Research On Resource Management And Orchestration In Service Oriented 5G Network

Recently,with the evolution and development of wireless communication networks and the improvement of the processing capabilities of intelligent terminals,mobile data traffic has a trend of exponential growth.As a result,network operators continually increase the density of base station deployment,and multiple heterogeneous networks are deployed at the same time,which makes the network resource management more complicated.Also,the demand of services has been diversified.Third-party applications with different network capability requirements appear explosively.These make the contradiction between closed network architecture and the requirement of open network management capability more prominent.In addition,the scenarios of wireless networks are also being diversified so that how to provide customized and differentiated network services for multiple types of applications and scenarios is a big challenge for the wireless network.To deal with the challenges above,this thesis focuse on architecture design and resource management and orchestration in wireless network.Architectures for resource management and capability exposure of service oriented 5G network are designed based on micro-service idea.And virtual radio resource management,core network resource management and load balancing,end-to-end network slice management are studied respectively.The research results in this thesis can improve the efficiency and flexibility of 5G network and reduce the cost of network operation and maintenance.The main work and innovation can be summarized as follows:(1)The research and design on service oriented 5G network management architecture.To meet the requirements of different scenarios,a service oriented 5G network architecture is designed based on the idea of micro-service.The elements of core network and access network are reconstructed into lighter services respectively.For resource management of service oriented 5G network,a resource virtualization architecture of heterogeneous wireless network is proposed.By virtualizing the radio resources,baseband resources,computing resources and storage resources of heterogeneous networks,the underlying infrastructure resources can be shared among different operators.Furthermore,an SDN and NFV based architecture for resource management is proposed.In addition,a network resource management algorithm based on random learning and auction mechanism is proposed for the problem of network slice resource management under the proposed resource management architecture.In addition,a capability exposure architecture for service oriented 5G core network is proposed.We design the architecture and implementation scheme of API gateway,capability exposure interface protocol and interaction mechanism,and performance improvement scheme for the interface.The proposed capability exposure architecture enables operators to exposure and manage network capabilities based on different third-party application requirements.(2)The research on the virtual radio resources management and orchestration in service oriented 5G network.To cope with the problem of virtual radio management in proposed architecture,we propose a virtual radio resource allocation algorithm based on the resource competing in biological model.Firstly,the virtualized radio resource allocation problem for multiple mobile virtual network operators is modeled as the resource competition model of biological population.Then,the virtual radio resource allocation algorithm based on Lotka-Volterra model is proposed.The simulation results show that the proposed algorithm has good convergence and can achieve a better balance between total utility and fairness.In addition,system user and resource dynamics can be analyzed based on the biological model,which helps network operators to develop more efficient management strategies.(3)The research on resource management and optimization for service oriented 5G core network.To cope with the problem of resource management and load balancing in service oriented 5G core network,a load balancing and resource optimization algorithm based on the Proximal Jacobian ADMM is proposed.Firstly,the load balancing and resource optimization problem in the distributed core network architecture is modeled as a mixed integer linear programming problem with minimal system cost,considering network energy consumption,processing delay,cost and delay of backhaul and so on.Then,the optimization problem is transformed,and the Proximal Jacobian ADMM algorithm is used to solve the problem.Finally,a load balancing and resource optimization algorithm is proposed.The proposed algorithm can effectively deal with the challenges to the backhaul network and the core network,and reduce the operating cost of operator.The algorithm has good performance on convergence and information interaction,in addition,it has high operating efficiency thanks to parallel computing.(4)The end-to-end slice orchestration and management prototype system for the service oriented network.To deal with the challenges of multiple types of applications and scenarios,we design and implement an end-to-end slice orchestration and management prototype system.Firstly,the service oriented 5G network slice management and deployment architecture is proposed,and the energy-efficient network service migration scheme under the architecture is designed.Then,the end-to-end slice orchestration and management prototype system platform is designed and implemented.Through the prototype platform,network operators can provide customized and guaranteed network slicing services based on common hardware,and can realize dynamic network slice status monitoring.