Multi-Modal Sensing-Assisted Robust and Cost-Efficient 3D Underwater Acoustic Communication System for Underwater IoT

Abstract

In this study, we propose a robust and cost-efficient underwater acoustic communication system designed to support adaptive acoustic beam alignment between a surface vessel and an autonomous underwater vehicle (AUV) operating over a wide range of three-dimensional (3D) underwater environments. We first introduce a multi-modal sensing-assisted particle filter (PF) algorithm for AUV localization estimation performed at the surface vessel. This algorithm fuses low-cost inertial measurement unit sensor data with multi-modal acoustic modem observations, including signal-to-noise ratio (SNR), one-way travel time, and Doppler shift. A key innovation of our approach is the use of the real-time SNR to compute an adaptive process noise covariance, which intelligently guides the PF’s prediction phase to account for the fluctuating quality of the input data. This mechanism enables the PF to robustly handle non-linear uncertainties, achieving both high accuracy and the computational efficiency required for low-cost embedded systems. Subsequently, using the estimated AUV position, an adaptive acoustic beam alignment mechanism is implemented with a servo motor-driven pan–tilt unit to dynamically adjust the surface modem’s orientation toward the AUV, ensuring robust acoustic link quality despite the AUV’s 3D wide-ranging mobility. Both simulation and empirical water-tank experiments demonstrate that the proposed system significantly enhances localization accuracy and communication performance, while reducing system complexity and hardware costs relative to conventional approaches. These results underscore the potential of the proposed framework to enable scalable, reliable, and cost-effective underwater drone communications in future 3D underwater Internet of Things networks.