is shown that if the noise (after the deconvolution) is a white stationary process the TLS-ESPRIT derived returns to TLS-ESPRIT derived earlier for the stationary noise case. The computer simulations show that the derived method produces unbiased, high resolution time delay estimates.