Title :
Model reduction and wideband analysis [underwater acoustic application]
Author :
Le Cadre, Jean-Pierre ; Ravazzola, Patrice
Author_Institution :
IRISA, Rennes, France
Abstract :
Exact model reduction methods are extended to wideband analysis. After a review of the realization methods in the underwater acoustic context, it is shown that the optimal realization method is very well suited to the wideband analysis, because of the rational modeling. The basic steps of the proposed method are a multidimensional AR (autoregressive) modeling of the sensor outputs, spatial interpolation, and model reduction of a unique model for the frequency band of analysis. It is important to note that this method uses no prior assumption about source locations or source models. Moreover, this method inherits the robustness with regard to spatial noise coloration from its narrowband version. Simulation results demonstrate the resolution and detection improvements achieved by the method
Keywords :
acoustic signal processing; underwater sound; exact model reduction methods; multidimensional autoregressive modelling; realization methods; robustness; sensor outputs; signal modelling; spatial interpolation; spatial noise coloration; underwater acoustics; wideband analysis; Acoustic sensors; Context modeling; Frequency; Interpolation; Multidimensional systems; Noise robustness; Position measurement; Reduced order systems; Underwater acoustics; Wideband;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1990. ICASSP-90., 1990 International Conference on
Conference_Location :
Albuquerque, NM
DOI :
10.1109/ICASSP.1990.116077