Vincent E. Premus

A real-time method for autonomous passive acoustic detection-classification of humpback whales

This paper describes a method for real-time, autonomous, joint detection-classification of humpback whale vocalizations. The approach adapts the spectrogram correlation method used by Mellinger and Clark [J. Acoust. Soc. Am. 107, 3518-3529 (2000)] for bowhead whale endnote detection to the humpback whale problem. The objective is the implementation of a system to determine the presence or absence of humpback whales with passive acoustic methods and to perform this classification with low false alarm rate in real time. Multiple correlation kernels are used due to the diversity of humpback song. The approach also takes advantage of the fact that humpbacks tend to vocalize repeatedly for extended periods of time, and identification is declared only when multiple song units are detected within a fixed time interval. Humpback whale vocalizations from Alaska, Hawaii, and Stellwagen Bank were used to train the algorithm. It was then tested on independent data obtained off Kaena Point, Hawaii in February and March of 2009. Results show that the algorithm successfully classified humpback whales autonomously in real time, with a measured probability of correct classification in excess of 74% and a measured probability of false alarm below 1%.

Mode excision adaptive beamforming for source detection in an uncertain shallow‐water waveguide

Passive sonar detection is uniquely characterized by the fact that the acoustic clutter distribution is generally confined to the ocean’s surface. There is considerable evidence to support the hypothesis that surfaced and submerged sources are well separated in acoustic mode space, and that shallow‐water waveguide normal modes are relatively robust to imperfect environmental knowledge. In this work, the use of mode physics is explored for the purpose of identifying an improved adaptive subspace for submerged source detection in the presence of surface interference. The basic premise is to perform adaptive weight computation in a mode subspace that is weakly excited by the submerged source of interest, yet well coupled to the surface interference. The rationale is to excise as much of the target signature as possible from the sample covariance without excessively compromising the measurement of the interference spatial spectrum. This enables more aggressive nulling of the surface clutter spectrum for a giv...

High-resolution bathymetric simulations based on Kirchhoff scattering theory and anisotropic seafloor modeling

The bathymetric resolution of the seafloor map generated from multibeam or sidescan echo-sounder systems can be improved with various signal processing methods. In order to evaluate the accuracy and resolution performance of such techniques, the authors developed a realistic sonar simulation and seafloor modeling environment by merging the Kirchhoff scattering theory with the anisotropic seafloor parameterization of Goff and Jordan (1988). Multibeam sonar simulations utilizing the above environment, in combination with an eigenanalysis (e.g. MUSIC) power spectral estimation method for beamforming were carried out to obtain high-resolution bathymetry.<<ETX>>

A mode space energy detector for acoustic source depth discrimination

A likelihood ratio test based on a newly defined statistic, the modal scintillation index (MSI), is presented for acoustic source depth discrimination in a shallow‐water waveguide. The MSI is defined as the variance in the estimated modal excitation normalized by its expected value over some observation interval. The approach represents, in effect, a mode space energy detector. In a shallow‐water waveguide, the acoustic modes are nearly sinusoidal, and thus all share a zero‐crossing and maximum derivative at z=0. Consequently, surface source modal excitation levels will be very sensitive to low bandwidth source depth modulation due to surface wave interaction. The hypothesis herein is that a surface source will be characterized by a high scintillation index across all modes, while a source at depth will exhibit a low scintillation index for at least one acoustic mode. The test statistic is self‐normalizing, so knowledge of source level and source range is not required to separate the two source classes. E...

Maximum a posteriori probability estimation of seafloor microroughness parameters from backscatter spatial coherence

A technique is presented for the estimation of a set of parameters associated with a geologically motivated model for seafloor microroughness due to Goff and Jordan (1988). The method seeks to connect the spatial covariance of the backscattered acoustic field with the correlation properties of the seafloor by constructing the a posteriori probability density function (pdf) of the parameters that define the seafloor microroughness wavenumber spectrum. The processor maximizes the joint a posteriori probability density of the model parameter set. Due to the complexity of the probability surface, the method of simulated annealing is used to search for the globally optimum solution vector.

