Subband adaptive filtering: Oversampling approach

Fast FIR filtering technique is generalized to propose a new oversampled subband filtering structure. Complexity reduction of the order of (M/2) is achieved in filtering by restricting non-adjacent filters of PRFB non-overlapping. The channel filters of the proposed structure are shown to be related to the modelling system through the PRFB filters. The subband adaptive filtering algorithm is developed by making channel filters adaptive. An adaptive algorithm of NLMS type is derived for main channel filters by minimizing the mean-squared-error (MSE) criterion. The cross filters are derived from the main channel filters. Simulations reveal the modelling capability and convergence properties of the proposed algorithm