Study On Resource Management With Wireless Network Virtualization

With the development of wireless network technologies and the evolution of intelligent mobile terminals,the volume of mobile data traffic is increasing exponentially.At the same time,the wireless network presents a trend of node density,service diversification and heterogeneity with the topology becoming more complex and diverse.As a result,the operating expenses(OPEX)and capital expenses(CAPEX)of the mobile network operator(MNO)increase.Meanwhile,the resources are limited in the wireless network,which makes it challenging for resource management with a high utilization in the future wire-less network.Wireless network virtualization(WNV)is proposed to make the physical resources of the MNO and the services it provides decoupled.It en-ables different service providers to share the physical resources belonging to the same MNO,resulting in maximizing the network resource utilization.Network function virtualization(NFV)and wireless resource virtualization(WRV)are the key technologies to realize the WNV architecture.How to deploy the virtual functions and how to slice virtual resources are the key challenges to be solved.Based on these two aspects,this dissertation will investigate the resource man-agement under the WNV architecture.It will focus on the deployment of virtual functions in an intelligent vehicular network,and the wireless resource slicing in a multi-service coexisting wireless network.The detailed contributions are summarized as follows.(1)The virtual network function(VNF)deployment and service function chaining in NFV-based intelligent vehicular networks(VANETs):In the in-telligent VANETs enabling collaboration between devices,a VNF distribution model is established based on the popularity monitoring by using the NFV tech-nology.Then,by analyzing the request hit probability of various VNFs,a clus-tered VNF chaining scheme is proposed.The expressions of the average service time(AST)of the service function chain(SFC)in two scenarios are analyzed as well.The simulation results show that the VNF distribution parameters,cluster radius/cluster length and the number of VNFs in a SFC are the main factors affecting the AST,where an optimal cluster size can be obtained.Besides,the simulation results also show that the proposed scheme can reduce 68%and 36%AST compared to the random VNF deployment scheme and the none-clustered VNF deployment scheme,respectively.(2)The dynamic AST analysis of the service function chaining scheme in NFV-based VANETs:Based on the service function chaining scheme in(1),considering a network scenario with a varying topology which results from the random movement of vehicle nodes,the macroscopic vehicle mobility model and the microscopic vehicle mobility model are established.Then,by analyzing the variations of the vehicle number in a cluster under the macroscopic and microscopic motion models,the dynamic expressions of AST variations of the SFC are given respectively.A given case study is shown in detail to analyze the factors influencing AST of one SFC.The numerical results and analyses show that vehicle density and distance-headway variations are two factors influencing the AST of the SFC in one given cluster.Meanwhile,it is verified that the AST in one cluster can be guaranteed to maintain a relatively stable value over time by using the proposed scheme.(3)Delay-aware resource management for multi-service coexisting D2D underlay networks with WNV:In the multi-service coexisting D2D underlay networks,considering dynamic traffic arrivals,the queueing model is estab-lished.Then,a multi-objective-stochastic-optimization-problem-based wire-less network virtual resource allocation framework is established.Leverag-ing on weighted Tchebycheff approach and Lyapunov optimization technique,the multi-objective stochastic optimization problem can be transformed into a mixed-integer nonlinear programming(MINLP)problem.A distributed algo-rithm based on the alternating direction method of multipliers(ADMM)is pro-posed.The simulation results show that the proposed scheme performs com-parably with the optimal scheme by exhaustive search with a much lower com-putational complexity.Besides,a rate-delay tradeoff is achieved shown in the simulation results by regulating the control factor.(4)Two-time-scale-based hierarchical resource allocation in multi-service wireless networks with WNV:In the multi-service wireless virtualized network with different service requirements,a 3-slices system model by WNV is estab-lished.In order to guarantee the isolation between slices,a two-time-scale-based hierarchical resource allocation scheme is proposed.In the large time period,a multi-objective-optimization-based inter-slice resource preallocation problem is formulated.While in small time slot,the stochastic-optimization-based resource scheduling problem in each slice is formulated.For the inter-slice resource pre-allocation problem,an optimal resource pre-allocation scheme is obtained to make the performance of each slice balanced by the weighted Tchebycheff approach.Based on this pre-allocation decision,the correspond-ing joint spectrum and power scheduling algorithms are proposed for each slice.According to the simulation results,the proposed schemes perform compara-bly with the exhaustive search with a lower complexity.It can also guarantee a better slice isolation,especially showing the superiority in the case that the service request rate of users is increasing.