Research On Resource Optimization Strategy Based On Crosstalk Aware In Space Division Multiplexing Elastic Optical Networks

With the continuous emergence of various new requests,network traffic has grown exponentially,which leads to the transmission capacity of single-core elastic optical networks gradually approaching its physical limit.At the same time,this also makes the network require higher flexibility.Space division multiplexing elastic optical networks with multi-core fiber have become one of the most promising candidates for future optical transport networks,due to their high capacity and flexibility.However,space division multiplexing elastic optical networks also face challenges such as inter-core crosstalk and spectrum fragmentation.To this end,a fragmentation avoidance strategy of layered graph-auxiliary crosstalk-aware and an adaptive defragmentation strategy of crosstalkaware modulation degradation are proposed.In order to mitigate inter-core crosstalk and avoid spectrum fragmentation,a hierarchical graph-assisted crosstalk-aware fragmentation avoidance strategy is proposed in the planning stage of network resource allocation.Firstly,combining the adjacencies between multi-core fiber,the priorities of cores is divided,and a three-dimensional network resource model based on the priorities of cores is established.Then,to plan network resources reasonably,deep learning technology is applied to predict traffic request sequences in the future period of time.Finally,to perform resource scheduling in a three-dimensional network resource model adaptively,the occupation of network resources is quantitatively analyzed.Simulation results show that the proposed strategy can not only significantly reduces the network blocking probability but also improves resource utilization under alleviating inter-core crosstalk.However,due to the frequent releasing and building of lightpaths,the problem of fragmentation will become more prominent after the network has been running for a period of time,which results in a sharp increase in the network blocking probability.In addition,the impact of inter-core crosstalk cannot be ignored.To this end,an adaptive defragmentation strategy of crosstalk-aware modulation degradation is proposed.Firstly,in order to properly scheduling network resources,the deep learning prediction model is still used to predict the traffic request sequences in the future period of time.Then,the predicted traffic request information is used to achieve modulation format degradation to mitigate the inter-core crosstalk.Finally,combining with the network blocking probability,the defragmentation is adaptively performed to achieve reasonable scheduling for network resources.Simulation results show that the proposed strategy significantly reduces the network blocking probability under the premise of reducing defragmentation delay.