Designing Deep Learning Model for Accurate vNF Service Chain Pre-deployment in Inter-DC EONs

This work considers a provisioning framework for inter-datacenter elastic optical networks, which incorporates prediction and pre-deployment of virtual network function service chains (vNF-SCs), and designs a new deep learning (DL) model to realize accurate prediction. Simulation results confirm that the new DL model provides higher prediction accuracy than the existing approach, and with it, a service provider achieves better tradeoff between resource utilization and blocking probability.