Field Trial of Monitoring On-Demand at Intermediate-Nodes Through Bayesian Optimization

During 12/2017-03/2018, a field trial  was conducted by researchers (led by Bristol High Performance Network Group) to monitor on-demand function at network nodes which monitors intermediate-node OSNR performance with intelligent switching strategy. With Bayesian optimization on top of Gaussian processes, MoD saves up to 91% of the monitoring data while accurately predicting the worst OSNR performance of the link with as few monitoring trials as possible. This capability enables a self-learning “out-of-the-loop” monitoring process and potentially eliminates big data issues in SDN. This work was presented in the OFC conference. As shown in the picture, Fanchao Meng (HPN) presented the work on Machine learning for optical communications.