Research On Dynamic Resource Allocation Algorithm For 5G Wireless Network Virtualization

The booming 5G wireless network technology paves the way for changes in the emerging modes of social production,such as the Internet of Things,Internet of Vehicles and industrial Internet.However,due to the lack of wireless resource and increasingly complicated network system,wireless network technology is facing more severe challenges,which mainly reflected in the contradiction between isomerization and densification of wireless networks,diversification of application scenarios,dynamic business and traditional wireless network architecture.The 3G and 4G networks mainly focus on the improvement of peak rate and spectrum efficiency.5G network is designed on the basis of heterogeneous ultra-dense network deployment.Therefore,it is imperative to build a new network architecture that is flexible,efficient and scalable based on software implementation and virtualization technology.At present,the network virtualization technology has been proposed as an effective way to solve the issue of ossified network architecture in many literatures.Existing researches of network virtualization focus on the wired network,mainly concerning the virtualization of data center network and backbone network.Nevertheless,the research of virtualization technology in wireless access network is still in its infancy.In order to accelerate the innovation of wireless communication technology and meet the differentiated needs of emerging services,wireless network virtualization has become a research hotspot and priority in academia and industry.From the perspective of resource allocation in wireless network virtualization environment,this thesis establishes the wireless virtual network resource allocation model and the cross-domain virtual network resource allocation model based on the analysis of the existing resource allocation algorithms.The network centrality theory in social network and complex network is introduced to analyze the topological attributes of virtual network and physical network,and two efficient virtual network resource allocation algorithms are put forward.The successive generation of the virtual network requests in the wireless network environment leads to the fragmentation of resource allocation in the underlying physical network and the lower construction success rate of the subsequent service requests.Aiming at the issue,a wireless virtual network embedding algorithm based on node aggregate centrality(NACA)is designed in this thesis.By introducing the network centrality theory,the topological attributes of virtual network and substrate network are analyzed from local and global perspectives,the resource centrality ranking measurement models of virtual nodes and substrate nodes are established respectively combined with the local resource information of network.Meanwhile the link interference quantitative coefficient and the resource load of the wireless substrate links are calculated,so the substrate bearer paths with lower interference and well-resourced could be selected to deploy.From the simulation results,this algorithm achieves higher construction success rate of virtual network requests,increases the network embedding revenue to cost ratio,and effectively improves the resource allocation balance of physical network.Existing network architecture has become more and more ossified,and it is hard to meet the demands of the heterogeneity of access modes and the interoperability of multi technologies in the future 5G network.The cross-domain virtual network embedding issue is researched in this thesis based on the multi-domain converged network architecture which supports holistic network virtualization and multi-type services processing,and a cross-domain virtual network embedding algorithm based on topology awareness and resource attributes(TARA)is designed.Based on the local and global perspectives,the topological properties of nodes within virtual network and physical network are analyzed with network centrality theory.The multi-attribute evaluation models of nodes are established combined with the expand network resource.Based on the evaluation models,the principal component analysis method(PCA)is introduced to decrease indicator dimensions,then the deployment priority of nodes is measured with the technique for order preference by similarity to an ideal solution(TOPSIS).The link load distribution is considered with link resource price to deploy virtual links.From the simulation results,the algorithm achieves higher construction success rate of multidomain virtual network requests and better network revenue to cost ratio,effectively strengthens the resource allocation balance in substrate network,and can decrease the network embedding delay.