A new method for parameter estimation of autoregressive signals in colored noise

This paper presents a new method for parameter estimation of autoregressive (AR) signals from colored noise-corrupted observations using a damped sinusoidal model of the autocorrelation function of the noise-free AR signal. Unlike conventional correlation-based techniques, the proposed scheme first estimates the damped sinusoidal model parameters from the given noisy observations using a least-squares (LS) based method. The AR parameters are then directly obtained from the sinusoidal model parameters. Simulation results show that the proposed method performs better at low SNRs as compared to other existing methods.