The Study Of Adaptive Echo Cancellation Method In Communication Systems

People's requirements of voice communication quality have become increasingly enhancing as developing of the communications, digital signal processing and LSI technology. However, achieving satisfactory voice quality is not an easy task because of the presence of echo and other reasons. Adaptive echo canceller can offset the echo interference of the voice communication effectively, and improve the communication quality obviously. Echo canceller is generally consisted of adaptive filter, single tone signal detector, far signal detector, near signal detector, nonlinear processor of the remaining echo and comfortable noise generator and other accessories. Accurate and reliable double talk detection is very important for improving the performance of echo canceller, also is the difficulty of echo canceller design. Acoustic echo cancellation is more difficult than the electrical echo cancellation caused by some reasons.Several common and typical adaptive echo cancellation algorithm were discussed in this paper in detail, and they were given comparisons and analysis. LMS algorithm is the most basic adaptive algorithm, has a very important theoretical significance and application value, but its convergence speed is related to the power of the input signal. The convergence factor is normalized by NLMS algorithm using the input signal power, in fact NLMS algorithm is a normalized variable step-size algorithm.The use of new information of RLS algorithm is more fully than the LMS algorithm, and RLS algorithm has fine adjusting nature, so its convergence speed is faster, but the computation of RLS algorithm is very larger than the LMS algorithm. Affine projection algorithm can be adjusting flexibly between the convergence speed and computational complexity based on the actual application, can be minimizing the cost under the premise of meeting the performance requirements of the system.Traditional fixed step size LMS algorithm is simple and easy to deal with real-time, has small amount of calculation and good stability, but the demand between its convergence speed and steady-state error is contradictory, the problem was well resolved by the variable step size LMS algorithm. But variable step size LMS algorithm is more sensitive to noise, does not apply at low SNR environment. The paper proposes the concept of unrestricted and restricted variable step-size LMS algorithm, gives more analysis about an improved variable step size LMS algorithm, the algorithm has a good performance in low SNR environment.Echo cancellation systems based on fixed step size LMS algorithm, variable step size LMS algorithm and improved variable step size LMS algorithm are simulated and analyzed in this paper. The convergence speed and steady-state error of variable step size LMS algorithm have greatly improved than that of the fixed step-size LMS algorithm, the steady-state error of improved variable step-size LMS algorithm is reduced numerously. In this paper, the adaptive echo cancellation system simulation model and test model are setting up using Simulink platform, the system is simulated and demonstrated, simulation time and cancellation effect of LMS algorithm at different step size are analyzed. Simulation results confirm the validity of the method, the distortion is little and the dynamic process of the system is better reflected.Figure 40 table 1 reference 51...