Source Range Estimation Research Articles

Source Range Estimation Using Linear Frequency-Difference Matched Field Processing in a Shallow Water Waveguide

Matched field processing (MFP) is an established technique for source localization in known multipath acoustic environments. Unfortunately, in many situations, imperfect knowledge of the actual propagation environment and sidelobes due to modal interference prevent accurate propagation modeling and source localization via MFP. To suppress the sidelobes and improve the method’s robustness, a linear frequency-difference matched field processing (LFDMFP) method for estimating the source range is proposed. A two-neighbor-frequency high-order cross-spectrum between the measurement and the replica of each hydrophone of the vertical line array is first computed. The cost function can then be derived from the dual summation or double integral of the high-order cross-spectrum with respect to the depth of the hydrophones and the candidate sources of the replicas, where the range that corresponds to the minimum is the optimal estimation. Because of the larger modal interference distances, LFDMFP can efficiently provide only one optimal range within the same range search interval rather than some conventional matched field processing. The efficiency of the presented method was verified using simulations and experiments. The LFDMFP unambiguously estimated the source range in two experimental datasets with average relative errors of 2.2 and 1.9%.

Estimation of Source Range and Location Using Ship-Radiated Noise Measured by Two Vertical Line Arrays with a Feed-Forward Neural Network

Machine learning-based source range estimation is a promising method for enhancing the performance of tracking both the dynamic and static positions of targets in the underwater acoustic environment using extensive training data. This study constructed a machine learning model for source range estimation using ship-radiated noise recorded by two vertical line arrays (VLAs) during the Shallow-water Acoustic Variability Experiment (SAVEX-15), employing the Sample Covariance Matrix (SCM) and the Generalized Cross Correlation (GCC) as input features. A feed-forward neural network (FNN) was used to train the model on the acoustic characteristics of the source at various distances, and the range estimation results indicated that the SCM outperformed the GCC with lower error rates. Additionally, array tilt correction using the array invariant-based method improved range estimation accuracy. The impact of the training data composition corresponding to the bottom depth variation between the source and receivers on range estimation performance was also discussed. Furthermore, the estimated ranges from the two VLA locations were applied to localization using trilateration. Our results confirm that the SCM is the more appropriate feature for the FNN-based source range estimation model compared with the GCC and imply that ocean environment variability should be considered in developing a general-purpose machine learning model for underwater acoustics.

Performance analysis of passive acoustic ranging using a single hydrophone in shallow waters

Passive source range estimation is one of the important research fields on the sonar system. Recent research has carried out range estimation using single hydrophone. This study presents the result of analyzing the impact of environmental conditions on range estimation performance when estimating the range of a moving source with a single hydrophone. Range estimation of a moving source using a single hydrophone estimates source velocity from the cross-correlated fields of the signal measured at time intervals. This method of estimating the source range from estimated source velocity is used. Therefore, it is important to accurately estimate the source velocity for ranging, and the accuracy of estimating the source velocity is affected by the propagating mode characteristics in waveguide. This study investigates the changes in velocity estimation performance of a moving source according to various underwater environment and mode formation characteristics through simulation analysis and discusses ways to reduce velocity estimation error. [Work supported by KRIT (Contract No. 22-305-B00-001).]

Automatic detection and 2D localization using a network of unsynchronized passive acoustic sensors in a dispersive waveguide

The time-frequency positions of modal dispersion curves in shallow-water low-frequency impulsive signals are strongly dependent on source-receiver range, making them suitable for range-based localization. Here, we apply a temporal convolutional network (TCN) to spectrograms estimated from individual sensors in an array of unsynchronized hydrophones to simultaneously detect dispersive signals and produce source-range estimates. The TCN is trained on simulated signals generated over a spatial grid and various environmental parameters using the adiabatic approximation for normal modes. Assuming that the number of unique sources is unknown, range measurements from the same source across different sensors are simultaneously associated and used in localization. To accomplish this automatically, the proposed method considers all unique combinations of range measurements from every collection of k sensors. For every range measurement combination, if the location estimate generated using each subcombination of k-1 measurements is within a certain threshold of the remaining measurement, the whole collection is labeled as group-k consistent. All such groups of measurements are represented as neighboring nodes in a graph, and strongly connected components are used to calculate the final source location estimates. The whole detection/localization method is validated using both simulated and experimental marine data. [Work supported by NDSEG and ONR].

