L0 Norm Based Improved Proportionate Maximum Correntropy Criterion Algorithm for Sparse System Identification

To overcome the effect of impulsive noise on physical systems an algorithm was developed in the past using an improved proportionate maximum correntropy criterion (IPMCC) based on an adaptive filter incorporating l 1 -norm. This approach yielded satisfactory results for the identification of timevarying sparse physical systems but its execution is limited because it tends to draw the active and dominant coefficients to zero irrespective of their magnitude. To overcome the limitation of the l 1 -norm based IPMCC algorithm, this paper presents a l 0 -norm based IPMCC approach that significantly enhances the convergence rate of the inactive coefficients which influence the sparse system. Simulation findings prove that the suggested technique delivers a rapid convergence and lower steady-state error in contrast to the PMCC and l 1 -norm based IPMCC algorithm.