Research On Robust Affine Projection Adaptive Filtering Algorithm And Its Application

The adaptive filter has the prior knowledge of the input signal without the prior knowledge of the input signal,so it can dynamically adjust the advantages of the adaptive filter coefficients.Therefore,it has a wide range of applications in echo cancellation,system identification,channel equalization and so on.Adaptive filtering algorithm directly affects the effect of filter is good or bad which has always been the research hotspot of signal and information processing fields.With the continuous expansion of the application fields,the traditional adaptive filtering algorithms face new challenges: the input signals in the multimedia communication system have strong correlations,non-Gaussian noises widely exist in the real world,some of the communication system channels have long impulse response and sparsity,the choice of step size parameters has limitations between the algorithms convergence rate and the steady-state error and so on.Affine Projection Algorithm(APA)is a class of important adaptive filtering algorithm.It is developed from the Normalized Least Mean Square(NLMS)algorithm based on l2 norm optimization criterion.But compared with the NLMS class algorithm,more suitable for the situation of input signal has highly correlation.However,non-Gaussian noises widely exist in the real world,which seriously destroys the APA adaptive filtering algorithm based on l2 norm optimization criterion.In this thesis,based on the study of the traditional APA algorithm,considering the influence of non-Gaussian noises,an Affine Projection Sign Algorithm(APSA)based on the l1 norm optimization criterion is deduced.The APSA algorithm combines the good convergence characteristic of APA algorithm and the suppression capabilities of the Sign Algorithm(SA)for non-Gaussian noises.Furthermore,considering the sparse characteristics of hands-free calling system,in this thesis,variable step-size method and proportional matrix thought are combined together,and the step size function is developed.Finally,a robust affine projection sign adaptive filtering algorithm—the Variable Step-Size Improved Proportionate Affine Projection Sign Algorithm(VSS-IPAPSA)is proposed,which alleviates the contradiction between convergence speed and steady-state misadjustment,as well as increases its adaptability of the system to different degree of sparsity and noise characteristics.Theoretical analysis and simulation result demonstrate the effectiveness and robustness of the algorithm.