Research On Cooperative Management Of Network Slicing Based On Multilayer Complex Network Theory

With the development of various services from vertical industry such as au-tomatic drive,energy Internet,smart city,etc.,mobile communication networks should provide ubiquitous access at high rates for users while using limited net-work resources to provide differentiated and customized services.To achieve the above requirements,network slicing(NS)has been proposed as a con-cept for slicing a common infi-astructure network into multiple logical networks which are mutually isolated,managed independently and created on demand.In order to accelerate the deployment cycle of differentiated services,End-to-End(E2E)management,deployment and orchestration of NS are required to realize resource control covering radio access network(RAN),transport net-work and core network domains.Meanwhile,for supporting the demand of multi-tenants such as vertical industry participants to customize network slices,mobile network operators(MNOs)expand network capacity through coopera-tion and resource sharing among multiple infrastructure networks,thereby re-ducing capital expenditure and operational expenditure(CAPEX/OPEX)and improving the return on investment.Therefore,it has become a crucial issue to realize E2E slice management and the cooperation among multiple infrastruc-ture networks in order to achieve the demands of differentiated and customized services.The aims of this thesis are to provide users with differentiated and cus-tomized E2E services,improve the resource utilization efficiency of the infras-tructure networks,support new multi-tenant business model.Therefore,this thesis will focus on the deployment and management of E2E slices under the scenario of cooperation among multiple infrastructure networks based on multi-layer complex network theory.First,a cooperative management framework for NS is designed,and then the performance analysis of demand-drived coopera-tion among multiple infrastructure networks is accomplished.Finally,accord-ing to the established framework and the obtained analytical results,the service-oriented deployment policy and the dynamic resource management strategy for E2E NS are proposed in order to realize a new NS cooperative management mode driven by the customized service demands.The main research work and innovative achievements are as follows:1.The creation of cooperative management framework for NS based on multilayer complex network theoryThe existing management framework for NS provides a solution for the instantiation process and life-cycle management of core network slices,which promotes the realization of deploying,managing and orchestrating slices.How-ever,few existing frameworks can support E2E slice management,provide co-operation strategies among multiple infrastructure networks,and realize cross-domain deployment of E2E slice.This requires obtaining the topology of the infrastructure networks and describing slice deployment in a straightforward way from a global perspective.To deal with this problem,a cooperative man-agement framework for NS is introduced.The joint physical network and the E2E slices which are deployed in it are abstracted as a multilayer complex net-work.In this framework,the joint physical network is formed by multiple in-frastructure networks provided by several MNOs.Based on multilayer complex network theory,the topology of the joint physical network and the information of slice deployment are captured.Compared with the traditional framework,the network topology,deployment location and resource allocation are more clear and transparent,which is conducive to achieve optimal cooperation,rapid deployment and efficient management.2.Analysis of service demand oriented cooperation strategy among mul-tiple infrastructure networksIn the scenario of multi-tenant slicing,the existing cooperation strategies lack the consideration of the structural characteristics of infrastructure networks,and the joint physical network cannot adapt to the corresponding service de-mands of slices.To solve this problem,this thesis proposes a cooperation strat-egy which is driven by the demand of slices,and analyzes the influence of in-frastructure network structure characteristics on the results of cooperation.Tak-ing the delay-critical slices as an example,the demand of this type of slices is reducing the transmission delay and the joint physical network needs to support fast data packet transmission.Hence,this thesis uses four typical complex net-work models to imitate different infrastructure networks which have different structural properties,the selection of cooperation nodes and the optimization of cooperation locations are realized according to the demands of the slices.The structural properties are obtained according to their corresponding measures used in complex network theory,and the effect of these properties on the above two stages is analyzed.Finally,the simulation results of the joint physical net-work formed by using the proposed cooperation strategy show that the demand of slices is met.3.The cross-domain deployment policy for E2E NS with the differentiated service demand guaranteeThe deployment of slices includes the mapping of virtual network func-tions(VNFs)and the mapping of connections between VNFs.The existing de-ployment algorithms cannot realize the cross-domain deployment of E2E slices,and cannot meet the differentiated demands of multiple slice types for resources in different domains at the same time.The service-oriented deployment policy proposed by this thesis provides specific objectives and algorithms for three typical slices according to their demands.In order to realize fast deployment,the measures of structural properties used in complex network theory are com-bined to define the node importance for physical nodes.The mapping of virtual nodes is achieved with the breadth-first-serach(BFS)algorithm considering the node importance.The mapping of virtual links is achieved with the K-shortest paths(KSP)algorithn which is used to search the physical links.Considering that three types of slices have different requirements for networking resources,storage resources and computing resources from RAN,transport network and core network domains,the experimental results demonstrate that the proposed deployment policy can optimize the resource utilization and meet the demands of slices simultaneously.4.The elastic resource management strategy for E2E NS with the traffic performance guaranteeIn order to slove the problem of traffic congestion caused by resource com-petition among E2E network slices,this thesis establishes a data packet trans-mission model based on multilayer complex network.The measures of traffic performance are defined and the traffic performance of E2E slices is analyzed from two aspects.One aspect is analyzing the effect of resource allocation parameters including the weight of allocation for overall slice and the weight of allocation for single node.The other aspect is analyzing the effect of the total number of slices and the scale of each slice.On the basis of the analy-sis results,this thesis innovatively presents two strategies for elastic resources management.The first management strategy based on the part resource opti-mization avoids the traffic congestion by improving the resource capacity of several hub physical nodes.The second management strategy based on global resource cooperation improves the traffic performance by enlarging the scales of the infrastructure network.According to the experimental results,the traffic performance can be promoted largely at a lower cost and the quality of slices can be guaranteed with these two proposed resource management strategies.