Wall Localization in a Water Tank Using a Cooperative Source of Opportunity

Model-based underwater acoustic localization in confined spaces like water tanks typically depend on the geometry. This paper investigates the problem of water tank wall localization using an acoustic source of opportunity that may not offer sufficient bandwidth for time-of-arrival-based methods to perform satisfactorily. We propose that in these scenarios, approximate time-of-arrival-based methods are used followed by a ray-based differentiable model using a gradient-based optimization to obtain a more accurate estimate. We demon-strate the successful operation of the algorithm on data ob-tained from a tank simulator, whose parameters are adjusted to those of the Scripps Ocean Atmosphere Research Simula-tor (SOARS).