Genetic algorithm for amplifiers gain optimization in all-optical networks

In transparent, wavelength routed optical networks, the optimization of network amplifiers gains is a difficult task, because of the amplifier saturation effect and the dynamics of admission and termination of connections. In this paper, we propose and demonstrate a genetic algorithm for the enhancement of network performance by optimization of the gain of each amplifier in the network. The gain of each amplifier in the network is coded in a bit string. These strings are optimized through evolutionary processes such as crossover, mutation and selection, which is based on minimum network blocking probabilities. We show that this algorithm is able to improve considerably the overall network performance. Parameters such as the network load, output saturation power of amplifiers, initial population and number of generations used in the genetic algorithm are considered and their effects are analyzed.