Researches On Intelligent Control Technology Of Optical-Wireless Converged Access Network For 5G

In recent years,with the rapid development of the fifth generation of mobile communication(5G),the access network has shown a trend of broadband and diversification.In addition,in order to meet the ultra-low latency requirements of real-time services,the concept of edge computing came into being.The cloudification optical and wireless converged access has become an important networking architecture in 5G networks.The ever-increasing user requirements and application scenarios have placed new demands on the flexible bearer and intelligent resource management of the converged access network.In 5G era,services types are various,and network resources are diverse and complex.Some new services,such as ultra-reliable low-latency services,Internet of Things,and industrial control networks,require optical and wireless converged access networks to provide higher quality in specified performance and service guarantee.Therefore,how to use intelligent management and control technology to reasonably allocate resources and effectively provide more optimized network for 5G new services is a key issue to be solved in optical and wireless converged access networks.In addition,with the deployment of 5G networks,green and low-cost is also an important development trend.How to use intelligent resource orchestration to improve network resource utilization and reduce network resource consumption is also a challenge that needs to be solved in the converged access network.Therefore,this thesis focuses on intelligent control technology for the 5G optical and wireless converged access network.The main innovations are as follows:1)Aiming at the problem of poor network scalability and rigescent resource management in 5G mobile optical fronthaul network,a joint optical and wireless integrated resource scheduling scheme and a corresponding wavelength dynamic sharing algorithm for load balancing are proposed.The simulation results show that the proposed scheme can realize the joint scheduling of heterogeneous resources and the QoS guarantee of multi-services while improving the load balacing performance between multiple wavelengths.2)Aiming at the convergence of edge computing and optical access for 5G C-RAN,a hierarchical edge cloud architecture and a corresponding low-latency guarantee scheme based on traffic offloading are proposed to implement localized processing of local delay-sensitive traffic.The simulation results shows that the proposed scheme can effectively reduce the delay of delay-sensitive services reduce the jitter and the occupancy of the fronthaul bandwidth,in detail,the delay will decrease by 80%in heavy-load condition.3)Aiming at the ultra-low latency requirement in 5G network,a load-aware dynamic traffic migration scheme based on the the hierarchical edge cloud architecture is proposed,and a SDN network testbed based on OpenFlow is built for evaluating the proposed scheme.The experimental results show that the proposed scheme can realize on-demand dynamic orchestration of heterogeneous resources in the hierarchical edge cloud optical fronthaul network and achive the low latency of the delay-sensitive services.Moreover,the scheme is able to improve the resource utilization.Specifically,the delay can decrease by 50%-80%,and the jitter can decrease by 50%?75%.In low-load condition,the bandwidth usage can decrease by 38%?4)The existing "one size fits all" architecture can not provide flexible and customizable resources for various scenarios,in order to realize the flexibility of existing network architecture and dynamic resource supply,this paper proposes a hierarchical edge cloud-based optical fronthaul network architecture and a corresponding integrated network resource management based on MILP algorithm.The simulation results show that the proposed scheme and algorithm can effectively realize the flexible resource allocation and heterogeneous resources orchestration,while meeting the different QoS requirements between different network slices.In addition,the proposed schemes and algorithms can also reduce the load on the fronthaul network and improve network resource utilization.For example,to the uRLLC slice,the end to end delay can be maintained within 2ms?and the bandwidth resource release ratio can reach to 21%at most.5)The existing optical transmission network lacks intelligent decision-making ability and resource supply of which is not timely.To solve the problem mentioned above,a load-aware network slicing scheme using artificial intelligence technique and two corresponding traffic prediction algorithms based on artificial neiural network or LSTM are proposed for the future 5G metro-aggregation networks,experimental results show that the proposed artificial intelligence algorithms has excellent prediction accuracy on traffic prediction.Moreover,the proposed load-aware network slicing scheme can reduce 57%blocking probability and 18.2%energy consumption.