Joint Estimation of Underwater Target Location and Sound Speed Profile Based on Argo Data

Underwater wireless sensor network (UWSNs) have enormous potential for development in fields, such as ocean resource development, military security, and environmental monitoring. However, the traditional localization methods for underwater networks are hampered by the challenge of acquiring real-time sound speed profiles (SSPs), leading to substantial localization inaccuracies. The advancement of the Argo program in recent years has facilitated the global collection of millions of temperature and salinity profiles, making it possible to invert SSP using big data in conjunction with various methodologies. Nevertheless, the utilization of the Argo dataset confronts two issues: 1) the presence of outliers in historical data and 2) the potential for data to originate from multiple subspaces, both of which can adversely affect the fitting capability of the basis functions for the subspaces. To address these challenges, this article proposed a subspace shrinkage algorithm based on robust PCA via outlier pursuit (OP-RPCA) algorithm, enhancing the expressive capacity of the basis functions and thereby improving the estimation accuracy of the SSP and locations. Based on the proposed algorithm, we present a joint estimation method for the locations of underwater nodes and the SSP. This method integrates sound ray tracing and circular intersection techniques for the localization of underwater network nodes, employing an alternating minimization approach to solve for node locations and SSP alternately, thus estimating both the SSP and node locations simultaneously. Numerical experiments based on Argo data from the North Atlantic are conducted to validate our approach. The results indicate that compared to traditional methods, the root mean square error (RMSE) of the horizontal localization results obtained using our method was reduced by approximately 80%, with the maximum inversion error of the SSP kept within ±2 m/s.