Signal modeling with stochastic and deterministic data: a lossless inverse scattering approach

It is often desirable to obtain low-order models for signals and systems, which preserve stochastic (steady-state) properties such as autocorrelation coefficients as well as deterministic (transient) properties such as impulse response coefficients of the underlying true model. Recursive and computationally efficient algorithms are developed using the theory of lossless inverse scattering. Moreover, this approach enables one to solve a more general problem, namely, to interpolate the spectral factor and the scattering function associated with the process at complex points.<>