Ray-based Blind Deconvolution Algorithm Research Articles

Data driven source localization using a library of nearby shipping sources of opportunity.

A library of broadband (100-1000 Hz) channel impulse responses (CIRs) estimated between a short bottom-mounted vertical line array (VLA) in the Santa Barbara channel and selected locations along the tracks of 27 isolated transiting ships, cumulated over nine days, is constructed using the ray-based blind deconvolution algorithm. Treating this CIR library either as data-derived replica for broadband matched-field processing (MFP) or training data for machine learning yields comparable ranging accuracy (∼50 m) for nearby vessels up to 3.2 km for both methods. Using model-based replica of the direct path only computed for an average sound-speed profile comparatively yields∼110 m ranging accuracy.

Quantifying the influence of the range and motion of shipping sources on the ray-based blind deconvolution (RBD) algorithm

The ray-based Blind Deconvolution (RBD) can estimate the relative arrival-times of the channel impulse responses (CIRs) between a shipping source (acting as a shallow source of opportunity) and a remote receiver array (e.g., a short bottom-mounted vertical line array (VLA)) [Durofchalk & Sabra, JASA Vol. 147(3), p. 1927–1938, 2020]. Previous RBD studies primarily estimate the phase of the unknown source by using beamforming on the (most energetic) direct path, which is only observable at short ranges for a shipping source in typical downward refracting environments. Alternatively, when dealing with long range sources of opportunity and only bottom-interacting multipath arrivals can reach the VLA (i.e., exceeding the range when the direct path is observable), the general performance of the RBD algorithm remains to be evaluated. In this study, numerical simulations and experimental validation (using short bottom mounted VLAs) are used in conjunction to investigate the maximal detection range of shipping sources of opportunity per multipath arrivals as a function of frequency, Doppler effect and anisotropy of the shipping source, and the implications for the accuracy of the relative arrival time of the CIR estimated with RBD.

Quantifying the performance of ray-based Blind Deconvolution (RBD) algorithm for long-range shipping sources

The ray-based Blind Deconvolution (RBD) can provide an estimate of the relative arrival-times of the channel impulse responses (CIRs) between a shipping source (acting as a shallow source of opportunity) and a remote receiver array (e.g., a short bottom-mounted vertical line array (VLA)) [Durofchalk and Sabra, JASA 147(3), 1927–1938 (2020)]. Previous RBD studies primarily rely on beamforming on the (most energetic) direct path to estimate the phase of the unknown source; but in typical downward refracting environments the direct path is only observable at short ranges for a (very shallow) shipping source. Alternatively, the overall performance of the RBD algorithm when treating long range sources of opportunity remains to be evaluated in similar settings when only bottom-interacting multipath arrivals can reach the VLA, i.e., beyond the range when the direct path is observable. In this study, a combination of numerical simulations and experimental validation (using the SBCEx16 experiment conducted in the Santa Barbara shipping channel with short, bottom-mounted VLAs) is used to investigate the maximum detection range of shipping sources of opportunity via multipath arrivals as a function of frequency, bathymetry effects and anisotropy of the shipping source and the implications for estimating the CIRs using the RBD method.

Analysis of the ray-based blind deconvolution algorithm for shipping sources.

The ray-based blind deconvolution (RBD) technique for ocean waveguides estimates both the unknown waveform radiated by some source of opportunity and the channel impulse response (CIR) between the source and the receiving elements of an array of hydrophones using only measured signals, knowledge of the array geometry, and the local sound speed. Previous studies have investigated the applicability of this method for shipping sources in a shallow, nearly range-independent waveguide (∼200 m depth), but using a limited set of shipping vessels (typically only the research vessel itself) and operating within a small domain of RBD processing parameters (e.g., integration time and frequency band). This study systematically investigates the performance of the RBD method for estimating the CIR for a large set of shipping vessels recorded on short aperture, bottom-mounted, vertical arrays deployed in the Santa Barbara channel across different frequency bands and integration times, and also in comparison to CIR measured using active sources. Furthermore, the influence of the source motion on the RBD algorithm is quantified both numerically and experimentally.

Ray-based blind deconvolution of shipping sources using single-snapshot adaptive beamforming.

The original ray-based blind deconvolution (RBD) algorithm estimates the phase of a source of opportunity-e.g., a shipping vessel-using conventional wideband beamforming (CWBF) along a well-resolved ray-path to estimate the channel impulse response (CIR). However, for the case of an anisotropic radiator and when using a short-aperture receiving array, CWBF is unable to accurately beamform on weak multipath arrivals. Instead, a higher-resolution RBD algorithm based on single snapshot adaptive beamforming is introduced by using space-frequency smoothing of the array covariance matrix; and its performance is demonstrated using experimental recordings of a container vessel to estimate the bottom-bounce arrival of the CIR.

