Sparse Bayesian learning localization method with environmental perturbation constrains

Matched-Field Processing (MFP) passive localization method suffers from its sensitivity to the ocean environmental perturbations.A method named Sparse Bayesian Learning (SBL) for robust MFP (MFP-SBL) is proposed for source localization in an uncertain shallow water waveguide.By exploiting the acoustic field structure using the normal mode model with the existence of environmental mismatch,an ocean environmental perturbation model is established.Then,we reformulate the MFP localization as a linear sparse signal recovery problem to exploit the inherent sparsity of underwater localization problem.The proposed algorithm iteratively updates the target location and the model mismatch weights via performing Bayesian inference and convergences to the optimal sparse solution as the estimated location.Simulations and sea trial data in north Elba island are used to evaluate the performance of the proposed MFPSBL algorithm.It demonstrates that MFP-SBL can estimate the target location accurately even with the existence of environmental mismatch,and can distinguish two targets with a horizontal space of 100 m.Therefore,the proposed scheme can enhance the robustness and positioning accuracy of MFP via exploiting the acoustic field structure under ocean environmental mismatch and the inherent sparsity of underwater localization problem.Also,it can deal with the multi-source positioning problem.