Research of Adjustment Theory Based on Extended Correlation Least Mean Square Algorithm
|School||Central South University of Forestry Science and Technology|
|Course||Applied Computer Technology|
|Keywords||Echo cancellation Adaptive filtering algorithm Full-duplex environment Variable tap Variable step size|
With the development of science and technology , human step forward to see the era of information , including signal processing has become one of the hot topics of the research . The endless variety of means of communication , including the quality of voice calls more and more to get people's attention , for a long time , call echo and noise has always been the major problem affecting communication quality , especially in a variety of high - delay application of the means of communication , so that these The problem is more and more prominent . To this end , as represented by the minimum mean square algorithm (LMS) adaptive filter is started to be applied. LMS algorithm with a small amount of calculation , easy to implement real-time processing , fast development . However, in the duplex case , LMS can not work normally . Adaptive filtering algorithm based on the correlation function is designed for full-duplex environment designed a new adaptive filtering algorithm . However, the existing adaptive filtering algorithm based on the correlation function of the problem also exists . This article describes two algorithms : the relevant minimum mean square algorithm (CLMS) and expand the minimum mean square algorithm ( ECLMS ) by the experimental data can be seen in the duplex under the conditions of normal work . But the algorithm still exist shortcomings , such as the error signal convergence rate is not high, can not handle long - echo pulse response . Solve these problems , this paper carried out the research work , mainly as follows : ( 1) in the the ECLMS algorithm based on variable tap improvement . Proposed the the variable tap ECLMS algorithm ( TECLMS ) . According to the characteristics of its wiper in the process of operation of the algorithms , the error signal according to the conditions on the front and back twice compared to select the appropriate tap values ??such that the number of taps descending gradually changed , thereby reducing the computational complexity of the algorithm . (2 ) A variable step size the the algorithm of ECLMS ( SECLMS ) , select the exponential function as a step adjustment formula , making the step for a larger value of the error signal , in the hours of the wrong signal to the error signal for a smaller value . The experimental results can be seen the effect SECLMS experimental ECLMS . The new algorithm with existing adaptive filtering algorithm based on the correlation function of the same strain , it is entirely possible to better apply to full-duplex environment .