Virtual Network Embedding In Software Defined Networking

By combining software defined network(SDN)and network virtualization technology,virtual SDN(v SDN)technology allows many tenants to operate their own SDNs through controllers.These SDNs have different network protocol and topology,paralleling in a set of Shared physical SDN,which further enhances the network programmability and flexibility,and overcomes this dilemma traditional network faced.However,owning to the network forwarding devices and architecture of v SDN are essentially different from the traditional network,v SDN embedding problem needs to balance the load of multiple resource while ensuring the control latency lower than the threshold.In this paper,we first formulated the v SDN embedding problem into integral linear programming model,then proposed a heuristic resource-efficient and latency-sensitive v SDN embedding algorithm KCL-v SDNE to get the suboptimal solution.KCL-v SDNE algorithm screens out the subset of the physical nodes based on K-Means for each virtual node,and the map the nodes based upper bound of latency derived by network calculus.Finally,the k-shortest Path algorithm is used to map virtual links.Simulation results show that KCL-v SDNE improves the acceptance rate of while keeping lower control latency.Next,to improve the efficiency of v SDN embedding algorithm,we proposed the v SDN mapping algorithm based on Hopfield neural network: H-v SDNE.H-v SDNE extracted subset for all of the virtual nodes in v SDN by Hopfield neural network,and then measures the mean square deviation between virtual nodes and physical nodes in this subset to determine which one should be mapped on.Finally,the K-shortest Path algorithm is executed to map virtual links.Compared with the existing greedy algorithm,the H-v SDNE algorithm also improves the acceptance rate of the v SDN while guaranteeing the control latency.