Interference-structure-based passive source localization provides computational efficiency and physical interpret-ability but suffers performance degradation for weak acoustic targets,where low signal-to-noise ratios(SNRs)obscure the observable interference patterns necessary for reliable depth estimation.Narrowband time-window limitations and environ-mental mismatches further introduce blind zones and nonlinear distortions between the source depth and the measured interference structure.To mitigate these effects,this study proposes a hybrid depth estimation method that integrates arrival-angle interference features with a deep learning(DL)framework based on a residual network.Sound intensity in the beam domain is employed to exploit multidimensional and nonlinear relationships in the acoustic field,thereby enhancing robustness under weak target and low SNR conditions.Simulation and sea experiment results demonstrate that the proposed method achieves improved depth estimation accuracy compared with conventional interference-based techniques.The findings indicate that the hybrid interference-DL approach effectively extends the applicability of interference-structure-based localization to weak acoustic targets.

Vertical line arrays (VLAs) are widely used for passive ocean monitoring, but colored backgrounds, background power-scale drift, and short-window operation can make covariance-based detection unstable under fixed false-alarm constraints. This paper proposes AIRM-Shape, an affine-invariant Riemannian metric (AIRM) covariance-shape detector for steering-free short-window VLA alerting. Each sample covariance matrix is mapped to a dimensionless Hermitian positive definite (HPD) covariance-shape matrix through trace normalization and shape-space diagonal loading. A noise-only background center is learned by the AIRM/Karcher mean, and detection is performed using the squared AIRM distance with empirical H0 quantile calibration. KRAKEN-based simulations compare AIRM-Shape with raw-domain statistics, Euclidean and eigenvalue shape variants, a normalized Bartlett constant false-alarm rate (CFAR) detector, and H0-pre-whitened energy. The results show that AIRM-Shape is not a universal replacement for ideal known-steering detectors, but it provides more stable offline time-fixed threshold operation under colored noise, global background-scale drift, and steering mismatch. SWellEx-96 S5 continuous-wave (CW) data further demonstrate fixed-threshold covariance-shape monitoring using proxy-H0 frequency-bin calibration. The method is therefore positioned as a scale-robust, steering-free front end for short-window ocean alerting rather than as a complete localization or classification system.

Continuous wide-area detection and communication are vital for deep-sea applications. However, deep-sea acoustic shadow zones cause significant energy attenuation, severely limiting the effective coverage of underwater acoustic systems. Phased array emission combined with bottom reflection exhibits the potential to achieve acoustic shadow-zone ensonification; however, systematic experimental validation remains limited, and observational capabilities for large-scale acoustic field characterization are constrained. To address these issues, this paper conducts a theoretical analysis and sea trial verification of the phased array emission sound field, proposing a wide-area observation method utilizing an underwater glider as a passive reception platform. In a deep-sea experiment conducted in the South China Sea at a depth of approximately 4310 m, shadow-zone ensonification within a horizontal range of 14.5-32.5 km and a depth interval of 58-889 m was achieved and experimentally verified by steering the phased array emission angle. Experimental results indicate a distinct negative correlation between the optimal phased array emission angle and the acoustic energy focusing distance; simultaneously, an acoustic energy gain of approximately 30 dB was achieved at a near-field range of 14.5 km, and an energy enhancement exceeding 14 dB was maintained in far-field and greater-depth regions.

To address the high energy consumption of underwater acoustic communication (UAC) signal recognition and the difficulty of long-term operation on embedded chips, this paper proposes a low-power recognition method based on spiking neural networks (SNNs). By designing dual-branch architecture, the main branch is used for recognition, while the auxiliary branch provides reconstruction-based interference suppression during training, thereby improving the model’s robustness under noisy and reverberant conditions. Integer Leaky Integrate-and-Fire (ILIF) neurons are introduced to enable efficient training and low-power spike-driven inference. Next, a Channel Rearrangement (CR) module is used to reorganize the input spectrum by dividing it into several partially overlapping sub-bands, further reducing power consumption. Finally, an Integer Activation Random Dropping (IARD) method is employed to retain a portion of floating-point activations during training, thereby alleviating the optimization difficulty introduced by fully integer activations and improving the model's generalization capability. The lake trial results show that, compared to the single-branch Artificial Neural Network (ANN) network architecture, the dual-branch ILIF-SNN-CR-IARD method reduces power consumption by approximately 98.17% while achieving a recognition accuracy of 99.04%. In various low-SNR sea trial scenarios, it maintains recognition performance comparable to that of the ANN, demonstrating the feasibility and practical value of the dual-branch ILIF-SNN-CR-IARD architecture for efficient, low-power recognition in complex underwater acoustic environments.

