PASSIVE TRANSIENT LOCALIZATION USING SIGNAL AUTOCORRELATION MATCHING

摘要

The ability to localize the source position of broadband transient signals from data measured over short aperture arrays is of great practical importance in modern sonar systems. Previous work in the area of matched-field processing has focussed on narrowband signals measured across long vertical arrays with multiple frequencies being combined incoherently. In this work, the robustness of a simple, semi-coherent Bartlett-type processor based on matching broadband signal autocorrelation functions is investigated. Specifically, the autocorrelation of the time-domain signal and the autocorrelation of the frequency-domain response will be matched to received data (both synthetic and measured) at single hydrophone locations. Measures of robustness to be examined include the size of the localization footprint on the ambiguity surface and the peak-to-sidelobe levels in the presence of environmental mismatch. A full-wave PE model is used to produce broadband replicas. Model-generated synthetic signals with complete knowledge of the environment will provide baseline results. Explosive SUS data measured in a complex littoral region will also be analyzed. Some of the specific issues to be addressed are which autocorrelation function provides the most robust match and over what bandwidths, how much environmental mismatch can be tolerated at what level of SNR, what source/receiver geometries are most robust, and can source signal spectra be extracted from the estimated localization position? [Work supported by NUWC, Code 2121.]