Multi-armed bandit problem-based core selection method for multi-core fiber networks

In multi-core fiber networks, it is very important to consider a routing, core, and spectrum allocation (RCSA) problem to efficiently utilize bandwidth resources. This paper proposes a multi-armed bandit (MAB) problem-based core selection method for a dynamic RCSA problem where lightpath-setup requests are stochastically generated with time elapsed. The MAB problem estimates the probability distribution of rewards of objects by repeating trials. In the proposed method, for a newly established lightpath, a core at each link along the path is selected based on the MAB problem to suppress the generation of the fragmentation of bandwidth resources. By doing so, we expect to decrease the blocking probability of lightpath establishments. Through simulation experiments, we evaluate the performance of the proposed core selection method.