A new efficient and fast adaptive filtering structure based on wavelet packet transform

Signal processing has benefited a great deal from the wavelet transforms. The basic idea underlying the proposed method is to decomposes the spectral content of the signal under analysis into uniform frequency sub-bands, according to the desired frequency resolution, and then, to treat them individually in order to applying several adaptive algorithms for corresponding packets, simultaneously. Afterwards, unlike the other usual sub-band filtering algorithms, adaptive algorithms are applied independently to each pair of input and corresponding desired packets. Simulation Results verify the good performance of the new algorithm in terms of convergence speed and steady state error.