Parameter estimation of inhomogeneous AR model expanded with unknown basis

We proposed a new model for non-stationary time series analysis based on th e IAR (inhomogeneous autoregressive) model, and a method for model paramete r estimation when the set of basis is given [1], [2]. In this paper, we fur ther propose a method for parameter estimation including that of basis set: we set a new condition that power of the input sequence is concentrated in low-frequency domain, and developed an iterative estimation method. We fir stly select an initial set of basis, from which new sets are created in ord er to minimize the difference between the model and data. Among new sets of basis, we self ct a good one that gives minimum standard deviation of esti mated frequencies.