Abstract

Network function virtualization (NFV) provides cost-effective network service by means of service function chain (SFC) for users. Meanwhile, billions of terminals are connected to the Internet (Internet of Things), and the vast variability of bandwidth requirements of SFC requests from heterogeneous terminals poses challenges to the network management, then neither only electronic networks nor only optical networks can provide highly efficient transmissions for network services. In this paper, a hybrid optical-electronic network paradigm is considered to provide connection services for SFC requests to benefit from the large bandwidth of optical networks and the flexibility of electronic networks. A profit-based utility function is established, and a mathematical formulation of the SFC embedding problem with the $log$ - $sum$ - $exp$ approximation is provided. An approach based on Markov chain modeling is used to adaptively embed SFC requests with the aim to maximize the utility function. Simulation results demonstrate the quality and adaptivity of the proposed approach in terms of the utility function.