THE AVERAGED, OVERDETERMINED AND GENERALISED LMS (AOGLMS) ALGORITHM (ThuPmPO3)

Author(s) : Enrique Alameda Hernandez (School of Electronic and Electrical Engineering. University of Leeds, United Kingdom) Diego P. Ruiz (Department of Applied Physics. University of Granada, Spain) David Blanco (School of Engineering and Electronics. The University of Edinburgh, United Kingdom) Desmond C. McLernon (School of Electronic and Electrical Engineering. University of Leeds, United Kingdom) Maria C. Carrion (Department of Applied Physics. University of Granada, Spain)

Abstract : This contribution presents a new algorithm of the LMS family, derived from a novel orthogonality condition that holds for overdetermined problems that include an instrumental variable. This instrumental variable can be used to introduce higher-order statistics information. The convergence of the MSE for this new algorithm is theoretically studied, together with its superior performance when compared with other similar algorithms, under quite general hypotheses. The algorithm is then applied to the blind identification of moving average models; simulation results verify the analysis.

Menu