A Modified Variable Step-Size Continuous Mixed p-Norm Algorithm for System Identification

A modified variable step-size continuous mixed -norm (MVSS-CMPN) algorithm is introduced for system identification. The proposed algorithm is applied to serial input serial output (SISO) and auto regressive moving average (ARMA) systems in the presence of impulsive noise. The standard variable step-size continuous mixed p-norm (VSS-CMPN) algorithm employs a uniform weighting function _λn(p) whose value is assumed as one for the calculation of the variable step size, which makes the algorithm parameter dependent. To act on this constraint, a modified VSS-CMPN (MVSS-CMPN) algorithm is introduced where a time-varying weighting function is used in order to avoid the dependence of algorithm on predefined parameters controlling its proportionality and initialisation. From the simulation results, it is shown that the mean square error of the proposed MVSS-CMPN algorithm attains a better steady state and converges faster due to impulsive noise interference.