This paper proposes an energy-efficient spiking neural network (SNN) for underwater acoustic communication (UAC) signal recognition. First, in the designed artificial neural network (ANN) model, integer leaky integrate-and-fire (I-LIF) neurons are introduced, with integer activation realized through the extension of virtual time-steps. Spike-driven computations are performed during the inference phase. Additionally, a channel-wise reorientation (CR) module that reduces computational complexity by partitioning the input spectrum into multiple overlapping sub-bands is introduced. Furthermore, the integer activation dropping (IADP) method is proposed to alleviate the difficulty of optimization by retaining some floating-point activation, thereby increasing weight flatness and generalizability. Experimental results verify the effectiveness of the proposed method, providing a reference for achieving efficient and low-power communication signal recognition on resource-constrained platforms.

Passive vertical line arrays (VLAs) often require fixed-cardinality channel reduction, but under MVDR beamforming the resulting subarray design becomes highly sensitive to snapshot-limited covariance estimation and steering/replica mismatch. This paper develops an output-SINR-driven robust subarray selection framework for passive VLAs. The problem is formulated through an extraction-matrix representation of the selected VLA aperture, shrinkage-and-loading covariance regularization, and an objective that combines MVDR output SINR with a white-noise-gain safeguard. The binary cardinality-constrained design is relaxed to a bounded continuous space and mapped to a feasible M-element subarray by a deterministic Top-M projection, thereby avoiding exhaustive combinatorial search. To solve the resulting projected optimization problem under a fixed evaluation budget, an improved whale migration algorithm (ImWMA) is developed by incorporating differentiated leader perturbation, hybrid follower updating, diversity-triggered budget-preserving injection, and adaptive step-size control. The test results show that the proposed framework consistently improves upon the original WMA, remains competitive with representative deterministic selection baselines, and markedly reduces run-to-run variability. These results support robust and reproducible subarray design for passive VLA beamforming in uncertain underwater environments.

Continuous phase modulation (CPM), which is widely used in aviation telemetry and satellite communications, may help improve the performance of underwater acoustic (UWA) communication systems owing to its high spectral and power efficiency. However, applying conventional frequency-domain equalization (FDE) algorithms to CPM signals over time-varying UWA channels considerably degrades performance. Moreover, time-domain equalization algorithms often rely on excessive approximations for symbol detection, compromising overall reception. This study presents an iterative-detection–based time-domain adaptive decision feedback equalization (ID-TDADFE) algorithm that tracks channel variations through symbol-by-symbol detection. The symbol detection in ID-TDADFE fully considers the inherent coding gain of CPM signals can be cascaded with an adaptive equalizer, and enhances symbol detection performance by utilizing joint probability estimation. Numerical simulations with minimum-shift keying (MSK) and Gaussian MSK signals demonstrated that ID-TDADFE significantly improved communication performance over a time-varying UWA channel within one or two iterations. In a sea trial for experimental verification, ID-TDADFE reduced bit errors by 45.08% and 51.8% in the first and second iterations, respectively, compared to FDE.

This paper proposes a frequency hopping binary frequency shift keying underwater acoustic (UWA) communication system, where a denoising diffusion probabilistic model (DDPM) and a convolutional neural network (CNN) are sequentially used for signal reconstruction and signal demodulation, respectively. Unlike the deep transfer learning (DTL)-based system, this system employs a DDPM to process the received Mel-spectrogram, reconstructing the distorted Mel-spectrogram caused by UWA channel effects, and generating a spectrogram that approximates the transmitted signal, which is then demodulated by the CNN. The proposed system outperforms conventional systems and achieves performance comparable to DTL-based systems in simulation and experiment. DTL requires data samples from new scenarios to learn signal characteristics during deployment; in contrast, this method uses the generative capability of DDPM to enable direct deployment in dynamic underwater environments without additional adaptation processes, offering flexibility and suitability for complex and variable UWA propagation channels.

