Time delay estimates in passive sonar target localization

In passive sonar systems, a target is localized using time delays obtained from signals received at spatially separated points. The variance of the time delay estimates about the true time delay at low signal‐to‐noise ratio (SNR) depends on SNR, bandwidth, center frequency, and spectral slopes of signal and noise. The accuracy of time delay estimation can be enhanced using a pre‐shaping filter. We shall show in this paper that if signal and noise spectral fall off rates are close to each other (within 3 dB/oct), then the preshaping filter does not appreciably enhance the accuracy of time delay estimation. However, if the signal and noise spectral slopes are significantly different (about 6 dB/oct), then the preshaping filter increases the accuracy.