中文摘要：

针对SDN中控制器系统的单节点故障问题，兼顾系统成本和系统时延，应用N＋1冗余备份模型来提高SDN控制器部署的可靠性，并将其抽象为多目标优化问题。同时，提出了一种融合K—means聚类算法和遗传算子的多目标二进制粒子群算法——KMOBPS0算法，以求解SDN控制器高可靠性部署问题的解。仿真结果表明，所提算法具有求解精度高、分布均匀、沿Pareto前沿面覆盖广的特点，能够显著提高SDN中控制器部署的可靠性。

英文摘要：

N＋1 redundancy backup model was applied to controller placement problem in SDN in order to solve single node failure problem and to improve system reliability with taking system cost and latency into consideration, and the problem was abstracted as a multi-objective optimization problem. Meanwhile, an algorithm called KMOBPSO was proposed to calculate the result of high-reliability controller placement in SDN merged with K-means clustering algorithm and genetic operators. Simulations show that the proposed algorithm who has high accuracy, uniform distribution and wide coverage of Pareto front, can significantly improve the reliability of controller placement in SDN.