Phase Distortion Estimation by DIRECT and Spline Interpolation Algorithms

This paper presents a learning-based method for source localization in the presence of directional interference under reverberant and noisy conditions. The proposed method operates on the spherical harmonic decomposition of the spherical microphone array recordings to yield spherical harmonics coefficients as the features. An attention mechanism is incorporated through a binary mask that filters out the dominant undesired source components from the features before training. A convolutional neural network is trained to map the phase and magnitude of the filtered coefficients with the location class. Hence, the objective is to develop the binary mask followed by source localization.

Subspace-based higher-order compressive sensing algorithm for raypath separation in a shallow-water waveguide

When low-spline interpolation algorithm is adopted by super-resolution reconstruction for video images, there are some defects, such as saw tooth and blur edge, if the result image is magnified. In this paper, high-order spline interpolation algorithm is introduced and it is optimized. Firstly, the common low-spline interpolation algorithms are analyzed and their shortcomings are pointed out. Then cubic spline interpolation algorithm is discussed. If the image is rotated by cubic spline interpolation algorithm, the magnified image may be not correctly displayed and the image can not be registered in super-resolution reconstruction. Finally, the cubic spline algorithm has been improved. Experimental results show that the improved cubic spline interpolation algorithm can not only eliminate the edge blur and saw tooth, but also do registration in reconstruction when image is rotating.

Underwater Acoustic Source Localization and Sounds Classification in Distributed Measurement Networks

A generalized multidimensional Wiener filter for denoising is adapted to hyperspectral images (HSIs). Commonly, multidimensional data filtering is based on data vectorization or matricization. Few new approaches have been proposed to deal with multidimensional data. Multidimensional Wiener filtering (MWF) is one of these techniques. It considers a multidimensional data set as a third-order tensor. It also relies on the separability between a signal subspace and a noise subspace. Using multilinear algebra, MWF needs to flatten the tensor. However, flattening is always orthogonally performed, which may not be adapted to data. In fact, as a Tucker-based filtering, MWF only considers the useful signal subspace. When the signal subspace and the noise subspace are very close, it is difficult to extract all the useful information. This may lead to artifacts and loss of spatial resolution in the restored HSI. Our proposed method estimates the relevant directions of tensor flattening that may not be parallel either to rows or columns. When rearranging data so that flattening can be performed in the estimated directions, the signal subspace dimension is reduced, and the signal-to-noise ratio is improved. We adapt the bidimensional straight-line detection algorithm that estimates the HSI main directions, which are used to flatten the HSI tensor. We also generalize the quadtree partitioning to tensors in order to adapt the filtering to the image discontinuities. Comparative studies with MWF, wavelet thresholding, and channel-by-channel Wiener filtering show that our algorithm provides better performance while restoring impaired HYDICE HSIs.

This paper considers the problem of blind channel estimation of multi-channel FIR filters. This is a problem arising in, for example, mobile communication systems using digital signalling. By using the orthogonality property between the noise subspace and the channel matrix, it has been shown in earlier work that the channel matrix is identifiable up to a multiplicative constant. In this article, the asymptotic properties of a subspace method using this orthogonality property is presented. An asymptotically correct weighting matrix is derived, demonstrating an attainable lower theoretical bound using the subspace estimate.

Signal processing involving source localization and detection in underwater acoustics often depends on comparing covariance or cross-spectral density matrices (CSDMs) estimated for data, replica or noise source data from passively sensed acoustic fields incident on sensor arrays. Such comparisons often depend on a measure of similarity between matrix pairs and this typically involves a Euclidean metric. While Euclidean-based metrics are ubiquitous, convenient and useful, they are not fundamental to this task. By exploiting the facts that a CSDM is Hermitian and positive semi-definite, one can interpret such matrices as points constrained to a Riemannian, rather than a Euclidean manifold, with the implication that similarity between matrix pairs should be measured using a metric consistent with the manifold’s intrinsic structure. This geometric interpretation leads to alternative matched-field processors for source localization involving the Riemannian distance as a measure of similarity and can be used to solve this inverse problem. Some implications of this non-Euclidean approach in underwater acoustics, as well as a possible extension to source detection are discussed. [Work supported by the Office of Naval Research.]

