Matched Field Tracking Of Underwater Acoustic Source With Bernoulli Filtering

The problem of passive localization of acoustic sources in oceanic waveguide is of importance in community of underwater acoustic signal processing.Matched field processing(MFP)is a com-mon method to this problem.Its basic idea is to model the ocean channel,mesh the parameter space for source position,calculate replica for all possible parameter and match replica with field data to obtain the ambiguity surface.Therefore,range and depth of the acoustic source can be esti-mated by choosing parameter corresponding to the max value in the ambiguity surface.The main drawback of MFP is its dependence on the accuracy of the model,that is,when the parameters of the acoustic model is not accurate,the positioning accuracy is often reduced while the positioning error is increased.In addition,noise and interference often lead to the sidelobes in the ambiguity plane being higher than the mainlobe of the acoustic source of interest,which has a great influence on the location performance of the matched field method.In order to solve the above problems,this article is divided into the following sections:Firstly,the mainlobe and sidelobe distribution of the ambiguity plane of MFP in low signal-to-noise ratios(SNR)as well as environmental mismatch scenarios has been studied in this paper.The results show that even when SNR is quite low or environmental parameters is not accurate,we can still find a mainlobe or sidelobe near the location of the acoustic source in the ambiguity plane with quite a high probability.This feature then became the foundation for the subsequent studies.Secondly,in order to make full use of the distribution characteristics of the mainlobe and sidelobe distribution of MFP,the Bernoulli filter is introduced into the underwater acoustic source tracking problem.Bernoulli filter is a single target tracking algorithm based on the random finite set and Bayesian theory.The proposed MFP-Bernoulli filter framework can be used in scenarios where traditional MFP doesnot work well,such as low SNR or the environmental mismatch scenarios.The numerical simulation and experimental data processing results show that under the above circumstances,the MFP-Bernoulli filter method performs much better than MFP.Finally,in order to locate or track an acoustic source when there is a strong interference(such as moving ship),the labeled multi-Bernoulli(LMB)filter based on the random finite set and Bayesian theory is introduced into the underwater acoustic source tracking problem.The numerical simulations show that in the presence of strong motion interference,the underwater weak source can still be located and tracked accurately by using the MFP-LMB filter while the MFP method doesnot work in the same scenario.