A new recursive algorithm for estimating the adaptive function coefficients autoregressive (AFAR) models in impulsive noise environment

This paper proposes a recursive algorithm for estimating the adaptive function coefficients autoregressive (AFAR) models. Due to its recursive nature, its arithmetic complexity is relatively lower than conventional methods. Furthermore, a new M-estimation-based AFAR parameter estimation algorithm is developed to suppress the effect of impulsive outliers. Simulation results show that the M-estimation-based algorithm offers improved performance than the conventional LS-based algorithm