Sonar Processing Research Articles

SPHERICAL SUBSPACE TRACKING ON SYSTOLIC ARRAYS

The spherical subspace tracker is a low complexity adaptive algorithm that converges to the dominant (or subdominant) subspace. Spherical subspace updating is a fast way to monitor slowly time-varying subspaces, which arise in high resolution direction of arrival tracking algorithms. In real time sonar and radar array processing applications, it becomes necessary to find parallel algorithms and pipelined architectures that provide high speed solutions to these tracking problems. In this paper the spherical subspace tracker is parallelized and mapped onto systolic array architectures. The arrays are ideally suited for implementation with custom VLSI or networks of available parallel numeric processing chips such as iWarps and TMS320C40s. C-language like cell programs specify the functionality and timing of the systolic array in concrete terms.

Alternative neural network training methods [active sonar processing

Investigates three potential neural network training algorithms in processing active sonar returns. Although all three methods generate reasonable probabilities of detection and false alarm in discriminating between man-made objects and background events, the stochastic training methods of simulated annealing and evolutionary programming outperform backpropagation. >

Technical note Seafloor characterization through side scan sonar image processing

Abstract This paper presents the most suitable image processing techniques being used at present for recognising features either lying on the seafloor or floating in the water column. The preprocessing methodology is already incorporated in the sonar system, and as we cannot access the raw data, it is not included in this paper. The studied images correspond to the Cabrera Archipelago, in the Mediterranean Sea.

Three-Dimensional target recognition via sonar: A neural network model

A neural network was trained to recognize two three-dimensional shapes independent of orientation, based on echoes of ultrasonic pulses similar to those used by an echolocating bat, Eptesicus fuscus. Following supervised learning, the network was required to generalize and recognize echoes from the shapes at novel orientations. The representation of the echo was manipulated to explore how information about target shape may be encoded in sonar echoes. Three types of input representations were used: time domain (waveform and cross-correlation), frequency domain (power spectrum), and time frequency (spectrogram). The probability of correctly recognizing the novel echoes by chance was only 25%. The network using the spectrogram representation recognized 90% of the echoes from novel orientations. The bat Eptesicus fuscus uses a multiple echolocation cry, and we explored the relative contribution of low and high frequencies for carrying information about target shape. We presented the network with low-pass and high-pass filtered spectrogram representations, preserving sound energy in frequency bands roughly corresponding to either the first or the second harmonic of the bat's echolocation sounds. The network was able to recognize 90% and 95% of the novel echoes using only the first or only the second harmonic, respectively. We continued by examining whether the network could perform the task using only time domain or only frequency domain information. When presented with a time waveform representation, the network was unable to perform the task. Similar results were obtained with other time domain representations, the cross-correlations between the emitted sound and its returning echo. However, when using only frequency domain information, the network was able to recognize 70% of the echoes from novel orientations. Again, we explored the relative contribution of the frequency bands corresponding to the first and second harmonics used by the bat Eptesicus fuscus. We found that the network was able to recognize 70% of the novel echoes using the first harmonic, and only 55 % of the novel echoes using the second harmonic. The results are discussed in light of studies on echolocation of bats and models of sonar processing.

Localization accuracies for a moving source in an oceanic waveguide

The passive localization of a narrow-band target in the ocean is a standard procedure in sonar processing. In this paper Cramér–Rao lower bounds are derived for the accuracy with which target track parameters can be estimated using a horizontal array. The target, which may be moving, is observed over a long time period, so that parameters such as bearing rate and radial speed cannot be assumed to be constant. The acoustic propagation in the oceanic waveguide is modeled realistically, and the temporal coherence of the received signal is accounted for. Several computational examples are presented in which the parameter bounds are examined as a function of target position in the waveguide. The dependencies of the bounds on propagation characteristics, observation time, and signal coherence time are also examined. The bounds differ substantially from those computed using simple models that ignore the effect of propagation.

