A switched regressor algorithm for adaptive IIR filtering

In the present work an algorithm for adaptive IIR filtering is proposed, in which the updating of the coefficients can be switched between the Steiglitz-McBride identification method (SMM) and the recursive prediction error method (RPEM), or vice-versa, in order to reach the global minimum of the cost function or its proximity. We also analyse aspects related to the speed of convergence of the SMM and the RPEM, and aspects related to the minimization of the output error in the undermodelled case. This serves to justify the proposed algorithm.