Volume 118, Issue 3, September 2005
Index of content:
- ACOUSTIC SIGNAL PROCESSING 
118(2005); http://dx.doi.org/10.1121/1.1993149View Description Hide Description
In a shallow oceanwaveguide the acoustic field can be characterized by depth-dependent modes propagating in range with an associated propagating wavenumber. Though recently developed methods for determining the modes from recorded acoustic data alone without ocean or bottom modeling have shown promise, they are only applicable when the acoustic field is sampled over the entire water column. This paper presents a method for determining the acoustic modes from measured data alone when the field is sampled over only a portion of the water column. The method requires broadband sources at many ranges, e.g., a moving source, in order to construct the frequency-wavenumber (f-k) structure of the waveguide. Because modal propagation is dispersive, the modes are characterized by a discrete set of wavenumbers that vary continuously with frequency. Due to the discreteness of the modal wavenumbers, it is possible to isolate the modes in the f-k domain and extract them individually with a singular value decomposition (SVD). Because the modes are extracted individually the full-spanning and degeneracy limitations of the SVD are removed. Theory, simulation, and laboratory data confirm the process.
118(2005); http://dx.doi.org/10.1121/1.1993131View Description Hide Description
The high sidelobe level of the ambiguity surface has been a major drawback of matched field processing in oceanwaveguides compared to the free space beamforming. In this paper, a contrast-maximized optimization scheme that enhances the contrast between the acoustic power output of the source region and the power output of the total region of interest is presented. The method utilizes the signal vectors determined from the contrast-maximizing formulation. Through numerical simulation and oceanic data analysis, the presented algorithm is shown to be effective in improving the contrast between the target and the averaged background sidelobe level. It was also shown that, even when the resolution of the source region in conventional matched field processing is degraded due to severe bottom attenuation, the resolution of the source is not affected in the contrast-maximized method. The same principle and method can be applied to the time-reversal processing to maintain the focal size in an ocean environment with high bottom attenuation.
118(2005); http://dx.doi.org/10.1121/1.1979407View Description Hide Description
In solving nonlinear inverse-scattering problems an iterative approach is usually required, particularly if the nonlinearity is strong. A typical approach is to use a descent algorithm to minimize a global error norm, such as the mean-square error, to arrive at a set of parameters that best predicts the data based on an assumed model of the unknown scattering object. In minimizing the error norm, the possibility of converging to a local minimum can lead to false solutions. A different strategy is proposed that is not based on gradient descent which, at least in principle, should avoid local minima and converge to the true solution (the global minimum). A simple inverse-scattering problem is used as an illustration.