Awareness-based Research Of Network Virtualization Scheduling And Embedding

With the rapid development of network technologies,there is a tremendous growth in wireless traffic and services.The wireless networks is becoming much more heterogeneous and denser.Wireless network virtualization is one possible solution to simplify the management and allow the co-existence of different technologies.Network virtualization makes heterogeneous virtual networks able to run on the shared substrate networks,thus improving the utility of infrastructure and decreasing the CapEx and OpEx.Virtual network embedding is one of the key point of network virtualization,and it is related to whether network virtualization can move into engineering practice.In this thesis work,I first introduce the concept of network virtualization and virtual network embedding,and briefly analyzes the existing virtual network embedding algorithm in wired and wireless environments.Then for the wireless network scenario,in order to deal with that the objective indicators,like QoS,cannot truly reflect the QoE,a wireless network virtualization architecture that extends the concept of network virtualization to the user equipment is proposed.Taking user satisfaction as the optimization goal,a service-aware resource scheduling schema is proposed.Experimental simulation of the proposed user satisfaction model shows that the algorithm has better user satisfaction and fairness of resource allocation.Most of the recent works on VNE algrithem do not consider the impact of network topology.This thesis introduces the centrality of the nodes in the social network into the VNE algorithm,and proposes a mapping algorithm based on topology awareness.In the node mapping stage,the algorithm determines the node selection by the aggregation centrality of nodes,and then uses multiple commodity flow methods to perform link mapping.By defining Similarity of the VNE algorithm,it is possible to compare VNE algorithms' reproducibility of the virtual network topology and independence of the physical network topology.Simulation experiment shows that the proposed algorithm has a better virtual network request acceptance rate and revenue while keeping the cost unchanged.And for link mapping,the results have shorter paths,better virtual network topology reproducibility,and independence from physical network topology.