Evaluation of centralized solution methods for the dynamic optical bypassing problem

Due to increasing traffic volumes and access bandwidths, the power consumption of core networks will grow considerably. Adapting network configuration to traffic load is one counter-measure. Dynamic optical bypassing is a promising approach to reconfigure multi-layer networks: it adapts the virtual topology while keeping traffic on fixed paths in the physical topology. So far, research focused on distributed bypassing schemes. In this paper, we evaluate three centralized solution methods for the bypassing problem: one based on linear programming, one based on heuristic optimization, and a greedy heuristic. We find that all methods can achieve similar energy savings while limiting changes to the network configuration.