A new approach to variability of step size in LMS algorithm and boundary setting

The LMS algorithm has a standard fixed step size. This paper introduces a new dynamically varying step size and set its boundary. At the beginning, the newly introduced variable step size algorithm facilitates an option to select the highest possible value of the overall step size and thus increases the convergence rate dramatically. However, as the iteration progresses, it facilitates an option to choose the least possible value of the step size, which in turn reduces the instability problem of the classical LMS algorithm. A boundary has been defined in such a way that the variability of the overall step size changes from maximum to minimum.