Bilinear representation of non-stationary autoregressive time series

Abstract

This paper considers a class of non-stationary autoregressive systems in which non-stationarity is caused by time varying parameters of the system. Distinction between two or more non-stationary systems based on observation of the output signal only, is difficult and sometimes may be impossible. In this paper a bilinear approximation of non-stationary autoregressive model is proposed. This way, a model with time varying parameters is approximated by a constant parameters model, what can facilitate the distinction between systems.