Deep unsupervised adversarial domain adaptation for underwater source range estimation

In this study, an underwater source range estimation method based on unsupervised domain adaptation (UDA) is proposed. In contrast to traditional deep-learning frameworks using real-world data, UDA does not require labeling of the measured data, making it more practical. First, a classifier based on a deep neural network is trained with labeled simulated data generated using acoustic propagation models and, then, the adaptive procedure is applied, wherein unlabeled measured data are employed to adjust an adaptation module using the adversarial learning algorithm. Adversarial learning is employed to alleviate the marginal distribution divergence, which reflects the difference between the measured and theoretically computed sound field, in the latent space. This divergence, caused by environmental parameter mismatch or other unknown corruption, can be detrimental to accurate source localization. After the completion of the adaptive procedure, the measured and simulated data are projected to the same space, eliminating distribution discrepancy, which is beneficial for source localization tasks. Experimental results show that range estimation based on UDA outperforms the match-field-processing method under four scenarios of few snapshots, few array elements, low signal-to-noise ratio, and environmental parameter mismatch, verifying the robustness of the method.

Optimal sampling strategies for seabed classification and source localization with Gaussian processes and machine learning

Workshop ’97 data are employed for seabed classification and source range estimation. The data are acoustic fields computed at vertically separated receivers for various ranges and three different environments. Gaussian Processes are applied for denoising the data and predicting the field at virtual receivers, sampling the water column densely within the array aperture. The enhanced fields are then used in combination with machine learning in order to map the signals to one of 15 sediment-range classes (corresponding to three environments and five ranges). In prior work, the classification results after using Gaussian Processes for denoising were demonstrated to be superior to those when noisy workshop data are employed. Here, we explore optimal sampling strategies (e.g., nonuniform sampling, subsampling) for inducing sparsity in the correlation matrices that are based on hydrophone locations, and compare these with uniform sampling that was used in our prior work.

Understanding and applications of striation-based beamforming in underwater acoustics

Utilizing the shallow water acoustic interference phenomenon in the range-frequency domain, striation-based beamforming (SBF) was proposed to estimate waveguide invariant and enhance active sonar output. In the past decade, SBF has also been applied to enhance LOFARgrams to improve tracking performance. In this paper, the correlation theory is used to understand SBF, and some further applications of SBF in underwater acoustic problems are also discussed. It shows that SBF essentially selects the optimal signal components with the best correlation. During the process, the correlation of array signals is improved and the ability to converge signals arriving via different paths is achieved to ensure array gain. The level of gain improvement is related to the direction of striation-based processing. Hence, a cost function can be constructed to facilitate the implementations of SBF in tasks such as striation slope estimation and source range estimation. Additionally, the SBF technique can be further applied to estimate the Green's function by introducing the cross-correlation preprocessing. Utilizing the mathematical connection between SBF and CBF, SBF can also be naturally extended to other beamforming algorithms and improve their performance, such as the subspace algorithms. The above is explained using simulation and experimental results.

Source range and depth estimation of propeller cavitation bubble collapse transients in a multipath environment

An experiment is conducted in which a wide aperture array, which is proximate to the sea floor in water 20 m deep, records transits of two motorboats. Occasionally, acoustic transients, which are attributed to collapsing propeller cavitation bubbles, are observed to be superimposed on the continuous broadband radiated noise component of each of the array’s three widely spaced hydrophones. By measuring the time differences of arrival of the transient signals propagating via the direct path at adjacent pairs of hydrophones, passive ranging by wavefront curvature is used to estimate the range and bearing of the transient source. Next, the source depth is estimated by measuring the multipath delay of the surface reflected transient signal. Also, the spectrogram of a single hydrophone’s output is observed to display a Lloyd’s mirror interference pattern due to the direct propagation path of the broadband radiated noise combining with the indirect propagation path signal reflected from the sea floor. The reciprocal of the frequency difference between adjacent destructive interference fringe minima is equal to the multipath time delay. This multipath delay, which equates to the quefrency of the cepstrum’s first rahmonic, contains source range information. The results of the two passive ranging methods are presented.

