Research On Long-range Source Localization By Matched Field Processing In Shallow Water

Source localization by matched-field processing is a classical problem in underwater acoustics; many methods were presented in the past several decades. Because of the complexity of ocean environment, until now, there is no efficient method that can solve long-range source localization in time under real environmental condition. The first reason lies in that sound pressure is a time variant quantity, its amplitude and phase are fluctuating, single measurement or limited time measurements couldn't give the real character of the sound field completely, so it is difficult to localize a source accurately based on sound pressure only; the second reason lies in that ocean acoustic channel is three-dimensional, while almost all the current methods are only fit to range independent environment; the last reason lies in that the 3-D sound field calculation is time exhausted, there is no efficient algorithm that can calculate the sound field quickly. In order to realize long-range source localization in shallow water and to improve the accuracy of localization as well as the speed of matched-field processing, a new method is studied in this thesis. Not sound pressure but eigenray travel time is adopted. Using ray travel time, noise tolerance could be enhanced and the sensitivity to model mismatch may be degraded by neglecting the amplitude information; phase ambiguity could be eliminated by exploiting group delay of the arriving wave packets; ray travel time is a pseudo-linear function of sound speed and its fluctuation is much smaller than sound pressure's; eigenray calculation in 3-D environment is fast and accurate. With the above analysis of ray travel time, long-range source localization based on relative ray arrival time between two isolated hydrophones is studied in this dissertation.It is well known that source localization by matched-field processing consists of two parts. One is the establishment of acoustic model; the other is the construction of a suitable cost function. Establishment of an accurate model is the basis of the whole problem. Derived from the geometrical model, two 3-Deigenray calculation programs such as HARPO and RTPO are analyzed in detail. The results demonstrate that the RTPO program developed by Dr Tang Junfeng, a student of our group is more accurate and has higher calculating speed than HARPO. After that, two kinds of cost functions are constructed. The first is based on relative ray arrival time between two adjacent hydrophones; the second is directly based on ray travel time. With these two functions, mismatches of water depth, receiver location and sound speed have been simulated respectively. The simulation results show that the first cost function is sensitive to the mismatches of water depth and receiver location, but is insensitive to sound speed mismatch and can give high accuracy of depth estimation; the second cost function is insensitive to all three kinds of mismatches, but its resolution of depth estimation is very poor. In order to overcome the first function's sensitivities to the mismatches, a method of increasing the number of elements is used. Numerical calculation shows that this method degrades the sensitivity to the mismatch of receiver location and improves the accuracy of range and depth estimation, but is ineffective to ocean depth mismatch. For the purpose of analyzing the sensitivities to the mismatches of the cost functions, three formulas for calculating the ray travel time error or time delay error caused by the mismatches of receiving array location, ocean depth and sound speed are derived respectively. With them, the effects on source localization by matched field processing taken by the mismatches are verified, and the theoretical results agree with the simulations completely.As an important part of the thesis, time delay estimation in multipath environment is studied theoretically and experimentally. Based on the extraction of the signal's relative arrival time, source localization in different range is studied step by step. By computer simulations, we find that multi-elements method is effective when the space between two elements is large, but it can only localize the source in the range of 10km. In order to expand the efficient localization range, fractionization method is used. But it is not always feasible when the ocean bottom becomes flat. Under this condition, multipath time delay is needed. Usingthe fractionization method or the combination of multipath time delay with time delay between two adjacent phones, source localization of the range of 60km in different ocean environmental models is realized respectively. Especially when the ocean bottom is complicate, fractionization method becomes more efficient. In order to confirm the validity of the methods, ASIAEX data and South China Sea experimental data are used to do the research of source localization. The calculated results agree with the true source locations.With the above simulation analysis and experimental studies, it could be concluded that the new method introduced in the thesis is feasible to realize long-range source localization in shallow water, and it has high accuracy and fast speed. Ray tracing in 3-D sound channel is very fast and accurate. Besides that ray theory is helpful to calculate three-dimensional sound field. Using eigenray travel time, parameters inversion in real environment could be finished quickly. It is true that introducing ray travel time into matched field processing is an optimal choice.