We seek to improve the signal-to-noise (SNR) ratio of clicks recorded from Blainville's beaked whales (Mesoplodon densirostris). The proposed filter is based on subspace principles and projects the noisy speech vector onto signal and noise subspaces. An estimate of the clean signal is made by retaining only the components in the signal subspace. To test the filter, a detector using the filter output has been designed for the detection of beaked whale calls. Through simulations, the proposed detector is shown to be capable of detecting most of the desired clicks but is not able to differentiate other co-existing species such as Risso's dolphins and pilot whales' that is, it efficiently detects all clicks. By combining the proposed detector with the energy ratio mapping algorithm (ERMA; Klinck and Mellinger (in press), which measures energy differences between different species, higher detection accuracy for beaked whale clicks can be achieved. The filter can also be used to improve the SNR of other marine mammal acoustic signals.

This paper looks into the problem of localization of an acoustic source in shallow sea with a sloping bottom as found in the continental shelf area. A vertical array of sensors, either floating or secured to the sea bottom, is used to receive the acoustic signals from the source including those travelling by different paths. First, a spectral matrix of the array signal is computed and it is then subjected to eigenvalue decomposition. The space spanned by the eigenvectors is partitioned into two subspaces, namely, a signal subspace spanned by the eigenvector corresponding to largest eigenvalues and a noise subspace spanned by the eigenvectors corresponding to repeated eigenvalues. The signal subspace is also spanned by a linear combination of direction vectors corresponding to all multipaths These two subspaces are orthogonal. We exploit this property of orthogonality in the estimation of range, bearing and depth of a source. Some of the significant findings are as follows:(1) Unlike in the flat bottom channel it is now possible to measure the bearing with respect to upslope or downslope direction.(2) The azimuth scan shows a sharp peak at the correct location for steep slope(>1/spl deg/) (3) A study of finite data effect has shown that the background generally remains unchanged as the number of snapshots increases. However, the signal peak amplitude increases roughly at the rate of 3dB per doubling of the number of snapshots.

The problem of passive localization of an acoustic source is a classic area of research in underwater acoustics. While the three-dimensional problem presents difficulties, most past investigations have been able to make suitable approximations to reduce the localization to one in a plane involving two dimensions. However, acoustic source localization in anechoic tanks, pressure vessels, or other similar enclosures often requires a truly three-dimensional formulation. This paper presents the equations describing source localization in three dimensions, given time delays between four receiving channels. Numerical difficulties associated with their solution are discussed, particularly those involved in the nonlinear solution of sets of intersecting hyperboloids. An alternate formulation is presented which makes use of intersecting spheres rather than hyperboloids. This reduces the numerical complexity markedly since finding intersections of closed surfaces such as spheres is inherently more stable than for open surfaces like hyperboloids. Several possible algorithms for implementing this approach are presented, and the practical limitations of each for use with nonideal data are contrasted. Results indicate that the proposed method of intersecting spheres is numerically simpler and provides a more accurate solution for passive source localization than the method of intersecting hyperboloids. [Work partially supported by NAVSEA PMS 409 and by NAVSEA 63X5.]

A deep-learning approach is proposed for simultaneous source localization and seabed classification of underwater signals generated by SUS (Signal, Underwater Sound) MK64 explosives. The charges were deployed in two separate shallow water acoustic experiments on the New England Mudpatch region in 2017 (SBCEX17) and 2022 (SBCEX22) conducted under different oceanographic conditions. The proposed approach uses a multitask learning (MTL) methodology, and convolutional neural networks (CNNs) trained on the spectrogram representation of SUS signals in the 10–199 Hz frequency band where the time-frequency modal dispersion curves exhibit distinct patterns. CNNs were trained on simulated data generated using the normal modes model, ORCA. This synthetic dataset is composed of several sound speed profiles measured in the water column in both experiments and representative seabeds ranging from muddy to sandy sediments. CNNs were tested on the signals measured in the two experiments where the source localization achieved a low mean squared error (0.35–0.8 km), and the seabed classification closely matched existing inversion results in the Mudpatch area. Notably, our method effectively handled oceanographic variations, showcasing the utilization of modal dispersion for training deep learning algorithms. [Work supported by ONR Grant N00014-21-1-2760.]

