Entropy-based load-balancing for software-defined elastic optical networks

To support emerging traffic demands, metro optical networks evolve over time with the addition of new links to relieve network bottlenecks and congestion. These new resources need to be optimally located in the network, in order to minimize the cost of the new deployment. Spectrum-fragmentation and network-wide load imbalance impede the optimal allocation of network resources. With the recent convergence of advanced flexible and programmable optical devices with emerging software-defined network paradigms, it is possible to flexibly and dynamically defragment the spectrum and balance the load on the network. This study introduces an optimization strategy for software-defined elastic optical networks for fiber-load balancing across the network, while minimizing the cost of the resulting service disruption. An entropy-based metric is proposed for measuring load imbalance and used to design utility functions for the joint optimization, taking into consideration the optical and defragmentation constraints.