| In order to meet the ever-increasing diversity and innovative requirements of the service,in the future evolution of the network,an application-oriented new network architecture will be gradually built to virtualized resources and improve the layered control capability of the network.As the network component closest to the user,the access network should also adapt to the development trend of the future network,give full play to the advantages of being close to users,introduce SDN,NFV and cloud computing technologies,and combine with the edge data center to reconstruct the optical access network,as a result provides customized network services and dynamic real-time network resource scheduling.This paper focuses on the multi-dimensional resources(such as computation,bandwidth,storage,etc.)allocation problem in the soft-defined elastic optical access network,and expands network resource dynamic scheduling and service function chain to provide technical research on demand.The main research results are as follows:(1)Based on the WDM/OFDM-PON system of software defined elastic optical network architecture,we propose a traffic-aware joint scheduling algorithm for wavelength,subcarrier and modulation format.The algorithm performs wavelength and subcarrier allocation according to the ONU traf:fic demand and comprehensively considers the fairness of bandwidth allocation as well as the limitation of the physical modulation distance of each ONU relative to the OLT on the signal modulation format and spectrum efficiency.In the tidal effect network model scenario,the simulation results show that,compared with the traditional wavelength and subcarrier equalization allocation algorithm in a WDM OFDM-PON system with a peak of 240 Gbit/s,the weighted load balancing and subcarrier joint allocation algorithm proposed in this paper balances the request bandwidth satisfaction rate of all ONUs,provides 145.7 Gbit/s bandwidth during the peak period of the Internet,which increases the OLT throughput by 5%and guarantees a bandwidth request satisfaction rate of at least 93.4%,thereby improving user access bandwidth satisfaction and OLT throughput.(2)Aiming at minimizing the service delay,this paper proposes a delay-aware scheme for service chain resource allocation based on resource topology.According to the service demand of the service function chain and the resource(service and computation)occupancy of the node,the spine-leaf structure network topology of the edge data center is decomposed into resource connection topology.Taking the delay as the link and node cost,the Dijkstra shortest path algorithm is used to calculate the minimum delay service path,and the load condition is considered in the node cost to ensure the balance distribution of the load.The simulation results show that,in the edge data center spine-leaf structure networking topology,the algorithm can balance the resource usage of each service node.Compared with the random path resource allocation algorithm.The algorithm proposed can reduce the business latency by 60%(from 138.8ms to 55.3ms)under the maximum network load condition of 100%guaranteed user service function chain request,thereby effectively reducing the service delay and improving the service function chain.Service carrying capacity and user experience.(3)We build a service chain management and service provision experimental platform based on OpenDayLight open source controller,simultaneously,based on the northbound interface to provide customized service chain demand for users with a graphical service function chain customization window,receiving and parsing the intent of users and transforming into virtual resource configuration.Experiments show that the service function chain management platform can realize the functions of creating,modifying,querying and deleting service function chains according to user requirements,and also verify that the data traffic can be served through the assigned virtual network function according to the user's need. |
PDF Full Text Request