Blind Estimation of the Ocean Acoustic Channel by Time–Frequency Processing

A blind estimator of the ocean acoustic channel impulse response envelope is presented. The signal model is characterized by a deterministic multipath channel excited by a highly nonstationary deterministic source signal. The time-frequency (TF) representation of the received signal allows for the separation between the channel and the source signal. The proposed estimator proceeds in two steps: First, the unstable initial arrivals allow for the estimation of the source signal instantaneous frequency (IF) by maximization of the radially Gaussian kernel distribution; then, the Wigner-Ville distribution (WV) is sequentially windowed and integrated, where the window is defined by the previously estimated IF. The integral gives the channel impulse response envelope, which turns to be an approximation to the blind conventional matched filter (MF). The blind channel estimator (CE) is applicable upon the following conditions: that the multipath channel contains at least one dominant arrival well separated from the others, and that the IF of the source signal is a one-to-one function. Results obtained on real data from the INternal TIde Measurements with Acoustic Tomography Experiments (INTIMATE'96), where the acoustic channel was driven by an linear frequency modulation signal, show that the channel's envelope detailed structure could be accurately and consistently recovered, with the correlation of the estimates ranging from 0.796 to 0.973, as compared to the MF result     

Inversion of Lloyd Mirror Field for Determining a Source's Track

This paper describes a simple approach for inferring the depth and track of a sound source at short ranges by inversion of acoustic field data at a set of sea bottom hydrophones. At short ranges, the acoustic field consists of a dominant Lloyd mirror (LM) signal from the direct and surface-reflected ray paths and a series of bottom-reflected paths that modulate the LM signal. A computationally efficient propagation model based on the method of images is developed to calculate replica fields for the inversion. The matched field inversion method for inferring the source depth and track is demonstrated using data from an experiment carried out in shallow water off the east coast of Canada. The estimated values were in very good agreement with independent measurements taken during the experiment.     

Track Parameter Estimation from Multipath Delay Information

It is desired to track the location of an underwater acoustic source with range difference measurements from a stationary passive array. Many times, the array has only one or two sensors, and the multipath and intersensor range difference measurements are insufficient to localize and track a source moving along an arbitrary path [1]. Here, we propose to track sources with one- or two-sensur stationary passive arrays by making the simplifying assumption that the source's path can be described by a small set of so-called track parameters. Range difference information can then be used to estimate the track parameter set rather than the source location as a function of time. In this paper, we choose the track parameters to specify a straight-line constant-velocity constant-depth path. Cramer-Rao bounds are presented for estimating these track parameters from the time history of multipath and intersensor range difference measurements. It is shown that this track parameter set cannot be accurately estimated from the time history of a single multipath range difference without side information (an independent velocity estimate, for instance), although multipath and intersensor range difference measurements from a two-sensor array are generally sufficient to estimate the track parameter set. Computationally efficient techniques are presented which estimate track parameters from range difference measurements taken from one- and two-sensor arrays. Monte-Carlo simulations are presented which show that these techniques have sample mean-square error approximately equal to the Cramer-Rao bound when a single multipath range difference and an independent velocity estimate are available. The sample mean-square error is shown to be in the range of two to ten times the corresponding Cramer-Rao bounds when these techniques are applied to two-sensor range difference data.     

Sensor position estimation and source ranging in a shallow water environment

Underwater acoustic transient signals are generated mechanically at known positions along a wharf. These signals are received by a wide aperture planar array of four underwater acoustic sensors, whose positions relative to the wharf are unknown. A method is described that enables the positions of the sensors to be estimated from accurate differential time-of-arrival measurements (with 0.1 /spl mu/s precision) as the signal wavefronts traverse the array. A comparison of the estimated positions with the nominal positions of the first three sensors, which form a 20-m-wide aperture horizontal line array, reveals a 2-cm displacement of the middle sensor from the line array axis. This slight bowing of the line array results in overranging (bias error of 3%) when the wavefront curvature method is used with the nominal collinear sensor positions to locate a static source of active sonar transmissions at a range of 59.2 m. The use of the spherical intersection method coupled with the estimated sensor positions of the line array provides an order of magnitude improvement in the range estimate (within 0.3% of the actual value). However, systematic ranging errors are observed when the sound propagation medium becomes nonstationary. Next, the differences in the arrival times of the direct path and boundary-reflected path signals at the middle sensor of the wide aperture line array are estimated using the differential phase residue of the analytic signal at the sensor output. These multipath delays are used to estimate the range and depth of the source. Although the average value of the multipath range estimates is within 0.5% of the actual value, the variance of the range estimates is 50 times larger when compared with the results of the spherical intersection and wavefront curvature methods. The multipath delay data are also processed to provide a reliable estimate of the temporal variation in the water depth enabling the tidal variation to be observed.     

