Signal processing based on parameter structural modeling and separation of highly correlated signals of known structure

摘要: Results in the study of signal processing based on use parameter structural modeling (PSM) are presented. First, we introduce a special form time-series signal-dependent building blocks. Such is used design nestedform transversal structure, known as composite filter, shift-invariant finite impulse resonse (FIR) well infinite response (IIR) The newly proposed PSM model (CPSM) possesses unique feature, namely, its ability to suppress one given while at same time being ideally transparent another one. intrinsic property this CPSM enhanced insensitivity with respect noise fast track, contrast commonly linear line-enhancer conventional autoregressive moving average (ARMA), thus leading more practically sound short-duration signals. It shown that can be effectively applied towards separation superimposed signals heavily overlapping spectra. Next, parameter-invariant nonlinear representation annihilation operators (AO) described, and parameter-free (CPFSM) developed. Based model, two canonical forms null filters (PINF) presented, their suppression class signals, independently values theira priori unknown parameters, illustrated. paper also presents some simulation examples illustrating application CPFSM solving problems detection estimation presence highly non-Gaussian, mainly signal-like interferences.