Spatial and temporal variation of acoustic backscatter in the STRESS experiment

Acoustic backscatter measurements were made of the seabed with a bottom mounted, circularly scanning sonar. The placement was at 91 m depth, mid-shelf of Northern California (38° 34′N), site C3 of the experiment STRESS I (1988–1989). Our expectation was that sonar images (70 m radius, 12,000 m 2) would provide a means of observing, over a large field of view, changes in the bottom due to storm-induced sediment transport and due to bioturbation. This expectation was supported in part by towed sonar measurements at 35 kHz over a sandy area in the North Sea, where dramatic spatial variation in the level of the backseattered signal was observed during an Autumn storm on scales of a few km with no concomitant change in sediment grain size [ Jackson et al. (1986) The Journal of the Acoustical Society of America, 80, 1188–1199]. It appeared possible that storm-driven sediment transport might have been responsible for this patchiness, by altering bottom roughness and by redeposition of suspended material. At the California site, a conventional sonar processing of our data from the STRESS experiment reveals no such dramatic change in backscattered signal level due to storms. The sonar images contain random structures whose time evolution is subtle and difficult to interpret. A much clearer picture of temporal and spatial variations emerges from a processing scheme involving cross-correlation of time-separated acoustic views of the bottom. In effect, the sequence of correlation data images produces a movie in which patches of activity are seen to develop as functions of time. It appears that most of this activity is biological rather than hydrodynamic. A tentative explanation is two-fold. The bottom shear stress might have been considerably greater at the North Sea site (with depth only one-half of the California site). The seafloor at the California site was silty-clayey, and backscatter from such floor is less sensitive to the water-floor interface shape and roughness than it would be to the same parameters of a sandy bottom.

Acoustical applications of wavelets: Audio coding and sonar

Wavelets are a family of basis functions for the space of finite energy signals. In a wavelet decomposition, a signal is expanded in terms of translates and dilates of a single function. Wavelets provide a flexible decomposition of signals. This flexibility is the foundation of a new generation of advanced low-bit-rate, high-quality audio coding procedures. By adaptively choosing the wavelet representations of consecutive segments of the audio signal one can optimally exploit the masking effect in human hearing. The performance of the resulting coding techniques surpasses that of known nonwavelet-based approaches. Another important property of wavelet decompositions is the fact that one can relate the structure of the wavelet coefficients of a given signal to the local time and frequency domain characteristics of the signal. In sonar processing, this fact has been exploited by a new class of robust direction of arrival estimation algorithms that can deal with background correlated noise of an unknown correlation structure. These algorithms separate signal from noise by analyzing the structure of the wavelet coefficients of the array outputs.

Side-scan sonar image processing using thin plate splines and control point matching

The development of reliable software tools to aid human interpretation of complex digital images is of increasing importance. In particular, change detection requires that two or more images be carefully overlaid (registered) using digital techniques that compensate for picture distortions caused by sensor motion, ambient conditions, etc. We are primarily interested in small scale changes that occur in a given underwater scene observed at different times. In this paper we describe our use of thin plate splines in registering pairs of digital images composed of millions of pixels. We also review work on selecting and matching the control points needed for image registration.

Real time time-frequency active sonar processing: a SIMD approach

A paradigm for massively parallel processing of matched filters, replica correlators, ambiguity functions, and time-frequency distributions is presented, using a SIMD (single instruction stream, multiple data stream) programming methodology. It is shown that active sonar detection algorithms, as implemented by frequency domain processing, can be a natural match to a SIMD methodology, meeting the extensive computational needs of enhanced active sonar systems. The decomposition process is presented, and examples are given of the output of the computer program CMASP (Connection Machine Ambiguity Surface Processor). CMASP can provide real-time simultaneous multiple-beam, Doppler, and waveform replica correlations. Synthetic data are processed, and the corresponding CMASP outputs are displayed as three-dimensional ambiguity surfaces on networked graphic workstations. Because of efficient problem decomposition, other time-frequency processing can be exploited. Specifically, real-time instantaneous-like time-frequency distributions have been realized in which the data set is presented and processed as time-varying spectral representations. >

Cascaded frequency domain bandpass filters for multivernier spectra

Passive sonar processing generally consists of many concurrent verniers on each beam or channel which are very narrow bands centered at widely separated, independent frequencies sparsely placed across the total search band. Each is analyzed by a fast Fourier transform (FFT) with extremely narrow coefficients. A cascaded frequency domain filter structure that provides such verniers with fewer computations than other methods is described.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Signal design for Doppler processing

The Doppler sonar processing requires that the illuminating signal have a spectrum with low sidelobes over the range of frequencies at which Doppler shifts are expected. This is conventionally achieved by “windowing” the driving sinewave. Uncertainties in the amplitude response (threshold onset and linearity) of typical transducers make it desirable to have alternate means of obtaining low sidelobes. One approach is to transmit a series of nonuniformly spaced short pulses of constant duration and amplitude. Pulse start times are specified to yield the desired spectral shape using the approach of A. Ishimaru [IEEE Trans. Ant. Propag. AP‐10, 691‐702 (1962)] developed for space tapering antenna array elements. The resulting spectrum is not affected by nonzero threshold onset or limited linear range. The sidelobe reduction that can be achieved depends on the number of pulses, hence, on the amplifier bandwidth. Amplitude shading and varying pulse durations allow the design of a second type of signal, a windowed cw signal that does not depend on threshold onset and is less sensitive to amplifier bandwidth. The sidelobe reduction achievable with this second design depends on the linear range.