Geoacoustic inversion using ray-based blind deconvolution of shipping sources.

The ray-based blind deconvolution algorithm can provide an estimate of the channel impulse responses (CIRs) between a shipping source of opportunity and the elements of a receiving array by estimating the unknown phase of this random source through wideband beamforming along a well-resolved ray path. However, due to the shallow effective depth (typically <10 m) and low frequency content (typically less than a few kHz) associated with shipping sources, the interfering direct and surface arriving pair and subsequent bottom and surface-bottom arrival pair cannot always be resolved in the CIR arrival-time structure. Nevertheless, this study demonstrates that the bottom reflection loss can be inferred from the ratio of the magnitude spectra of these two arrival pairs if a frequency-dependent correction (which can be purely data based) is applied to correct for the dipole source effect. The feasibility of the proposed approach is demonstrated to invert for the geoacoustic parameters of a soft-layer covering the ocean floor using a nonlinear least-square algorithm.

Analysis of a ray-based blind deconvolution algorithm on ships of opportunity in the Santa Barbara Channel

The ray-based blind deconvolution (RBD) technique for ocean waveguides estimates both the Channel Impulse Response (CIR) and the unknown waveform radiated by some source of opportunity measured on an array of hydrophones using only knowledge of the array geometry and the local sound speed. This method has been shown to be well suited for estimating CIRs in environments with water column depths of ~100 m and the estimated CIRs have been successfully used in conjunction with other techniques, such as matched field processing or waveguide invariant based approaches, for source localization, bottom parameter inversion, and other related studies. With the recent 2016 at-sea passive noise experiment in the Santa Barbara Channel (SBC), plenty of processing remains to be done; however, the RBD’s performance and robustness in such an environment (~580 m depth, downward refracting profile) is not well documented. Here, we seek to investigate the feasibility and effectiveness of the RBD method in estimating CIRs from ships of opportunity in the SBC in terms of the assumptions of the technique, application to data, and numerical simulation.

Sequential bottom parameter estimation using blind deconvolution of sources of opportunity in ocean waveguide for Santa Barbara channel experiment

This paper investigates the performance of sequential bottom parameter estimation based on ray-based blind deconvolution (RBD) [Sabra et al., JASA EL42-7 (2010)] of sources of opportunity using the 2016 Santa Barbara Channel (SBC) experimental recordings of shipping noise. The RBD algorithm relies on estimating the unknown phase of the source of opportunity through wideband beamforming along a well-resolved ray path to approximate the environment’s channel impulse responses (CIR) between the source and the VLA elements. The corrected power ratio of the direct and bottom-bounced arrivals is processed to infer the bottom reflection loss and is utilized to invert for the bottom parameters. Sequential parameter estimation uses a state space model for predicting and correcting the bottom parameters as the estimated bottom reflection loss values become available. Inversions results for the SBC experiment were also performed with conventional active sources to validate the inversion obtained with RBD of sources of opportunity.

Ray-based blind deconvolution of shipping sources using multiple beams separated by alternating projection.

This article presents a method for improving the performance of the ray-based blind deconvolution (RBD) algorithm, which was first proposed by Sabra, Song, and Dowling [J. Acoust. Soc. Am. 127(2), EL42-EL47 (2010)]. In order to retrieve the channel impulse response (CIR), the original RBD algorithm uses the source signal phase from a selected single beam output. However, when the impinging multipath signals have low coherence, the channel estimate from a selected beam may not show all paths correctly. In this research, the maximum likelihood estimator, which is called the alternating projection, is applied to separate multipath signals. Then the multiple CIRs obtained from those separated signals are coherently combined. This results in more robust detection of existing multipaths. The performance of the proposed method is verified using Noise09 sea experiment data, where the proposed method better resolves the multipath arrival structure.

Sequential bottom parameter estimation using blind deconvolution of sources of opportunity in ocean waveguide

This paper investigates the performance of sequential bottom parameter estimation based on ray-based blind deconvolution (RBD) of sources of opportunity using both numerical simulations and the 2016 Santa Barbara Channel (SBC) experimental recordings of shipping noise for ranges up to several kilometers. The RBD algorithm relies on estimating the unknown phase of the random sources to approximate the source-to-array channel impulse responses (CIR) by wideband beamforming along a well-resolved ray path [Sabra et al., JASA, 2010, EL42-7]. The power ratio of the direct and bottom-bounced arrivals is processed to infer the bottom reflection loss and is utilized to invert for the bottom parameters. Usually, the estimated bottom reflection loss is not accurate enough as the estimated CIR is noisy. Sequential parameter estimation uses a state space model for predicting and correcting the bottom parameters as the estimated bottom reflection loss values become available. This approach is a robust estimation tool employing predictions from previous estimates and corrects stemming from models that relate bottom parameters to the bottom reflection loss. Inversions results for the SBC experiment were also performed with conventional active sources to validate the inversion obtained with RBD of sources of opportunity.

