基于近邻情景认知的软件定义网络多域协同控制机制

针对软件定义网络（software-defined networking，SDN）分布式多控制器部署中存在的控制器负载不均衡问题，提出了一种基于近邻情景认知的多域协同控制机制。首先，通过改进现有的近邻传播（affinity propagation，AP）聚类算法，以跳数为原则，对网络中节点进行聚类操作，形成SDN子域并在聚类中心部署控制器；然后基于控制器 —交换机之间的协同映射对网络连接关系进行优化，以增强网络的稳定性。仿真表明，相比其他方法，该机制能够实现网络区域的合理划分，控制器负载均衡率至少提高了26.7%。

Affinity situated cognition-based multi-controller cooperation mechanism in software-defined networking

In SDN, there are controller loads imbalance in the distributed multi-controller environment. In order to solve this problem, this paper introduced an affinity situated cognition-based multi-controller cooperation mechanism. Firstly, it improved the existing affinity propagation clustering algorithm and clustered the nodes in the network based on the hop count principle to form the SDN subdomain, and deployed controllers in the cluster centers. Then, it optimized the controller-switch connection through the cooperative mapping between controllers and switches to enhance the network stability. Results show that, compared with other methods, this mechanism can divide the network into multi-domains reasonably and improve the controller load balancing rate by 26.7% at least.