Edmund J. Sullivan

Extended towed array processing by an overlap correlator

A method of extending towed array measurements that provides an aperture greater than that of the physical array is presented. Such a technique can be used by matched‐field estimators to obtain information about the range and depth of a source and in other towed array applications requiring a very large aperture. The approach is to combine coherently the acoustic signals arriving at a moving array of hydrophones by making proper compensation through a factor that corrects for considerable fluctuations in phase irregularities in the tow path of the physical array as well as fluctuations in amplitude experienced during the coherent integration time. In this manner, the finite aperture of the physical array is exploited in a process that synthesizes the extended aperture of the method. The concept is based on an algorithm that we call an ‘‘overlap correlator,’’ which provides the phase correction factor by correlating overlapping space samples of the acoustic signal received at successive moments by the movi...

Estimation and detection issues in matched-field processing

A conceptual framework in which the model-based, space-time acoustic signal processing procedure known as matched field processing (MFP) can be handled in a consistent manner is established. A framework for strong-signal MFP based on standard statistical estimation theory, in which MFP is regarded as essentially an estimation problem in the strong-signal regime, is developed. In the weak-signal case, the necessary requirement of detection dictates that MFP then be considered a joint detection-estimation task. It is demonstrated that, generally, MFP is essentially a space-time processing problem rather than simply an array processing (spatial processing only) procedure. An overview of the processing schemes used to date in MFP is given, showing how these methods relate to the optimal space-time structure. Weak-signal detection and estimation scenarios relevant to MFP are then noted. Present methods for dealing with the inherent instability of MFP algorithms (mismatch) are discussed. The current status of MFP is summarized, and recommendations for future research are made. >

Space-time array processing: The model-based approach

A method of space–time array processing is introduced that is based on the model-based approach. The signal and measurement systems are placed into state-space form, thereby allowing the unknown parameters of the model, such as signal bearings, to be estimated by an extended Kalman filter. A major advantage of the model-based approach is that there is no inherent limitation to the degree of sophistication of the models used, and therefore it can deal with other than plane-wave models, such as cylindrically or spherically spreading propagation models, as well as more sophisticated representations such as the normal mode and the parabolic equation propagation models. Since the processor treats the parameters of interest as unknown parameters to be estimated, there is no explicit beamformer structure, and therefore no accuracy limitations such as fixed beam bin sizes and predetermined number of preformed beams. After a theoretical exposition of the underlying theory, the performance of the processor is evalu...

Passive localization in ocean acoustics: A model-based approach

A model‐based approach is developed to solve the passive localization problem in ocean acoustics using the state‐space formulation for the first time. It is shown that the inherent structure of the resulting processor consists of a parameter estimator coupled to a nonlinear optimization scheme. The parameter estimator is designed using the model‐based approach in which an ocean acoustic propagation model is used in developing the model‐based processor required for localization. Recall that model‐based signal processing is a well‐defined methodology enabling the inclusion of environmental (propagation) models, measurement (sensor arrays) models, and noise (shipping, measurement) models into a sophisticated processing algorithm. Here the parameter estimator is designed, or more appropriately the model‐based identifier (MBID) for a propagation model developed from a shallow water ocean experiment. After simulation, it is then applied to a set of experimental data demonstrating the applicability of this approach.

Maximum‐likelihood passive localization using mode filtering

The maximum‐likelihood estimator for passive range and depth estimation of an acoustic point source in a shallow‐water waveguide is presented. The data from a vertical array of hydrophones are passed through a modal filter, the output of which is the set of complex modal amplitudes associated with the normal‐mode model of acoustic propagation. The range and depth estimates are then found by a maximum‐likelihood estimation procedure that uses these modal amplitudes as inputs. This technique is compared to the matched‐field procedure and is shown to have better signal‐to‐noise and sidelobe behavior for a given scenario. Results are given for both synthetic and real data. The results with the real data demonstrate the importance of the mode‐filtering property of the maximum‐likelihood estimator presented in this work.

Broadband passive synthetic aperture: Experimental results

Bearing estimation using an acoustically short towed array can be enhanced by the incorporation of a realistic signal model. By casting the problem as a joint estimation of bearing and source frequency, the bearing estimation performance, as measured by the variance of the bearing estimate, can exceed that of the conventional array processor. Experimental results based on the radiated noise of a ferry are shown. Bearing estimation results using an array of acoustic length of approximately two wavelengths are compared to results using the same data but with a conventional frequency domain beamformer. A significant improvement in performance is demonstrated.

Model-based processor design for a shallow water ocean acoustic experiment

Model‐based signal processing is a well‐defined methodology enabling the inclusion of environmental (propagation) models, measurement (sensor arrays) models, and noise (shipping, measurement) models into a sophisticated processing algorithm. Depending on the class of model developed from the mathematical representation of the physical phenomenology, various processors can evolve. Here the design of a space‐varying, nonstationary, model‐based processor (MBP) is investigated and applied to the data from a well‐controlled shallow water experiment performed at Hudson Canyon. This particular experiment is very attractive for the inaugural application of the MBP because it was performed in shallow water at low frequency requiring a small number of modes. In essence, the Hudson Canyon represents a well‐known ocean environment, making it ideal for this investigation. In this shallow water application, a state‐space representation of the normal‐mode propagation model is used. The processor is designed such that it...

Limitations on the overlap-correlator method imposed by noise and signal characteristics

Limitations on the performance of the overlap-correlator method of forming a passive synthetic aperture are derived. The technique uses the overlap of the array in sequential positions to estimate a series of phase correction factors that compensate for the motion of the array over time. It is of primary interest to optimize this overlap with respect to the effects of random noise. By minimizing the variance of the estimates of the set of phase correction factors, it is found that the optimal overlap is one-half the length of the physical array. Using this optimal overlap, the bounds on the usable spatial response are then determined as a function of signal-to-noise ratio and the number of hydrophones in the physical array. The ability of the overlap-correlator algorithm to synthesize a coherent aperture is investigated for the case of multiple sources in the absence of noise. >

Sonobuoy for forming virtual vertical sensing arrays

A sonobuoy for forming a plurality of virtual vertical arrays, the sonobuoy having means for receiving and expelling water to cause said sonobuoy automatically to descend and rise in a water environment, whereby to form sequentially a plurality of virtual vertical arrays.

Model-based passive ranging

Passive localization by use of acoustic propagation models, sometimes called ‘‘matched field processing’’ is usually carried out in three steps. First, some appropriate model is selected. Then, model parameters, usually taken from archival data or from auxiliary measurements, are introduced into the model. Finally, acoustic measurements of the field radiated by the source to be located are made that, in combination with the properly parametrized model, allow a solution for the source coordinates to be carried out. Here, such a model‐based approach is used in conjunction with a normal‐mode model. By coupling the procedure with a parameter estimation/identification scheme and using a horizontal (towed) array instead of the usual vertical array, it is shown that the model parameters need not be known a priori in order to carry out the solution. This is in contrast to the standard approach in which the modal functions must be computed directly from the model (wave equation) in order to solve the problem. It i...