Research And Implementation Of Adaptive Acoustic Echo Cancellation Algorithm

Echo cancellation needs to be applied into the communication system for there is inevitable echo in the communication system; with the continuous development of communication and the distance of communication is getting farther and farther, the echo sounds more obvious and obvious, and the requirement for the real-time signal is getting higher and higher. As the echo is changing with the time, so the adaptive technology is widely applied into the echo cancellation technology for adaptive filter has the characteristic of tracing the changing input signals.The performance of the adaptive algorithm can directly affect the effect of the echo canceller. When we select a certain adaptive algorithm, we must not only consider the speed of the convergence of the algorithm, the stability of the whole system, but also the less complexity of the algorithm to insure that it is realizable. There are many scholars inside and outside our country have advanced many adaptive echo cancellation algorithms and the corresponding ameliorated algorithms, such as Least Mean Square (LMS) Algorithm, Normalize Least Mean Square (NLMS) Algorithm, Block update (BLMS) Algorithm, Recursive Least Square (RLS) Algorithm, Subband Filter, Frequence Field Adaptive Filter, Affine Projection Algorithm, LMS/Newton Algorithm and Conjugate Gradient (CG) Algorithm, etc.This paper gives an in-depth research and discussion on the RLS and LMS algorithms, analyses their merit and demerit. Considering the realization of the algorithm, this paper suggests a new algorithm–Modified Normalize and Block Update Least Mean Square (MNBLMS) Algorithm and implements it on the TMS320C5410 DSP, optimizes the algorithm and gives tests on the performance of this algorithm.