Extracting tomographic arrival time information from ships of opportunity in the Santa Barbara channel using blind deconvolution

This paper summarizes the ongoing investigations surrounding the use of a ray-based blind deconvolution algorithm to recover arrival time information from sources of opportunity, such as shipping vessels, recorded on vertical line arrays (VLAs) in ocean waveguides. The deconvolution is primarily performed by using an estimate of the unknown source phase, obtained through wideband beamforming, to essentially match filter the VLA recordings and recover the channel impulse response (CIR). This paper will focus on results from an experiment performed in 2016 in the Santa Barbara shipping channel (water depth ~550 m). Four VLAs, with both short (~15 m) and long (~56 m) apertures, were deployed between the north and south bound shipping lanes and continuously collected acoustic data during one week. With the ultimate goal of passive acoustic tomography in mind, this paper aims to discuss (1) the robustness of the algorithm to extract differential arrival times along VLA elements using ships as sources of opportunity, (2) the achievable accuracy of blind arrival time measurements in comparison to the time-of-flight precision required for tomographic inversions, and (3) the ideal parameters (e.g., frequency bandwidth, snapshot duration, beamforming methodology...) for which to perform this ray-based blind deconvolution method in SBC-like ocean environments.

Performance of a ray-based blind deconvolution algorithm for shipping sources in the Santa Barbara channel

This paper investigates the performance of a ray-based blind deconvolution (RBD) algorithm for sources of opportunity, such as shipping noise, in an ocean waveguide recorded on a vertical line array (VLA). The RBD algorithm [Sabra et al., JASA,2010, EL42-7] relies on estimating the unknown phase of the source through wideband beamforming along a well-resolved ray path to approximate the environment’s Green’s Functions (or Channel Impulse responses) between the source and the VLA elements, as well with recovering the unknown radiated source signal. The RBD algorithm is tested here for shipping sources recorded in the Santa Barbara shipping channel (water depth ~550 meters). The four VLAs, with short (~ 15 meter) and long (~56 meter) apertures, were deployed between the north and south bound shipping lanes and collected acoustic data throughout a week in mid-September, 2016. Here, we discuss (1) the performance of conventional beamforming and adaptive (MVDR) beamforming when estimating ray-arrivals, (2) the ability of the RBD algorithm to deconvolve multiple VLAs using the only the same source phase estimated from a single VLA. The performance of the RBD results will be discussed in terms of accuracy of travel times and of the estimated Green’s Functions.

Passive characterization of underwater sound channel using ray-based blind deconvolution algorithm

Ray-based blind deconvolution (RBD) algorithm is a blind channel estimation method, which can estimate both the channel impulse response and the original signal broadcast from an unknown source. It is based on high-frequency ray assumption and requires only the elementary information of array geometry and sound speed at the array position. Its capability of estimating the channel in a passive manner can be used to infer the sound channel characteristics from a source of opportunity such as passing surface ship whose positioning information is available from geo-referencing systems like the Automatic Identification System. Recently, we applied the RBD algorithm to the shipping noise data recorded on vertical line arrays at multiple locations in shallow water and confirmed its applicability for channel estimation and performing environmental inversion. This presentation shows the channel impulse responses estimated from a variety of source-receiver geometry and discusses the characteristics of sound propagation as well as the environmental property such as bottom loss which can be observed using the RBD algorithm.

Improving ray-based blind deconvolution of random shipping sources with short arrays in an ocean waveguide using adaptive beamforming

This paper introduces an improved ray-based blind deconvolution(RBD) algorithm for sources of opportunity, such as shipping noise, recorded on a short vertical line array (VLA) by using adaptive beamforming. The original RBD algorithm [Sabra et al., JASA, 2010, EL42-7] relies on first estimating the unknown phase of the random source using broadband beamforming along a well-resolved ray path to estimate the full channel impulse responses (CIR) between the unknown source and the VLA elements (up to an arbitrary time-shift) as well as recovering the radiated signal by the random source. Hence, RBD performance is limited by the VLA ability to separate adjacent ray path, especially for the case of small aperture. To address this limitation, broadband MVDR is used as a high resolution estimator to improve the RBD performance when using single snapshot recordings and short VLAs. To do so, smoothing techniques are used to increase the rank of cross-spectral data matrix since the ray arrivals emanating from the same random source are naturally correlated in an ocean waveguide when using single snapshot recordings. The improvement of this proposed RBD algorithm will be demonstrated using both numerical simulation and at-sea shipping noise data recorded in shallow water.