To address increased time delay estimation errors and reduced positioning accuracy caused by relative motion between the target and reference array elements, this paper proposes an improved underwater acoustic positioning method that integrates real-time Doppler estimation with motion and distance constraints in a joint adjustment framework. Firstly, a Doppler estimation method based on dual linear frequency modulation (LFM) combined signals is proposed. Using a decision feedback two-step estimation approach, the phase ambiguity search space is effectively reduced, enabling fast and accurate Doppler factor estimation. In complex, time-varying shallow water channel conditions, the method reduces estimation error from 10⁻3 to approximately 10⁻4. The short transmission period of the combined signal enhances its practicality. Secondly, based on the Doppler estimation results, a further improvement was proposed by using the nonlinear least squares Levenberg-Marquardt method for joint adjustment with velocity and range constraints to optimize the positioning results. The lake trial results show that the velocity estimation error of this method is approximately 0.02 m/s. After Doppler compensation, the positioning error is reduced from 5 cm to 2 cm. By conducting joint adjustment improvement with range constraints, the positioning error is further reduced to 1 cm. The experiments proved that this method can ensure the accurate positioning of underwater dynamic targets.

The prediction of ocean ambient noise is crucial for protecting the marine ecosystem and ensuring communication and navigation safety, especially under extreme weather conditions such as typhoons and strong winds. Ocean ambient noise is primarily caused by ship activities, wind waves, and other factors, and its complexity makes it a significant challenge to effectively utilize limited data to observe future changes in noise energy. To address this issue, we have designed a multi-modal linear model based on a “decomposition-prediction-modal trend fusion-total fusion” framework. This model simultaneously decomposes wind speed data and ocean ambient noise data into trend and residual components, enabling the wind speed information to effectively extract key trend features of ocean ambient noise. Compared to polynomial fitting methods, single-modal models, and LSTM multi-modal models, the average error of the relative sound pressure level was reduced by 1.3 dB, 0.5 dB, and 0.3 dB, respectively. Our approach demonstrates significant improvements in predicting future trends and detailed fittings of the data.

Existing research has referenced polar code construction methods in wireless channels to estimate the instantaneous reliability for underwater acoustic (UWA) polarized sub-channels. However, it overlooks potential mismatches between the instantaneous codebook and the actual channel state caused by complex time-varying characteristics of UWA channel, and lacks sufficient study on the evolution of sub-channel reliability. This paper extends the foundational channel polarization theory to propose a new framework for parallel UWA channel polarization based on Orthogonal Frequency Division Multiplexing (OFDM). By integrating the time-frequency statistical characteristics of UWA channels, the reliability theoretical model of polarized channels was established. And the intrinsic relationship between its evolution and channel parameters as well as system performance was deeply analyzed. Moreover, through simulation and lake trials, this paper investigates the time-frequency variation of polarized sub-channel reliability under incomplete channel polarization with finite code length. Additionally, it analyzes the impact of UWA channel parameters, such as Doppler shift and time-variation, on channel polarization distribution and polar-coded OFDM system performance. This paper provides a reference for subsequent practical polar coded OFDM communication, and streamlines the code construction process by avoiding the repeated assessments of polarized sub-channel's reliability.

