DyAGO: Dynamic Adaptive Gain Optimization for Multi-Pump Raman Amplifiers in Ultra-Wideband Optical Transmission Systems

Flexible and accurate gain providing is urgently needed in ultra-wideband transmission systems. Raman amplifiers (RAs) are widely used for their potential to generate arbitrary gain profiles. However, the adjustable pump parameters in commercial RAs are limited, resulting in the degradation of gain optimization accuracy. Traditional methods statically perform gain optimization at all signal channels without considering which channels have precise gain requirements under the hybrid full-load transmission of service-carrying waves and filling waves, exacerbating the accuracy degradation caused by RA limitations. Therefore, we propose a “service channel-only” dynamic pattern to perform gain optimization just at service channels. The challenges of implementing this pattern include the unfixed objective function caused by different channel working conditions, the complex channel coupling due to cross-channel interference, the lack of real-time feedback and the absence of support for re-optimization after objective function changing. We proposed a dynamic adaptive gain optimization (DyAGO) framework to overcome these challenges. In DyAGO, adaptive objective function adjustment is conducted by a constructed module to cope with various channel working conditions. A linear regression module with two pre-calculated matrices is designed to assist the optimization initialization. A machine learning-based online dual symmetric module is constructed to provide real-time feedback for optimization. The whole optimization process is performed by an improved genetic algorithm, which adaptively adjusts the population size to accelerate the optimization process. Both simulation and experimental results demonstrate the superiority in gain optimization accuracy of the proposed DyAGO over two static gain optimization methods. Short runtime and great stability of DyAGO are also indicated by these results.