An automatic step-size adjustment algorithm for LMS adaptive filters, and an application to channel estimation

We propose a least-mean-square adaptive filter with automatic step-size adjustment (ASSA). At each time instant when a new observation of the input signal arrives, a new step-size parameter is chosen such that the sum of the squares of the measured estimation errors up to that current time instant is minimized. This step size, after being normalized by the power of the current tapped filter input, is used to update the filter weights for the next time instant. The filter weights are thus updated automatically without the aid of any preset control parameters. When applied to channel estimation, simulation results show the performance advantage of the ASSA algorithm over the existing step-size adjustment algorithms under different wireless channel environments.