An LMS adaptive algorithm with a new step-size control equation

In this paper, we introduce a new variable step-size LMS (VSSLMS) adaptive algorithm. The algorithm step-size equations estimate an optimal derived step-size and are controlled by only one parameter. Mean-square performance analysis is provided for zero-mean stationary Gaussian input signal, and a simple expression that predicts the algorithm steady state misadjustment is derived for small step-size fluctuations. The algorithm is compared with other well-known VSSLMS algorithms through simulation experiments, which demonstrate the performance advantages of the proposed algorithm over these algorithms.