Prediction of Acoustic Communication Performance in Marine Robots Using Model-Based Kriging

George P. Kontoudis, Daniel J. Stilwell

Abstract: In this paper, we present a data-driven iterative algorithm for accurate prediction of underwater acoustic communication performance at unvisited sites. The prediction algorithm consists of two steps: i) estimation of the covariance matrix; and ii) prediction of the communication performance. The importance of the covariance estimation is highlighted with a multi-stage, model-based iterative methodology that produces unbiased and robust results. The efficiency of the framework has been validated with synthetic data.