Research On Adaptive Filtering Method For Nonlinear Acoustic Echo Cancellation

The continuous development of communication technology has overcome the regional obstacles.People's communication mode has ushered in a great change,which can be instant communication,more convenient and fast.However,acoustic echo is inevitable which will affect the normal communication between the two sides of the call and adversely affect the user experience.At present,adaptive filtering echo cancellation is an effective echo cancellation method.However,in the face of nonlinear echo,the traditional adaptive filtering method can't meet the needs.For the problems in nonlinear echo cancellation,this thesis makes some improvements based on the existing adaptive filtering methods as follows:(1)Aiming at the nonlinear echo that is difficult to be eliminated by the basic linear echo canceller in single channel echo cancellation,a pipelined functional link adaptive filtering echo cancellation method based on convex combination is proposed.The nonlinear part of the method is composed of multiple pipeline function link filters,and the linear part is composed of tapped delay line filters.The output signals of the two parts are convex combined,which has the advantages of cascade nonlinear module and convex combination.(2)Aiming at the problem of large attenuation of echo cancellation ability of traditional echo cancellation methods in multi-channel echo cancellation,a nonlinear dual channel echo cancellation method based on Cooperative spline adaptive filter is proposed.The filter structure consists of two parallel collaborative spline adaptive filters,and each collaborative filter includes two branches of linear and spline adaptive filters.The two input signals pass through their respective collaborative filters,and then the four output signals are combined together to update the weight of the corresponding filter with different error signals.(3)Aiming at the echo cancellation performance of the algorithm based on the minimum mean square error is greatly reduced in the non-Gaussian noise scene,a dual channel collaborative functional link adaptive filtering nonlinear echo cancellation method based on the minimum deviation criterion is proposed.This method uses the cooperative combination adaptive filter composed of linear adaptive filter and functional link adaptive filter to replace linear adaptive filter,and updates the filter weight with the minimum dispersion criterion to achieve better echo cancellation effect in the context of non-Gaussian noise.