Gegenbauer Graph Neural Networks for Time-Varying Signal Reconstruction.

Reconstructing time-varying graph signals (or graph time-series imputation) is a critical problem in machine learning and signal processing with broad applications, ranging from missing data imputation in sensor networks to time-series forecasting. Accurately capturing the spatio-temporal information inherent in these signals is crucial for effectively addressing these tasks. However, existing approaches relying on smoothness assumptions of temporal differences and simple convex optimization techniques that have inherent limitations. To address these challenges, we propose a novel approach that incorporates a learning module to enhance the accuracy of the downstream task. To this end, we introduce the Gegenbauer-based graph convolutional (GegenConv) operator, which is a generalization of the conventional Chebyshev graph convolution by leveraging the theory of Gegenbauer polynomials. By deviating from traditional convex problems, we expand the complexity of the model and offer a more accurate solution for recovering time-varying graph signals. Building upon GegenConv, we design the Gegenbauer-based time graph neural network (GegenGNN) architecture, which adopts an encoder-decoder structure. Likewise, our approach also uses a dedicated loss function that incorporates a mean squared error (MSE) component alongside Sobolev smoothness regularization. This combination enables GegenGNN to capture both the fidelity to ground truth and the underlying smoothness properties of the signals, enhancing the reconstruction performance. We conduct extensive experiments on real datasets to evaluate the effectiveness of our proposed approach. The experimental results demonstrate that GegenGNN outperforms state-of-the-art methods, showcasing its superior capability in recovering time-varying graph signals.

Introduction to the special issue on ocean acoustics in the changing arctic.

This paper introduces the Special Issue of The Journal of the Acoustical Society of America on Ocean Acoustics in the Changing Arctic. The special issue includes papers on ocean (and in one case atmospheric) acoustics. Changes in both the ice cover and ocean stratification have significant implications for acoustic propagation and ambient sound. The Arctic is not done changing, and papers in this special issue, therefore, represent a snapshot of current acoustic conditions in the Arctic.

Automatic detection and classification of bearded seal vocalizations in the northeastern Chukchi Sea using convolutional neural networks.

Bearded seals vocalizations are often analyzed manually or by using automatic detections that are manually validated. In this work, an automatic detection and classification system (DCS) based on convolutional neural networks (CNNs) is proposed. Bearded seal sounds were year-round recorded by four spatially separated receivers on the Chukchi Continental Slope in Alaska in 2016-2017. The DCS is divided in two sections. First, regions of interest (ROI) containing possible bearded seal vocalizations are found by using the two-dimensional normalized cross correlation of the measured spectrogram and a representative template of two main calls of interest. Second, CNNs are used to validate and classify the ROIs among several possible classes. The CNNs are trained on 80% of the ROIs manually labeled from one of the four spatially separated recorders. When validating on the remaining 20%, the CNNs show an accuracy above 95.5%. To assess the generalization performance of the networks, the CNNs are tested on the remaining recorders, located at different positions, with a precision above 89.2% for the main class of the two types of calls. The proposed technique reduces the laborious task of manual inspection prone to inconstant bias and possible errors in detections.

Seabed classification from merchant ship-radiated noise using a physics-based ensemble of deep learning algorithms.

Merchant ship-radiated noise, recorded on a single receiver in the 360-1100 Hz frequency band over 20 min, is employed for seabed classification using an ensemble of deep learning (DL) algorithms. Five different convolutional neural network architectures and one residual neural network are trained on synthetic data generated using 34 seabed types, which span from soft-muddy to hard-sandy environments. The accuracy of all of the networks using fivefold cross-validation was above 97%. Furthermore, the impact of the sound speed and water depth mismatch on the predictions is evaluated using five simulated test cases, where the deeper and more complex architectures proved to be more robust against this variability. In addition, to assess the generalizability performance of the ensemble DL, the networks were tested on data measured on three vertical line arrays in the Seabed Characterization Experiment in 2017, where 94% of the predictions indicated that mud over sand environments inferred in previous geoacoustic inversions for the same area were the most likely sediments. This work presents evidence that the ensemble of DL algorithms has learned how the signature of the sediments is encoded in the ship-radiated noise, providing a unified classification result when tested on data collected at-sea.

Seabed type and source parameters predictions using ship spectrograms in convolutional neural networks.

