Local vs. Global Optimization for Optical Line System Control in Disaggregated Networks

Setting the operating point of optical amplifiers of optical line systems (OLS)s within transparent, disaggregated and reconfigurable networks is a crucial task that determines the optical transmission performance of the specific infrastructure. In this work, four optimization strategies for OLS control are compared through a simulation campaign, where a realistic physical layer is replicated using a machine-learning model derived from an experimental dataset on commercial devices for the Erbium-doped fiber amplifiers (EDFA)s and a characterized set of fiber spans. In particular, two distinct objective functions are evaluated, both at the end of the line (global approach), and, in turn, at the end of each single span (local approach).