Average Service Time Analysis of a Clustered VNF Chaining Scheme in NFV-Based V2X Networks

The recent advancement in intelligent transportation system makes it crucial to equip vehicles with intelligent onboard system (IOS). With the help of IOSs, vehicles can obtain varieties of convenient services from both remote service centers and other vehicles. However, the deficiency of the existing IOS makes the upgrade process complex and costly for the reason of its closed architecture. In order to solve this problem, network function virtualization (NFV) is introduced to make the IOS open by virtualizing services as software applications that can be executed on standard IT platforms. Furthermore, the NFV-based IOS can build an ordered service chain, consisting of virtualized network functions (VNFs), among different IOSs, and support VNF reuse. Thus, it is imperative to optimize the chaining scheme to obtain a shorter average service time (AST). In this paper, we propose a clustered VNF chaining scheme, which deploys VNFs in clusters according to the cluster head of each vehicle clusters, and derive the expression of the AST in a relative static scenario. Based on this scheme, the dynamic AST of a given cluster in a moving scenario is also analyzed on the grounds of both macroscopic mobility model and microscopic mobility model. The analyses show that both of the vehicle density and distance-headway variations have influence on the AST of one given cluster, which is verified by numerical results. In addition, the experiment results also show that the proposed scheme has a better performance compared with other two schemes, which reduces the AST 61% and 43%, respectively.