Research On Topology Aggregation And Routing Algorithms In Hierarchical Software Defined Network

With the continuous development of internet,various emerging network businesses require the network to provide QoS guaranteed network services.However,due to the rigid traditional network structure,models and mechanisms based on the traditional network can hardly provide satisfactory services.In this context,software defined network(SDN)emerged and attracted the attention of researchers from both domestic and abroad.SDN decouples the control plane with the forwarding plane,and network control functions are transferred to controller in the form of software.SDN is especially suitable for the research of new network mechanisms and algorithms,since network functions can be easily developed and tested by programming in the controller.However,in large scale networks,SDN controller encounters the problem of performance bottleneck,hierarchical network architecture is considered to be an important measure to address the scalability issue.In hierarchical networks,the research of topology aggregation(TA)algorithms and routing algorithms has great significance.In this paper,we first study the Qo S metric aggregation algorithm.Based on the research of QoS parameter aggregation theory,the concepts of effective service region and represent path are presented,after analyzing the represent path ladder,RMP-LS,which is a hybrid QoS aggregation algorithm based on regular multiple polyline fitting and line segments fitting,is proposed.RMP-LS can reduce the storage complexity of logical link to the constant level.Simulation results based on the SDN controller reveal that RMP-LS has a lower relative distortion with respect to other QoS parameter aggregation algorithms.Secondly,the routing algorithms of SDN hierarchical network are researched.Aiming at the issues encountered by particle swarm optimization(PSO)in solving routing problems,HawkPSO is proposed based on the improved PSO.In HawkPSO,the actual meaning and representation method of particle are redefined,and the position-velocity updating formula is discretely improved.Besides,HawkPSO introduced clone operation to achieve local proliferation of the best particle and mutation operation to increase the randomness and variousness of the particles.Finally,simulation is conducted in the SDN environment to prove that algorithms put forward in this paper have higher routing efficiency and success rate relative to other algorithms.