5G Network Architecture Research Based On SDN

Advances in software and hardware technology have set higher and higher demands on network bandwidth.Multi-terminals and large traffic have become the main application scenarios for future mobile networks.However,the existing LTE mobile network has a lot of problems and it is hard to meet the future needs.5G program gradually be put on the agenda.SDN(Software Defined Networking)has decoupled architectural features that provide a solution for designing 5G network architectures.Designing 5G network architectures based on SDN has gradually become a mainstream practice.The existing 5G network architectures based on SDN have various problems,such as the difficulty in implementing and deploying,the poor scalability,etc.Based on the principles of the future network architecture advocated by the IMT-2020(5G)promotion group,this article participates in the project of the Institute of Information Engineering of the Chinese Academy of Sciences and jointly designs an SDN-based 5G network evolution structure called SEA.This article has completed the development of the most important SDN controllers in SEA,including flow table delivery control,link abnormality recovery,etc.At the same time,this paper builds and tests successfully the SEA experimental system and LTE comparison system.SEA reforms the LTE/EPC system and retains the main service logic of LTE,so that it has excellent service performance.Besides,SEA uses the GTP-enabled SDN switch to forward traffic,making it easier to deploy on a large scale.Experiments show that the SEA demo and LTE system take the same time to build the bearing.Meanwhile,both of them can respond to the data request of the user equipment(UE)within 30-60ms.In addition,SEA can recover data path within 6ms.Besides,given the large-scale deployment of SEA,load imbalance caused by uneven traffic distribution is a characteristic issue in the underlying SDN network for distributed deployment.Therefore,we aim at the distributed deployment of SEA architecture,build a suitable business load model,and propose the model supporting business scheduling and load balancing strategy.Experimental simulation shows that as the amount of traffic increases,the load balancing gradually approaches zero,which proves the effectiveness of the load balancing strategy.