Quality of service provisioning and energy minimized scheduling in software defined flexible optical networks

The over-provisioning of capacities in optical networks is not a sustainable approach in the long run. In this paper, we propose a software defined networking scheme for quality of service provisioning through energy efficient assignment of optical transponders, employing bandwidth variable distance adaptive modulation and coding. Our scheme enables avoiding over-provisioning of transponder capacity as well as short-term major changes in equipment allocation for networks with dynamic traffic. We make use of the seasonal auto-regressive integrated moving average model to forecast the statistics of network traffic for an arbitrary time span based on the requirements and the constraints of the service provider. The quality of service measure is defined as the probability of congestion at the core router ports. A stochastic linear programming approach is used to provide a solution for energy efficient assignment of optical transponders and electronic switching capacity while ensuring a certain level of quality of service to core routers. The scheduling of optical lightpath capacities is performed for the entire duration of time under consideration, whereas the scheduling of electronic switching capacities is performed based on the short-term dynamics of the traffic. Numerical results show up to 48% improvement in the energy efficiency of optical networks and 45% reduction in the number of optical lightpaths through the implementation of the proposed technique, compared to a design based on employing conventional fixed optical transponders and no traffic rerouting, where both schemes satisfy the congestion probability requirements.