Dynamic Source Localization via Finite-Element Underwater Acoustic Field Estimation

In this paper we address the problem of dynamic source localization in spatially distributed systems governed by Partial Differential Equations (PDEs) which aims to detect and localize a mobile source from a passive array of acoustic sensors. We consider an underwater environment where the space-time dynamics of the source-induced field is modeled by a finite-element (FE) approximation of the full acoustic wave PDE. Based on recent advancements in large-scale state estimation of PDE systems, we present a novel Multiple Model (MM) filtering approach to underwater dynamic source localization. The proposed framework sequentially estimates the acoustic field and the source location by running in parallel a bank of FE-based field estimators, each conditioned to the source being placed in a given element of the FE mesh. We adopt the Ensemble Kalman Filter (EnKF) implementation for computationally efficient estimation of the large-scale acoustic field. The effectiveness of the proposed Finite-Element Multiple Model Ensemble Kalman Filter (FE-MM-EnKF) is demonstrated via simulation experiments in the underwater acoustic environment.