Research On Routing And Spectrum Assignment Mechanism In Elastic Optical Networks

Traditional WDM networks are constrained by fixed grid conditions,which makes it unable to meet novel servce requirements such as artificial intelligence,big data.Elastic Optical Network(EON)adopts OFDM technology to further divide the network spectrum resources into spectrum units with smaller granularity.EON can allocate appropriate adjacent spectrum units according to actual service requirements dynamically and effectively,thereby achieving the purpose of improving resource utilization and service transmission efficiency in the network,respectively.Through theoretic comparison,modeling and simulation methods,this thesis has conducted in-depth research on routing and spectrum allocation,spectrum defragmentation in elastic optical network.This thesis first briefly introduces the elastic optical network from four aspects: basic concepts,network architecture,key technologies,and routing and spectrum allocation issues.Secondly,by analyzing the problem of spectrum fragmentation generated during the operation of the elastic optical network,this thesis proposes the concept of Consilience between Services and Fragments(CSF),which measures the degree of spectrum fragmentation on links according to the value.Different from traditional spectrum fragmentation evaluation parameters that only consider the current fragmentation on the link,the core idea of CSF is to better and more accurately assess the degree of fragmentation from the three dimensions of all fragmentation,single fragmentation,and arriving services.Based on CSF,this thesis proposes an RSA algorithm based on consilience between services and fragments.Numerical simulation results show that in three typical network topologies including NSFNET,CERNET and USNET,the proposed algorithm can improve 34.38%,8.68%,28.51%,and 9.45%,9.35%,11.23% in terms of bandwidth blocking probablity and resource utilization compared to Dijkstra+First Fit algorithm,while 45.83%,18.06%,30.89%,and 11.09%,17.29%,12.4% improvement over Dijkstra+Random Fit algorithm,respectively.At the same time,this thesis proposes an EON defragmentation algorithm based on resource saving strategy.The core idea of the algorithm is to traverse all spectrum blocks and filter according to the first-in first-out principle when defragmentation,so that avoiding new and smaller fragmentation after the defragmentation is completed.In the three network topologies as mentioned above,the improved algorithm can improve the bandwidth blocking probablity and resource utilization by 15.83%,8.54%,10.01%,and 5.88%,6.7%,9.79% compared to FBNDA algorithm;Compared with Dijkstra+First Fit algorithm,the improvements are 56.18%,13.61%,33.06% and 20.83%,12.84%,and 19.81%,respectively.