On the exponential assumption for the time-to-repair in optical network availability analysis

The time-to-repair of network elements is usually modelled as an exponential random variable, as is the time-to-failure. This allows the use of Markov chains to perform availability analysis of connections routed in the optical network. However, the times-to-repair may be better described as Weibull distributed random variables, thus preventing the use of Markov chains. This paper investigates the impact of the time-to-repair distribution on the interval availability of a connection in an optical network. The exponential and the Weibull assumptions are compared. By computer simulations, and considering the case study of a connection routed in an American network, we identify the cases where the Weibull and exponential assumptions yield similar or distinct results.