Research On Intelligent Routing Algorithms Based On Combination Of SDN And NFV

With the rapid development of the Internet of Things,the diversity of services has made their demands for the network more and more diverse.A key issue driving the development of the Internet of Things is to create a performance-optimized network for users to guarantee the quality of services and provide customized services.The rise and rapid development of SDN and NFV have aroused widespread concern in the acdemia and industry.SDN can dynamically adjust network-wide traffic to meet changing needs.NFV uses virtualization to eliminate dependencies on dedicated servers,switches or storage,thereby reducing the cost.However,with large-scale services and highly dynamic network loads,it is a challenge for current and future networks to construct complex service-driven dynamic network.Therefore,this thesis studies the algorithms of intelligent routing in the context of multi-service requirements and mainly carries out the following three aspects:(1)Focusing on the deployment of service function chaining,NFV and SDN are combined to optimize resources.Considering the forwarding cost and the traffic balance in the network,and a joint optimization model of virtual network function placement and service routing is given.In the solution phase,two heuristic algorithms have been proposed according to the order of VNFs placement and traffic flow routing.The simulation results show that the two proposed algorithms have different advantages in reducing the forwarding cost and balancing the load.However,compared with the greedy placement algorithm based on the shortest path,the two proposed algorithms can effectively balance the traffic in the network.(2)In view of the shortcomings of traditional routing algorithms,such as slow convergence speed and difficulty in real-time perception of network changes,Q-learning algorithm is introduced to solve the SDN routing problem in this thesis.In the established Markov decision process,the reward function includes bandwidth,delay as well as other parameters such as packet loss.Compared with other algorithms,the Q-routing algorithm takes into account the link state and shows obvious advantages in improving link utilization.It can better balance link bandwidth utilization and transmission delay,thus ensuring service QoS.(3)With respect to the routing system based on Ryu controller,SDN management and control platform is designed to display dynamic information such as real-time network dynamic topology,switches ports statistics.During the testing phase,the routing algorithm based on bandwidth utilization is compared with the shortest path algorithm.Simulation results demonstrate that the routing algorithm based on bandwidth utilization plays an important role in balancing the traffic load in the network.