Solution And Implementation Of SDN Multi-controller Deployment And Disjoint Path Method Based On FPGA

In such a developed era of network technology,the internet has already become an indispensable part of our lives,such as surfing the internet,online shopping,mobile payment and other aspects,which have subtly changed our lifestyle.As network technology continues to affect people’s lives,it also spawns more innovative network services,which are also bringing a series of challenges to traditional network architectures.Because the invention of the network serves the national needs,the initial network business requirements are relatively simple,and designers have not considered that such complicated network business requirements will arise in the future.Therefore,there must be some historical limitations in the design of traditional network architecture.For example,the defect that the network cannot be deployed flexibly has caused it to fail to meet the needs of emerging network services.SDN is expected to become a substitute for the traditional network structure.Because the SDN network has programmable features,it can make network configuration and management easier.In addition,the decoupling feature of the SDN control layer and forwarding layer greatly improves network flexibility and further improves network efficiency.In the early stage of SDN research,single controller was the mainstream network structure.With the deepening of research,a single controller has encountered a performance bottleneck in large-scale SDN networks.When controller is under heavy load,it often fails to respond to transmission requests from SDN switches in a timely manner.In order to solve this problem,the multi-controller SDN network was proposed.Although the multi-controller SDN network can reduce the load of each controller,it also brings some problems,such as multi-controller deployment,network topology collection and communication path selection.With further research,the researchers found that the deployment position of the controller in the network can directly affect network performance.Existing methods are mostly based on a controller software platform to implement a deployment algorithm.Aiming at improve the algorithm efficiency,this paper proposes a method of multi-controller deployment in SDN based on FPGA.Firstly,a cooperative model of multi-controller based on FPGA is designed,which can take advantage of FPGA parallel computing and improve the consistency of multi-controller.By running the SDN multi-controller deployment algorithm in this model,a reasonable controller deployment solution is obtained.Then,in order to improve the communication efficiency of the network,a disjoint path algorithm based on FPGA is designed.In order to solve the above problems,the innovative work of this article can be summarized into the following two points:1)Aiming at the problem of overloading the controller in SDN environment,a multi-controller deployment algorithm of SDN network based on FPGA is proposed.In this paper,the goal of minimizing the delay between a controller and switches,and use the characteristics of hardware parallel computing to improve the efficiency of the algorithm.Firstly,a cooperative model of multi-controller based on FPGA is designed.by By using this model,the multi-controller deployment has been discussed,and the multi-controller deployment strategy based on the improved K-means clustering algorithm has been proposed.2)To solve the problem of network congestion,a disjoint path algorithm based on FPGA is designed.Through FPGA,the improve disjoint path algorithm based on Dijkstra is hardware accelerated,and the improved algorithm can realize the calculation of the shortest path that do not intersect the optimal and suboptimal paths on the specified network topology.Experiments show that this method has lower latency and higher execution efficiency.