Study On Traffic Scheduling And Spectrum Reconstruction In Elastic Optical Networks

With the development of broadband services and widespread adoption of smart terminal equipment, the network services have been gradually integrated into various sectors of society. At the same time, the types of traffic tend to be more diverse. The traditional optical network uses WDM technology to achieve channelized "transparent" transmission for the end-to-end traffic. The network performance of carrying multi-rate traffic is improved. However, the fixed-grid mode makes efficient, dynamic and flexible demand of the future network unable to meet. The novel elastic optical network breaks the fixed-grid pattern using variable bandwidth optical transceiver and switch. The spectrum granularities are divided into more detailed so that the spectrum resource utilization and network flexibility can be improved. Meanwhile, the elastic optical network also faces complex optimization problems including traffic scheduling, resource allocation and spectrum fragmentation, etc. This thesis mainly finished the research on traffic scheduling mechanism and spectrum reconstruction of the elastic optical network. The major contributions of this thesis are as follows.(1) The adaptive traffic scheduling algorithm based on the bandwidth, service duration, delay and other specific characteristics and requirements for elastic optical network, is proposed. The simulation shows the scheduling ratio under different criteria with priorities and time window as well as the effectiveness of the algorithm.(2) The least hops first and frequency slot packet routing strategies are presented, as well as the node spectrum allocation algorithm based on fairness between links. Simulation results show that the proposed algorithms can effectively reduce the number of optical links and improve the imbalance spectrum between links. Moreover, traffic in each link can access to spectrum equitably and the spectrum fragmentation is reduced.(3) The spectrum defragmentation model and two frequency slot allocation strategies are presented. The GA-TS spectrum defragmentation algorithm combing advantages of genetic algorithm and tabu search is proposed. The simulation results show that the algorithm can automatically select a reasonable spectrum reconstruction mode according to the traffic distribution. Therefore, the network performance can be improved effectively.