Underwater Signal Research Articles

Computing the environment of an artificial neuronal network for underwater signal classification

An environment for ANN signal processing has been designed and developed, which is written with Turbo Vision’s CLASS in Borland C++ 3. Algorithms for mapping a BP network on a machine and task assign tactics are devised to implement ANN parallel algorithms. To use the PC as controller of the whole parallel system, some problems have to be solved. So the floating-point conversion between TMS320C30 and IEEE formats, executable program transforms, as well as data-loading techniques, are developed and presented in this paper. This application system has been used to quickly classify three kinds of underwater signals with a three-layer BP network in the ANN. Results show that the above classification can be finished in 28 μs. The FFT algorithm with 1024 points is also simulated on the system and it takes 2.3 ms. The other advantage of this software is that it is easily expanded to include new functions.

Nonlinear analysis of environmental distortions of continuous wave signals in the ocean

Data from an underwater continuous wave signal are analyzed using methods derived from the study of chaotic systems. Distortions by the environment effectively add three degrees of freedom (dimensions) to the signal. The dimension required to reconstruct and analyze the received data is five in contrast to a pure tonal or sine wave that only requires two dimensions. The data are examined with a chaotic data analysis toolkit that includes determination of degrees of freedom using local and global false-nearest-neighbor statistics, average mutual information, correlation dimension, and local Lyapunov exponents. These results are important for the development of methods for correction of propagation distortions and nonlinear noise reduction algorithms. Application of this knowledge should lead to improvements in the detection and classification of underwater signals.

Descriptors of impulsive signal levels commonly used in underwater acoustics

Impulsive acoustic waveforms are characterized by a different set of derived quantities than are continuous waveforms. This note presents commonly accepted definitions, units, and symbols used to describe the magnitude of impulsive underwater signals.

Timespread modeling of propagating pulsed and transient underwater signals

Many Navy standard and other widely used underwater acoustic propagation models were developed through an evolutionary process to address applications featuring narrow‐band or single‐frequency cw signals. A consequence of this development process is that, although the models may handle the originally intended problem very well, they are not necessarily able to describe the propagation of high resolution pulses or transients, where time‐domain or multi‐frequency FFT approaches seem called for. This paper presents the results of an evaluation of several propagation models to characterize short duration signals. The models tested represent a variety of physics‐based approaches: parabolic equation, rays and Gaussian beams, modes, and hybrid approaches. The evaluation criteria include an ability of the modeling system to accommodate real‐world, in‐situ environments, accuracy of the results in the context of intended applications, and reasonable execution time.

Experimental results of localization of moving underwater signal by adaptive beamforming

The problem of weak moving signal localization and tracking in the presence of single motionless strong interference is investigated using real data of an underwater experiment in the Baltic sea (Sept. 1990) with a horizontal receiving array of 64 hydrophones and with two independent powerful narrowband sources imitating the signal and interference. Three simple adaptive beamforming methods were employed for the experimental data processing. The first one is based on the well-known projection approach to adaptive beamforming, the second method uses the adaptive canceler approach (also termed the dipole pattern method), and the third method combines these approaches. The signal-to-interference power ratio (SIR) threshold of the signal localization and tracking is evaluated by a special technique, which allows examination of the considered algorithms with change of the SIR in consecutive order. The results of the data processing show the high possibilities of signal localization in the presence of strong interference. The combined method performs better than the methods considered and enables localization of the signal source up to an SIR/spl sime/-25 dB. >

Active sonar detection in multipath: A New bispectral analysis approach

Detecting active sonar returns in multipath media is a central underwater signal processing problem. This paper studies a new approach to active sonar detection using bispectral analysis of sonar data. Its sensitivity to non-stationarities is used to develop a threshold detector that can be applied to broad classes of signals and noise. Precise statistical descriptions of the underwater medium and noise are not required. Theoretical analyses predicting its performance as a function of signal-to-noise ratio and time-bandwidth product are presented. Computer simulation experiments verify the results and show that its performance compares favorably to that of conventional detectors.

Low-frequency hearing in California sea lions and harbor seals

Studies on pure-tone detection thresholds were conducted on two female California sea lions and on a harbor seal. The older sea lion and the harbor seal were trained to wear custom-fitted headphones in order to determine minimum audible pressures in a binaural listening task. All three animals were trained to respond to underwater signals at frequencies ranging from 100–1600 Hz at a depth of about 1.5 m. Results were very reliable, owing to a combination of psychophysical threshold determining measures. Sensitivity to low-frequency sounds by both species were 25–30 dB better underwater than in air. The low-frequency hearing of the harbor seal was 2–25 dB better than the older seal lion. At 100 Hz, the sensitivity of the harbor seal was 17 dB superior to that of the younger seal ion, and 23 dB superior to the older seal lion. Results at low frequencies support the notion that the harbor seal (phocid) ear is more water adapted than the sea lion (otariid) ear.

