Delay-Sensitive and Resource-Efficient VNF Deployment in Satellite-Terrestrial Networks

Large-scale low earth orbit (LEO) satellite communication systems are integral to future 6G communication networks. Due to the scarcity and mobility of satellite network resources, it is necessary to integrate software defined network (SDN) and network function virtualization (NFV) into the large-scale LEO satellite-terrestrial network. In this paper, we propose a centralized and distributed software-defined satellite network architecture. Under this architecture, we focus on the virtual network function (VNF) deployment and flow scheduling problem (VDS). Considering user delay performance and limited satellite network resources, we formulate the VDS problem as a nonlinear binary programming (NLBP) problem, which includes 0-1 fractional and piecewise function programming. To address this challenge, we linearize the NLBP problem and propose a Mixed Integer Linear Programming (MILP)-based exact method to obtain optimal solutions for small-scale scenarios. Furthermore, we propose a resource-efficient VNF deployment and flow scheduling algorithm (RE-VDS) to efficiently obtain suboptimal solutions for large-scale scenarios. Simulation results demonstrate that our proposed algorithm closely approximates the optimal solution in small-scale scenarios and exhibits favorable performance in terms of system resource utilization, load balancing, and trade-off user performance and operator profit in large-scale scenarios.