Generalized Variable Step Size Continuous Mixed ${p}$ -Norm Adaptive Filtering Algorithm

To further improve the performance of the variable step size continuous mixed p-norm (VSS-CMPN) adaptive filtering algorithm in the presence of impulsive noise, a generalized VSS-CMPN algorithm is proposed in this brief. Instead of assuming the probability density-like function λ(p) to be uniform, a linear function is proposed for λ(p) to control the mixture of various error norms. The influence of the selection of the regulating factor (slope of the linear function) is discussed. Besides, the computational complexity as well as the mean-square convergence analysis is presented. Simulations conducted in the system identification scenario demonstrate the superiority of the proposed algorithm over known algorithms.