A New Method for Matched Field Localization Based on Two-Hydrophone

The conventional matched field processing (MFP) uses large vertical arrays to locate an underwater acoustic target. However, the use of large vertical arrays increases equipment and computational cost, and causes some problems such as element failures, and array tilting to degrade the localization performance. In this paper, the matched field localization method using two-hydrophone is proposed for underwater acoustic pulse signals with an unknown emitted signal waveform. Using the received signal of hydrophones and the ocean channel pulse response which can be calculated from an acoustic propagation model, the spectral matrix of the emitted signal for different source locations can be estimated by employing the method of frequency domain least squares. The resulting spectral matrix of the emitted signal for every grid region is then multiplied by the ocean channel frequency response matrix to generate the spectral matrix of replica signal. Finally, the matched field localization using two-hydrophone for underwater acoustic pulse signals of an unknown emitted signal waveform can be estimated by comparing the difference between the spectral matrixes of the received signal and the replica signal. The simulated results from a shallow water environment for broadband signals demonstrate the significant localization performance of the proposed method. In addition, the localization accuracy in five different cases are analyzed by the simulation trial, and the results show that the proposed method has a sharp peak and low sidelobes, overcoming the problem of high sidelobes in the conventional MFP due to lack of the number of elements.     

Underwater acoustic channel models for 4800 b/s QPSK signals

Two computer models are presented, one for short-range and one for long-range propagation of acoustic signals through an underwater channel from a transmitter to a receiver. In the short-range model, the received signal is due to a direct path (steady component) and a random path (diffused component) that could be the result of boundary scattering. For the long-range case, the received signal is the superposition of a number of time-delayed, randomly propagated components arriving by different paths. Both models assume perfect transmitter-receiver synchronization but use realistic channel time delays. They demonstrate the time-varying characteristics of underwater acoustic channels and are used in simulations to evaluate the performance of the detection technique     

Source localization in shallow water in the presence of sensorlocation uncertainty

The problem of source localization in shallow water in the presence of sensor location uncertainty is considered. The Cramer-Rao Bound is used to carry out a feasibility study for the joint source and sensor location problem when the multipath propagation channel is modeled as a known, deterministic waveguide. Unlike the free-space propagation channel, the boundedness of the shallow-water waveguide along its vertical axis provides the key to joint determination of the source and sensor location parameters. It is seen that, when a set of intuitive identifiability conditions are satisfied, numerical examples indicate that, for the scenarios considered, the resulting loss in accuracy with which the source location can be estimated due to sensor location uncertainty may be tolerable     

Adaptive Surface Interference Suppression for Matched-Mode Source Localization

Localizing a quiet submerged target in the presence of loud interfering surface ships is an important problem for matched-field processing (MFP) in shallow water. Typically, a data-driven interference suppression scheme is employed which requires neither prior information of the interferer's location nor filter design optimization and iterative estimation. However, the target and the interferers are usually in motion resulting in spreading or mixing of signal energies in their subspaces, thus making it difficult to determine the interference subspace dimension. In this paper, we exploit the difference in modal amplitudes for surface and submerged sources by eigenanalysis of the modal cross-spectral density matrix (CSDM). Simulation and experimental data results show that the interference subspace can be estimated adaptively and the beam output for the target is enhanced.      