To scan the horizontal range of the deep-sea shadow zone with high accuracy and efficiency, enabling active target detection, a horizontal detection range estimation method based on phased sound focusing technology is presented. The model of the phased emission sound field is derived, along with the theoretical estimation formula for the horizontal detection range, by combining the half-power beamwidth of the beam curve at the phased-emitter end with the ray propagation law of deep-sea acoustics. The corresponding relationships among each discrete phased angle, the horizontal distance, and the detection range of the focused beam boundary are detailed. The estimation formula for the horizontal detection range is then improved to account for the sea surface reflection effect, and angle optimization is applied to the transmitted beam curve to counteract the boundary error accumulation effect of the focused sound beam as the sound ray propagates. To address the issue of reduced scanning efficiency at large phased angles, the grating lobe-focused sound beam generated by phased emission array parameters is used to achieve angle scanning within the close deep-sea sound shadow zone, increasing the phased angle scanning efficiency. Finally, the deep-sea phased emission active target detection experiments were carried out, and the echo signal generated by the phased focused sound beam and received by the vertical receiving array was recorded. Numerical simulation and echo signal processing show that the high-efficiency angle scanning of the entire deep-sea sound shadow zone is achieved by the proposed estimation method.

In this study, a transdimensional particle filtering method is proposed to simultaneously estimate a geoacoustic model and its associated parameters, whose computational efficiency is improved through a mechanism of parallel calculations.The proposed method is first applied to seabed bottom loss derived from vertical array data collected in the Yellow Sea in 2016.The thickness and number of estimated sediment layers are close to those obtained in previous studies.Moreover, particle filtering has an intrinsic advantage in sequential processing, making it a straightforward method for mapping range-dependent geoacoustic properties of a horizontal array towed from a moving ship.Range-dependent inversion is conducted for a spatial varying environment, and the data are collected from an experiment conducted in the South China Sea in 2022.The inversion results suggest that the sediment layer and associated geoacoustic parameters can be effectively tracked and that the thickness and number of estimated sediment layers are consistent with those obtained by reversible jump Markov chain Monte Carlo method.

Acoustic scattering modulation caused by an undulating sea surface on the space-time dimension seriously affects underwater detection and target recognition. Herein, underwater acoustic scattering modulation from a moving rough sea surface is studied based on integral equation and parabolic equation. And with the principles of grating and constructive interference, the mechanism of this acoustic scattering modulation is explained. The periodicity of the interference of moving rough sea surface will lead to the interference of the scattering field at a series of discrete angles, which will form comb-like and frequency-shift characteristics on the intensity and the frequency spectrum of the acoustic scattering field, respectively, which is a high-order Bragg scattering phenomenon. Unlike the conventional Doppler effect, the frequency shifts of the Bragg scattering phenomenon are multiples of the undulating sea surface frequency and are independent of the incident sound wave frequency. Therefore, even if a low-frequency underwater acoustic field is incident, it will produce obvious frequency shifts. Moreover, under the action of ideal sinusoidal waves, swells, fully grown wind waves, unsteady wind waves, or mixed waves, different moving rough sea surfaces create different acoustic scattering processes and possess different frequency shift characteristics. For the swell wave, which tends to be a single harmonic wave, the moving rough sea surface produces more obvious high-order scattering and frequency shifts. The same phenomena are observed on the sea surface under fully grown wind waves, however, the frequency shift slightly offsets the multiple peak frequencies of the wind wave spectrum. Comparing with the swell and fully-grown wind waves, the acoustic scattering and frequency shift are not obvious for the sea surface under unsteady wind waves.

The intensity characteristics of deep-sea ambient noise are important parameters for the evaluation of the operating distance and signal-to-noise ratio of underwater equipment. This study conducted research on the prediction method of ambient noise intensity based on long short-term memory network. The prediction was conducted on measured deep-sea data in two scenarios, medium-long time scale and short time scale. The results show that under medium-long-time scale conditions, the temporal trend of predicted noise intensity is consistent with the real trend, but in some frequency bands, considerable errors and time delays exist. The mean of root mean square error (RMS) in the frequency band of 20 Hz-5 kHz is 4.31 dB. Under short time scale condition, the average RMS between the prediction and measured is 0.73 dB within the same frequency range. At the same time, the correlation coefficient between the predicted noise intensity curve with frequency and the true value curve is 0.96.

