Research On SDN Load Balancing Based On Ant Colony Optimization Algorithm

Due to the emergence and development of various emerging network technologies,the increase in internal traffic of data center networks has led to more and more serious congestion of network links,which has greatly affected the performance of communication networks.The study finds that the current traditional network architecture can not adapt to the current network service requirements,and it is difficult to achieve efficient and flexible scheduling of data traffic.Software defined network in such a historical background came into being,this new type of network architecture completely changed the traditional network data traffic scheduling mode,it realizes the separation of network traffic control and forwarding,it has the characteristics of global network awareness and centralized network control,and provide a new idea and platform for achieving a more superior network load balancing strategy.The research content of the paper is mainly from the software defined network architecture of the link load balancing problem of data center network.In view of the traffic collision problem of traditional network traffic scheduling,the paper proposes a software defined network load balancing strategy based on the improved ant colony optimization algorithm,which can combine the characteristics and advantages of software defined network to achieve good dynamic load balancing of data center network.Based on the analysis and summary of traditional network load balancing problems,this paper designs and implements a load balancing strategy based on ant colony optimization algorithm under software defined network architecture.The paper initializes the differential link pheromones according to the load conditions of the network link,and redefines the heuristic function of the state transition probability in combination with the link load conditions,so that the network state can be combined with the better path selection.The paper also proposes a global optimal worst pheromone update rule,and in order to ensure the reasonable operation of the algorithm in the execution process,the pheromone volatilization factor is taken in stages,which improves the problems of ant colony algorithm in the actual network application,such as local optimal solution and unreasonable convergence speed.In addition,the pheromone content is also limited to ensure the continuous scheduling path search ability of the algorithm.Finally,based on the ant colony optimization algorithm,considering the characteristics of data traffic size differentiation,a software defined network load balancing strategy is designed to achieve more intelligent and efficient load balancing of network traffic.The designed load balancing strategy is then implemented through the deployment of software defined network controllers,and a complete software defined network load balancing system is built.The system mainly includes functional modules such as global topology awareness,network information collection,routing decision-making,and flow table installation.The global topology awareness module is responsible for obtaining the overall network structure in the early stage of system operation,while the information collection module collects the real-time status and traffic information of the network on the basis of the global network topology knowledge,and provides relevant information for the routing decision module to calculate the scheduling path.The routing decision module is equipped with the load balancing strategy based on improved ant colony optimization algorithm to achieve efficient and accurate load balancing.The flow meter installation module performs the flow table issuance according to the calculated optimal scheduling path to complete the reasonable scheduling of the data traffic,so as to realize the unified and centralized management of network traffic through the software defined network controller.The paper uses the Mininet network simulation tool to simulate the data center network and different network communications,and analyzes the performance of the software defined network load balancing proposed in the paper under each communication mode,which verifies the feasibility and effectiveness of the load balancing research in the paper.