Research On Resource Allocation Algorithm In Virtual Optical Network

Under the background of the rapid development of cloud computing and Internet technologies,the elastic optical networks(EONs)virtualization technology improves the utilization of network resources by allowing multiple virtual optical networks(VONs)that are logically isolated to share the underlying optical network resources.This thesis presents new dynamic virtual optical networks embedding(VONE)algorithms from two different perspectives for how VONs can efficiently use the underlying optical network resources.First,this thesis proposes a dynamic VONs Mapping algorithm that cooperates with nodes and links(VMCNL).The computational capacity of the node,the spectrum resources and the relatioanship between the nodes to be mapped and the nodes mapped are considered during node embedding in VMCNL algorithm when virtual nodes are mapped into the physical nodes,link embedding is considered.The virtual nodes around a virtual node are mapped into the nodes around a physical node.The VMCNL algorithm uses cooperation between nodes and links to avoid long-hop physical routes during link embedding,and reduces the waste of network spectrum resources.At the same time,this thesis proposes a algorithm that of link free spectrum continuity maximum block fist(SCMBF).The SCMBF algorithm first finds all free spectrum blocks that can meet traffic request on the path that needs spectrum allocation.Then,the traffic is put into different spectrum blocks,and the spectrum block with maximum free spectrum continuity is calculated to complete the spectrum allocation.This SCMBF algorithm reduces the generation of network spectrum fragmentation to a certain extent,achieves a reasonable allocation of spectrum resources.Second,this thesis proposes a dynamic VONs Mapping algorithm based on K-means clusterings(VMKC).The VMKC algorithm first divides the physical network into K clusterings dynamically according to the VONs traffic request conditions.The partitioned clustering guarantees that the distances between nodes in the clustering are relatively close,and the number of nodes in the clustering cannot be less than the number of the nodes of virtual networks,and then directly in a clustering to complete the VONE.Because the nodes within the clustering are relatively close to each other,the joining of long-hop paths is reduced when link embedding is carried out,which reduces the consumption of network resources and increases the number of VONs that can be mapped.At the same time,a VONs Mapping algorithm with Improved K-means Clusterin(VMIKC)is proposed,which is a dynamic VONs mapping algorithm in which the center nodes of the physical network can be divided into different clusters.When clustering is divided,if a certain physical node is close to different clusterings center,such physical nodes can be added to different clusterings,so that the distance of the divided clustering nodes is relatively closer,the performance of VONE is further enhanced.Simulation experiments show that the proposed algorithm has significantly improved with blocking ratio and link utilization performance.