ABSTRACT
This chapter presents methods to improve the echo detection and time delay estimation in a noisy multipath channel. The composite function z(t) is not spatially separable by the receiving system. The autocorrelation and cepstrum have been used to detect and estimate the time delay between the signal and its echo. The choice of sampling rate relative to the signal bandwidth has an effect on the quality of the cepstrum. The sampling rate determines the frequency range over which the summation is performed. The echo detection threshold depends upon several measures including signal-to-noise (S/N) ratio, echo strength, number of samples in discrete Fourier transform N, and effective measure of relative S/N bandwidths. The time delay tracking problem is discussed, and Kalman filtering is applied to provide the quality measures on the time delay estimates and to adjust the processing parameters accordingly.
