为降低软件定义光网络对单控制器的依赖,并避免多控制器冲突,有效提升控制平面的生存性,该文提出基于最小点覆盖的控制平面生存性设计策略。该策略结合集中控制约束条件,以最小点覆盖理论为基础,建立可靠的分级管控模型,设定控制器的管控优先级:全局控制器具有最高管控优先级,对全网进行集中管控;本地控制器次之,只对本地业务进行集中管控;权威交换机的管控优先级最低,用于完成局部波长粒度的光层快速管控。在此基础上,基于跨层信息模型为控制信道路由和资源分配进行生存性冗余设计。仿真表明,该策略能够满足网络对控制时延的要求,使控制平面的故障概率降低了30%,有效提升了网络在恶劣环境下的生存性。
In order to lower the reliance on single controller in Software Defined Optical Network(SDON), avoid the conflict of different controllers, and improve effectively the survivability on the control plane, the survivable design of SDON control plane based on minimum point covering is proposed. Combined with the constraint of centralized control, the algorithm based on the minimum point covering establishes reliable hierarchical control model, and sets control priority to controllers. The master controller with the highest priority centralized controls the whole net, and the next comes the regional controller, which only intensively controls regional traffic, besides the authority switch which is introduced to control local wavelength in the optical layer owns the lowest level. Meanwhile, the model based on cross-layer information designs survivability redundancy for routing and allocates resource for the control channel. Simulation results show that the proposed strategy can satisfy the request on control delays, and lower the failure probability in the control plane by 30%, thus promoting the network survivability under critical environment.