Research On The Delay Optimization Algorithm In SDN Control Plane

The electric power communication network is the foundation support platform of the smart grid,which has great challenges in meeting the key functional requirements of new business.Traditional network architecture urgently needs to be updated.Software Defined Networking(SDN)has the characteristics of logically separating the control plane from the data plane,and the network is programmable.It has become an option for updating the network architecture.The SDN control plane is the core of the SDN architecture,its performance has an impact on the reliable and real-time operation of the SDN architecture.Delay is a key factor affecting timeliness.On the control plane of SDN,delay include static transmission delays and dynamic processing delays.The former is affected by the placement of the controller during SDN construction and varies with the number and location of the controllers.The latter is caused by the load imbalance of the control plane when the SDN network is running.In this thesis,we focus on reducing the above delay of the control plane of SDN.Both reliable and real-time need to be considered together when building an SDN architecture for power communication networks.A controller placement model considering both transmission delay and reliability is established in this thesis.In order to find the optimal solution for a function of multiple objectives including the reduction of network loss,the reduction of information transmission delay and the reduction of synchronization delay among controllers,an improved Multi-objective Sorting Harmony Search(MSHS)solving algorithm combining population sorting is proposed in this thesis.Finally,the effectiveness of this method in the thesis is verified by simulation experiments.Network flow unevenness and load imbalance are the main factors for the increase of dynamic processing delay.In order to reduce the processing delay,a Multi-target Optimization Switch Migration(MOSM)is proposed in this thesis.MOSM takes into account the costs incurred during migration and the global load of the control plane when selecting the target controller for migration.Finally,by comparison with traditional switch migration methods after simulation,which can indicate that MOSM can effectively reduce the dynamic processing delay.