A Temporal Network Calculus Approach to Service Guarantee Analysis of Stochastic Networks

Many computer networks such as wireless networks are stochastic in nature. In order to perform performance guarantee analysis of such networks, a theory, called stochastic network calculus, has evolved. In the stochastic network calculus literature, most results are based on space-domain traffic and service models where the arrival process and the service process are respectively characterized by the cumulative amount of arrival and the cumulative amount of service. Recently, a novel approach called time-domain approach to stochastic network calculus (SNC) has been proposed, where the traffic and service models are defined based on the cumulative inter-arrival times and the cumulative service times respectively.

In this paper, we concretize the time-domain SNC traffic and service models by linking some well-known stochastic processes to them. In addition, we exemplify the temporal analysis approach by investigating the delay performance of a Gilbert-Elliott channel. The results show that the delay bound can be improved under the independence condition. Furthermore, a comparison between the temporal and the spatial analysis results reveals that the two analytical approaches essentially yield close results.