Research On Resource Allocation Strategy In Space Division Multiplexing Elastic Optical Networks

Recently,with the drive of cloud computing,internet of things and ultra-bandwidth,the scale of video services and internet of things terminals has exploded.It makes the channel bandwidth consumption of optical fiber communication has increased exponentially and rapidly.In addition,the transmission capacity of elastic optical networks based on single-core optical fiber has approached its physical limit,which puts forward higher requirements for optical network capacity.In order to achieve higher communication transmission capacity,elastic optical networks based on space division multiplexing emerge.Space division multiplexing elastic optical networks have the advantages of large capacity,fine resource granularity and flexible resource allocation.However,it also brings some prominent problems such as spectrum fragmentation and inter-core crosstalk.To this end,a crosstalk-aware spectrum fragmentation avoidance strategy and a blocking-aware bidirectional spectrum defragmentation strategy in space division multiplexing elastic optical networks are proposed.In order to adopt appropriate strategy to prevent spectrum fragmentation before resource allocation and to ensure crosstalk performance,a crosstalk-aware spectrum fragmentation avoidance strategy is proposed.It first divides the spectrum resource on each core into two sub-areas,and then the resource in each area between the adjacent cores is alternately used to avoid the inter-core crosstalk.In addition,in order to serve more requests with limited resource,the resource allocation process is transformed into a two-dimensional rectangular packing filling problem.To make the filling more effective,the Elman neural network in machine learning is used to predict the attributes of requests.Then the filling priority of requests is determined according to the predicted results,and finally the requests are filled into the two-dimensional resource pool according to the determined priority.The simulation results show that the proposed strategy can achieve higher resource utilization and lower blocking probability while ensuring inter-core crosstalk performance.Although spectrum fragmentation avoidance strategy can reduce the generation of fragmentation to a certain extent by rationally planning network resource.New spectrum fragmentation will inevitably generate due to the dynamic setup and release of requests after a period of network operation.Therefore,a blocking-aware bidirectional spectrum defragmentation strategy is proposed.Firstly,the traffic in the future is predicted,and the blocking probability in the next period is calculated based on the blocking probability and traffic in the previous period.And then compare the predicted blocking probability and the preset blocking probability threshold to complete whether to trigger the spectrum defragmentation.Taking the multi-core characteristics into consideration in the space division multiplexing elastic optical networks in the defragmentation process,the strategy divides all the cores into three categories,then adopts adaptive defragmentation for requests on different cores.The simulation results show that the proposed strategy reduces the network blocking probability and has low defragmentation delay while fully considering the multi-core characteristics.