Symptom-aware hybrid fault diagnosis algorithm in the network virtualization environment

Abstract

As an important technology in next-generation networks, network virtualization has received more and moreattention. Fault diagnosis is the crucial element for fault management and it is the process of inferring the exact failurein the network virtualization environment (NVE) from the set of observed symptoms. Although various traditional faultdiagnosis algorithms have been proposed, the virtual network has some new characteristics, which include inaccessiblefault information of the substrate network, inaccurate network observations, and a dynamic embedding relationship.To solve these challenges, a symptom-aware hybrid fault diagnosis (SAHFD) algorithm in the NVE is proposed in thispaper. First, a multifactor Bayesian hierarchical model is proposed to denote the relationships between multiple factorsin different layers. Second, the contribution degree is improved to locate the faults in the virtual layer and the activedetection algorithm is introduced to filter some spurious faults in virtual layer fault diagnosis. Then, in substrate layerfault diagnosis, the active detection algorithm is introduced to solve the problem of incomplete network observations.Finally, a heuristic greedy algorithm is proposed to select appropriate actions based on minimum weight set coveringmethod with minimum cost. Simulation results show that, compared with other algorithms, the SAHFD algorithm hasa higher accuracy rate, lower false positive rate, and better environmental adaptability in the NVE.