机构地区: 华南理工大学电子与信息学院电子与通信工程系
出 处: 《系统工程与电子技术》 2004年第4期542-543,559,共3页
摘 要: 相关函数递推最小二乘(CRLS)算法在回波消除中双方对讲情况下是有效的,但其计算复杂度较高。把相关函数最小二乘准则中的遗忘因子视为时变的遗忘因子,应用最速下降法使当前时刻的方向矢量正交于前一时刻的方向矢量,从而获得时变的遗忘因子的表达式,得到一种新的相关函数自适应滤波算法。该算法的计算复杂度比相关函数递推最小二乘算法的要低。计算机数值仿真结果表明,新算法的收敛性能和相关函数递推最小二乘算法的收敛性能相当。 The correlation function recursive least square algorithm (CRLS) is effective for double-talk echo canceling, but the algorithm has high computional complexity. The steepest descent method to the correlation function least squares criterion is applied. A time varing forgetting factor and a new correlation function adaptive filtering algorithm are proposed by orthogonalizing the previous direction vector to the current direction vector. The computional complexity of the proposed algorithm is less than that of the correlation function recusive least square algorithm(CRLS). It is shown through the computer numerical simulation that the proposed algorithm's convergence performance is as good as that of CRLS.