Based On The Theory Of Dual System Coupling Research Of Adaptive Filtering Algorithm

With the development of communication technology, adaptive filter as the filter of amodern, widely used, it can automatically adjust the iteration parameters satisfy certaincriteria in the case of optimal filtering, and these guidelines is the adaptive filteralgorithm. The quality of the effect of adaptive filtering is the key to the merits offiltering algorithms. Researchers have been in the pursuit of fast convergence, stability,error is small, low computational complexity and good robustness, and adaptivefiltering algorithm theory.In this paper, based on a large number of domestic and foreign literature in the relatedfields, First, the paper describes the background and significance of adaptive filtertheory, and its four main application areas, as well as the history of the development ofadaptive filter theory and research status. Theoretical basis for the subsequent filteringproblem is given, including wide, strictly stationary random process and the Wienerfiltering problem, Paper outlines the basic principles of adaptive filtering andclassification structure. In the paper of a typical adaptive algorithm to compare thetheoretical study and gives and the corresponding simulation, including the least meansquare error (LMS) algorithm. The article profound analysis of its convergence rate,steady-state performance parameters of the algorithm is not synchronized effects of longand gives a comprehensive evaluation. The paper verifies the theoretical analysis byrepeating computer simulations.For a fixed step size factor in the convergence speed and steady-state error on thedrawbacks to the adaptive algorithm. This paper further describes the normalized leastmean square error (NLMS, normalized leas-mean-square) algorithm, LMS algorithm ofthe coupling method (CLMS, An Adaptive Combination of Two LMS Algorithms),adding momentum term LMS algorithm (MLMS, Momentum least-mean-square), etc.and associated performance analysis. MLMS algorithm for the momentum factor tofocus on analysis, taking into account the size of the momentum factor for the algorithm’s convergence speed and steady-state error. Learning CLMS step size factorin the coupling method, proposed an improved algorithm, coupling the momentumfactor. The new algorithm improves the algorithm convergence rate; steady-state errorperformance is better, tracking performance and stability, without increasing thecomputational complexity. Then the coupling method is applied to the momentumnormalized least mean square error (MNLMS, Momentum normalized least meansquare) algorithm, which is equally good results. With adaptive channel equalization,the computer simulation experiments verify the theoretical results of this improvedalgorithm.