The role of fluctuations in the interpretation of sonar detection performance

The predictive probability of detection (PPD) is a metric that accounts for uncertainty in sonar detection performance due to random fluctuations in transmission loss, noise level, and source level [P. Abbot and I. Dyer, Impact of Littoral Environmental Variability on Acoustic Predictions and Sonar Performance , 2002]. It is well known that a significant portion of Ira’s career was dedicated to understanding the role fluctuations play in interpreting acoustic measurements. Building on this foundation, we now embrace the notion that a useful statement of sonar system performance is one linked with a probabilistic description of the acoustic environment’s intrinsic variability. In this paper, we discuss the impact of fluctuations on passive sonar performance, including how PPD facilitates the interpretation of sonar system recognition differential. Data from a recent field test conducted in August, 2011, on the New Jersey continental shelf, will be used to illustrate the methodology and interpret measured detection performance in the presence of a cold pool duct and variable ambient noise conditions.The predictive probability of detection (PPD) is a metric that accounts for uncertainty in sonar detection performance due to random fluctuations in transmission loss, noise level, and source level [P. Abbot and I. Dyer, Impact of Littoral Environmental Variability on Acoustic Predictions and Sonar Performance , 2002]. It is well known that a significant portion of Ira’s career was dedicated to understanding the role fluctuations play in interpreting acoustic measurements. Building on this foundation, we now embrace the notion that a useful statement of sonar system performance is one linked with a probabilistic description of the acoustic environment’s intrinsic variability. In this paper, we discuss the impact of fluctuations on passive sonar performance, including how PPD facilitates the interpretation of sonar system recognition differential. Data from a recent field test conducted in August, 2011, on the New Jersey continental shelf, will be used to illustrate the methodology and interpret measured d...

Receiver operating characteristic for a spectrogram correlator-based humpback whale detector-classifier

This paper presents recent experimental results and a discussion of system enhancements made to the real-time autonomous humpback whale detector-classifier algorithm first presented by Abbot et al. [J. Acoust. Soc. Am. 127, 2894-2903 (2010)]. In February 2010, a second-generation system was deployed in an experiment conducted off of leeward Kauai during which 26 h of humpback vocalizations were recorded via sonobuoy and processed in real time. These data have been analyzed along with 40 h of humpbacks-absent data collected from the same location during July-August 2009. The extensive whales-absent data set in particular has enabled the quantification of system false alarm rates and the measurement of receiver operating characteristic curves. The performance impact of three enhancements incorporated into the second-generation system are discussed, including (1) a method to eliminate redundancy in the kernel library, (2) increased use of contextual analysis, and (3) the augmentation of the training data with more recent humpback vocalizations. It will be shown that the performance of the real-time system was improved to yield a probability of correct classification of 0.93 and a probability of false alarm of 0.004 over the 66 h of independent test data.

Robust adaptive beamforming for hydrodynamic self‐noise rejection

White noise gain (WNG) is a metric widely used in the radar community for limiting the degree of adaptivity of an adaptive beamformer (ABF). Constraining adaptation of an ABF algorithm has two key effects: (1) protection against self‐nulling associated with steering vector mismatch, and (2) limitation of white noise gain associated with squinting a beampattern to place a null on a mainlobe interferer. Hydrodynamic self‐noise, or cable strum, commonly manifests itself as a source of mainlobe interference for passive acoustic towed horizontal line arrays. Cable strum is the result of vibrations excited in the array body by vortex shedding in the presence of flow. Strum is particularly detrimental on beams near endfire and typically exhibits very high dynamic range. As such, it requires an aggressive adaptation strategy for its effective removal. In this work, a white noise gain constraint for the rejection of cable strum is derived based on the scaled orthogonal projection approach of Cox et al. [IEEE Trans. ASSP 35 (1987)]. Tradeoffs involved in balancing strum rejection performance against mismatch‐induced signal self‐nulling are examined. [Sponsored by the Dept. of the Navy, under Air Force Contract No. F19628‐00‐C‐0002. Opinions, interpretations, conclusions, and recommendations are those of the authors and not necessarily endorsed by the U.S. Air Force.]

Cable strum self‐noise cancellation for sonar towed arrays

Nonacoustic self‐noise observed on marine seismic streamers and towed sonar arrays represents a serious problem for acoustic source detection at low frequency. Towed array self‐noise, also known as cable strum, consists of mechanical vibrations induced by vortex shedding. Transverse vibrations in the array body subject each hydrophone pressure head to local accelerations. The resultant acoustic response can be several orders of magnitude stronger than the water‐borne acoustic signals of interest. In this paper, a beamspace, time‐domain adaptive signal processing architecture for the coherent rejection of broadband nonacoustic self‐noise is presented. The approach is based on the recognition that most vibrational modes of a towed array propagate at phase speeds substantially less than those of acoustic signals in the water column. This property supports the formation of a signal‐free, strum reference using the same sensor that samples the acoustic data. The approach removes the need for additional measurem...

Mode scintillation and surface/submerged discrimination in SwellEx‐96

It has been proposed that the short time‐scale scintillation of normal mode amplitudes may be an effective discriminant for passive surface/submerged acoustic source classification in a shallow waveguide [V. Premus, J. Acoust. Soc. Am. (submitted)]. The approach is based on a two part hypothesis which states: (1) platform motion in the vertical due to surface or internal wave action induces observable temporal fluctuations in normal mode amplitudes, and (2) the probability densities of mode amplitude scintillation for surface and submerged sources are well separated under modest depth fluctuation conditions. In this work, results from the application of the mode scintillation classifier to narrow‐band vertical array data collected during SwellEx‐96 Event S5 are reviewed. Classifier output in the form of a thresholded mode scintillation interest image provides empirical evidence to support the classification hypothesis for the lowest order resolvable modes. Limitations imposed by the non‐fully spanning nat...