Digital twin-enabled multi-step strategies for autonomous power equalization in optical networks

Abstract

This paper proposes and experimentally evaluates digital twin techniques, incorporating multi-step lookahead and dynamic step-size adjustments for per-channel power equalization in optical networks. Digital twins, which are software replicas of physical systems, are utilized to monitor, analyze, and predict the network behavior, thereby enhancing decision-making processes before implementing any physical adjustments. The study focuses on optimizing the signal-to-noise ratio (SNR) through per-channel launch power equalization, addressing challenges, such as nonlinear inter-channel interference and power transfers across multiple optical multiplex sections. The proposed methodology leverages a digital twin to simulate and predict SNR variations using multi-step lookahead, ensuring the monotonous SNR improvement without service disruptions. Additionally, parallel adjustment and dynamic step-size methods significantly enhance efficiency. Experimental validation on a C-band meshed optical network testbed demonstrates substantial reductions in power errors, improved SNR performance, and decreased commissioning time, highlighting the practical feasibility and efficiency of the approach. The findings underscore the transformative potential of digital twins in advancing autonomous optical network management.