A neural network model of erbium-doped photonic crystal fibre amplifiers

The evaluation of the general evolution equations that describe the longitudinal propagation of pump, signal, forward and backward amplified spontaneous emission in rare-earth-doped optical fibre amplifier could be computationally expensive. In this paper, to reduce the computational time, a neural network approach for the modeling of erbium-doped photonic crystal fibre amplifiers is proposed. A number of simulations have been performed to investigate the characteristics of the proposed approach. The numerical results show good agreement between the neural network approach and the conventional algorithm based on the solution of the power evolution equations. The proposed approach exhibits attractive performance in terms of flexibility, accuracy and computational time.