A practical VME‐based sonar system development workstation

This paper describes present capability and furture development plans of a practical low‐cost sonar system‐development work station based on off‐the‐shelf VME equipment. The work station allows system engineers and algorithm designers to quickly develop sonar processing algorithms and systems. The work station can be used to (a) simulate and analyze the characteristics of real time environments for system and algorithm development, (b) concurrently develop final implementation software, and (c) generate analog test data for system test and operator instruction. The work station overcomes many of the difficulties experienced with developing sonar acoustic signal acquisition and processing systems: high design and maintenance costs of custom equipment; lack of controllable test data; inability to correlate simulation data with final system performance; and high software development costs, particularly with parallel, pipelined signal processing equipment. The sonar development work station hardware consists ...

Alternate schemes in synthetic aperture sonar processing

Suboptimal processing schemes, the application of which is not widespread in synthetic aperture sonar processing, are described with reference to seafloor imaging. It is shown that their application can result in a significant increase of the azimuth resolution of the sonar system with respect to the resolution due to its physical beamwidth, without imposing unreasonable constraints on the sonar platform trajectory.

Control Ordered Sonar Hardware (COSH)—a hardware based signal processing graph implementation

Current generation sonar acoustic signal processing systems operate at throughputs comparable to those predicted for fifth generation computer systems. Programmable sonar processors are currently deployed that operate with throughputs in excess of 100 million operations per second. These processing speeds are achieved using distributed networks of programmable hardware based primitives which execute the basic signal processing functions. The network of primitives is configured by simple control software to implement the signal processing flow graph that defines the entire sonar processing function.This paper outlines the architecture and development of the processing primitives and their use in high throughput multiprocessor signal processing flow graph systems.

Passive sonar processing for noise with unknown covariance structure

The piedmont dry forest of northwestern Argentina has been under intensive and unplanned forest logging focused on 12 tree species, without any attempts having been made to ensure their regeneration or long-term conservation. In this study, we assessed the conservation status of these timber species in the piedmont dry forest of northwestern Argentina and compared our results with the IUCN assessment. We considered an inadequate conservation status if the species: (1) occurred in less than 50 % of the sampled plots, (2) had a density of large trees (more than 40 cm DBH) lower than one in. ha−1, or (3) had a density of saplings (more than 30 cm in height and less than 9.9 cm in DBH) lower than 30 in. ha−1. Our results showed an inadequate conservation status for eight of the 12 studied timber tree species. Additionally, only seven species were previously assessed and categorized on the IUCN Red List, with our categorization agreeing with only four of them. The data provided in this work can serve as a baseline to monitor the population trends of these species; information can also help to prioritize conservation efforts, which is essential considering the high number of tree species that have potential extinction risk in the short-term.

Passive Sonar Processing for Noise with Unknown Covariance Structure

Passive processing is interpretation of the noise and noise like signals received by an array of hydrophones. Our model assumes that there is an important noise component about which very little is known, that the data are stationary, that a long data sample is available, that signal wavefronts are planar, and that the response of the array to plane waves is known. We propose a processor and compare it with high-resolution frequency-wavenumber spectrum analysis. Let F̂ denote an estimate of the spectral density matrix, and let H denote any spectral density matrix that corresponds to a noise field composed of uncorrelated plane waves. A reasonable estimate of the spectral matrix of the signals and ambient noise is a value of H for which F̂-H is positive semidefinite and trace H is maximized. The rationale for this estimate is closely related to a rationale for the estimate obtained by high-resolution frequency-wavenumber spectrum analysis (which is also called adaptive beamforming). The difference is that the former extracts the contribution from all directions at once, whereas the latter extracts the contribution from each direction separately. The former estimate rejects more unwanted noise than the latter.

Monte-carlo simulation of an active sonar

This paper describes the analysis of a non-linear sonar processing system by simple Monte-Carlo methods. The target detection capability with isotropic background noise and spatially coherent reverberation is found with attention being paid to the effects of frequency filters in the input channels from the transducers.The resolution and detection of multiple and diffuse targets is assessed under noise free conditions and with the assumption of target scintillation due to forward scattering or medium inhomogeneity.