Multiobjective optimisation model and algorithm for data center scheduling in elastic optical networks

Elastic optical network (EON), as one of the most promising tools recently, enables users to flexibly schedule network resources on demand. But how to properly assign data centers in EONs is a particularly important issue, which determines both the quality of service (QoS) and the operational expenditures of data centers. First, we propose a multiobjective optimisation model for data center scheduling (DDEA/RSS) and design an evolutionary algorithm based on NSGA-II to solve it. Then, we design three heuristics to construct routes between a pair of nodes. Finally extensive experiments are performed to analyse the performance of three heuristics and validate the efficiency of DDEA/RSS on two network topologies by comparison with other existing algorithms. The results indicate that DDEA/RSS produces the lower call request blocking probability under the same number of data centers and the smaller number of data centers under the same call request blocking probability.