A Hierarchical Reservation Adaptive Evolutionary Algorithm Based on NSGA-II for Multiobjective Optimization of Networks
Qing Yu, Fan Yang, Wenchao Song, Chaoran Ying · 2023
In network scenarios where 5G and cloud computing are deeply applied, reasonable deployment of VNFs, rational routing and spectrum allocation in elastic optical networks are all conducive to optimizing the resource allocation of the entire network. The multi-objective optimization model obtains a set of service function chain deployment and link mapping strategies by considering multiple factors to provide a comprehensive solution. In this paper, a five-objective optimization model is established to minimize node resource consumption, link maximum frequency slot number, node load difference, unreliability of deployed nodes and mapped links, and service source-to-destination propagation delay. On this basis, a hierarchical reservation adaptive evolutionary algorithm based on the NSGA-II is proposed to solve the proposed optimization model. The algorithm adopts a combined method to jointly code the node selection scheme and the link mapping scheme, and integrates a hybrid initialization method to generate an initial population that consider both diversity and convergence. To improve the global optimization ability of the algorithm, we add a two-stage crossover mechanism, a dominated solutions hierarchical retention mechanism, and a last-ranking elimination mechanism when the solutions are not dominated by each other, as well as a neighborhood correction mechanism for illegitimate solutions and a two-stage flip-flop strategy for jumping out of the local optimum. The simulation results verify the effectiveness of the proposed algorithm.