On bias compensation estimation for noisy AR process

This paper focuses on bias compensation estimation of autoregressive (AR) process in the presence of white noise. It is known that bias compensation principle (BCP) based method requires the estimate of unknown noise variance to compensate the bias of least-squares (LS) estimate to provide consistent AR parameter estimate. In this paper, estimation of noise variance in BCP based methods for noisy AR process estimation is discussed from a unified point of view. It is found that some BCP based methods can be explained in a unified form. Computer simulations are also presented to compare these BCP based methods.