Link-layer modeling of a wireless channel using stochastic network calculus

Network calculus (NC) is an elegant theory for performance evaluation of QoS-based queueing networks. By applying the stochastic NC, we present a modular dynamic approach which effectively models the behaviour of a finite-state Markov wireless channel, observed from the link level. This is sufficient for extending the theory to wireless networking, by allowing us to analyse a complete data network containing wireless links, and to evaluate its end-to-end QoS metrics, e.g. delay, backlog, and packet loss, within probabilistic constraints. These bounds are expressed with equations that provide the likelihood of attaining these bounds within a desired probability. With a practical example, the effectiveness of the proposed approach is demonstrated.