Forward Scattering Observation With Partially Coherent Spatial Processing of Vertical Array Signals in Shallow Water

In this paper, we address the problem of detecting an inhomogeneity in shallow water by observing changes in the acoustic field as the inhomogeneity passes between an acoustic source and vertical line array of receivers. A signal processing scheme is developed to detect the perturbed field in the presence of the much stronger primary source signal, and to estimate such parameters as the time when the inhomogeneity crosses the source-receiver path, its velocity, and its size. The effectiveness of incoherent, coherent, and partially coherent spatial processing of the array signals is evaluated using models and data obtained from experiments in a lake. The effect of different bottom types is also considered, and it is shown that partially coherent processing can have a significant advantage depending on the bottom type. Estimates of the minimum input signal-to-noise ratios (SNRs) for which the diffracted signal can be observed are presented.     

Transiting aircraft parameter estimation using underwater acousticsensor data

Sound from an airborne source travels to a receiver beneath the sea surface via a geometric path that is most simply described using ray theory, where the atmosphere and the sea are assumed to be isospeed sound propagation media separated by a planar surface (the air-sea interface). This theoretical approach leads to the development of a time-frequency model for the signal received by a single underwater acoustic sensor and a time-delay model for the signals received by a pair of spatially separated underwater acoustic sensors. The validity of these models is verified using spatially averaged experimental data recorded from a linear array of hydrophones during various transits of a turboprop aircraft. The same approach is used to solve the inverse time-frequency problem, that is, estimation of the aircraft's speed, altitude, and propeller blade rate given the observed variation with time of the instantaneous frequency of the received signal. Similarly, the inverse time-delay problem is considered whereby the speed and altitude of the aircraft are estimated using the differential time-of-arrival information from each of two adjacent pairs of widely spaced hydrophones (with one hydrophone being common to each pair). It is found that the solutions to each of the inverse problems provide reliable estimates of the speed and altitude of the aircraft, with the inverse time-frequency method also providing an estimate that closely matches the actual propeller blade rate     

Interrelations between matched-field processing and airborne MTIradar

In matched field processing (MFP) the spatial characteristics of a dispersive wave field are exploited to estimate certain parameters of the acoustic field, such as source location or characteristics of the acoustic channel including environmental parameters. In airborne MTI (AMTI) radar, interfering echoes (clutter) are Doppler colored due to the platform motion. Optimum clutter suppression requires space-time or space-frequency processing. Some thoughts concerning cross-fertilization between these two areas are put forward. In particular, the idea of space-time MFP is stressed. A processor for space-time power estimation is proposed     

Effects of sensor placement on acoustic vector-sensor arrayperformance

We consider the role played by the sensor locations in the optimal performance of an array of acoustic vector sensors, First we derive an expression for the Cramer-Rao bound on the azimuth and elevation of a single far-field source for an arbitrary acoustic vector-sensor array in a homogeneous wholespace and show that it has a block diagonal structure, i.e., the source location parameters are uncoupled from the signal and noise strength parameters. We then derive a set of necessary and sufficient geometrical constraints for the two direction parameters, azimuth and elevation, to be uncoupled from each other. Ensuring that these parameters are uncoupled minimizes the bound and means they are the natural or “canonical” location parameters for the model. We argue that it provides a compelling array design criterion. We also consider a bound on the mean-square angular error and its asymptotic normalization, which are useful measures in three-dimensional bearing estimation problems. We derive an expression for this bound and discuss it in terms of the sensors' locations. We then show that our previously derived geometrical conditions are also sufficient to ensure that this bound is independent of azimuth. Finally, we extend those conditions to obtain a set of geometrical constraints that ensure the optimal performance is isotropic     

Matched-field processing for broad-band source localization

In the Hudson Canyon experiment, a sound source moved at a constant depth in 73 m of water while transmitting four tonals. The signal was received on a vertical array of hydrophones that spanned the water column. The data set from this experiment has become a standard test case for studying source tracking using matched field processing. As part of that process it was important to first determine a suitable environment model and demonstrate the feasibility of matched-field processing. In this paper, we provide the background on the original data processing that was done to accomplish this. Several interesting results emerged from that study. Frequency averaging was demonstrated to be extremely beneficial when used with the Bartlett processor. However, the popular Minimum Variance processor performed poorly. Finally we discuss a very simple approach to combining the energy coherently that provided significantly improved results.     

