EM-based Underwater Localization in Stratified Medium

Acoustic waves in an underwater environment do not necessarily travel in straight lines due to sound speed variations, which poses a set of challenges for underwater acoustic localization. In this paper, we consider acoustic localization in a stratified underwater medium based on time of arrival (TOA) measurements. We assume that the sound speed profile (SSP) only depends on depth and is stratified in vertical. A multi-layer depth-dependent SSP model is considered in this paper. However, which layer originates the source is uncertain in practice. An expectation-maximization (EM) based underwater localization approach is proposed, where uncertainty in measurement origin is handled. This approach largely simplifies data association and reduces the complexity of localization induced by sound speed variations. In addition, the Cramér-Rae lower bound (CRLB) of this problem is derived. We illustrate the effectiveness of the proposed algorithm by locating several sources in different layers.