Marine science and Internet of Underwater Things applications rely significantly on collecting data from underwater sensors. Data collection using long-distance underwater acoustic communications consumes a lot of energy in underwater sensor nodes, which are powered by batteries. To achieve low-energy consumption, we can use the autonomous underwater vehicle (AUV) to move close to sensor nodes and exploit the short-range and high-rate communications. Most of the existing AUV-based data collection schemes consider the scenarios having the knowledge of node positions, where the cruising trajectory can be computed before the AUV’s departure. These schemes cannot apply to some scenarios such as turtle tracking for a certain sea area having no position information. To this end, we first propose a planning-while-detecting approach to dynamically detect the sensors on turtles and adjust the AUV cruising direction to collect data. To further improve data efficiency under the energy limit of the AUV, we group the sensors that can share the same trajectory using their detected directions. A grouping-based dynamic trajectory planning (GDTP) is then proposed to determine the next cruising direction that can visit the group of sensors having the largest amount of data and demanding the least cruising energy at the risk of detection errors. Simulation results show that GDTP achieves significantly higher data collection efficiency than the existing trajectory planning algorithms in dynamic scenarios, and as the communication range increases, it can even outperform the existing algorithms with node locations.

It is challenging to design an equalizer for the complex time-frequency doubly-selective channel. In this paper, we employ the deep unfolding approach to establish an equalizer for the underwater acoustic (UWA) orthogonal frequency division multiplexing (OFDM) system, namely UDNet. Each layer of UDNet is designed according to the classical minimum mean square error (MMSE) equalizer. Moreover, we consider the QPSK equalization as a four-classification task and adopt minimum Kullback-Leibler (KL) to achieve a smaller symbol error rate (SER) with the one-hot coding instead of the MMSE criterion. In addition, we introduce a sliding structure based on the banded approximation of the channel matrix to reduce the network size and aid UDNet to perform well for different-length signals without changing the network structure. Furthermore, we apply the measured at-sea doubly-selective UWA channel and offshore background noise to evaluate the proposed equalizer. Experimental results show that the proposed UDNet performs better with low computational complexity. Concretely, the SER of UDNet is nearly an order of magnitude lower than that of MMSE.

This paper considers an autonomous underwater vehicle (AUV) equipped with a sensor array to detect the presence of moving targets and estimate their relative locations. We propose a track-before-detect labeled multi-Bernoulli filter, which directly utilizes the signals received from the sensor array as the measurement in the form of a sample-covariance matrix (SCM). Using the unthresholded sensor data avoids creating bearing measurements and thus alleviates the challenges in the presence of multiple targets and high levels of background noise. The SCM measurement is affected by multiple targets in an additive fashion, which causes multi-dimensional integrals in measurement updating. We employ a change of variables and complex inverse-Wish art approximations to derive the closed form solutions, which ensures the filter is computationally tractable. Moreover, sequential Monte Carlo implementation is provided, complemented by an adaptive birth procedure based on the current SCM measurement. Simulation results demonstrate the improved performance of the proposed filter in challenging scenarios of low signal-to-noise ratio, small number of snapshots, trajectory crossing, and fluctuating received signal power.

The long propagation delay in underwater acoustic channels has attracted tremendous attentions in designing Medium Access Control (MAC). The low acoustic propagation speed and wide area of the acoustic communication range led to a wide range of variations in the propagation delay. This paper identifies an important characteristic of two-scale delay variations by field test results. We carry out simulations to study the impact of delay variations on MAC, and the results suggest a slot length adaptation scheme for the handshake and slotting based MAC. We further model an absorbing Markov chain to derive the closed-form equation for the throughput of MAC with adaptive slot length. Both the analytical and simulation results show that our proposed slot length adaptation improves significantly the throughput of MAC in underwater acoustic networks. Particularly, the Slotted-FAMA with an adaptive slot length achieves more than double the throughput than the Slotted-FAMA with a fixed slot length in a network with six nodes.

The classification of high-frequency marine mammal vocalizations often faces challenges due to the limitations of acoustic features, which are sensitive to mid-to-low frequencies but offer low resolution in high-frequency ranges. Additionally, single-channel networks can restrict overall classification performance. To tackle these challenges, we introduce MT-Resformer, an innovative dual-channel model with a multi-scale framework designed for classifying marine mammal vocalizations. Our approach introduces a feature fusion strategy that combines the constant-Q spectrogram with Mel filter-based spectrogram features, effectively overcoming the low resolution of Mel spectrograms in high frequencies. The MT-Resformer model incorporates two key components: a multi-scale parallel residual network (MResNet) and a Transformer network channel. The model employs a multi-level neural perceptron (MLP) to dynamically regulate the weighting of the two channels, enabling flexible feature fusion. Experimental findings validate the proposed approach, yielding classification accuracies of 99.17% on the Watkins dataset and 95.22% on the ChangLong dataset. These results emphasize its outstanding performance.