Improved efficient proportionate affine projection algorithm based on l 0 ‐norm for sparse system identification

Abstract

A new improved memorised improved proportionate affine projection algorithm (IMIPAPA) is proposed to improve the convergence performance of sparse system identification, which incorporates l 0 ‐norm as a measure of sparseness into a recently proposed MIPAPA algorithm. In addition, a simplified implementation of the IMIPAPA (SIMIPAPA) with low‐computational burden is presented while maintaining the consistent convergence performance. The simulation results demonstrate that the IMIPAPA and SIMIPAPA algorithms outperform the MIPAPA algorithm for sparse system identification.