Broadband spectrograms from surface ships are employed in convolutional neural networks (CNNs) to predict the seabed type, ship speed, and closest point of approach (CPA) range. Three CNN architectures of differing size and depth are trained on different representations of the spectrograms. Multitask learning is employed; the seabed type prediction comes from classification, and the ship speed and CPA range are estimated via regression. Due to the lack of labeled field data, the CNNs are trained on synthetic data generated using measured sound speed profiles, four seabed types, and a random distribution of source parameters. Additional synthetic datasets are used to evaluate the ability of the trained CNNs to interpolate and extrapolate source parameters. The trained models are then applied to a measured data sample from the 2017 Seabed Characterization Experiment (SBCEX 2017). While the largest network provides slightly more accurate predictions on tests with synthetic data, the smallest network generalized better to the measured data sample. With regard to the input data type, complex pressure spectral values gave the most accurate and consistent results for the ship speed and CPA predictions with the smallest network, whereas using absolute values of the pressure provided more accurate results compared to the expected seabed types.

Temporal and spatial dependence of a yearlong record of sound propagation from the Canada Basin to the Chukchi Shelf.

The Pacific Arctic Region has experienced decadal changes in atmospheric conditions, seasonal sea-ice coverage, and thermohaline structure that have consequences for underwater sound propagation. To better understand Arctic acoustics, a set of experiments known as the deep-water Canada Basin acoustic propagation experiment and the shallow-water Canada Basin acoustic propagation experiment was conducted in the Canada Basin and on the Chukchi Shelf from summer 2016 to summer 2017. During the experiments, low-frequency signals from five tomographic sources located in the deep basin were recorded by an array of hydrophones located on the shelf. Over the course of the yearlong experiment, the surface conditions transitioned from completely open water to fully ice-covered. The propagation conditions in the deep basin were dominated by a subsurface duct; however, over the slope and shelf, the duct was seen to significantly weaken during the winter and spring. The combination of these surface and subsurface conditions led to changes in the received level of the sources that exceeded 60 dB and showed a distinct spacio-temporal dependence, which was correlated with the locations of the sources in the basin. This paper seeks to quantify the observed variability in the received signals through propagation modeling using spatially sparse environmental measurements.

Underwater acoustic energy fluctuations during strong internal wave activity using a three-dimensional parabolic equation model.

The three-dimensional Monterey-Miami parabolic equation model is used to simulate a nonlinear internal wave (NIW) crossing the sound field in a shallow water environment. The impetus for this research stems from acoustic measurements taken during the Shallow Water '06 (SW06) field experiment, where a NIW traversed the water column such that soliton wavecrests were nearly parallel to the source-receiver path. Horizontal refraction effects are important in this scenario. A sound speed profile adapted from experimental SW06 data is used to simulate the NIW, assuming variations along the wavecrests (e.g., curvature) are negligible. Broadband and modal energy metrics show acoustic fluctuations due to internal wave activity. Repeated model runs simulate the NIW crossing the parabolic equation (PE) field over space and time. Statistical analysis shows the PE data are best fit by a lognormal distribution but tends to an exponential distribution during certain scenarios. Small angle differences between the acoustic track and the propagating NIW cause substantial differences in energy distribution throughout the PE field. While refraction effects due to the leading edge of the NIW's arrival are important in all cases, the impacts of focusing and defocusing in the perfectly parallel case dominate the field fluctuations. In the non-parallel case, the strong fluctuations introduced by the passage of the NIW are of similar order to the refraction off the leading edge.

Azimuthal and temporal sound fluctuations on the Chukchi continental shelf during the Canada Basin Acoustic Propagation Experiment 2017.

The Shallow Water Canada Basin Acoustic Propagation Experiment was conducted on the Chukchi Sea continental shelf from October 2016 to November 2017. The experimental goals were to access (1) long-range (basin-scale) and (2) short-range (shallow-water) spatial and temporal energy variation. This letter focuses on a 20-dB energy change of acoustic signals in the frequency band 700-1100 Hz from June to August 2017 occurring along two shallow-water tracks from a common source, correlated with the occurrence of an oceanographic event in the top 150-m water column due to a Pacific Water outflow from the Bering Sea and retreat of the Marginal Ice Zone.

The Airy phase of explosive sounds in shallow water.

