Sequential Monte Carlo Methods for Shallow Water Tracking Using Multiple Sensors with Adaptive Frequency Selection

We propose a matched-field processing framework for tracking problems in shallow water environments where the conventional plane-wave assumptions do not hold. Multiple passive acoustic sensors are employed to collect observation data, and sequential Monte Carlo techniques are used for tracking due to the high nonlinearity in the dynamic state formulation. In order to enhance the tracking performance, we design a frequency selection algorithm which adaptively chooses the optimal observation frequency for the sensors at each time instant. The improved tracking performance is demonstrated using simulations.