Adaptive Laguerre network realization

Abstract An adaptive scheme is developed to realize the set of linear discrete-time (DT) systems by the Laguerre networks using gradient search or least mean-squares and steepest descent (LMS/SD) algorithms. Both Laguerre parameter and Laguerre coefficients are optimally evaluated by the adaptive scheme. Two suboptimal procedures called approximate and modified methods are addressed and introduced, respectively, to initialize the Laguerre parameter in its adaptive evaluation. The Laguerre coefficients are also evaluated by two adaptive algorithms presented as bilinear and direct methods. This adaptive scheme simply realizes the linear DT systems either in time or frequency domain. The application of linear DT modeling and filter design is illustrated via some examples.