Subband Adaptive Filtering Algorithm And Application Research

The subband adaptive filter has a wide range of applications in the fields of communication system identification,speech echo cancellation,etc.The basic idea is to reduce the correlation of the input signal by subband segmentation extraction,thereby improving the convergence speed.However,with the development of its application field,the traditional subband adaptive filtering algorithms are facing new challenges,for example,the non-Gaussian impulse noise,which is ubiquitous in underwater acoustic communication and other practical applications,leads to the performance degradation or even failure of the Subband Adaptive Filter?SAF?algorithm based on l2 norm.In addition,wireless communications,video calls and other systems often show sparse characteristics.Many existing subband adaptive filtering algorithms do not take full advantage of this feature,which leads to the increase of complexity and slow convergence speed when the algorithm is applied to sparse systems.In this thesis,the Sign Subband Adaptive Filter?SSAF?algorithm which is very effective in suppressing non-Gaussian noise,is studied thoroughly and comprehensively.Its performance limitations for sparse systems are discussed through theoretical analysis and simulation experiments.Based on the sparse signal processing theory,firstly,zero attraction factor is introduced into the cost function of the Sign Subband Adaptive Filtering algorithm to increase the l₁ norm constraint of the weight vector in the updating process.Secondly,the idea of proportional matrix is merged,and the proportional step size is set according to the coefficient of the system to be estimated.Finally,the Zero Attractor?-law Proportionate Sign Subband Adaptive Filter?ZAMPSSAF?algorithm is obtained.The algorithm can ensure that the zero coefficient or smaller coefficient of the main part of the sparse system converges faster,while the large part of the large coefficient converges to the optimal value.A large number of simulation experiments show that the proposed algorithm has better convergence performance and steady state performance under different background noise environments in different sparse characteristics systems.At the same time,the new algorithm proposed in this thesis is applied to the echo feedback cancellation system of digital hearing aids.The simulation results also show that the proposed algorithm can effectively suppress the acoustic feedback phenomenon under non-Gaussian and Gaussian backgrounds.