Provisioning of 5G services employing machine learning techniques

摘要

This study proposes a modeling framework for optimal online 5G service provisioning, based on low computational complexity machine learning techniques such as Neural Network (NNs). NNs are trained to take optimal decisions adopting an offline Integer Liner Programming (ILP) model. This framework is used to solve the generic joint Fronthaul (FH) and Backhaul (BH) service provisioning problem over a converged high capacity and flexibility optical transport aiming at minimizing the overall energy consumption of the 5G infrastructure. Our modeling results indicate that the proposed approach adopting NN based real time service provisioning can provide very similar performance to the one derived adopting the high complexity but accurate ILP approach.