A Hybrid Discrete Grey Wolf Optimization Approach for Spectrum Assignment in Elastic Optical Networks

Abstract

Future optical networks are expected to face increasing dynamism, heterogeneity, and unpredictability while supporting the growing demand for bandwidth-intensive services. Elastic Optical Networks (EONs) enable flexible and efficient spectrum utilization through fine-grained resource allocation. However, the Routing and Spectrum Assignment (RSA) problem, particularly spectrum allocation in large-scale topologies, remains computationally challenging because of its NP-hard nature and the rapid growth of the search space with network size. This motivates the development of scalable optimization techniques capable of maintaining efficient spectrum utilization in complex environments. To provide a demanding and representative evaluation, this work considers both realistic backbone networks and large-scale synthetic topologies with progressively increasing sizes and different connectivity characteristics for scalability assessment. A hybrid metaheuristic combining the Grey Wolf Optimizer (GWO) with Simulated Annealing (SA) is proposed for spectrum allocation in EONs. The hybridization aims to balance global exploration and local exploitation while mitigating premature convergence. The method is evaluated across benchmark topologies of varying sizes, with the Germany50 and Nobel-EU networks serving as the principal realistic benchmarks. Experimental results demonstrate that GWO–SA reduces the maximum number of frequency slots (FSs) required, improves spectrum utilization, and enhances network capacity, confirming its effectiveness and scalability for large-scale EON spectrum allocation problems.