Wavelet Domain Proportionate Adaptive Filtering Algorithms for Impulsive Noise Environments Under Correlated Input Conditions

This paper proposes wavelet domain adaptive filtering algorithms for sparse system identification under impulsive noise and correlated input conditions. Different algorithms are obtained by first transforming the input to wavelet domain and then combining proportionate adaptation with error nonlinear adaptive filtering techniques. A comprehensive comparative study of the derived algorithms is carried out considering two important parameters which are steady state mean square deviation (MSD) and rate of convergence to show that the proposed algorithms significantly outperform their time domain counterparts.