Parameter Estimation of Fractional Low Order Time-Frequency Autoregressive Based on Infinite Variance Analysis

ABSTRACT

The Parameter model analysis algorithms include autoregressive (AR) model, moving average (MA) and autoregressive moving average (ARMA) model. The existing TFAR model is improved, the new fractional low order timefrequency autoregressive (FLO-TFAR) model and the concept of generalized TF-Yule-Walker equation are proposed, fractional low-order covariance is preferred instead of autocorrelation in the model; The parameter estimation of the model is derived, spectrum estimation algorithm based on the FLO-TFAR model is presented, and the steps of the algorithm are summarized. The detailed comparison of the FLO-TFAR SαS model based on fractional low order moment (FLOM) and the Gaussian TFAR model based on autocorrelation is done. Simulation shows that the proposed FLO-TFAR algorithm can carry out high-resolution spectrum estimation, provides better performance than the TFAR algorithm, and is robust.

Keywords:

Generalized TF-Yule-Walker equation d index, Stable distribution, time-frequency autoregressive, time-frequency spectrum.