Cooperative Anomaly Detection With Transfer Learning-Based Hidden Markov Model in Virtualized Network Slicing

Abstract

Network slicing can partition a shared substrate network into multiple logically isolated virtual networks to support diverse service requirements. However, one anomaly physical node (PN) in substrate networks will cause performance degradation of multiple network slices. To realize the self-organizing management of network slices, a cooperative anomaly detection scheme is designed in this letter through utilizing the transfer learning-based hidden Markov model (TLHMM). The PNs are first classified into four different states. Then, the hidden Markov model (HMM) is used to capture the current states of PNs based on the measurements of virtual nodes (VNs). Finally, according to the learned knowledge of networks and the similarity between PNs, the concept of transfer learning is introduced into HMM to propose a cooperative anomaly detection algorithm. Simulation results demonstrate that the proposed TLHMM-based cooperative anomaly detection algorithm cannot only speed up the learning process, but also achieve an average detection accuracy of more than 90%.