An enhanced proportionate NLMF algorithm for group-sparse system identification

A novel adaptive filtering algorithm is devised and derived for group-sparse system identification. To adequately make use of the group-sparsity in satellite communication and network echo channels, we integrate a mixed-norm constraint into the proportionate normalized least mean fourth (PNLMF) algorithm, which is referred as mixed-norm constrained PNLMF (MNC-PNLMF) algorithm. The MNC-PNLMF algorithm is derived and analyzed in detail. Serval experimental experiments are constructed to validate the effectiveness of the MNC-PNLMF. The experimental results demonstrate that the MNC-PNLMF outperforms the NLMF, PNLMF, zero-attraction NLMF (ZA-NLMF), and reweighted ZA-NLMF (RZA-NLMF) for group-sparse system identification.