A Laguerre-CPWL Adaptive Filter

J.L. FIGUEROA; J. E. COUSSEAU

Rio Cuarto

Congreso; XI Reunión de Procesamiento de la Información y Control; 2005

RPIC

An adaptive nonlinear filter wasstudied in the context of the nonlinearWienersystem representations. The realization assumesa Laguerre polynomial description ofthe linear part and a Canonical Piece-WiseLinear (CPWL) approximation for the nonlinearpart. The resulting structure requires fewerparameters than others realizations foundin the literature with comparable modelingcapabilities. As a consequence, the proposednonlinear filter, in addition to its simple description,has low computation and implementationcosts. Some results related to convergenceproperties and implementation of theadaptive algorithm associated to this new realizationare presented. The performance ofthe proposed filter is illustrated through simulationexamples.