Dependability modelling and analysis of networks as taking routing and traffic into account

The existing work on dependability modelling and analysis of networks have main focuses on the, structural properties. However, the communication networks also have dynamic characteristics and behavioral properties, such as changes of topology, routes, and traffic loads, which should necessarily be reflected in the modelling. We regard the dependability modelling and analysis of IP networks with an intra-domain routing and traffic handling. This yields to an increased complexity in the modelling and analysis, due to the increase of the state space. Several techniques have been proposed to deal with the growth of the state space. To an analyst, an interesting point is, whether he/she can make a trade-off to the factors under consideration, i.e. the model representativeness vs. modelling complexity, and computation efforts vs. result accuracy. To meet these requirements, we propose a modelling approach where failures and repairs of network components as well as routing and traffic handling are described by a set of stochastic activity networks (SAN). The SANs are designed to restrict the generation of the state space to the most probable states, which enables an analysis without creation of an excessively large state space and achieving a sufficient accuracy of the result. A case study based on the Norwegian university IP backbone network demonstrates the applicability of this approach