Diffusion generalized spline nonlinear adaptive filters under Maximum Correntropy Criterion

Due to complex nonlinearities in general, linear adaptive filter is not suitable, the nonlinear adaptive filter using splines based on minimum mean square error criteria is proposed to identificate nonlinear systems in additive Gaussain noise environment. To address the issues of the more general nonlinear system structure and the addictive non-Gaussain noise environments disturbance, this paper proposes generalized spline nonlinear adaptive filters under maximum correntropy criterion (GSNAF-MCC). Meanwhile, GSNAF-MCC is extended to diffusion networks, resulting in the diffusion GSNAF-MCC (D-GSNAF-MCC). The proposed D-GSNAF-MCC decreases the steady-state error and improves the convergence speed. The simulations demonstrate that the proposed algorithms have better performance compared with related SAFs algorithms.