Variable Step-Size LMS Algorithm With a Gradient-Based Weighted Average

This paper proposes a new variable step-size least-mean-square (VSLMS) algorithm with an approach in which a gradient-based weighted average is used to improve the weakness of previous VSLMS for application to unknown channel estimation or system identification in low-SNR or colored input environments. The proposed scheme leads to a faster convergence rate and a lower misadjustment error.