Dynamic resource management for neighbor-based VNF placement in decentralized satellite networks

By introducing software-defined networking (SDN) and network function virtualization (NFV), low-earth-orbit (LEO) satellite networks can facilitate virtual network function (VNF) placement, which will provide computing services for satellite applications on-demand. In this paper, we study the VNF placement problem in a decentralized LEO satellite network due to the requirements for real-time processing and network resilience, where our aim is to jointly optimize end-toend service delay and network bandwidth cost in a dynamic environment. To this end, a decentralized LEO satellite network architecture is first implemented for resource management by establishing the neighboring sub-network for each satellite. Then we formulate the VNF placement problem as an integer non-linear programming problem with multiple constraints of network resources and service requirements. A neighbor-based VNF placement (N-VNFP) approach is proposed to address the optimization problem. Finally, we conduct the experiments to evaluate the performance of the proposed N-VNFP approach in a Walker constellation with 66 LEO satellites. The simulation results show that the proposed N-VNFP approach provides an effective solution for resource management in a decentralized LEO satellite network and also outperforms the two centralized baselines, i.e., Viterbi and Greedy, in terms of end-to-end service delay and network bandwidth cost.