The Airy phase is identified in the received signals from explosive charges deployed in a shallow water acoustic experiment conducted in the New England Mudpatch region during the spring of 2017. Measured and modeled time-frequency dispersion curves are compared and a geoacoustic sensitivity study utilizing marginal probability distributions for the sound speed in five sediment layers is performed. The analysis suggests that inclusion of the Airy phase frequency and arrival time in a geoacoustic-inversion method could lower the uncertainty of sound speed parameter estimation in a multi-layer sediment as compared to methods that do not include the Airy phase structure.

Geoacoustic inversion using low frequency broadband acoustic measurements from L-shaped arrays in the Shallow Water 2006 Experiment.

When using geoacoustic inversion methods, one objective function may not result in a unique solution of the inversion problem because of the ambiguity among the unknown parameters. This paper utilizes acoustic normal mode dispersion curves, mode shapes, and modal-based longitudinal horizontal coherence to define a three-objective optimization problem for geoacoustic parameter estimation. This inversion scheme is applied to long-range combustive sound source data obtained from L-shaped arrays deployed on the New Jersey continental shelf in the summer of 2006. Based on the sub-bottom layering structure from the Compressed High-Intensity Radiated Pulse reflection survey at the experimental site, a two-layer (sand ridge overlaying a half-space basement) range-independent sediment model is utilized. The ambiguities of the sound speed, density, and depth of the sand ridge layer are partially removed by minimizing these objective functions. The inverted seabed sound speed over a frequency range of 15-170 Hz is comparable to the ones from direct measurements and other inversion methods in the same general area. The inverted seabed attenuation shows a nonlinear frequency dependence expressed as αb=0.26f1.55(dB/m) from 50 to 500 Hz or αb=0.32f1.65(dB/m) from 50 to 250 Hz, where f is in kHz.

Arrival-time fluctuations of coherent reflections from surface gravity water waves.

Arrival time fluctuations of coherent reflections from surface gravity waves are examined. A two-dimensional ray model with an evolving rough sea surface is used to explain the mechanism and formation of the deterministic striation patterns due to the surface reflection. Arrival time predictions from the ray model match qualitatively well with the measurements from bidirectional acoustic transmissions in a water depth of 100 m.

Three-dimensional mapping of evolving internal waves during the Shallow Water 2006 experiment.

Detailed knowledge of sound speed profiles and the sound speed profile's spatial and temporal variability resulting from internal waves (IWs) are indispensable to investigating significant acoustic field fluctuations in shallow water. A strategy to obtain a time-varying, three-dimensional (3D) IW temperature field is presented. It uses two types of simultaneous measurements: dense observations from a farm of thermistor strings and IW surface expressions from a ship-based radar. Using data from the Shallow Water 2006 experiment, the temperature field, over multiple kilometers in range, was reconstructed and, fed to a 3D acoustic model to demonstrate IW impacts on acoustic propagation.

Coherent reflection from surface gravity water waves during reciprocal acoustic transmissions.

During a recent experiment in Kauai, Hawaii, reciprocal transmissions were conducted between two acoustic transceivers mounted on the seafloor at a depth of 100 m. The passage of moving surface wave crests was shown to generate focused and intense coherent acoustic returns, which had increasing or decreasing delay depending on the direction of propagation relative to the direction of surface wave crests. It is shown that a rough surface two-dimensional parabolic equation model with an evolving sea surface can produce qualitative agreement with data for the dynamic surface returns.

Time reversal acoustic communication for multiband transmission.

In this letter, multiband acoustic communication is proposed to access a relatively wide frequency band. The entire frequency band is divided into multiple separated sub-bands, each of which is several kilohertz in width. Time reversal decision feedback equalizers are used to compensate for inter-symbol interference at each sub-band. The communication scheme was demonstrated in a shallow water acoustic experiment conducted in Kauai, Hawaii during the summer of 2011. Using quadrature phase-shift keying signaling at four sub-bands over the frequency band of 10-32 kHz, a data rate of 32 k bits/s was achieved over a 3 km communication range.

Frequency dependent beating patterns and amplitude increase during the approach of an internal wave packet.

A frequency-dependent beating pattern in the spectrogram of broadband signals transmitted during the approach of an internal wave packet to an acoustic propagating path is reported. An analytical expression relating the acoustic signal measurements and environmental parameters under certain conditions is obtained. Three-dimensional parabolic equation modeling results compare well with Shallow Water 2006 experiment data.

