Reliable Virtual Network Mapping Algorithm with Network Characteristics and Associations

Abstract

In the existing research on reliable virtual network mapping, the relationship between network parameters is insufficiently analyzed, the network topology and historical mapping data are not fully utilized, resulting in a low virtual network mapping success rate and low utilization of the underlying network resources. To resolve this problem, this article proposes a reliable virtual network mapping algorithm based on network features and associations. Firstly, the network features related to reliable virtual network mapping are sorted out, and the underlying node reliability matrix is established based on historical data. Secondly, a Bayesian network-based inference model is constructed. Finally, based on the underlying node reliability matrix and virtual network features, two algorithms named NFA-TS and NFA-LR about reliable virtual network mapping were proposed. Compared with two algorithms named SVNE and VNE-SSM, the proposed algorithm shows the competitive performance on virtual network mapping success rate and the underlying network resource utilization.