Photonic device programmability in support of autonomous optical networks

The emergence and consolidation of increasingly programmable optical devices such as transceivers, amplifiers, multiplexers, or ROADMs—which allow their remote configuration and control by adopting software-defined networking principles such as model-driven development—is enabling the evolution toward gradually more autonomous networks. Such networks leverage device programmability and are able to adapt and react to traffic and network condition changes, e.g., changing modes of operation or reconfiguring the network state, paving the way for the increased adoption of AI/ML models in support of enhanced network operation. In this paper, after a short review of some key elements in the control and orchestration systems of optical networks in support of autonomous networking, we present in detail a proof-of-concept validation of autonomous, closed-loop dynamic adaptation of transceiver operational modes. This includes (i) the design and development of an SDN agent of a multi-band sliceable bandwidth variable transceiver, based on extended OpenConfig terminal device data models; (ii) an SDN controller that performs discovery and management of transceivers’ operational modes and maps to transport API (TAPI) profiles enabling efficient physical layer impairment-aware path computation; (iii) a dedicated externalized path computation element/digital twin that performs adaptation recommendations; and (iv) an MQTT-based telemetry platform for publisher/subscriber based state synchronization between the control plane functional entities to avoid systematic polling.