D-Schedule: Dependency-Aware VNF Scheduling in Satellite-Terrestrial Networks

The deployment of Virtual Network Functions (VNFs) to network nodes involves the interdependent task of scheduling the VNFs at each node. A VNF scheduling strategy determines the execution time of the VNFs when multiple VNFs are mapped to a node. The computation of an optimal VNF scheduling strategy is difficult due to diverse service demands, dynamic traffic load, and dependency between VNFs. When applied to Satellite-Terrestrial Networks (STNs), this problem is even more challenging due to time-varying topology. In this work, we propose a priority-based VNF scheduling scheme that is updated dynamically based on network dynamics. The proposed scheme, named D-Schedule, schedules the VNFs mapped to each node based on revenue and delay budget after executing correlated VNFs. The simulation results demonstrate that D-Schedule achieves 63.98% more profit and 14.13% more node utilization compared to the benchmark.