A stochastic model for a new robust NLMS algorithm

We present a stochastic model for a new recently proposed robust NLMS algorithm. Under very standard and reasonable assumptions we show that the algorithm converges to the true unknown system in a mean square sense. With the aggregate of more restrictive, but standard, assumptions we can build a model for the transient behavior of the algorithm. The model can take into account the presence of impulsive noise. Finally we also present simulations results which show the excellent agreement with the model.