An Efficient Genetic Algorithm for Solving Spectrum Assignment Problem in Elastic Optical Networks

With the rapid expansion of the Internet, optical transmission and network technologies are moving towards the objective of greater flexibility. The Elastic Optical Network (EON) could allocate just enough bandwidth to satisfy each request effectively. In the EON, the spectrum slots allocated to two requests must be separated by a guard band to avoid intrusions and minimize security risks that could have entered the physical layer. In this paper, we are interested in the spectrum allocation problem in EON using guard bands of different sizes, which is an NP-hard problem. To solve this problem, we have proposed an approach based on a genetic algorithm using edge assembly crossover to obtain a global minimum of frequency slots used in the network. Our simulation results indicate that the proposed method yields nearly optimal results for the majority of problem instances.