Dynamic Resource Allocation Algorithm For Multi-access Edge Computing Enhanced Fiber-wireless Access Networks

Fiber-wireless(Fi Wi)networks combine the advantages of optical and wireless access networks,which can support the emerging business applications of 5G networks efficiently and flexibly.Multi-access Edge Computing(MEC)enhanced Fi Wi network combines Fi Wi with MEC organically by deploying MEC server at Optical Network Unit-Base Station(ONU-BS)to bring computing resources closer to users,which is a promising solution for implementing MEC.Meanwhile,its high-capacity and highly reliable optical backhaul network can meet the high bandwidth and low latency requirements of mobile terminal’s computing offloading tasks.In addition to solving the offloading decision and computing resource allocation,the computing offloading of the MEC enhanced Fi Wi network also involves the Dynamic Bandwidth Allocation(DBA)problem of the optical backhaul network,which requires reasonable authorization of bandwidth resources for the coexisting broadband access services and computing offloading services in the network to meet the differentiated requirements of different users.First,this thesis introduces the background and significance of MEC enhanced Fi Wi network,outlines the basic concepts of Fi Wi network,MEC and MEC enhanced Fi Wi network,summarizes the key issues and research status of MEC enhanced Fi Wi network,and analyzes the typical application scenarios.On this basis,this thesis further analyzes some typical offloading strategies and dynamic resource scheduling algorithms of MEC enhanced Fi Wi networks.Secondly,an offloading aware dynamic bandwidth allocation algorithm is proposed.In each polling cycle,the mobile device offloads the computing tasks that exceed its computing capacity to its associated ONU-BS through the wireless channel.ONU-BS will offload the bandwidth requirement information of overloaded computing tasks and traditional services to the Optical Line Terminal(OLT)according to the computing resource occupation and computing task load of the MEC server.Through the collaborative scheduling of offloading decision and DBA algorithm,the OLT selects the appropriate light-loaded ONU-BS to complete the offloading computing tasks and returns the computing results according to the offloading strategy that maximizes the offloading task under the constraint of computing delay.With a comprehensive consideration of the coexistence of broadband access services and computing offloading services,the algorithm makes full use of the bandwidth resources of the optical backhaul network and the computing resources of the MEC server configured at local or adjacent ONU-BSs to satisfy as much as the demand for broadband services and computing services of mobile users.The algorithm can not only increase the completion rate of computing tasks and reduce their delay,but also improve the broadband service’s throughput and reduce its end-to-end delay.Additionally,a cooperative offloading aware dynamic wavelength and bandwidth allocation algorithm is proposed,which adopts online and offline hybrid scheduling scheme.To reduce the delay of computing results and improve channel utilization,the algorithm returns computing results in the channel gap at the beginning of the polling cycle by online scheduling.The offline sub-cycle is divided into the computing offloading sub-cycle and broadband services transmission sub-cycle,and preferentially uses the earliest available wavelength channel to transmit computing offloading tasks in the computing offloading sub-cycle to reduce the queuing delay of the computing tasks,and subsequently transmit the traditional broadband services in the broadband services transmission sub-cycle by using the remaining bandwidths and slots.During the offloading decision,in order to fully meet the computing requirement of overloaded ONU-BS,the OLT preferentially selects appropriate light-load ONU-BS as the offloading target for the computing tasks of overloaded ONU-BS by using the horizontal computing collaborative offloading strategy,and then uses the remote centralized cloud to meet the remaining computing requirements of overloaded ONU-BS.The algorithm can make full use of the computing resources of MEC servers and remote centralized cloud to improve the completion rate and reduce the completion delay of computing task,and also reduce the adverse impact on the traditional broadband service and core network.Finally,the MEC enhanced Fi Wi network platform is built based on OPNET simulation software.Then the proposed algorithms are simulated and analyzed by comparing with several typical algorithms.The simulation results verify the effectiveness of the proposed algorithm.