An Efficient Implementation of Forward-Backward LMS Adaptive Line Enhancer

Abstract

An efficient implementation of the forward-backward least mean square (FBLMS) adaptive line enhancer is presented. Without changing the characteristics of the FBLMS adaptive line enhancer, the proposed implementation technique reduces multiplications by 25 percent and additions by 12.5 percent in two successive time samples in comparison to that of the direct implementation in both prediction and weight control. It can be applied to acquisition and tracking in both stationary and nonstationary environments.