SIGNAL-CONTROLLED TIME-SERIES MODELING BASED ON ARMA BLOCKS, AND SEPARATION OF SUPERIMPOSED, OVERLAPPING SPECTRA SIGNALS

This paper introduces signal-controlled time-series modeling based on arbitrarily chosen building blocks. Such modeling is used in the desig n of a nested-form transversal structure based on Almost-Symmetrical A RMA (AS-ARMA) building blocks. This structure can operate in the trans ient mode, in contrast to the commonly used linear line-enhancers base d on an conventional ARMA, leading to practically sound processing of short-duration signals. It is shown that the proposed time-series mode ling can be effectively applied towards the separation of superimposed signals of heavily overlapping spectra.