Fast stable normalised least-mean fourth algorithm.

A fast stable normalised least-mean fourth (FSNLMF) algorithm is proposed. The major drawback of the stable normalised least-mean fourth (SNLMF) algorithm is its poor convergence rate, especially in the presence of high signal-to-noise ratios (SNRs). Considering the effect of high SNRs, a corresponding constant is introduced in SNLMF algorithm to offer a fast convergence. Simulations on system identifications demonstrate that the proposed FSNLMF algorithm achieves a faster convergence speed than the SNLMF algorithm. [ABSTRACT FROM AUTHOR]