An improved Markov chain description for fading processes

A new approach to model a fading process (e.g., as generated by Jakes´ simulator) is proposed and investigated. This is based on a finite-state Markov chain, with a two-dimensional state given by the quantized values of the amplitude and its speed of variation. This results in better accuracy with respect to classical one-dimensional schemes. The validity of the proposed approach is shown by first comparing the autocorrelation function of the resulting process with that of the simulated process. Furthermore, a simple go-back-N protocol has been used for the same purpose, in order to verify the impact of modeling accuracy on upper layers. Some investigations have also been performed to evaluate the sensitivity to parameter selection. Finally, the widely used model with only two values of envelope amplitude has been studied, in the case of both one-dimensional and two-dimensional Markov chain states.