Improved affine projection sign algorithm for sparse system identification

Abstract

Proposed is a novel affine projection sign algorithm with L 0-norm cost to improve the convergence rate in a sparse system. The proposed algorithm is robust to impulsive noise due to L 1-norm minimisation. It also ensures improved performance in terms of convergence rate owing to the L 0-norm cost. The simulation results demonstrate that the proposed algorithm improves the filter performance of sparse system identification. ？ 2012 The Institution of Engineering and Technology.