| Based on the advantages of softwarization brought by Software Defined Network(SDN)and Network Function Virtualization(NFV),network slicing enables the decoupling of network function from dedicate hardware to provide flexible,cost-effective and tailored services,but at the same time loses the high reliability guaranteed by dedicated hardware.Self-healing techniques have received a lot of attention in order to ensure high reliability of network slicing on common hardware platform.Self-healing technology,or automated network fault management,is designed to enable automatic detection,diagnosis and recovery of network faults,thereby ensuring the high performance required for network slicing.Network anomalies are considered as the presymptom of network fault,this paper focuses on anomaly detection and service function chain healing in network slicing,the main research content and innovation points are summarized below.1.To timely detect physical node anomalies and thus ensure high performance requirements for network slicing,this paper proposes a distributed online physical node anomaly detection method based on support vector data description.First,a distributed physical node anomaly detection model based on support vector data description is built;second,to calculate the kernel function in distributed data storage scenario,a stochastic approximation function is introduced to realize the distributed processing of the observation data;finally,in order to solve the model aging problem caused by offline training,an online physical node anomaly detection algorithm based on stochastic gradient descent method is proposed to ensure dynamic model update and mitigate the model performance degradation problem caused by anomaly data.Simulation results show that the method can effectively detect physical node anomalies while implementing distributed processing of Virtual Network Function(VNF)observation data.2.To ensure high reliability of network slicing,achieve prompt healing after Service Function Chain(SFC)anomalies,and improve the low network resource utilization of backup SFC healing method,this paper proposes a resource preconfiguration SFC healing algorithm.First,a VNF healing overhead model based on VNF reconfiguration and state migration overhead is established,and a VNF healing algorithm with minimal healing overhead is proposed,aiming to migrate VNF to normal physical nodes;second,to reduce the frequent migration caused by instantaneous anomalies,a node anomaly index is introduced,and a SFC healing algorithm based on resource preconfiguration is proposed,thereby reducing the overall healing overhead of network slicing.Simulation results show that the algorithm can effectively reduce the number of VNF migrations,thereby reducing the overall healing overhead of the network,while having a higher resource utilization. |