Range estimation in a weakly range-dependent waveguide in the South China Sea using single hydrophone and multi-task trained deep neural network

In this paper, an underwater source range estimation method using a multi-task trained deep neural network is proposed. Unlike existing machine learning frameworks, in this case, the physical quantities utilized for the training and testing are not identical. The training data parameter is the phase of the complex sound pressure, which is independent of the source and receiver depth and can be generated easily by an acoustic propagation model with only a few environmental parameters, decreasing computational costs significantly. The testing data parameter is the realistically-measured complex sound pressure. Instead of simply using the range as the only supervision signal, two other parameters — the phase mode and the bottom sound speed — are applied as auxiliary labels during the training process, embedding additional information about the sound field into the deep model. The projection between the sound field features and the range extracted from the training data by the deep model is applied to range estimation task on testing data. The effectiveness of the proposed approach is verified for a weakly range-dependent waveguide in the South China Sea by localizing several upsweep signals within a frequency band of 20–200 Hz. Testing results illustrate that the proposed method outperforms the classical matched impulse response approach and deep-learning-based single-label trained models, with the mean absolute percentage error (MAPE) decreasing by 24.63% and 59.46%, respectively. The mechanism of promotion is explained by comparing visualized hidden features of the neural network using t-distributed stochastic neighbor embedding (t-SNE) between the single-task and multi-task approach, showing that the latter helps to construct a more reasonable data manifold in high-dimensional space and thus ensures better performance of the deep model. We also fine-tune the network with a few realistic samples and further improve the performance by up to 25.34% against the baseline.

Prefiltered Striation-Based Beamforming for Range Estimation of Multiple Sources

The element–frequency acoustic intensity of a horizontal line array with a sufficient aperture exhibits interference striation patterns, which can be used for source range estimation without prior environmental information. Under multisource scenarios, the interference striations of the sources overlap with each other, leading to great difficulty in utilizing the information of striations. In this paper, the wavenumber filtering method is applied to each sensor of the horizontal line array to extract the surface-reflected–bottom-reflected modes and reconstruct the recognizable interference spectrogram for each source. Then, via beamforming along the striations, the source ranges can be estimated individually with little prior environmental information and without the long-time observation of moving sources. The required sensor spacing is analyzed, and the spatial filtering capabilities for a single source from different bearings and two sources for which azimuth angles are close to each other have also been investigated. The simulation results indicate that the proposed algorithm can estimate ranges of multiple sources within 25 km, with relative errors of less than 4%.

Source Range Estimation with Horizontal Arrays Using Frequency Difference Matched Field Processing in Shallow Water

Due to the modal interference, the ambiguity function of conventional matched field processing (CMFP) exists periodic solutions. To solve this problem, a frequency difference matched field processing (FDMFP) method with horizontal linear array (HLA) for estimating the source range is proposed. The method calculates the ambiguity function of two-neighbor-frequency between the measurement and the replica of each hydrophone, and then computes the high-order cross-spectrum. The range corresponding to the maximum of the ambiguity function is the optimal estimation. Because of the larger modal interference distances, the FDMFP can efficiently provide one-only optimal range instead of some of CMFP.