Time reversal multiple-input/multiple-output acoustic communication enhanced by parallel interference cancellation.

Multiple-input/multiple-output (MIMO) techniques can lead to significant improvements of underwater acoustic communication capabilities. In this paper, receivers based on time reversal processing are developed for high frequency underwater MIMO channels. Time reversal followed by a single channel decision feedback equalizer, aided by frequent channel updates, is used to compensate for the time-varying inter-symbol interference. A parallel interference cancellation method is incorporated to suppress the co-channel interference in the MIMO system. The receiver performance is demonstrated by a 2008 shallow water experiment in Kauai, Hawaii. In the experiment, high frequency MIMO signals centered at 16 kHz were transmitted every hour during a 35 h period from an 8-element source array to a wide aperture 16-element vertical receiving array at 4 km range. The interference cancellation method is shown to generate significant performance enhancement, on average 2-4 dB in the output signal-to-noise ratio per data stream, throughout the 35 h MIMO transmissions. Further, communication performance and achieved data rates exhibit significant changes over the 35 h period as a result of stratification of the water column.

Acoustic multipath arrivals in the horizontal plane due to approaching nonlinear internal waves.

Simultaneous measurements of acoustic wave transmissions and a nonlinear internal wave packet approaching an along-shelf acoustic path during the Shallow Water 2006 experiment are reported. The incoming internal wave packet acts as a moving frontal layer reflecting (or refracting) sound in the horizontal plane. Received acoustic signals are filtered into acoustic normal mode arrivals. It is shown that a horizontal multipath interference is produced. This has previously been called a horizontal Lloyd's mirror. The interference between the direct path and the refracted path depends on the mode number and frequency of the acoustic signal. A mechanism for the multipath interference is shown. Preliminary modeling results of this dynamic interaction using vertical modes and horizontal parabolic equation models are in good agreement with the observed data.

Delay and Doppler spreads in underwater acoustic particle velocity channels.

Signal processing and communication in acoustic particle velocity channels using vector sensors are of interest in the underwater medium. Due to the presence of multiple propagation paths, a mobile receiver collects the signal with different delays and Doppler shifts. This introduces certain delay and Doppler spreads in particle velocity channels. In this paper, these channel spreads are characterized using the zero-crossing rates of channel responses in frequency and time domain. Useful expressions for delay and Doppler spreads are derived in terms of the key channel parameters mean angle of arrival and angle spread. These results are needed for design and performance prediction of systems that utilize underwater acoustic particle velocity and pressure channels.

Impact of source depth on coherent underwater acoustic communications.

A recent paper [Song et al., J. Acoust. Soc. Am. 123, 856-865 (2008)] investigated ocean variability impact on coherent underwater acoustic communications (8-16 kHz) for a single near-seafloor transmitter in shallow water during an extended period (27 h). This letter extends that investigation to various source depths and receiver subarrays. Specifically, the middle water column source, which is either in or out of the thermocline, experiences performance variability of 6-7 dB in terms of output signal-to-noise ratio. Further, the source below the thermocline consistently outperforms the source above the thermocline when the receiver subarray is located below the thermocline.

Effect of reflected and refracted signals on coherent underwater acoustic communication: results from the Kauai experiment (KauaiEx 2003).

The performance of a communications equalizer is quantified in terms of the number of acoustic paths that are treated as usable signal. The analysis uses acoustical and oceanographic data collected off the Hawaiian Island of Kauai. Communication signals were measured on an eight-element vertical array at two different ranges, 1 and 2 km, and processed using an equalizer based on passive time-reversal signal processing. By estimating the Rayleigh parameter, it is shown that all paths reflected by the sea surface at both ranges undergo incoherent scattering. It is demonstrated that some of these incoherently scattered paths are still useful for coherent communications. At range of 1 km, optimal communications performance is achieved when six acoustic paths are retained and all paths with more than one reflection off the sea surface are rejected. Consistent with a model that ignores loss from near-surface bubbles, the performance improves by approximately 1.8 dB when increasing the number of retained paths from four to six. The four-path results though are more stable and require less frequent channel estimation. At range of 2 km, ray refraction is observed and communications performance is optimal when some paths with two sea-surface reflections are retained.