Abstract:In order to improve the performance of sparse system identification, the l0 norm constraint LMS algorithm is studied and improved in this paper. Firstly, the convergence of the algorithm is accelerated by the introduction of a step size control method based on the status information provided by mean square estimation error. Secondly, the zero attraction item is reweighted by the absolute estimation error to reduce the steady-state misalignment. Then the parameters in the proposed algorithm, which control the convergence and steady-state misalignment, are discussed qualitatively. Finally, the simulations demonstrate that the proposed algorithm significantly outperforms l0-LMS and several other existing sparse system identification algorithms.