In this paper, a new information theoretic algorithm is proposed for signal enumeration in array processing. The approach is based on predictive description length (PDL) that is defined as the length of a predictive code for the set of observations. We assume that several models, with each model representing a certain number of sources, will compete. The PDL criterion is computed for the candidate models and is minimized over all models to select the best model and to determine the number of signals. In the proposed method, the correlation matrix is decomposed into two orthogonal components in the signal and noise subspaces. The maximum likelihood (ML) estimates of the angles-of-arrival are used to find the projection of the sample correlation matrix onto the signal and noise subspaces. The summation of the ML estimates of these matrices is the ML estimate of the correlation matrix. This method can detect both coherent and noncoherent signals. The proposed method can be used online and can be applied to time-varying systems and target tracking.

It is known that classical subspace-based direction-of-arrival (DOA) estimation algorithms are not straightforwardly applicable to scenarios with unknown spatially nonuniform noise. Among the state-of-the-art solutions, this problem is tackled by iterative subspace estimation algorithms to incorporate subspace-based approaches or nonlinear optimization routines to bypass the direct identification of subspaces. In this paper, the problem of DOA estimation in nonuniform noise is revisited by devising two computationally efficient proposals. It is proved herein that, if the signals are uncorrelated, the signal and noise subspaces can be directly obtained from the eigendecomposition of a reduced array covariance matrix. On the other hand, when the signals are correlated, the estimation of the noise covariance matrix is formulated into a rank minimization problem which can be approximately solved by semidefinite programming. In both cases, the signal and noise subspaces are easy to compute without iterations. Consequently, classical subspace-based algorithms can be employed to determine the DOAs. Numerical examples are provided to demonstrate the performance and applicability of the proposed methods.

This paper addresses the problem of estimating direction of arrival (DOA) and polarization parameters for vector-sensor array in MA noise based on quaternion. First, the computation method of right eigenvalue decomposition of quaternion matrix (REVDQ) is introduced in detail. Then, duplicate auto-correlation function (that's auto-correlation function of auto-correlation function) of quaternion signal is defined, which can not only express signal parameters but also restrain MA noise. Finally, Toeplitz matrix based on duplicate auto-correlation function is constructed, and right eigenvalue decomposition of the Toeplitz matrix is used to part signal subspace and noise subspace. Thus, signal parameters are estimated with Quaternion-MUSIC (Q-MUSIC) taking advantage of orthogonality between signal subspace and noise subspace. Some simulations illustrate our idea.

Traditional orthogonal matching pursuit (OMP) algorithms for direction of arrival (DOA) estimation suffer from poor angular resolution and noise suppression. In this paper, we analyze the reason why the OMP algorithm has difficulties in resolving closely separated DOAs and conclude that it lies in the rules of support detection. Moreover, we propose a solution to this problem via developing the connection between the sparse reconstruction class algorithm and the subspace algorithm from the structure of the redundant dictionary. Based on the framework of the matching pursuit (MP) algorithm, the effective information of the signal and noise subspaces is integrated, and a noise subspace reprojection orthogonal matching pursuit (NSRomp) algorithm for DOA estimation is proposed. By adopting signal subspaces to reconstruct the original signal, the proposed NSRomp can reduce both the influence of noise on the selection of the support set and the computing time. By implementing the minimum norm method to optimize the noise subspace into a vector, which corrects the selection rules of the support set during each iteration, the angular resolution of the proposed algorithm is improved. From the simulation results, when the signal to noise ratio (SNR) is lower than or near 0, the angular resolution can be improved by > 15° using OMP algorithms to by > 5° using the proposed NSRomp algorithm.