Laser induced underwater acoustic source for remote sea-bed sensing

The remotely driven by means of CO 2 -laser pulse underwater near-surface optoacoustic source can be used for ocean medium monitoring. In the present paper this «unusual» pulse underwater signal was used for remote sea-bed sensing in shallora water regions of the Black Sea. Ultra wide band sounding pulse provided a set of narrow band reverberation responses. An appropriate sea-bed scattering model is developed provided consistence of calculated reverberation with experimental data and estimation of some sea-bed parameters as well

Audible signalling system for divers

An audible signalling system is provided for us by divers or the like to provide a relatively loud audible signal which can be used to indicate diver position, to signal distress, etc. The preferred system utilizes an air horn carried by the diver and adapted for connection to a pressurized air supply carried by the diver for breathing purposes. When the air horn is connected to the air supply, the horn can be operated to produce a loud audio signal of substantial range which can be heard relatively easily over ambient noise such as wind, water and/or boat noise. In one form, a sonic oscillator for use in underwater signalling is also included.

Underwater transmitter apparatus

First Page

Fourth‐order spectra of Gaussian amplitude‐modulated sinusoids

Propagating underwater signals are sometimes amplitude modulated by the medium or other physical effects. Since this is a multiplicative phenomenon, the resulting spectrum is a convolution of the spectra of the modulation and the desired signal. A new method based on the spectrum of a special case of the fourth‐order cumulant is proposed to extract the desired signal. It is shown that the special case of the fourth‐order cumulant is independent of the covariance of a Gaussian modulating function, and therefore it can extract the desired signal even when the modulating process is Gaussian white noise. Convergence equations are derived which show that the fourth‐order cumulant and its special case will converge to their asymptotic forms as the data length increases. The spectrum of the special case of the fourth‐order cumulant is also derived at the output of a low‐pass filter. To demonstrate these theoretical results, an experiment was conducted. A sinusoid was modulated by white Gaussian noise and transmitted through the water and received on an omnidirectional hydrophone. The second‐order spectrum, the spectrum of the special case of the fourth‐order moment and the spectrum of the special case of the fourth‐order cumulant were estimated from the filtered data. The experiment corroborated the theoretical results by showing that the second‐order spectrum could not extract the sinusoidal frequency, but the spectrum of the special case of the fourth‐order moment and the spectrum of the special case of the fourth‐order cumulant could. Simulations are included which further corroborate the theoretical results.

Estimator-correlator array processing: Theoretical underpinnings and adaptive implementation

We develop a generalized singular value decomposition methodology for implementing the estimator-correlator array processor. The usefulness of this structure stems from two key properties examined in this paper. First, the structure is canonical, in that it applies to a broad class of underwater signal processing problems. Second, it is robust, not critically dependent on a precise characterization of the medium in question. Numerical simulation experiments justifying this claim are presented.

Fourth‐order spectra of intensity‐modulated sinusoids

Often propagating underwater signals are intensity modulated by inhomogeneities in the medium. Since this is a multiplicative phenomenon the resulting spectrum is a convolution between the modulation and the desired signal. However, it has been recently shown that a special case of the fourth‐order cumulant spectrum can extract the frequency of an intensity‐modulated sinusoid without distortion from convolution. The special case of the fourth‐order cumulant spectrum is also independent of the covariance of the modulation and only depends on the variance. In order to test the theoretical results an experiment was conducted. An amplitude‐modulated sinusoid was transmitted through the water and received. Both the spectrum and the fourth‐order cumulant spectrum were estimated from the received data. The experiment corroborated the theoretical results.

Measurement accuracy of MVDR adaptive beamformers

Minimum variance distortionless response (MVDR) adaptive beamformers minimize noise under the constraint of maintaining a known signal amplitude response in the steered direction of the beam. These criteria make the MVDR beamformer attractive for measuring underwater signals from a known location in a noise background consisting of a number of point source interferences. This paper investigates the accuracy of measuring signals in such an environment by using a frequency domain implementation of a MVDR adaptive beamformer to process simulated and at-sea data. Examples will be shown of measurements using a 32-hydrophone planar array. Of special concern was the effect on measurement accuracy due to beam pointing error, adaptation time, and adaptive filter errors introduced by MVDR calculations. Also of interest was the trade-off of adaptation before or during the presence of the signal. Results are shown for three types of data scenarios: (1) a simulated sinusoidal signal and interference point noise sources in spatially uncorrelated noise, (2) a projected underwater sinusoidal signal in ambient noise, recorded off the eastern coast of Vancouver Island, Canada, and (3) an injected, simulated sinusoidal signal combined with recorded at-sea noise. Results showed that the MVDR beamformer significantly increased signal-to-noise ratios by canceling interfering noises, but did so at the expense of the measurement accuracy.

