Subband Adaptive Filter And Its Application

Adaptive filters use an adaptive algorithm to adjust the tap-weights of their digital filter to estimate unknown systems. Least-mean-square(LMS) algorithm is one of the most popular adaptive filtering algorithms due to its low computational complexity and ease of implementation. However, correlated input signals tend to deteriorate its convergence rate. In order to increase convergence rate, an affine projection algorithm(APA) was proposed by scholars, but its computational complexity is high. In order to increase convergence rate and to maintain low computational complexity, subband adaptive algorithms were proposed. This paper firstly proposes a shrinkage variable regularization matrix normalized subband adaptive filter(NSAF) to solve the problem that fixed regularization parameter NSAF must take a trade-off between fast convergence rate and low steady-state misalignment. Then, the NSAF is introduced into two-dimensional adaptive filtering and a two-dimensional NSAF and its variable step-size variants are presented. Lastly, the affine projection normalized subband adaptive algorithm is used in adaptive networks and a diffusion affine projection sign normalized subband adaptive algorithm is developed.