Mimetic wavelet-packet transform based adaptive algorithm for sparse response identification

This paper proposes a new wavelet-packet transform based adaptive algorithm for sparse response identification. The distinctive features of the new algorithm are the on-line adaptability of the discrete wavelet packet transform (DWPT) and an efficient weight deactivation/activation schedule. The new algorithm, called mimetic wavelet-packet based (MWPB) algorithm, generalizes the WPB algorithm presented in. The MWPB algorithm presents better estimation and tracking performances than existing wavelet-based sparse response identification algorithms. Monte Carlo (MC) simulation results compare the performances of the four most recently proposed algorithms in this class.