一种新的变步长LMS算法

在对基本LMS算法分析的基础上,通过构造步长因子μ与误差信号e（n）之间的非线性函数,提出一种新的变步长最小均方误差（LMS）算法,并且分析了参数的取值对算法性能的影响。该算法通过调整步长参数,使权向量达到最优,有效改善了收敛速度与稳态误差的性能。理论分析和仿真结果表明,与基本LMS算法以及部分同类变步长LMS算法相比,该算法具有更快的收敛速度和更小的稳态误差,进一步验证了新算法优于这里所述其他算法。

A New Variable Step-Size LMS Algorithm

Based on brief discussion of basic LMS,by constructing a nonlinear function between the step factor μ and the error signal e（n） ,a new variable step-size LMS algorithm is proposed,along with the performance analysis with regard to different parameters.The algorithm,though adjusting the step size parameters,makes the weight vector optimal,thus improves effectively the convergence rate and steady-state error performance.The theoretical analysis and simulation results shows that this algorithm,as is of compared with the basic LMS algorithms and some similar variable step-size algorithms,faster convergence speed and smaller steady-state error.This further indicates that this new algorithm is superior to other algorithms.