Underwater acoustic signal analysis with wavelet process

A recently developed mathematical method (wavelet transform) of signal processing has provided potential new options for underwater signal study. It is a time-frequency mathematical process that analyzes high-frequency signals. Signal frequency instabilities are identified, calculated, and compared with the chosen wavelet. The mathematical process is applied to underwater acoustic signals received by sonar. It is shown how this mathematical process analyzes: (1) a pulse emitted by a source with part of the return signal; (2) a sonar signal that travels through important subsurface biological sea realms; and (3) an echo coming from a target [a CTD (conductivity, temperature, depth) Guildline] located at a depth of 800 m.

Underwater signaling device

Underwater signaling device

Uncertainty tolerance in underwater acoustic signal processing

Because of the difficulties in accurately characterizing the statistical behavior of underwater acoustic channels, tolerance to uncertain statistical modeling is an important property for underwater acoustic signal processing procedures to possess. This paper provides an overview of a number of techniques developed in recent years that can be applied to provide uncertainty tolerance in underwater signal processing applications. In addition to a discussion of general concepts of uncertainty tolerance, specific methods for attaining tolerance to uncertainty in temporal/spatial statistics for procedures such as Wiener and matched filtering, time-delay estimation, sonar system design, and signal prediction are reviewed. Tolerance to uncertainty in amplitude statistics is also a key issue in underwater channels, and techniques for achieving this goal are discussed in the contexts of signal estimation, identification, and detection procedures.

Expansions of the likelihood ratio and applications

The likelihood ratio receiver, although optimal, is oftentimes difficult to realize. To overcome this difficulty, suboptimal approximations have been studied in the past. One of the simplest approaches is to expand the likelihood ratio and truncate this expansion under the weak signal assumption. In this paper, we first study the relations between the Taylor expansion and the Volterra expansion. Under a weak signal assumption, we emphasize the links between the Taylor expansion with respect to the SNR and the detection criteria on one hand, and with robustness on the other. In particular, using two criteria derived from deflection, we show that the minimax robust filter is the matched filter. Finally, an application to underwater signal detection is studied, showing that the degradation resulting from the use of the locally optimal receiver is small.

Measurements of the source levels of shallow explosive charges

Experiments have been carried out to measure the source levels of standard 0.82 kg SUS (Signal Underwater Sound) charges at nominal depths of 18, 91, and 180 m. The source levels are reported in this paper for 1/3 octave bands from 5 to 500 Hz. Approximately 30 charges for each depth were included in the analysis, and the standard deviation was less than 1 dB for each frequency band. The source levels estimated from these data were compared to the previous results of Gaspin and Shuler which were based on hand drawn pressure waveforms. Their values are generally within 2 or 3 dB of the measured source levels but there are some greater differences. Reasons for the disagreement between the two sets of values are suggested. In addition to the measured source levels, source levels for 0.82 kg SUS at 244 and 610 m, and for 22.7 kg (50 lb) charges at 100 and 200 m have been estimated using a scaling technique [R. Hughes, in Underwater Acoustics and Signal Processing (Reidel, New York, 1981), pp. 87–91]. The spec...

Measurement of the waveform parameters of shallow explosive charges

Measurements of the properties of shallow underwater explosions have been obtained in recent experiments with 0.82-kg SUS (signal underwater sound) charges at depths from 23.5 to 194.5 m. The data were recorded in the farfield with a high-resolution digital recording system to ensure good reproduction of both the high- and low-frequency components of the waveform. The measurements provide new results for the pressures, impulses, durations, and time constants of the shock wave and the first bubble pulse, and in addition, extend the available information to include the properties of the subsequent bubble pulses and the negative phases. Comparison of the data with previously published semiempirical relationships based on experiments with deeper charges indicates that those expressions generally do not provide adequate descriptions of the shallow charge data. In particular, there is significant disagreement with the measurements of the shock wave decay constant, and with the measurements of all of the bubble phase properties. The effect of vertical migration is appreciable for shallow charges and was observed in the measurements of bubble pulse periods, pressures, and impulses. Semiempirical relationships were derived from the data for the properties of both the shock wave and the bubble pulse series. These expressions account for the depth dependence due to migration observed in the bubble phase properties and consequently provide parameter values appropriate for modeling the pressure waveform of shallow charges.