Abstract:
A new nonlinear method of processing digital time series data is presented. The method is sensitive to fluctuations in the amplitude of the signals, attenuating them more severely as the amplitudes of the fluctuations increase. In the case of a background noise that has high amplitude fluctuations and a signal that has small amplitude fluctuations, the new method results in significant enhancements in signal-to-noise ratio (SNR) from 5--15 dB and other highly desirable characteristics including increased spatial and spectral resolution, unalerted detection and identification of stable tonals from submerged sources, and reduction in tonal clutter. Results are presented for two fluctuation sensitive processors designated WISPR and AWSUM. It is concluded that through exploiting amplitude fluctuations, these new processors access a new source of gain and are thus able to achieve higher levels of performance than had previously been imagined. Furthermore, they do it with very little additional processing overhead. [Work supported by ONR and NRL.]