PARAMETER-ESTIMATION OF LINEAR-SYSTEMS WITH INPUT-OUTPUT NOISY DATA - A GENERALIZED LP NORM APPROACH

In this paper, the parameter estimation of linear systems with input-output noisy data is considered. The system input and output are supposed to be corrupted by measurement noises, and the noise distributions are assumed to be unknown. For achieving an efficient parameter estimation, an l(p) norm iterative estimation algorithm (1 < p < infinity) is proposed. Furthermore, a method for achieving the consistent estimation is presented. Since the exponent p of the l(p) norm estimation algorithm is essentially sensitive to the noise distribution, based on the sample kurtosis of the residual, an adequate exponent p can be selected to achieve an efficient parameter estimation at each iteration step. Finally, several simulation results are presented to illustrate the proposed generalized l(p) norm iterative estimation algorithm, and we find that the proposed algorithm is a good approach to the system parameter estimation problem with unknown in-put-output measurement noise distributions.