Neural Network-Based Control of TDFA

The complexity of controlling optical amplifiers with multiple pumps, like Thulium-Doped Fiber Amplifiers (TDFAs), stems from the need to select the optimal pumping to ensure proper operation (e.g., maintaining a constant target output power). Therefore, an experimental demonstration of neural network-assisted SDN control of TDFA configuration is presented. The proposed optimization enables efficient automated TDFA control, resulting in an RMSE of 0.23 dBm between target and actual output power.