自适应回波抵消中变步长NLMS算法

Variable Step-Size NLMS Algorithms in Echo Cancellation

在对变步长归一化最小均方误差(Variable step size normalized least mean square,VSS-NLMS)的几种算法以及各个算法在远端和双端通话模式下的性能分析比较的基础上,对NEW-NPVSS(NEW non-parametricVSS)算法进行了改进。在双端通话的情况下改进算法具有更好的收敛性;然后提出了基于滤波器系数梯度的变步长新算法,当滤波器系数梯度小于门限值时,采用固定步长更新滤波器系数。反之,则停止更新滤波器系数,并且用远端模式下的系数替代当前系数。仿真结果表明所提出的算法在远端通话模式下比其他VSS-NLMS算法具有更好的收敛性,在双端情况下具有比固定步长NLMS(Normalized least mean square)和SVSS(Simple VSS)更好的收敛性。

By analyzing of variable step size variable step size normalized least mean square （VSS-NLMS） algorithms and comparing their performance in far-talk and double-talk mode, an improvement is implemented in the NEW non-parametric VSS（NEW-NPVSS） algorithm. Then a new variable step-size algorithm is proposed based on the filter coefficient gradient. When the filter coefficient of gradient is less than threshold value, the fixed step-size is used to update the filter coefficients. Conversely, it stops updating the filter coefficients, and uses the coefficient in far-talk mode to replace the current coefficient. The simulation results show that the proposed algorithm has better convergence than the other variable step-size NLMS algo- rithms in far-talk mode, as well as the fixed step normalized least mean square（NLMS） and simple VSS（SVSS） in double-talk mode.