Research On Spectrum Resource Optimization For Resource Reservation Services In Elastic Optical Networks

The emergence and development of high-traffic emerging applications such as cloud computing and data backup have promoted a dramatic increase in network capacity demand,making bandwidth resources particularly important.At the same time,users’ requirements for service quality are also increasing.Elastic optical networks have the advantages of fine granularity and large capacity and can carry more services through spectrum resource optimization.In addition,based on the immediate reservation service,the advance reservation service based on the resource reservation control mechanism can further adapt to the mixed transmission of services with different service quality parameters in the actual network scenario.However,under the three constraints,with the establishment/removal of light-paths by requests in the network,there are many discontinuous free resources in the network that cannot accommodate service requests,that is,spectrum fragmentation.The existence of fragmentation reduces the availability of spectrum resources,improves the blocking probability,and seriously degrades network performance.Therefore,in the elastic optical networks for resource reservation services,the issue of spectrum fragmentation has become a key issue for resource optimization.In this context,this thesis considers the time and spectrum dimensions to analyze the fragmentation issue.Firstly,a time-spectrum fragmentation-aware spectrum allocation algorithm based on the preventive spectrum resource allocation method is proposed to suppress the generation of fragmentation in the resource allocation process.Secondly,a maintenance defragmentation algorithm based on time-spectrum shift of resource reservation service is proposed,which reconfigures some service requests to reduce the fragmentation and blocking probability.The main research contents are as follows:(1)From the perspective of spectrum allocation,to address the issue of incomplete fragmentation avoidance in the spectrum allocation process,this thesis first adds the consideration of relative-flexibility of services in the resource scheduling phase to determine the resource allocation order;constructs a time-spectrum fragmentation metric model by comprehensively analyzing the fragmentation impact in terms of both resource connectivity and resource matching degree.Based on this,a time-spectrum fragmentation-aware spectrum allocation algorithm is proposed to finally select the resources with the least fragmentation impact for connection requests.The algorithm aims at minimizing the amount of fragmentation in the network and reducing the network blocking probability.The simulation results show that in the NSFNET environment,when the traffic load is 154 Erlang,the spectrum utilization is increased by 4.6 % and the fragmentation rate is reduced by 21.2 %.(2)From the perspective of maintenance-based defragmentation,to address the complexity of the defragmentation process,this thesis considers finding available resources for connection requests that are about to block by performing parallel shifts in the time-or spectrum domain for advance reservation services that have been locally reserved resources in the network but for resource allocation.Based on this,a maintenance-based defragmentation algorithm is proposed based on time-spectrum shifting of resource reservation services.When a connection request is blocked,the defragmentation algorithm is triggered.The simulation results show that in the NSFNET environment,when the traffic load is 130 Erlang,the spectrum utilization is increased by3.0 %,and the fragmentation rate is reduced by 5.1 %.