Research On Spectrum Defragmentation In Elastic Optical Networks

With the development of network technology and the increase in scale,and widespread adoption of smart equipment,the network services have been gradually integrated into various sectors of society,and the types are more abundant and the number is increasing.The traditional optical network using WDM technology unable to meet the requirements of the current service for high flexibility,large capacity and high speed of the transmission network,therefore,elastic optical networks come into being.In an elastic optical network,services routing and spectrum allocation need to meet the constraints of spectrum consistency and adjacency.Dynamically setting up and tearing down services generates a large number of small spectrum blocks.These small spectrum blocks are neither aligned on the link nor continuous in the spectrum called spectral fragments.Spectrum fragmentation will reduce the utilization rate of the spectrum resource,thereby reducing network performance.In order to make full use of network spectrum resources,reduce spectrum fragmentation and improve network performance,spectrum defragmentation is researched in this thesis.Firstly,this thesis comprehensively considering the factors of time domain and frequency domain proposes a fragmentation index method and use it to determine whether to trigger the network to defragment.To meet different network performance requirements,one defragmentation algorithm triggered by fragment index(DTFI)and another triggered by fragment index and periodic(DTFIP)are proposed.This thesis studies the effects of different trigger thresholds on the defragmentation performance of the two algorithms.The simulation shows that under a certain network load,there is a suitable trigger threshold,which makes the network profit-loss ratio larger,and the larger the network load,the larger the appropriate trigger threshold.At the same time,the DTFIP algorithm can avoid frequent defragmentation when the network fragmentation is large,reduce the service damage rate,and increase the profit-loss ratio.Secondly,in order to achieve non-disruptive defragmentation,we proposed the procedure for sequential spectrum defragmentation,and considers not only the bandwidth,the occupied location of the resources,but also the remaining time of the service consolidation to determine the order of defragmentation.The simulation shows that compared with the traditional consideration of the service bandwidth and the location of the occupied resources,the service order determined by the remaining time of the service combined can reduce the service blocking rate and improve the network performance when the service damage rate is basically the same.To solve the problem of insufficient space for optimizing spectrum fragmentation caused by the reduction of idle resources when the network is overloaded,a spectrum defragmentation algorithm in the network cutover process is proposed.When the network load is too heavy,the defragmentation cannot meet the requirements of network optimization.Therefore,you need to consider network expanding and cutover.In order to improve network performance,consider performing spectrum defragmentation in network cutover.The simulation shows that defragmentation in network cutover can effectively reduce the network fragmentation index and have a lower traffic blocking rate in the subsequent operation of the network.