Machine Learning for future optical network automation

Machine Learning (ML) tools have recently been adopted for a wide range of automated operations in optical networking, moving fundamental steps towards the paradigm of zero-touch management of optical-network infrastructures. Such network systems are capable not only of self-monitoring, but also of self-reconfiguration and self-repair, thanks to advanced capabilities such as autonomous learning, root-cause analysis and optimized decision-making. This chapter provides an overview of the application of ML-based methods for a variety of autonomous transmission and network-related tasks, including optical amplifiers tuning, lightpath QoT estimation, fault management, resource allocation in metro/core topologies and in inter/intra-datacenter optical networks.