个别时变步长LMS自适应算法

本文提出一种步长自适于输入信号与预测误差相关的个别时变步长最小均方(IVLMS)自适应算法。给出步长的物理解释,导出算法的收敛条件、权二阶矩矩阵差分方程和稳态失调量。将个别时变步长的思想应用于归一化LMS算法和分块LMS算法,均获得良好效果。

An Individual Variable Step LMS Adaptive Algorithm

An individual variable step LMS adaptive algorithm which varies itsstep size adaptively to the correlation between the input signal and prediction erroris proposed. A physical explanation of the step size is presented. The conditionof convergence, a matrix difference equation for the second moment of the weightand the misadjustment of the IVLMS algorithm are derived. The noise power ofthe adaptive weights and the tracking capability of the new algorithm are eviden-tly superior to the conventional LMS algorithm both in stationary and nonstationaryinputs. It is shown that the concept of individual variable step size is also avai-lable to the normalized LMS and block LMS adaptive algorithms.