Heuristic Optimization Strategies for Reliable Amplifier Reconfiguration in Autonomous Optical Networks with Field-trial Validation

Dynamic optical power optimization plays a crucial role in improving the quality of transmission (QoT) in autonomous optical networks (AON). While previous research widely investigates launch power optimization and optical amplifier (OA) control-based link power optimization, the reliability of optical power optimization processes in real network deployments remains largely unexplored. Specifically, due to the controller-to-device latency and distinct response time of devices, OAs are reconfigured sequentially in real networks. During this process, the system performance might fluctuate, and undesired QoT degradation in existing services might occur. To prevent such QoT degradation and ensure reliable optical power optimization, in this paper, we propose OrderOPT, a heuristic-based order optimization scheme for OA reconfiguration processes in AON. By integrating a heuristic algorithm and a digital twin (DT) system, OrderOPT efficiently searches for the optimal OA reconfiguration order and effectively prevents Q-factor degradation in various experimental scenarios. In comparison with random reconfiguration orders, OrderOPT outperforms 98.5% of the cases in terms of minimum and average Q-factors. Additionally, to address scenarios with large adjustment steps, we further develop the enhanced OrderOPT (E-OrderOPT). E-OrderOPT pre-processes the reconfiguration sequence by refining the adjustment step sizes before the heuristic search. Its effectiveness is further validated through field trials conducted on a loop link with spans of different lengths. In all evaluated scenarios, E-OrderOPT successfully prevents Q-factor degradation of up to 1 dB during the OA reconfiguration processes. E-OrderOPT also achieves superior Q-factor performance compared to randomly generated reconfiguration orders.