Unbiased maximum likelihood estimator for underwater DOA estimation

In shallow water the propagation of sound wave over large distances is associated with numerous reflections from the bottom and great losses of acoustic energy. In this context, the plane-wave assumption of arriving signals generally leads to biased estimates. To solve the problem, an array model based on normal mode sound field is put forward and maximum likelihood estimator is derived for DOA estimation. This proposed method is called NM-ML(Normal Mode based Maximum Likelihood method). Statistical analysis and lake experiment results show that the NM-ML method could effectively eliminate the bias of DOA estimates caused by multimode acoustic propagation and reach asymptotically unbiased DOA estimates of sources located in shallow water.