AbstractThe performance of Doppler velocity logs (DVLs) in terms of velocity estimate error is directly linked to the geometry of the beam and the pulse transmitted. Beyond a specific transmitted bandwidth, the phase‐shift beamformer can introduce significant errors in velocity estimation. To delineate the operating mechanism of phase‐shift errors within a phased array of acoustic DVLs, the correlation between bottom echo and velocity distribution, in conjunction with the power‐weighted function, was initially examined predicated on spectral estimation theory. Subsequently, numerical and analytical models of the Gaussian‐shaped Doppler spectrum were formulated. The models are employed to evaluate the velocity estimation inaccuracies attributed to phase shifts in extant DVLs, and the comparative results with field experiments corroborate the model's efficacy in forecasting errors. The theoretical findings evaluate the performance limitations of the current phased array transducer design and provide insights for developing new designs. Pool experimental results show that this design effectively reduces the velocity estimation error caused by phase shift under static conditions and in the presence of Doppler frequencies to a level of almost complete elimination of the error compared to conventional configurations.

The rapid development of human society has brought much noise pollution, which damage our quality of life seriously. However, it is difficult to design a material with broadband sound absorption performance. Herein, we explored composite aerogels composed of cyclodextrin, CNC and siloxane. The resulting CCA aerogels have broadband sound absorption performance. The bandwidth with a sound absorption coefficient greater than 0.8 can be over 5000 Hz (750-1250, 1750-6400 Hz), and the best sound absorption coefficient appears at the low frequency (around 768 Hz, 0.95). The CCA aerogels also show good elasticity and their mechanical properties are better than most sound-absorbing materials. Furthermore, CCA aerogels exhibit stable broadband sound absorption performance in humid environment or after compression treatment.

A vertical line array (VLA) deployed at the seabed bottom captures the arrival angle interference structure in the frequency-beam domain resulting from the Direct (D) and Surface-Reflected (SR) arrivals of a broadband source. This interference structure, sensitive to the source's depth, serves as a basis for depth estimation. In order to address limitations related to bandwidth and nonlinear errors stemming from environmental differences, and to enhance the applicability of interference structures, a hybrid source localization method based on deep learning is proposed. This method employs an optimized residual network (ORN) to effectively extract and evaluate features from the frequency-beam domain sound intensity matrix. Simulated and experimental datasets are used to test the performance of the proposed method, and results suggest that the performance of the ORN model is much better than those of multi-Fourier transform approach (MSTDE), matching field processing (MFP) and traditional Convolutional Neural Network (CNN) models.

Gaussian minimum shift keying (GMSK) stands as a specialized form of continuous phase modulation, offering inherent benefits such as a constant envelope and inherent coding gain, which enhance both spectrum utilization and power efficiency within underwater acoustic (UWA) communication. To improve symbol detection performance, a novel iterative detection algorithm is introduced and integrated with frequency-domain equalization (FDE) and time-domain adaptive decision feedback equalization (TDADFE). ID-FDE employs the correlation function to assess the joint probability of both forward and backward information, which becomes particularly valuable within the first two iterations, culminating in a substantial enhancement of symbol detection performance. ID-TDADFE handles interference caused by time-varying UWA channels through symbol-by-symbol detection and second-order digital phase-locked loops. Moreover, it rectifies the input symbols of the feedback filter via the correlation function and prior information. Using the time-varying underwater acoustic channel model for simulation, the results show that for the time-invariant channel, iterative detection can significantly reduce the bit error rate (BER) in 2 iterations, which is about 1dB at a BER of 1 × 10−4. For the time-varying channel, after third iteration, ID-TDADFE displayed better BER performance compared to ID-FDE by approximately 1.7dB at a BER of 1 × 10−4.

The underwater mobile formation network is widely used for the exploitation of the deep-ocean resources. However, the topology of the underwater mobile formation network is always changing, and it severely affects the communication quality between nodes. To solve this problem, a new Self-Organization and Routing Protocol (SORP) approach that aims to find right route paths for both broadcast packet and private packet is proposed. The SORP approach focuses on the initialization process of the underwater mobile formation network while the passing record is used to control the range of the broadcast packets and avoid the routing loops of private packets. Meanwhile, the SORP approach does not require precise clock synchronization among nodes. The OPNET software is used to evaluate the performance of the SORP approach. Simulation results reveal that the performance of the SORP approach is better in comparison with that of the legacy flooding broadcasts in terms of energy consumption and network efficiency.