Statistical characterization of a composite-feature directed deconvolver

The purpose of the composite-feature directed deconvolver is recovery of information from an additive combination of an intermittent signal-of-interest (SOI) and a stationary background signal-not-of-interest (SNOI). This device exploits specific a priori knowledge of the properties that characterize narrow-band, frequency-agile communications signals. This study is intented to establish performance benchmarks for neural-network based adaptive signal processors under development at the Office of Naval Research sponsored Center for Neural Engineering at Tennessee State University. Specifically, we examine SNR improvements in the -12 to 12 dB regime for on-off keying (OOK) signals against backgrounds consisting of bandlimited Gaussian noise, wideband FM, and combinations of these. We desire to find the best signal recovery scheme using conventional signal detection and linear filtering methods