Researches On Intelligent Control And Resource Optimization Of Mobile Optical Fronthaul Network

Recently,with the rapid development of 5G network technology,the future network services presents a trend of diversification and complexity.Under the development of increasing communication demand and system scale expansion,the current optical network is exposed to the limitation of network scale expansion caused by closed and rigid network operation and management.Low degree of network intelligence,rigid resource planning and pre-configuration resource allocation result in inefficient utilization of network resources,poor dynamic adjustment ability and other problems.All these problems result in the lack of intelligent differentiated management and dynamic and flexible service guarantee capabilities of optical networks.To conquer the scientific problems,optical network needs to develop from closed and rigid operation management to flexible,open,intelligent and differentiated management and control,and from rigid resource planning to dynamic and flexible resource scheduling.We focus on the study of intelligent management control and resource optimization strategies to effectively meet the requirements of services in terms of delay,bandwidth,connection reliability and other aspects,so that the network has the ability of intelligentization,differentiated management,scale expansion,and resource reconstruction.Therefore,this paper focuses on the intelligent control and resource optimization technology of mobile optical fronthaul network,analyzes the challenges of QoS and low latency requirements of services,as well as the orchestration and reconstruction of computing and storage resources faced by intelligent optical network from optical network management and control mechanism and resource optimization strategy,mainly including:network tidal effect resulting in low utilization rate of base stations,the large and extremely uneven geographical location distribution of users,orchestration and management optimization of fog resources,and high energy consumption caused by a large number of base stations.The main innovations are as follows:1)In 5G mobile fronthaul network,aiming at network tidal effect resulting in low utilization rate of base stations,an intent-based traffic grooming scheme are proposed.First,present an AGCN-GRU network to learn the temporal and spatial dependencies of traffic patterns of cell sites to provide accurate traffic predictions.Then,the predicted results are input to the proposed AI-assisted intent-based traffic grooming scheme to realise automatic and intelligent cell sites clustering.Finally we set up software defined network(SDN)-based 5G optical fronthaul network platform,and the scheme can optimize network resource allocation,improve the utilization rate of resources,and reduce the delay and packet loss rate.2)In view of the large and extremely uneven geographical location distribution of users,the range of conventional passive optical networks(PONs)is 20 km,which are difficult to accommodate increasing number of users and coverage areas.Therefore,a Long-Reach PON(LR-PON),to expand access network to hundreds of kilometers,is proposed.And in order to improve the resource utilization of network architecture,throughput and capacity,guarantee efficient data-reliable transmission for all types of services,considering the cross layer optimization based on the MAC and PHY layer,a resource allocation scheme of transmission of quality(QoT)-aware adaptive modulation format is proposed.The simulation results show that the algorithm can reduce the average latency,significantly improve the network throughput and quality of service,and meet the QoS requirements of various services as well as the increasing number of users and coverage area.3)Aiming at the challenge of orchestration and management optimization of fog resources,firstly,a cloud-fog based fronthaul network architecture is proposed,and then a containerized resource management mechanism is introduced to coordinate the computing and storage resources of fog computing network.On this basis,aiming at the problem that container resources cannot be occupied or released in time,an edgeintelligence-driven containerization resource pre-orchestration scheme is proposed to effectively allocate computing resources.A fog computing network testbed based on embedded devices is established and containerization technology is relised on the testbed.The experimental results show that the scheme can on-demand pre-orchestrat container resources,to ensure that the container resources can be occupied and released in time,significantly reduce the time delay.4)Aiming at the problem of high energy consumption caused by a large number of base stations in 5G mobile optical fronthaul network,an energy-efficient vBBU pre-migration strategy(EBPM)based on deep reinforcement learning(DQN)is proposed.According to the predicted results of traffic load,some vBBU resource pools can be shut down in advance.The services connected to this vBBU resource pools are migrated to the specified vBBU resource pool according to the policy result to reduce the number of enabled vBBU on the network and reduce power consumption.The proposed energy-efficient vBBU pre-migration strategy(EBPM)based on deep learning(DQN)can save energy while ensuring service quality,minimize energy consumption under different service loads,and generate satisfactory solutions for large-scale service paradigm.