Traffic-Aware Resource Management in SDN/NFV-Based Satellite Networks for Remote and Urban Areas
Ilora Maity, Giovanni Giambene, Thang X. Vu, Chandrakanth Kesha, Symeon Chatzinotas
Abstract
This paper proposes distributing Virtual Network Requests (VNRs) in Non-Geostationary Orbit (NGSO) satellite networks based on diverse traffic demands from remote and urban areas. VNRs have allocated node resources, but resource management is challenging due to topology dynamics and limited availability. Provisioned on-board resources in remote or urban areas require customized VNR management solutions for satellite networks. We address these challenges by designing an End-to-End (E2E) resource management solution that considers network topology and payload capabilities to route VNRs optimally based on resource availability. This solution includes modules for topology prediction, traffic prediction, and Virtual Network Function (VNF) mapping and resource allocation. The topology prediction module uses supervised learning to predict satellite-to-gateway link availability and stability. The traffic prediction module classifies the gateway into urban and remote and predicts the traffic load between gateway pairs. Based on the predicted data, the VNF mapping and resource allocation module allocates substrate networks to the VNRs by assigning optimal routing paths customized to the traffic generation area (remote or urban). Simulation results based on realistic topology, parameters, and datasets show that the proposed solution achieves promising performance regarding link stability prediction and VNR mapping accuracy. The proposed solution reduced E2E delay by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$11.17 \%$</tex-math></inline-formula> for remote traffic compared to the benchmark VNE-TEG.