Abstract:  In shallow-water acoustic channel，multipaths caused by reflections of ocean surface and bottom arc very serious.The correlated multipath returns cause unwanted peaks in the cross-correlation output of receivers.There peaks introduce performance degradation of passive localization estimator
based on time-delay estimation. Multipath signal becomes the dominant interference in traditional sonar signal processing. But if surface path or bottom path to direct path time delay could be estimated from received signal，and ocean acoustic parameters such as sound speed profile(SSP) and sea depth were
obtained，then it is possible to estimate source range and depth, In this paper, an effective method based on cepstrum is proposed which using a single hydrophone to separate surface or bottom multipath time- delay out from the cepstrogram of received signal.There multipath time-dclays are then used to estimate
sonrce rank and depth.To analyze the localization algorithm performance theoretically, slant rank bias and root-mean-square error expressions are derived. It can be found from the expressions that range estimation hias and root mean-square error are the function of surface delay, bottom delay, ocean depth，
source depth and receiver depth, It is difficult to find out the relationship between rank and SSP directly from the expression.To explore the SSP influence on localization algorithm, computer simulations are carried out. Four negative SSP gradicnt values varying from 0. 1 s^-1 to 0.4s^-1 are considered. From the
simulation results it is shown that rank bias relics strongly on SSP gradient value. At the same time root-mean-square error of slant rank is calculated statistically by Monte-Carlo simulation under different  multipath time delay estimation accuracies.It can be seen that root-mean-square error of slant range is
mainly influenced by multipath time delay estimation error. Sea trial was conducted to verify the effectiveness of the multipath passive localization method proposed in this paper. Acoustic target is a wide hand random noise source with spectrum level of about 140 dB@4 kHz(re:1µPa). A single
hydrophone collected the radiated noise and stored data in a computer for further processing. Acoustic ambient parameters during sea trial such as SSP，ocean depth and hydrophone depth were obtained.The UPS values of target and hydrophone were recorded simultaneously with acoustic data.The result of sea
trial shows that acoustic source range and depth arc estimated successfully by the method proposed in this paper.