Temporal resolutions of time-reversal and passive-phase conjugation for underwater acoustic communications

In this paper, we study the temporal resolution of a time-reversal or passive-phase conjugation process as applied to underwater acoustic communications. Specifically, we address 1) the time resolution or the pulse width of a back-propagated time-compressed pulse as compared with the original transmitted pulse; 2) the effectiveness of temporal focusing as measured by the peak-to-sidelobe ratio of the back-propagated or phase-conjugated pulse (both pulse elongation and sidelobe leakages are causes of intersymbol interference and bit errors for communications); 3) the duration of temporal focusing or the temporal coherence time of the underwater acoustic channel; and 4) the stability of temporal focusing as measured by the phase fluctuations of successive pulses (symbols). Binary phase-shift keying signals collected at sea from a fixed source to a fixed receiver are used to extract the above four parameters and are compared with simulated results. Mid-frequency (3-4-kHz) data were collected in a dynamic shallow-water environment, exhibiting high temporal fluctuations over a scale of minutes. Despite this, the channel is found to be highly coherent over a length of 17 s. As a result, only one probe signal is used for 17 s of data. The bit error rate and variance of the symbol phase fluctuations are measured as a function of the number of receivers. They are of the same order as that calculated from the simulated data. The agreement suggests that these two quantities could be modeled for a communication channel with high coherence time. The phase variance can be used to determine the maximum data rate for a phase-shift keying signal for a given signal bandwidth and a given number of receivers.     

Bayesian Sonar Detection Performance Prediction With Source Position Uncertainty Using SWellEx-96 Vertical Array Data

Predicting sonar detection performance is important for the development of sonar systems. The classical sonar equation cannot accurately predict sonar detection performance because it does not incorporate the effect of ocean environmental and source position uncertainty. We propose an analytical receiver operating characteristic (ROC) expression that characterizes the performance of the optimal Bayesian detector in the presence of ocean environmental and source position uncertainty. The approach is based on a statistical model of the environment and a physical model of acoustic propagation, which translates ocean environmental and source position uncertainty to signal wavefront uncertainty. The analytical ROC expression developed in this paper is verified for source position uncertainty due to source motion using both simulated data and real data collected during the Shallow Water Evaluation Cell Experiment (SWellEx-96). The results showed that the primary effect of source position uncertainty on optimal sonar detection performance is captured by the rank that corresponds to the significant eigenvalues of the signal matrix, an ensemble of replica signal wavefronts (normalized acoustic pressure vector) at the receiving array. The results also showed that the proposed ROC expression provides a realistic detection performance prediction for the Bayesian detector for source position uncertainty using real data. The proposed approach to sonar detection performance prediction is much simpler and faster than those using conventional Monte Carlo approaches.     

A simulation tool for high data-rate acoustic communication in ashallow-water, time-varying channel

The paper discusses the development of a simulation tool to model high data-rate acoustic communication in shallow water. The simulation tool is able to generate synthetic time series of signals received at a transducer array after transmission across a shallow-water communication channel. The simulation tool is suitable for testing advanced signal processing techniques for message recovery. A channel model has been developed based on the physical aspects of the acoustic channel. Special emphasis has been given to fluctuations of the signal transmission caused by time-varying multipath effects. At shorter ranges, the temporal variations are dominated by acoustic scattering from the moving sea surface. Therefore, the channel model produces a coherence function which may be interpreted as a time-varying reflection coefficient for the surface scattered acoustical path. A static, range-independent ray model identifies the significant multipaths, and the surface path is modulated with the time-varying reflection coefficient. The advantages and limitations of the channel model are discussed and assumptions necessary to overcome the limitations are emphasised. Based on the assumptions, an algorithm has been developed and implemented to model how a binary message will be modulated when transmitted by a transducer, is distorted in the channel and finally is received by a transducer array     

