An ANFIS based method for design optimization of soliton links using lumped EDFA repeaters

In this paper, an adaptive neuro-fuzzy inference system (ANFIS), based method is used to optimize the design of soliton links. An ANFIS is trained using data obtained from mathematical models by numerical simulations. The performance improvements in terms of the transmission distance, bit rate, amplifier spacing and bit error rate are considered. Using the ANFIS based model, the performance of links with non-uniform repeater spacing is also analyzed. This is often necessary because a performance improvement always demand the deployment of new amplifiers without distributing existing ones. The proposed model reduces the computational time and complexity with a mean square error of 1times10^-4.