Accurate and data-lite EDFA model for dynamic pump configurations

As a critical foundation for large-capacity transmission in long-haul systems, the Erbium-doped fiber amplifier (EDFA) spectral gain model has drawn significant attention recently. However, variations in pump power lead to severe accuracy degradation in existing analytical models, resulting in considerable estimation errors for output power and the optical signal-to-noise ratio (OSNR). To address this issue, an accurate and data-efficient EDFA spectral gain model tailored for dynamic pump scenarios is proposed. To reduce data requirements, singular value decomposition is employed to analyze cross-sections and erbium population inversion. Leveraging the underlying physical mechanisms, the model characterizes the EDFA's operational state to enhance its predictive capability. Experimental results demonstrate that the proposed model achieves superior prediction accuracy compared to the affine-law-based model, yielding a root mean square error (RMSE) of 0.11 dB with only 10 flat input measurements. By accounting for the inhomogeneous gain induced by the spectral hole burning (SHB) effect, the model's accuracy is further improved, achieving an RMSE of 0.07 dB. Moreover, the model maintains high prediction accuracy across a dynamic pump current range, from 360 mA to 880 mA. The proposed model facilitates more autonomous network operation by enhancing the adaptability of dynamic pump control in long-haul transmission systems.