Efficient Planning and Provisioning of Flexible-Grid Optical Networks with Adaptive Genetic Algorithms
Zuqing Zhu,Xiang Zhou,Wei Lu,Long Gong
2015-01-01
Abstract:In this paper, we discuss adaptive genetic algorithms (GA) that can solve the routing, modulation-level, and spectrum assignments (RMSA) for both static and dynamic optical orthogonal frequency-division multiplexing (O-OFDM) networks, with high efficiency. Based on network status and traffic information, the GAs encode the RSMA of lightpaths as genes and optimize them with a process that mimics the natura l evolution. Both the crossover and mutation in the GAs operat e adaptively according to the fitness of individuals. The simulation results show that compared with several existing algorithms, the adaptive GA achieves better static network planning with more efficient spectrum resource utilization in both the NSFNET and US Backbone topologies. Meanwhile, the GA converges within 80 generations. In dynamic network provisioning, multi-objective optimization is incorporated to improve spectral efficiency and reduce blocking probability simultaneously. For low traffic cases when there is no blocking, the GA focuses on minimizing the maximum number of slots required on any fiber in the network. Otherwise, it tries to minimize the blocking probability when the traffic load is high. The simulation results also indicat e that the GA achieves the best provisioning performance in terms o f blocking probability.