A method of underwater sound source range estimation without prior knowledge based on single sensor in shallow water

Introduction: The lack of prior knowledge of the marine environment increases the difficulty of passive ranging of underwater sound sources by using a single hydrophone. The dispersion curve of the normal mode contains extensive marine environmental information, which can be extracted without prior knowledge, but the characteristics of dispersion curves of different modes vary, and the mode order cannot be determined from the received data.Methods: Herein, a method based on a single hydrophone that can jointly identify the mode order and estimate the propagation range in unknown marine environment is proposed. The method uses Bayesian theory as the main methodology and is applicable to broadband pulse sound sources in shallow seas with long-range propagation. The dispersion curves extracted from the data and those calculated by the dispersion formula are the input signal and the replica of the methods, respectively. Accurate identification of the normal mode order and estimation of the propagation range can be achieved by establishing the joint cost function.Results: In the case of unknown a priori knowledge of the marine environment, the method enables rapid inversion, is tolerant to environmental parameter mismatch, and is low cost and practical.Discussion: The simulation and measured data analysis results demonstrate the accuracy and validity of the method. The measured data contains linear frequency modulation impulse source signal and explosion sound source signals, and the mean relative error of range estimation is less than 5%.

Semi-supervised Learning Method for Source Range Estimation in Shallow Water

Data-driven machine learning methods in recent years have shown great vitality in underwater acoustic, especially in the localization of underwater sound sources. Meanwhile, there exists a great difficulty in the acquisition and processing of underwater acoustic data, which caused the lack of label data. This paper proposes a two-step semi-supervised learning-based range estimation method for shallow ocean sound sources, which utilizes semi-supervised learning in deep learning to alleviate the problem of limited access to underwater acoustic label data and relatively abundant unlabelled data in practical scenarios. The performance of the method has been validated on SWellEx-96 experiment data.

Machine-learning-based simultaneous detection and ranging of impulsive baleen whale vocalizations using a single hydrophone.

The low-frequency impulsive gunshot vocalizations of baleen whales exhibit dispersive propagation in shallow-water channels which is well-modeled by normal mode theory. Typically, underwater acoustic source range estimation requires multiple time-synchronized hydrophone arrays which can be difficult and expensive to achieve. However, single-hydrophone modal dispersion has been used to range baleen whale vocalizations and estimate shallow-water geoacoustic properties. Although convenient when compared to sensor arrays, these algorithms require preliminary signal detection and human labor to estimate the modal dispersion. In this paper, we apply a temporal convolutional network (TCN) to spectrograms from single-hydrophone acoustic data for simultaneous gunshot detection and ranging. The TCN learns ranging and detection jointly using gunshots simulated across multiple environments and ranges along with experimental noise. The synthetic data are informed by only the water column depth, sound speed, and density of the experimental environment, while other parameters span empirically observed bounds. The method is experimentally verified on North Pacific right whale gunshot data collected in the Bering Sea. To do so, 50 dispersive gunshots were manually ranged using the state-of-the-art time-warping inversion method. The TCN detected these gunshots among 50 noise-only examples with high precision and estimated ranges which closely matched those of the physics-based approach.

Learning based method for near field acoustic range estimation in spherical harmonics domain using intensity vectors

Near-field acoustic range estimation is considered one of the least explored research problems in digital signal processing under noise and reverberant conditions. This letter develops a new learning-based range estimation technique utilizing the spherical harmonics intensity (SH-INT) coefficients. The conventional range estimation in the spherical harmonics (SH) domain relies on the pressure coefficients. However, at high frequencies, these coefficients of different order and range overlap and hinder the accuracy of range estimation. On the contrary, the SH-INT coefficients are well distinguished at high frequencies for various orders and ranges, making these features favorable for accurate range estimation using learning algorithms. Since the SH-INT coefficients in the radial direction are independent of the source signal and vary with range, a convolutional neural network (CNN) model has been adopted to map the SH-INT coefficients with the range classes. The performance of the proposed spherical harmonic intensity (SH-INT) features in the context of near-field range estimation is validated by conducting exhaustive experiments on simulated and real data. Further, the error in near-field source range estimates is characterized using root mean square error (RMSE) criteria. The results are impactful and encourage the use of this method for practical near-field source range estimation applications.

