Robust multi-model mobile target localization scheme based on underwater acoustic sensor networks

Abstract

Underwater mobile target localization based on underwater acoustic sensor networks (UASNs) is a critical application in marine fields. However, the mobility of underwater targets, asynchronous reception of localization information, stratification effects, and unknown measurement losses make UASN-assisted mobile target localization difficult. To address these issues, this study proposes a robust multi-model mobile target localization scheme (RMML) based on UASNs. The RMML suggests a simplified approach for optimal localization reference node selection based on the Cramér-Rao lower bound. This approach improves the localization accuracy of a mobile target by periodically selecting nodes that provide high-quality localization references. Using the localization reference provided by the selected nodes, a robust mobile target localization algorithm is developed based on an interacting multiple model and unscented Kalman filter. In this algorithm, a multipoint prediction method and ray tracing method are jointly proposed to improve the target state estimation accuracy under the asynchronous reception of localization information and the stratification effect. Maximum a posteriori probability estimation is used to estimate measurement loss, and pseudo-residuals are constructed based on the estimated measurement loss to improve the localization accuracy and robustness of the algorithm. Finally, extensive simulations and experiments are performed to verify the effectiveness of the RMML.