当前的智能光网络缺乏主动应对网络故障的能力, 为了进一步提高智能光网络的生存性, 文章提出了一种基于贝叶斯模型的智能光网络主动告警机制。该机制能够预先判断网络中潜在的故障, 并具有高度的准确率以有效减少误告警; 此外, 该机制对不同样本采用不同的算法来进行网络学习, 采用概率推理估计故障的预测概率, 从而为主动告警机制建立一个可推理的预测模型。仿真结果表明, 该机制能极大地提高智能光网络的生存性, 缩短业务连接的恢复时延, 优化资源利用率。

Current intelligent optical networks can not actively deal with failure in network. Aimed to improve the survivability of intelligent optical network, this paper proposes a Bayesian model based active fault alarm mechanism in intelligent optical network. The proposed mechanism is able to judge of potential failure in advance with high accuracy and to reduce the false-alarm efficiently. Using different stylebooks and their corresponding algorithms, the fault prediction probability is rational estimated through reasoning, which builds a rational Bayesian model for active fault alarm. Simulation results show that this proposed approach can greatly improve the survivability ability of intelligent optical networks, in terms of connection recovery time and resources utilization rate.