Abstract: LMS（Least Mean Square） algorithm has been studied for the echo problem of IP call center. The study shows that there is a contradiction between the convergence rate and steady-state error. In other words, speeding up the convergence rate, the steady-state error will be increased; reducing steady-state error, the convergence rate will be slow. The purpose of study is to improve the contradiction and better eliminate echo. A new LMS algorithm is proposed. The algorithm is simulated by using computer, and echo is eliminated by using DSP. Result is that this algorithm has good convergence and steady-state performance. It can better improve the contradiction between the convergence speed and steady-state error, and the elimination result is good.

Key words: call center, echo cancellation, Least Mean Square（LMS） algorithm

摘要： 针对IP呼叫中心系统中的回声问题，对算法LMS（Least Mean Square）进行了研究。经研究发现，回声消除算法LMS在收敛速度与稳态误差之间始终存在着矛盾，即加快收敛速度，则稳态误差随之加大；减小稳态误差，则收敛速度随之减慢。研究的目的就是能够使两者之间的矛盾得到改善，能够更好地消除回声。研究的方法是在现有算法的基础上，提出了一种改进的LMS算法，通过计算机对该算法进行了仿真，以及利用DSP进行了回声消除，结论表明该算法具有良好的收敛性能和稳态性能，更好地改善了收敛速度与稳态误差之间的矛盾，消除效果较好。

关键词: 呼叫中心, 回声消除, 最小均方误差算法