Joint Topology Design and Mapping of Service Function Chains in Network Function Virtualization

Network Function Virtualization (NFV) is promising to lower the network operator's capital expenditure and operational expenditure by replacing proprietary hardware-based network equipment with software-based virtual network functions that can be consolidated into telecom clouds. In particular, NFV provides an efficient way to deploy network services using service function chains that consist of a set of virtual network functions interconnected by virtual links. A practical and yet theoretically challenging issue related to NFV Management and Orchestration is how to jointly optimize the topology design and mapping of multiple service function chains, which is called the JTDM problem. In this paper, we develop an Integer Linear Programming (ILP) model to formulate the JTDM problem with the objective of minimizing the bandwidth consumption in the physical substrate. We propose a novel heuristic algorithm, namely Closed-loop with Critical Mapping Feedback (CCMF), to efficiently address this problem. Through comprehensive simulations, we demonstrate that the CCMF algorithm is efficient in terms of the bandwidth consumption in various scenarios, and can achieve a bandwidth consumption that is close to the minimum obtained by ILP.