Research On Control Architecture And Resource Management Algorithm In Sdoan

With the development of information technology in modern society,a variety of intelligent terminals and network services are emerging.As the last kilometer of the information superhighway,the access network has drawn more and more attention in recent years for its flexibility,stability and service quality.However,the traditional access network is not flexible enough,and the adjustment and expansion of the network architecture are not convenient enough.The scheduling of network resources is also inefficient,which is far from satisfying the development of future network services.In order to solve the above problems,in recent years,people are gradually focusing on software-defined network research and trying to apply this innovative network architecture solution to the access network,so as to enhance network flexibility and collect network global information for more efficient resource scheduling.In addition,in view of the delay of network information collected by the controller during network operation,some network traffic modeling theories have been put forward and some traffic prediction algorithms have been proposed according to related theories to perform bandwidth scheduling in real time.These studies aim to optimize network architecture and improve network performance,which has important practical significance and value.Based on the research of existing optical access network architecture and resource allocation scheme,this dissertation aims to optimize the network structure,and improve the network performance.In view of the current network intelligent and personalized management requirements,this dissertation focuses on the network architecture and resource management algorithms for the current intelligent and personalized management needs of the network and verified the performance of the proposed scheme through simulation.The specific outcomes are expressed as follows:(1)A software-defined time division multiplexing passive optical network(TDM-PON)architecture based on distributed control is proposed.This architecture divides the network into three layers using the idea of software-defined and programmable control,and use a distributed control cluster of three controllers at the control level.The simulation results show that this architecture improves network flexibility and network management efficiency while also improving the reliability of the control level.(2)Based on the study of network traffic characteristics,a BP neural network based traffic flow prediction model for access network is proposed.The simulation results show that the BP neural network model has the characteristics of fast convergence and good prediction.It is suitable for application traffic prediction in the access network.(3)Based on the above distributed control software-defined TDM-PON architecture and traffic prediction algorithm,an adaptive dynamic bandwidth allocation(ADBA-NNP)algorithm based on neural network prediction is proposed,which not only reduces the impact of control message transmission delay on the network performance through traffic prediction,but also fully analyzes the global information of the network and performs accurate and adaptive scheduling on the network in real time.The MATLAB simulation shows that the proposed algorithm reduces network average packet delay by 61%and improves network bandwidth utilization by 29%.