A new class of packet- and bit-level models for wireless channels

Stochastic channel error models are an important part of performance evaluations of wireless protocols. In fact, for a given protocol often the performance measures of interest seen in simulations depend on the choice of the error models, often simpler models give poorer quality performance predictions. This paper presents a new class of Markovianbased channel models, called bipartite models. They allow the user to freely choose the desired model complexity and therefore model accuracy. We demonstrate through simulations of an example system that this model class gives much more accurate predictions of performance parameters than other popular channel models.