Fast Basis Function Estimators for Identification of Nonstationary Stochastic Processes

The problem of identification of a linear nonstationary stochastic process is considered and solved using the approach based on functional series approximation of time-varying parameter trajectories. The proposed fast basis function estimators are computationally attractive and yield results that are better than those provided by the local least squares algorithms. It is shown that two important design parameters - the number of basis functions and the size of the local analysis interval - can be selected on-line in an adaptive way.