LAUSR

The affine projection sign subband adaptive filter (AP-SSAF) was designed by utilizing the ${M}$ most recent input samples to further enhance the performance of the traditional SSAF. However, the abovementioned algorithms only concentrate on the noise interferences that exist in the output of system under system identification application. Actually, the input of adaptive filter could contain the noises as well. Thereby the performance of the above algorithms will inevitably deteriorate. Based on the excellent performance of AP-SSAF, in this brief, through incorporating unbiasedness criteria into the AP-SSAF, the bias-compensated AP-SSAF (BC-AP-SSAF) is devised for diminishing the impact of input noises via introduced bias compensation term, with some necessary assumptions. And the convergence is provided. Monte-Carlo simulations certify that the proposed BC-AP-SSAF holds stability under impulsive interferences and decreases the estimated bias markedly.