Estimating Amplifier Cascade Output Signal Using an Artificial Neural Network and Considering Tilted Signals

We adapt and optimize an output power estimator based on an artificial neural network (ANN) to work in a scenario with multiple amplifiers (cascade). The optimization is made considering the ANN number of hidden layers and the number of neurons in these layers. The optimized ANN is applied in three link scenarios with a different number of amplifiers and different power tilt profiles. Results show that the model achieved less than 2.5 dB of absolute error median considering the scenario with 20 amplifiers and the worst power tilt profile.