Autonomous and Softwarized Management of Disaggregated Open Optical Networks

Given the continuous growth of data traffic demand, the optical network management of current infrastructures have become a key factor: Internet pervasiveness has enlarging fast with fiber-to-the-home (FTTH), 5G and future 6G technologies. In the last decade, software-defined network (SDN) concepts has been applied within the optical communication world, making optical networks dynamic and programmable. This has been achieved thanks to the adoption of a single network operative system and the virtualization of the network elements (NEs) forming the physical layer (PHY). The latter can be achieved by means of device disaggregation, vendor-neutral control and multi-vendor inter-operability: Internet service providers and network operators are now interested in these solutions, avoiding vendor lock-in and cutting down on capital expenses. Network virtualization and disaggregation lay the foundation for the creation and the manipulation of a virtual object called digital twin (DT) of the network: a data structure modelling and emulating the behavior of the real system. In the wake of the Yet Another Next Generation (YANG) data model, other implementations were born aiming to unlock vendor-neutral control and independent device virtualization. Cooperating with the standardization bodies (e.g. IETF, ITU-T, IEEE, OIF), the largest part of Internet Service Providers (ISP) promoted various activities in order to achieve agreements for network disaggregation and automation. In particular, we have recently seen the birth of various consortia specialized in improving the efficiency of various aspects of the management of an optical network. In this work, firstly, a SDN architecture is defined in a context of open and disaggregated optical networks. Consequently, an implementation for a centralized network controller managing open and disaggegated optical networks has been developed. The main goal is to enable the independent management of control and data planes, engaging dynamic operations and expanding the possibilities of optimization degrees. The conceived SDN architecture has three main actors: the optical network controller (ONC), each optical line controllers (OLCs) and the PHY-DT. The communication among the actors is performed defining a software orchestrator called open optical transport network controller (OOTNC), operating as dispatcher and integrating additional functionalities in terms of network management and control. Different data models, frameworks and protocols are investigated: Network Configuration Protocol (NETCONF) as a network management protocol, from IETF, YANG model for NE virtualization, from OpenConfig, reconfigurable optical add & drop multiplexers (ROADM) models, from OpenROADM, open network operating system (ONOS) as a SDN controller, from Open Networking Foundation (ONF), and GNPy for quality of transmission (QoT) estimation, from Telecom Infra Project (TIP). Using a custom representational state transfer (REST) application, ONOS is able to retrieve information regarding the network topology and the network status. Furthermore, additional endpoints allow to set the estimated LP modulation format. GNPy is integrated within the PHY-DT as QoT estimator. The developed software framework has been conceived also to promptly react in case of link or node failures, firstly evaluating the amount of lost traffic and then recovering the issue.