Effect of Traffic Uncertainities on the Design of Mixed Line Rate (MLR) Optical Networks

Abstract

In the existing studies on Mixed Line Rate (MLR) optical networks, the network design methodology is based on the assumption of deterministic traffic, and hence, the effect of traffic uncertainty on the design of an MLR network remains an open problem of research. In this study, we upgrade our previously proposed cost-efficient mixed integer linear program (MILP) formulation for an MLR network, which considered a specific mean traffic for every network source-destination pair. Our upgraded model employs an optimization technique to account for the traffic uncertainties that an actual MLR optical network may encounter. Our simulation results show that (i) if the MLR network is cost-optimized under the assumption that approximately 10-20% of the demands are at their maximum (or peak) value then, the network demonstrates robustness to traffic peaks in approximately all the other demands, and (ii) the saturation of network cost for a number of source-destination pairs is network topology dependent.