Event enhancement of reflections from subsurface layered media

The problem of obtaining amplitude and delay parameters associated with the response of subsurface layered sedimentation to impulse-type acoustic excitations is considered. We use a linear lossless model which characterizes the subsurface in terms of layerwise homogeneous segments. The parameters of the model are the time delays associated with the wave propagation in the various layers and amplitude parameters which are functions of the reflection coefficients at the interfaces. With knowledge of the travel time in a sediment of rock layer, its thickness can be estimated if the medium velocity is known. This paper extends theory developed by the authors. In particular, a procedure to enhance the detection of reflection events is presented. The procedure utilizes a balance property shown by the derivative of the input waveform signal to obtain an improved display of reflection-event portions of the received signals. Computational results are given in the paper to illustrate the effectivess of the procedure.     

台风“天秤”对两种深海声道下声传播的影响

将Argo浮标资料与卫星遥感再分析数据相结合,调用基于抛物方程算法的RAM(Range-dependent Acoustic Model)模型,研究了2012年第14号台风“天秤”对不完整深海声道(3 000 m)和完整深海声道(5 500 m)两种水深条件下声传播特性的影响。结果显示:台风对海水的影响局限于表层水体,具体为混合层加厚,混合层内温度梯度接近于零,声速在混合层内正梯度分布;混合层下方一定深度的水体增温,相应的声速也增大。声源在混合层内时,主要对海表层的声传播产生影响,两种水深条件下均出现表面波导声传播模式以及泄漏模式。声源在混合层以下时,不完整深海声道条件下台风使得会聚区向着声源方向靠近;完整深海声道条件下台风对会聚区的位置影响不明显,但声波的翻转深度增加近500 m。     

Remote sensing of ocean sound speed profiles by a perceptron neuralnetwork

A method is developed to estimate ocean sound speed profiles through synthesis of remotely measured environmental data and historical statistics of sound speed obtained at a remotely sensed location. Sound speed profiles are represented by an expansion of empirical orthogonal functions (EOF) of the historical sound speed variation, while the remotely sensed environmental data provide real-time information to determine the expansion coefficients. Environmental inputs are limited to sea surface temperature available from satellite infrared sensors, acoustic time-of-flight and ocean bottom temperature measurable from bottom mounted acoustic and thermal transducers. A multilayer perceptron neural network is implemented to learn the functional transformation from the measured environmental input to the desired EOF coefficient output on a set of representative sound speed profiles. Sea surface temperature, time-of-year, and time-of-flight from the acoustic multipath that maximally samples the vertical sound speed are found to be the dominant inputs. The trained network is computationally efficient and produces estimates for untrained environmental inputs with a mean error of 1.1-4.4 m/s     

Using matched-field processing to estimate shallow-water bottomproperties from shot data taken in the Mediterranean Sea

It is extremely difficult to determine shallow ocean bottom properties (such as sediment layer thicknesses, densities, and sound speeds). However, when acoustic propagation is affected by such environmental parameters, it becomes possible to use acoustic energy as a probe to estimate them. Matched-field processing (MFP) which relies on both field amplitude and phase can be used as a basis for the inversion of experimental data to estimate bottom properties. Recent inversion efforts applied to a data set collected in October 1993 in the Mediterranean Sea north of Elba produce major improvements in MFP power, i.e., in matching the measured field by means of a model using environmental parameters as inputs, even using the high-resolution minimum variance (MV) processor that is notoriously sensitive and usually results in very low values. The inversion method applied to this data set estimates water depth, sediment thickness, density, and a linear sound-speed profile for the first layer, density and a linear sound-speed profile for a second layer, constant sound speed for the underlying half space, array depth, and source range and depth. When the inversion technique allows for the array deformations in range as additional parameters (to be estimated within fractions of a wavelength, e.g., 0.1 m), the MFP MV peak value for the Med data at 100 Hz can increase from 0.48 (using improved estimates of environmental parameters and assuming a vertical line array) to 0.68 (using improved estimates of environmental parameters PLUS improved phone coordinates). The ideal maximum value would be 1.00 (which is achieved for the less sensitive Linear processor). However, many questions remain concerning the reliability of these inversion results and of inversion methods in general