A fast convergent algorithm for identification of noisy autoregressive signals

A fast convergent algorithm for unbiased identification of noisy autoregressive (AR) signals is presented. This algorithm is developed based on a bias correction procedure, but makes use of more autocovariances to estimate the variance of the corrupting noise which determines the noise-induced bias in the least-squares estimates of the AR parameters. Since better estimates of this corrupting noise variance can be attained at earlier stages of the iterative process, the proposed algorithm can achieve a faster rate of convergence. Simulation results are included that illustrate the good performances of the proposed algorithm