基于深度动态贝叶斯网络的服务功能链故障诊断算法

针对5G端到端网络切片场景下底层物理节点出现故障会导致运行在其上的多条服务功能链出现性能异常的问题，该文提出一种基于深度动态贝叶斯网络(DDBN)的服务功能链故障诊断算法。首先根据网络虚拟化环境下故障的多层传播关系，构建故障与症状的依赖图模型，并采用在物理节点监测其上多个虚拟网络功能相关性能数据的方式收集症状。其次，考虑到基于软件定义网络(SDN)和网络功能虚拟化(NFV)的架构下网络症状观测数据的多样性以及物理节点和虚拟网络功能的空间相关性，引入深度信念网络对观测数据特征进行提取，使用加入动量项的自适应学习率算法对模型进行微调以加快收敛速度。最后，利用故障传播的时间相关性，引入动态贝叶斯网络对故障根源进行实时诊断。仿真结果表明，该算法能够有效地诊断故障根源且具有良好的诊断准确度。

Fault Diagnosis Algorithm of Service Function Chain Based on Deep Dynamic Bayesian Network

To solve the problem of the abnormal performance of multiple service function chains caused by the failure of the underlying physical node under the 5G end-to-end network slicing scenario, a service function chain fault diagnosis algorithm based on Deep Dynamic Bayesian Network (DDBN) is proposed in this paper. This algorithm builds a dependency relationship between faults and symptoms based on a multi-layer propagation model of faults in a network virtualization environment. This algorithm first builds a dependency graph model of faults and symptoms based on the multi-layer propagation relationship of faults in a network virtualization environment, and collects symptoms by monitoring performance data of multiple virtual network functions on physical nodes. Then, considering the diversity of network symptom observation data based on Software Defined Network (SDN) and Network Function Virtualization (NFV) architecture and the spatial correlation between physical nodes and virtual network functions, a deep belief network is introduced to extract the characteristics of the observation data, and the adaptive learning rate algorithm with momentum is used to fine-tune the model to accelerate the convergence speed. Finally, dynamic Bayesian network is introduced to diagnose the root cause of faults in real time by using the temporal correlation between faults. The simulation results show that the algorithm can effectively diagnose the root cause of faults and has good diagnostic accuracy.