Differential evolution optimization applied to the routing and spectrum allocation problem in flexgrid optical networks

Flexible optical network (FON) architectures are considered a very promising solution where spectrum resources are allocated within flexible frequency grids. Routing and spectrum allocation (RSA) in FON is an NP-complete problem. So far, this problem has been optimally solved for small instances with integer linear programming and has been suboptimally solved for more realistic instances by heuristic strategies. In this paper, we introduce the application of differential evolution (DE) to the off-line RSA problem in flexible optical networks. To the best of our knowledge, our work is the first application of a DE algorithm to the RSA problem. We develop two DE permutation-based algorithms named DE general approach (DE-GC) and DE relative position index (DE-RPI). Comparative studies show that in many cases, DE outperforms many other well-known evolutionary computational approaches. Furthermore, the method typically requires few control parameters. An illustrative example of the application of the DE-based algorithms is presented, and then, different heuristics are compared against the DE-RSA algorithms. Algorithms are evaluated in different test bench optical networks, such as the NSFnet and the European optical network, and for networks up to 40 nodes, such as the USA and Japan networks. Moreover, the DE-based algorithms save up to 37 % of spectrum utilization for the NSFnet and 13 and 15 % for the USA and Japan networks, respectively.