A novel adaptive step-size algorithm of blind source separation based on non-linear function

The conventional fixed step-size algorithm of blind source separation (BSS) cann't adaptively chang in the process of original signals separation, so both the convergence speed and the steady-state misadjustment aren't simultaneously satisfactory. To solve the problem, a novel step-size algorithm of BSS based on non-linear function is proposed in this paper, it makes the step-size value match the dynamics of the input signals and unmixing matrix adaptively at each iteration, its performance is investigated through simulations. The results obtained show that superior performance can he achieved relative to that of an existing fixed step-size algorithm.