Data Flow Scheduling Strategy In Software Defined Networking

Software defined network(SDN)considered as a new network architecture has attracted attentions of academia and industry.The core idea of software defined network is to separate the control function and the forwarding function of networks and then constructs a new network architecture that consists of data plane,control plane and software defined application plane.This architecture greatly improves the flexibility of network control and the efficiency of data forwarding.With the rapid development of the Internet,the network scale becomes larger,which leads to the difficulties of the network management.Thus,the development of software defined network aims to enhance the management level of network by improving the operation efficiency of the network.At the same time,due to the higher demand of services quality of network required by users,effective data flow scheduling strategies should be made in the network to provide the lower data transmission delay for users.In addition,it is widely common that data flow conflicts appear in the large-scale networks,which is because the current data flow scheduling strategies are based on local network status information and further lead to the local optimal.Then,these strategies will greatly degrade the operation efficiency of the network.Therefore,the study of the manage network resources and the design data flow scheduling strategies through software defined control technologies have great realistic and practical value.Based on the latest researches of software defined network,this paper mainly studies the application of software defined control technology in data center networks,wireless sensor networks and intelligent transportation networks.The main works and contributions of this paper can be descried from the following three aspects:1.The data flow scheduling problem of software defined wireless sensor networks is studied.Considering the data flow minimization(TLM)issues of the software defined wireless sensor networks(SDWSN).From the perspective of different data packet types,data flow transmission load is proposed as the target of optimization function.And the Levenberg-Marquardt algorithm is used to solve the network load optimization problem and it can be proved that the algorithm is converged.In addition,we propose the flow splitting optimization(FSO)algorithm based on the optimal relay sensor node selection and the optimal split data flow.Finally,the simulation results verify that the FSO algorithm can perform the higher efficiency of data flow transmission in the software defined wireless sensor network.2.The multicast data flow scheduling strategy of software defined fat-tree data center network is studied.It should be taken into consideration that the load imbalance and the burst link congestion will seriously affect the performance of fat-tree data center network(DCN).From the perspective that the software defined control technology can manage the global network,the proposed topology of software defined fat-tree data center network(DCN)can avoid the multicast data flows that are not interfered with others bandwidth requests.In addition,based on the idea of uniform distribution of the multicast data flows over the effective link,the multicast data flow scheduling(MSaSDN)algorithm is proposed.The MSaSDN algorithm improves network performance through reducing the multicast data flow transmission congestion.It can be proved that the performance analysis of the MSaSDN algorithm can achieve the asymptotically minimum congestion probability of network and lower time complexity.Finally,the simulation results verify that the MSaSDN algorithm has the relatively higher transmission efficiency of multicast data flow.3.The multicast data flow scheduling strategy of the software defined intelligent transportation network is studied.It should be considered that the intelligent transportation network has the higher requirement for the delay of real-time multicast flow.Therefore,the software defined control technology is introduced to establish the topology of the software defined intelligent transportation network and the multicast congestion model.The multicast data flow scheduling(MSaMC)algorithm based on the markov chain is proposed.The MSaMC algorithm can effectively schedule real-time data flows of the intelligent transportation network.In addition,the analysis of the MSaMC algorithm shows that it has the lower probability of network congestion in the software defined intelligent transportation network.Finally,the simulation results verify that the MSaMC algorithm has the relatively lower transmission delay of multicast data flow,and the intelligent transportation system can perform the relatively higher traffic management capability on the basis of real-time data flow transmission.