Robust Constrained Exponential Hyperbolic Cosine Adaptive Filter

Recently, Constrained Adaptive algorithms based on Mean Square Error (MSE) remain prevalent due to low complexity. However, their performance tends to degrade under non-Gaussian noise conditions. Adaptive filtering methods employing correntropy and those utilizing hyperbolic cosine functions demonstrate improved robustness in such environments. Despite their strengths, these algorithms still exhibit steady-state misalignment. In this work, we introduce a robust constrained Exponential Hyperbolic Cosine Function (CEHCAF) algorithm that mitigates misalignment and withstands impulsive noise spikes. Furthermore, we analyze the proposed method’s computational overhead and convergence performance against the Constrained Least Mean Square (CLMS) and Constrained Maximum Versoria Criterion (CMVC) algorithms through both theoretical complexity estimates and simulations.