The performance of the fixed-point least mean kurtosis and noisy inputs

Since the optimal solution of the least mean kurtosis (LMK) is selected to minimize the negative of the kurtosis of the error signal, the noise that has symmetrical probability density function (PDF) does not affect the optimal solution. The LMK algorithm has been studied and proved to outperform the widely used LMS algorithm. Extending the previously proposed analyzing method to predict the performance of the LMK algorithm, in this paper, we present the steady state mathematical analysis of the fixed-point least mean kurtosis method and noisy input signal.