Seabed classification and source localization with Gaussian processes and machine learning

Workshop ’97 data are employed for seabed classification and source range estimation. The data are acoustic fields computed at vertically separated receivers for various ranges and three different environments. Gaussian Processes are applied for denoising the data and predicting the field at virtual receivers, sampling the water column densely within the array aperture. The enhanced fields are then used in combination with machine learning in order to map the signals to one of 15 sediment-range classes (corresponding to three environments and five ranges). The classification results after using Gaussian Processes for denoising are superior to those when noisy workshop data are employed.

Seabed classification and source localization with Gaussian processes and machine learning.

Workshop '97 data are employed for seabed classification and source range estimation. The data are acoustic fields computed at vertically separated receivers for various ranges and different environments. Gaussian processes are applied for denoising the data and predicting the field at virtual receivers, sampling the water column densely within the array aperture. The enhanced fields are used in combination with machine learning to map the signals to one of 15 sediment-range classes (corresponding to three environments and five ranges). The classification results after using Gaussian processes for denoising are superior to those when noisy workshop data are employed.

An overview of array invariant for source-range estimation in shallow water.

Traditional matched-field processing (MFP) refers to array processing algorithms, which fully exploit the physics of wave propagation to localize underwater acoustic sources. As a generalization of plane wave beamforming, the "steering vectors," or replicas, are solutions of the wave equation descriptive of the ocean environment. Thus, model-based MFP is inherently sensitive to environmental mismatch, motivating the development of robust methods. One such method is the array invariant (AI), which instead exploits the dispersion characteristics of broadband signals in acoustic waveguides, summarized by a single parameter known as the waveguide invariant β. AI employs conventional plane wave beamforming and utilizes coherent multipath arrivals (eigenrays) separated into beam angle and travel time for source-range estimation. Although originating from the ideal waveguide, it is applicable to many realistic shallow-water environments wherein the dispersion characteristics are similar to those in ideal waveguides. First introduced in 2006 and denoted by χ, the dispersion-based AI has been fully integrated with β. The remarkable performance and robustness of AI were demonstrated using various experimental data collected in shallow water, including sources of opportunity. Further, it was extended successfully to a range-dependent coastal environment with a sloping bottom, using an iterative approach and a small-aperture array. This paper provides an overview of AI, covering its basic physics and connection with β, comparison between MFP and AI, self-calibration of the array tilt, and recent developments such as adaptive AI, which can handle the dependence of β on the propagation angle, including steep-angle arrivals.

Adaptive array invariant for source-range estimation in shallow water

The array invariant (χ) developed for source-range estimation in shallow water is based on the broadband dispersion characteristics in ideal waveguides, summarized by the waveguide invariant, β = cos2θ. This approach relies on the waveguide invariant being constant (e.g., β = 1), valid for small propagation angles, but introduces a significant range error as the propagation angle increases. In this talk, the standard array invariant is extended to fully incorporate the angle dependence of the waveguide invariant, referred to as the adaptive array invariant, denoted by χβ = χ/β. The remarkable performance enhancement is demonstrated (1) using a 56.25 m vertical array for a surface ship (200–900 Hz) in approximately 100-m shallow water and (2) using a 2.8 m vertical array for a broadband source (0.5–3.5 kHz) towed on a continental slope where the water depth varies from 87.5 to 55 m over a 5-km range.

The waveguide invariant for a Pekeris waveguide

A closed-form waveguide invariant β for a Pekeris waveguide is derived. It is based on the modal Wentzel–Kramers–Brillouin (WKB) dispersion equation and implicit differentiation, in conjunction with the concept of the angle-dependent “effective boundary depth,'' ΔH(θ). First, an explicit expression for β(m,n) between mode pairs is derived assuming an ideal waveguide of the effective waveguide depth H+ΔH(θ), which provides an excellent agreement with the exact value calculated for the Pekeris waveguide of depth H using KRAKEN. Then, a closed-form expression for a group of adjacent modes is derived: β = (H + ΔH)/(H/cos2θ − ΔH), which reduces to β = cos2θ as ΔH/H ≪ 1, the analytical expression for an ideal waveguide. For applications to source-range estimation, the waveguide invariant β is used to derive the theoretical beam-time migration (i.e., dispersion curve) in a Pekeris waveguide.